Manipulating electrolyte solvation structures to build high-voltage concentration batteries for efficient energy storage

Abstract

Concentration batteries represent a distinctive type of Galvanic cells that use the same redox couple at both electrodes but generate voltage from differences in electrolyte compositions. However, their applications have been limited because ten times concentration difference typically produces only small voltages (<0.059 volts, V). Here we show that tailoring electrolyte properties can overcome this limitation and enable metal-based (zinc and copper) concentration batteries with higher voltages. By using a highly concentrated catholyte electrolyte, the redox potential of the positive electrode is increased, while a strongly complexing anolyte electrolyte lowers the free ion concentration and decreases the redox potential of the negative electrode. This asymmetric electrolyte design produces a cell voltage of 0.6-0.7 V, which is higher than the preconceived notion of 0.059 V. Furthermore, this approach can be combined with common positive electrodes to fabricate aqueous full cells with higher voltages (2.2-2.5 V), therefore boosting their voltages and specific energy densities.

Concentration batteries typically suffer from low voltages (>0.059 V) and are considered not suitable for energy storage. Here, authors tackle this challenge by strategically manipulating electrolyte properties, demonstrating high-voltage concentration batteries (0.7 V).

Introduction

Batteries play an essential role in electrochemical energy storage and utilization^1,2. Over the past decade, there has been an ever-increasing interest in exploring alternative battery chemistries beyond traditional lithium, starting from monovalent ions (sodium, potassium, ammonium, proton)^3–6 and expanding to multivalent metal systems (magnesium, calcium, zinc, iron, copper, nickel, manganese, aluminum, indium, etc.)^7–15. However, no matter which ion systems studied, all these batteries universally operate on the conventional Galvanic battery mechanism¹⁶ (Fig. 1a): the cathode and anode utilize two distinct redox couples (cathode: O_y/R_y; anode: O_x/R_x; O and R represent the oxidized and reduced species, respectively) and their electrochemical potential difference to generate a battery voltage (Fig. 1b), whereas the sole electrolyte provides commutable ionic charge carriers and unites these two redox reactions. Note that in this work, cathode refers to the positive electrode, whereas anode refers to the negative electrode. This definition is consistent with most battery publications^3–15.

Fig. 1. Schematics of two types of Galvanic batteries.

a The conventional Galvanic battery configuration, which utilizes two different redox couples (cathode: O_y/R_y, anode: O_x/R_x) and one electrolyte. b Scheme of the electrochemical potential and battery voltage, the latter of which stems from the potential gap in two redox couples. c The concentration battery configuration, which utilizes one redox couple (Mⁿ⁺/M) and two different electrolytes. d Scheme of the electrochemical potential and battery voltage, the latter of which stems from the potential deviation caused by the electrolyte concentration and solvation differences. In this figure, O and R represent the oxidized and reduced species in a redox couple, respectively. Mⁿ⁺/M denotes the redox couple of metal ions (Mⁿ⁺) and metal (M).

Using lithium-ion batteries as an example, the cathode and anode work on the distinct redox couples of Li_(1-x)CoO₂/LiCoO₂ (3.8 V vs. Li⁺/Li; LiCoO₂ represents the lithium cobalt oxide) and C₆/LiC₆ (0.1 V vs. Li⁺/Li; C₆ represents the graphite), respectively, yielding a battery voltage of 3.7 V¹⁷. To achieve higher voltages, we must develop advanced electrode materials with larger electrochemical potential gaps (cathode: O_z/R_z; anode: O_w/R_w; Fig. 1b). For example, pairing a LiNi_0.5Mn_1.5O₄ cathode (4.7 V vs. Li⁺/Li, LiNi_0.5Mn_1.5O₄ is lithium nickel manganese oxide) with a Li metal anode (0 V vs. Li⁺/Li) can fundamentally alter the battery chemistry and elevate the voltage to 4.7 V¹⁸. This approach hinges on the development of tailored electrodes rather than alternative electrolytes to increase the battery voltage, which has been established as a default method in conventional Galvanic battery systems.

In stark contrast, a concentration cell (or battery) represents a vastly different yet often overlooked variant of Galvanic batteries. Unlike conventional batteries with two distinct redox couples, the concentration battery exploits the same redox couple for both cathode and anode reactions, and it leverages the electrolyte concentration and/or solvation structure differences to generate a nominal voltage^19–21. Figure 1c illustrates a metal-based concentration battery that operates on the identical Mⁿ⁺/M (M is short for metal) redox couple in two different electrolytes (catholyte and anolyte; catholyte is the electrolyte on the cathode/positive electrode side, anolyte is the electrolyte on the anode/negative electrode side). According to the classical Nernst equation²², the Mⁿ⁺/M couple yields distinct potential due to the electrolyte concentration and/or solvation differences:

$$\mathbf{Redox}\mathbf{reaction}:{\mathrm{M}}^{\mathrm{n}+}\left(\mathrm{aq}\right)+{\mathrm{ne}}^{-}=\mathrm{M}\left(\mathrm{s}\right)$$

$$\mathbf{Cathode}:{\mathrm{E}}_{+}={\mathrm{E}}^{\circ }+\left(0.059/\mathrm{n}\right)\times \log \left(\mathrm{c}{+}^{\cdot }\gamma +\right)$$ 1

$$\mathbf{Anode}:\mathrm{E}-={\mathrm{E}}^{\circ }+\left(0.059/\mathrm{n}\right)\times \log \left(\mathrm{c}{-}^{\cdot }\gamma -\right)$$ 2

$$\mathbf{Cell}\mathbf{voltage}:\mathrm{E}=\mathrm{E}+-\mathrm{E}-=\left(0.059/\mathrm{n}\right)\times \log \left[\left(\mathrm{c}+\cdot \gamma +\right)/\left(\mathrm{c}-\cdot \gamma -\right)\right]$$ 3

In these equations, E₊ (or E₋), E°, n, c, and γ denotes the practical electrode potential, the standard electrode potential, the number of electron transfer, the Mⁿ⁺ ion concentration, and the Mⁿ⁺ ion activity coefficient, respectively. Assuming a tenfold concentration difference (c₊/c₋ = 10) and an identical activity coefficient (γ₊/γ₋ = 1) in two electrolytes, this concentration battery will generate a low voltage of 0.059/n volts²², which is far insufficient for practical applications. As a result, concentration batteries are often introduced as a basic concept in many chemistry textbooks^20,21, instead of a feasible device for electrochemical energy storage.

Recently, the battery field has witnessed substantial progress in electrolyte studies, and various elaborate electrolytes have emerged to consolidate the battery performance, such as concentrated electrolytes^23,24, localized high concentration electrolytes^25–27, “Water-in-Salt” (WiS) electrolytes^28,29, and diluted electrolytes^30,31. These electrolytes bring significant inspirations for us to devise high-performance concentration batteries. Nevertheless, prior electrolyte studies focus on the enhancement of electrode/electrolyte interfacial stability and cycling life instead of regulating the Mⁿ⁺/M redox potential, creating a significant knowledge gap. To build a meaningful concentration battery, there are two critical questions to be answered: (1) How can we regulate the electrolyte properties to dramatically raise the Mⁿ⁺/M potential for the cathode reaction? (2) How can we manipulate the electrolyte properties to efficiently decrease the Mⁿ⁺/M potential for the anode reaction?

In this work, we used zinc (Zn) as a demonstrative paradigm and systematically investigated how electrolyte properties (concentrations, cations, anions, and ligands) affected the Mⁿ⁺/M potential. We propose the catholyte to be a concentrated or “WiS” solution, which significantly elevates the Mⁿ⁺/M potential through a dominant concentration/solvation effect (Fig. 1d). Conversely, we formulate the anolyte as a metal-complex solution, which exerts a strong complexation effect and reduces the Mⁿ⁺/M potential (Fig. 1d). Integrating these two effects, the concentration battery achieves a notable voltage of ~0.7 V, higher than the standard (0.059/n V) by more than 20 times. Moreover, this concentration battery design can efficiently adapt to various cathodes (Prussian blue, metal oxides, sulfur), increasing their voltages by 0.7 V and enabling high-voltage aqueous batteries (2.2–2.5 V) with 36–228% energy enhancement. Our work has the potential to open further opportunities for concentration batteries in the energy storage context.

Results

Considering the capacity and cost advantages of Zn, we utilized Zn to showcase the fabrication of a high-performance aqueous concentration battery. However, we should note that the research rationale and experience can be translated to other metal batteries with similar parameter refinements.

Concentrated electrolytes to increase the Zn²⁺/Zn redox potential

In aqueous electrolytes, a variety of parameters can influence the Zn²⁺/Zn redox potential, including anion species, Zn²⁺ ion concentration, and additional cations solvated in electrolytes^32–34. As illustrated in Fig. 2a, Zn²⁺ ions are typically coordinated with six water molecules in the octahedral configuration, constituting the primary solvation shell. Beyond this first coordination sphere, a more diffuse secondary solvation shell forms, comprising additional water molecules and adjacent anions that interact via hydrogen bonding and electrostatic attraction. Notably, in concentrated or WiS electrolytes, some anions can enter the primary shell, displace water molecules, and competitively coordinate with Zn²⁺ ions²⁸. The reorganization of the solvation structure can significantly alter Zn²⁺ ion solvation energies and activities. The tertiary solvation shell is usually loosely defined, and it includes more distant water molecules and spectator ions (such as cations) that interact weakly through long-range Coulombic forces.

Fig. 2. Optimization of catholyte electrolytes for a feasible Mⁿ⁺/M cathode reaction.

a Scheme of solvation structures of Mⁿ⁺ ions in aqueous electrolytes (the grey, orange, and green spheres represent Mⁿ⁺ ions, anions, and additional cations, respectively). b Scheme of three-electrode M‖M symmetrical cells. c The calculation of the average Mⁿ⁺/M potential based on the plating-stripping potential. d The Zn²⁺/Zn potential in 1 m Zn²⁺ electrolytes with various anions. e The Zn²⁺/Zn potential in halide electrolytes with various concentrations. f The Zn²⁺/Zn potential in ZnCl₂ and ZnBr₂ electrolytes with various cations. g The Zn plating efficiency in ZnCl₂ and ZnBr₂ electrolytes (condition: 1 mA cm⁻² for 0.5 mAh cm⁻²). h The anodic stability window of ZnCl₂ and ZnBr₂ electrolytes (scanning rate: 1 mV s⁻¹).

The intricate solvation environment has a pronounced impact on the Zn²⁺/Zn redox potential. To accurately determine the potential, we propose a feasible three-electrode symmetrical cell setup (Fig. 2b), which introduces an external reference electrode (such as Ag/AgCl, silver/silver chloride) to the conventional Zn‖Zn symmetrical cells. Consequently, the Zn plating/stripping potential in each cycle can be easily measured, which enables the calculation of average Zn²⁺/Zn potential (Fig. 2c). For convenience, we further convert the Zn²⁺/Zn potential to the standard hydrogen electrode (SHE) system.

To identify the optimal Zn²⁺ catholyte electrolyte, we systematically investigated the effect of anions, concentrations, and cations on the redox potential. We first fixed the Zn²⁺ concentration at 1 mol kg⁻¹ (1 m) and compared five common zinc salts with different anions, including zinc chloride (ZnCl₂), bromide (ZnBr₂), iodide (ZnI₂), acetate (Zn(Ac)₂), and sulfate (ZnSO₄). Despite their substantial differences in elements, ionic sizes, and charge density, these anions yield comparable redox potential that fluctuates around −0.76 V (Fig. 2d and Supplementary Fig. 1), which approaches the standard Zn²⁺/Zn electrode potential. This phenomenon suggests that concentration plays a more significant role in Zn²⁺/Zn redox behavior than anion types.

Motivated by this observation, we studied the impact of salt concentration on the redox potential. Considering that ZnSO₄ and Zn(Ac)₂ salts are moderately soluble in water (<3.6 m, Supplementary Table 1), we focused on highly soluble salts of ZnCl₂, ZnBr₂, and ZnI₂, whose maximal solubility can reach 30, 20, and 15 m, respectively. As shown in Fig. 2e, increasing salt concentration can efficiently raise the Zn²⁺/Zn potential, with 0.25–0.3 V voltage elevation achieved for these salts (Supplementary Figs. 2–4). More importantly, there is an obvious trend between salt concentration and Zn²⁺/Zn potential in the tested range, which reinforces the pivotal role of concentration in governing the redox potential. In addition to the concentration effect, anions exert a secondary influence. For instance, 20 m ZnBr₂ exhibits a 25 mV higher potential (−0.445 V, Fig. 2e) than 20 m ZnCl₂ (−0.47 V). However, the very high concentration of ZnCl₂ (30 m) offsets the potential gap and reduces the potential difference to 15 mV, which is minor compared to the overall potential shift caused by the concentration (~300 mV).

Based on the above experiments, we identify 20 m ZnBr₂ and 30 m ZnCl₂ as two of the most promising electrolytes. Then, we explored the effect of cation on the Zn²⁺/Zn redox to seek even higher potential shifts. It has been reported that saturated electrolytes could dissolve additional salts with similar ions and further increase of the overall salt concentration^35–37. For instance, Wang et al. introduced 7 m LiOTF (lithium triflate) to a saturated 21 m lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) solution and obtained a unique “water-in-bisalt” electrolyte³⁸. Therefore, we attempted adding LiBr to 20 m ZnBr₂ and adding LiCl to 30 m ZnCl₂. The 20 m ZnBr₂ electrolyte can only dissolve 2 m LiBr, slightly increasing the potential from −0.445 to −0.43 V (Fig. 2f and Supplementary Fig. 5). By contrast, 30 m ZnCl₂ can dissolve more LiCl salts (2.5 and 5 m), progressively raising the potential from −0.46 to −0.42 V (Fig. 2f and Supplementary Fig. 6). Again, this observation highlights the important role of supporting salt concentration in impacting redox potential, since 2.5 and 5 m LiCl salts are more concentrated than 2 m LiBr. Additionally, we found that 30 m ZnCl₂ could dissolve 5 m NaCl, 5 m KCl, and 5 m NH₄Cl as well, and all these electrolytes show improved potential (Fig. 2f and Supplementary Fig. 7). There is an interesting potential shift trend of Li⁺ > Na⁺ > K⁺ > NH₄⁺, indicating the supplementary role of cations in regulating Zn redox potential. Overall, the 30 m ZnCl₂ + 5 m LiCl electrolyte exhibits the highest Zn²⁺/Zn potential (−0.42 V) and the largest potential shift (+0.34 V).

Compared with 20 m ZnBr₂ + 2 m LiBr, the 30 m ZnCl₂ + 5 m LiCl electrolyte offers additional advantages for concentration batteries. Coulombic efficiency (CE) is an essential parameter that measures the metal plating reversibility. Notably, the 30 m ZnCl₂ + 5 m LiCl electrolyte enables a high CE of 99.84% and is higher than that of 20 m ZnBr₂ + 2 m LiBr (Fig. 2g and Supplementary Fig. 8), which facilitates the stable operation of Zn plating reactions. Indeed, symmetrical Zn‖Zn cells demonstrate an extraordinary calendar life of 3400 h (4.7 months, Supplementary Fig. 9). Furthermore, according to their standard redox potential (Cl^-/Cl₂: +1.36 V vs. SHE; Br⁻/Br₂: +1.066 V vs. SHE)³⁹, Cl⁻ anions are more resistant to oxidization than Br⁻ anions, benefiting the broadening of the electrochemical stability window in electrolytes. As shown in Fig. 2h, the ZnBr₂ electrolyte suffers from drastic electrolyte oxidization at +1.1 V vs. Ag/AgCl, but the ZnCl₂ electrolyte resists oxidization even at +1.5 V. Besides, the chlorine element is much more abundant than bromine, which contributes to the lower cost of ZnCl₂. Collectively, the 30 m ZnCl₂ + 5 m LiCl electrolyte demonstrates higher potential, better efficiency, wider window, and lower cost, which is identified as the optimal catholyte solution. We also note that chloride electrolytes are generally corrosive to metallic substrates, and this drawback will restrict the choice of current collectors. Currently, anti-corrosion titanium has to be used to investigate concentration batteries. Despite this challenge, the ZnCl₂ salt offers several merits that are not available to other candidates, including low cost (Supplementary Table 2), high solubility (30 m), large potential shift effect (+0.34 V), and high plating efficiency (99.84%), which justifies the utilization of ZnCl₂ in this work.

The +0.34 V potential shift in ZnCl₂ electrolytes is significant and larger than the classical expectation of 0.059/n volts for a tenfold concentration change. To quantify the concentration dependence, we measured the Zn²⁺/Zn potential in a wider range of ZnCl₂ concentrations, from diluted solutions (0.1 and 0.5 M, M: mol L⁻¹) to concentrated electrolytes (1–30 m, Supplementary Table 3). Because the Nernst equation is related to molarity (M) other than molality (m), we converted the molality to molarity (Supplementary Table 4) and plotted the molarity-potential relationship in Supplementary Table 5. Specifically, the molarities for 1, 2, 3, 4, 5, 6, 10, 15, and 30 m ZnCl₂ correspond to ~0.98, 1.88, 2.74, 3.51, 4.26, 4.89, 7.30, 8.92, and 12.78 M, respectively.

At low concentration (0.1 to 1.88 M), the Zn²⁺/Zn potential (E) exhibits a good linear dependence on the logarithm concentration (logC, Fig. 3a), which is described by E = 0.0312 × logC – 0.7594 (R² = 0.997). This behavior is consistent with the conventional Nernst form (Fig. 3b): E = E° + (0.059/n) × log(c·γ), where E° equals to −0.76 V, and the slope is close to 0.059/n (0.0295 for n = 2). The linear relationship also indicates that in diluted electrolytes, it is the Zn²⁺ concentration that dictates the Zn²⁺/Zn redox potential, whereas the activity coefficient (γ) plays a minor role.

Fig. 3. The explanation of the redox potential shift in the ZnCl₂ catholytes.

a The logarithm relationship between molarity (M) and Zn²⁺/Zn potential. b The analysis of redox reactions and Nernst equations for Zn plating. c The activity coefficient of ZnCl₂ under various molality concentrations. d The activity of water under various molality concentrations. e The comparison between the experimental and calculated Zn²⁺/Zn potential. f The properties of ionic radius, hydrated radius, and hydration energy of Li⁺, Na⁺, K⁺, and NH₄⁺ cations. In this figure (e), the shading area highlights the good linear relationship between the experimental value and the calculated value.

In contrast, in concentrated electrolytes (2.74–12.78 M), the Zn²⁺/Zn potential deviates markedly from linearity and increases rapidly (Fig. 3a). A prevalent explanation for this behavior is the profound increase in the activity coefficient γ(Zn²⁺), as proposed by several influential WiS electrolyte studies^28,40. For example, Wang et al. reported a 0.14 V potential increase in Li-ion electrodes when increasing LiTFSI concentration from 5 to 21 m. By assuming γ(Li⁺) = 1 at 5 m, they estimated that γ(Li⁺) increased by a factor of 70–80 at 21 m²⁸. Similarly, Yamada et al. demonstrated a more concentrated hydrate-melt electrolyte of Li(TFSI)_0.7(BETI)_0.3·2H₂O (27.8 m; BETI: bis(pentafluoroethylsulfonyl)imide), which produced a 0.25 V potential shift in Li-ion electrodes. They proposed that γ(Li⁺) had increased by 1.29 × 10⁶ compared with the standard electrolyte.⁴⁰ Following this activity coefficient approach, we can calculate that the γ(Zn²⁺) in 30 m ZnCl₂ should increase by 1.16 × 10⁹ to achieve the 0.3 V potential shift (Supplementary Fig. 10). However, such an extraordinary increase is physically implausible and unlikely to represent a chemically meaningful electrolyte system. This inconsistency indicates that the conventional activity coefficient framework is insufficient to explain the potential shift in concentrated electrolytes, which motivates us to re-examine the redox reaction and the corresponding Nernst equation for more convincing explanations^41,42.

In fact, the more accurate description of the Zn²⁺/Zn redox reaction in aqueous electrolytes should be:

$${\left[\mathrm{Zn}\left({\mathrm{H}}_2\mathrm{O}\right)\mathrm{n}\right]}^{2+}+2\mathrm{e}=\mathrm{Zn}\left(\mathrm{s}\right)+\mathrm{n}{\mathrm{H}}_2\mathrm{O}$$ 4

In this process, hydrated [Zn(H₂O)_n]²⁺ ions receive two electrons, are reduced to Zn metal, and release n water molecules into the bulk electrolyte as free and uncoordinated water (Fig. 3b). In diluted electrolytes, water is typically treated as a pure solvent with a default activity of 1⁴³. However, in concentrated electrolytes, the water content is greatly reduced, and water can no longer be regarded as an inert solvent; instead, it must be explicitly considered as a reactant or product in the redox process. Accordingly, the Zn²⁺/Zn potential should be expressed in the following Nernst form (Fig. 3b):

$$E=E\genfrac{}{}{0ex}{}{\circ }{}+\left(\frac{0.059}{2}\right)*\log \frac{\alpha \left[Zn{\left(H_{2}O\right)}_n^{2+}\right]}{{\left(\alpha \left[H_{2}O\right]\right)}^n}$$ 5

This expression shows that the practical Zn²⁺/Zn potential depends not only on the activity of hydrated Zn²⁺ species but also on the activity (α) of water α(H₂O). Because the water activity enters the equation with a power of n, we can expect that the reduced water activity plays a more dominant role in enhancing the Zn²⁺/Zn potential.

A remaining question is how to obtain reliable values of γ(Zn²⁺) and α(H₂O) in concentrated ZnCl₂ electrolytes. Fortunately, R. N. Goldberg experimentally determined γ(Zn²⁺) and α(H₂O) over a wide range of ZnCl₂ solutions (0.001–23.193 m; Supplementary Table 6)⁴⁴. Using these data points, we replotted the results in Fig. 3c, d. As shown, in concentrated electrolytes (3–23 m), the mean γ(ZnCl₂) increases markedly from 0.2787 to 3.5965, due to the intensified ion-ion interactions. Moreover, there is a clear linear relationship between ZnCl₂ molality and γ(ZnCl₂) in the 10–23 m range, which allows us to extrapolate the γ(ZnCl₂) value as 5.0955 in 30 m ZnCl₂. However, if we consider the increment in γ(ZnCl₂) only, the calculated Zn²⁺/Zn potential reaches −0.7065 V (Supplementary Table 7), which is far below the experimental value of −0.46 V in 30 m ZnCl₂ (Fig. 3e). The discrepancy indicates that the Zn²⁺ activity coefficient alone cannot account for the large potential shift, and the suppressed water activity must be considered.

As shown in Fig. 3d, increasing salt concentration leads to an exponential decrease in water activity, and we can extrapolate a low water activity of 0.021 in 30 m ZnCl₂. Assuming a hydration number of six for Zn²⁺ (n = 6) and using C(ZnCl₂) = 12.78 M, γ(ZnCl₂) = 5.0955, and α(H₂O) = 0.021, the Zn²⁺/Zn potential is calculated to be −0.41 V (Fig. 3e and Supplementary Table 7), which is close to the experimental value (−0.46 V). Using the same approach, the theoretical Zn²⁺/Zn potential in 10 and 15 m ZnCl₂ is calculated to be −0.66 and −0.59 V, respectively (Supplementary Table 7). Although these values show minor deviations from experiments, the overall trend aligns well with the measured data, thereby validating the necessity of incorporating both Zn²⁺ activity and water activity into the Nernst equation analysis.

Another interesting phenomenon is that the Zn²⁺/Zn potential in 30 m ZnCl₂ + 5 m ACl (A⁺ = Li⁺, Na⁺, K⁺, and NH₄⁺) mixed electrolytes is consistently more positive than the pristine 30 m ZnCl₂ (Fig. 2f), although these electrolytes exhibit the same Zn²⁺ molality (30 m) and comparable Zn²⁺ molarity (12.3–12.8 M, Supplementary Table 8). Moreover, the Zn²⁺/Zn potential decreases in the order of Li⁺ > Na⁺ > K⁺ > NH₄⁺, which can also be explained by the revised Nernst equation. The addition of extra salts further increases the overall concentration and ionic strength of electrolytes, thereby intensifying ion/ion interactions. As a result, the γ(Zn²⁺) is enhanced, whereas the water activity α(H₂O) is further suppressed, both of which contribute to a higher Zn²⁺/Zn potential than the pristine 30 m ZnCl₂. However, the ability to suppress water activity α(H₂O) decreases in the order of Li⁺ > Na⁺ > K⁺ > NH₄⁺, which stems from the progressive reduction in cation charge density and hydration strength (Fig. 3f and Supplementary Table 9)⁴⁵. Particularly, Li⁺ possesses the highest charge density and exhibits the largest hydration shell (3.82 Å, Ångström) and most negative hydration energy (−520 kJ mol⁻¹; kilojoules per mole), enabling it to interact most strongly with surrounding water molecules and thereby achieve the maximal suppression of water activity. Consequently, the lowest α(H₂O) in the 30 m ZnCl₂ + 5 m LiCl electrolyte leads to the largest Zn²⁺/Zn potential elevation among these alkali salts. Overall, the revised Nernst equation provides a consistent framework to rationalize the potential shifts in various electrolytes.

The potential shift stems from the significantly enhanced Zn²⁺ activity coefficient and suppressed water activity in concentrated electrolytes, both of which are closely related to changes in microscopic solvation structures. To gain mechanistic insights into solvation structures, we conducted large-scale all-atom molecular dynamics (MD) simulations of 1, 5, 15, 30 m ZnCl₂, and 30 m ZnCl₂ + 5 m LiCl electrolytes (Supplementary Fig. 11). The MD production was performed in the NPT ensemble (the number of particles (N), pressure (P), and temperature (T) were kept as constants) for 20 ns with a time step of 2 fs with Particle Mesh Ewald (PME) method. Then, the radial distribution function (RDF) and coordination number (CN) of Zn²⁺-Cl and Zn²⁺-O(H₂O) within electrolytes were derived. The initial and final configurations of all electrolytes from MD simulations are given in Supplementary Data 1.

In 1 m ZnCl₂ (Fig. 4a), Zn²⁺ exhibits an overall CN of 6 in the first solvation shell, with the Zn²⁺-Cl and Zn²⁺-O(H₂O) coordination of 0.528 and 5.472, respectively. This suggests that in diluted electrolytes, Zn²⁺ primarily exists in the form of highly hydrated [Zn(H₂O)₆]²⁺, with only a small fraction of ZnCl⁺ species. Such a solvation environment is characteristic of dilute electrolytes, where sufficient free water molecules stabilize Zn²⁺ ions and limit strong ion-ion interactions, resulting in relatively small activity coefficients.

Fig. 4. Characterizations of solvation structures in ZnCl₂ electrolytes.

a–e RDF and coordination number of Zn²⁺-Cl⁻ and Zn²⁺-O(H₂O) pairs in 1 m ZnCl₂, 5 m ZnCl₂, 15 m ZnCl₂, 30 m ZnCl₂, and 30 m ZnCl₂ + 5 m LiCl, respectively. The inset figures are corresponding MD solvation structures. f The water diffusion coefficients derived from MD calculations. g, h Raman results. i The influence of Zn²⁺/Zn energy states on their redox potential (referred to SHE). In this figure (a–e), the white, red, gray, green, and yellow spheres represent H, O, Zn, Cl, and Li atoms, respectively.

With the elevation of concentrations, Zn²⁺ exhibits a trend to decrease its overall CN, which is 5.999, 5.725, and 5.133 for 5, 15, and 30 m, respectively (Fig. 4b–d). Close examination reveals that the Zn²⁺-O(H₂O) bond markedly loses its coordination in the first shell, which is 4.254, 2.935, and 1.714 for 5, 15, and 30 m, respectively. By contrast, the Zn²⁺-Cl interaction progressively increases its coordination, rising from 0.528 to 3.419. This transformation can be rationalized by the high ZnCl₂/H₂O molar ratio (30:56, <1:2) in 30 m electrolytes, since H₂O molecules are insufficient to fully hydrate Zn²⁺ ions; therefore, some Cl^- anions competitively enter the primary solvation shell for coordination, yielding ion pairs, aggregates, or even [ZnCl₄]²⁻ anions. The structural evolution reflects the transition from solvent-separated ions to extensive ion pairing and aggregate formation in highly concentrated electrolytes. Such strong Zn²⁺-Cl⁻ associations reduce the effective dielectric screening and enhance ion-ion interactions, which drive the increase of cation activity coefficients in concentrated electrolytes⁴⁰. Concurrently, the strong coordination of water molecules to ions significantly disrupts the bulk hydrogen bonding network, thereby reducing the activity of free water, α(H₂O), which is consistent with the thermodynamic discussions on Nernst equations above.

Adding 5 m LiCl to 30 m ZnCl₂ renders the solvation environment more crowded, which further strengthens Zn²⁺-Cl⁻ interactions and weakens Zn²⁺-O(H₂O) coordination (Fig. 4e and Supplementary Fig. 11). Specifically, compared with 30 m ZnCl₂, the introduction of 5 m LiCl increases the Zn-Cl coordination from 3.419 to 3.434 and decreases the Zn-O coordination from 1.714 to 1.631. For clarity, the detailed populations of different ionic species are summarized in Supplementary Table 10. Such changes indicate the enhanced ion interaction and more strongly bound water molecules, leading to increased Zn²⁺ activity coefficient and further suppressed water activity.

Additionally, water diffusion coefficients can be extracted from MD simulations. Although the water diffusion coefficient is not equivalent to the thermodynamic water activity, it reflects the degree of water mobility and provides complementary insights into water confinement in concentrated electrolytes. As the concentration increases, the water diffusion coefficient decreases monotonically and dramatically from 1 to 30 m ZnCl₂ and further reduces upon the addition of 5 m LiCl (Fig. 4f). We also performed MD simulations to evaluate the water diffusion coefficients in 30 m ZnCl₂ + 5 m ACl electrolytes (Supplementary Fig. 12). Notably, the water diffusion coefficients decrease in the order of 30 m ZnCl₂ > 30 m ZnCl₂ + 5 m NH₄Cl > 30 m ZnCl₂ + 5 m KCl > 30 m ZnCl₂ + 5 m NaCl > 30 m ZnCl₂ + 5 m LiCl (Supplementary Fig. 13). This trend is fully consistent with the suppression of water activity inferred from the revised Nernst equation analysis. Together, these solvation features and water mobility characteristics provide complementary microscopic evidence for the enhanced γ(Zn²⁺) and suppressed α(H₂O), thereby corroborating the observed Zn²⁺/Zn potential shifts.

Raman spectroscopy verifies the significant change in solvation structures and ionic species. In 1 m ZnCl₂, there is a weak peak near ~220 cm⁻¹, which is attributed to the Cl···O–H interactions and Zn-Cl bonds in ZnCl^+46,47. With the elevation of salt concentrations, Raman peaks at ~240 and ~290 cm⁻¹ have significantly increased their intensities (Fig. 4g and Supplementary Fig. 14), suggesting the formation of Zn-Cl aggregates and [Zn(H₂O)₂Cl₄]²⁻, respectively^48,49. This observation corroborates the theoretical simulations, where the competitive Zn²⁺-Cl interaction progressively substitutes the Zn²⁺-O(H₂O) interactions^50,51. Such enhanced ion pairing and aggregation reduce dielectric screening and strengthen ion-ion interactions, providing a microscopic origin for the increased Zn²⁺ activity coefficient in concentrated electrolytes.

Meanwhile, H₂O molecules also experience obvious vibrational changes in the high-frequency range (Fig. 4h). In 1 m ZnCl₂, the O–H vibration comprises a larger peak at ~3450 cm⁻¹ and a smaller peak at ~3250 cm⁻¹, which corresponds to the weak and strong hydrogen bond, respectively. With the increment of salt concentrations, the weak H-bond has significantly enhanced its intensity and shifted to higher positions, whereas the strong H-bond has dramatically weakened its intensity (Supplementary Fig. 15). These spectral changes indicate that water molecules are increasingly coordinated by Zn²⁺ (and Li⁺), which disrupts the hydrogen bonding network and reduces the population of free and bulk-like water. This progressive sequestration of water molecules is consistent with the suppressed water activity in the previous Nernst equation discussions.

In addition to the Nernst equation analysis, the potential shift can also be understood from a thermodynamic energy perspective. In electrochemical systems, the battery voltage is directly related to the Gibbs free energy change through the relation ΔG = −nEF (ΔG, n, E, and F represents the Gibbs free energy change, number of electron transfer, battery voltage, and Faraday constant, respectively)⁴³. If one considers a hypothetical H⁺/H₂ (SHE)‖Zn²⁺/Zn cell, the battery voltage depends solely on the Gibbs free energy of the Zn²⁺/Zn couple (Supplementary Fig. 16). Therefore, an increase in the Gibbs free energy of solvated Zn²⁺ reduces the energy gap between the H⁺/H₂ and Zn²⁺/Zn couples, leading to a higher Zn²⁺/Zn potential vs. SHE. However, it is challenging to directly calculate the Gibbs free energy of Zn²⁺ in concentrated electrolytes, due to the complexity of ionic species and solvation structures. As an alternative descriptor, we propose using the system potential energy derived from MD simulations as a qualitative proxy to reflect the relative energy states of solvated Zn²⁺ species. With increasing ZnCl₂ and LiCl concentrations, the average ion-ion and ion-solvent distances decrease, resulting in more frequent short-range Lennard-Jones contacts and stronger long-range Coulombic interactions. In parallel, the enhanced Zn²⁺-Cl⁻ association and reduced dielectric screening further intensify electrostatic interactions, collectively elevating the energy states of the electrolytes (Supplementary Fig. 17). As illustrated in Fig. 4i, the progressive increase in the Zn²⁺/Zn energy level reduces the energy gap between the H⁺/H₂ and Zn²⁺/Zn couples, leading to the elevation of the Zn²⁺/Zn potential vs. SHE. Therefore, the energy-based perspective provides a complementary interpretation of the potential shift in concentrated ZnCl₂ electrolytes.

Metal-complex electrolytes to decrease the Zn²⁺/Zn redox potential

To reduce the Zn²⁺/Zn anode potential, it is imperative to decrease the Zn²⁺ activity, as predicted by the Nernst equation. However, it is inappropriate to utilize low-concentration electrolytes (such as 1 × 10⁻⁵ M ZnCl₂), due to the minimized ionic conductivity under such conditions. Herein, we leverage the complex formation to decrease the free Zn²⁺ concentration⁴³, while maintaining a “nominal” Zn²⁺ concentration at reasonable values (such as 0.5 M). This complex formation approach has been widely utilized in redox flow batteries to adjust the Mⁿ⁺/M potential^52,53, but its impact on the metal plating chemistry in aqueous metal batteries remains relatively underexplored. Considering the wide variety of complex agents, we compared some representative candidates, including chloride (Cl⁻), ammonia (NH₃), and hydroxide (OH⁻). The target is to achieve optimal redox potential and cycling stability.

Figure 5a shows Galvanostatic charge/discharge (GCD) curves of Zn‖Zn batteries. The Cl⁻ ligand leads to relatively stable cycling but insufficient potential drop (from −0.76 to −0.84, <0.1 V), likely due to its modest coordination ability (logβ < 1, β: cumulative complex formation constant, Supplementary Table 11, and Supplementary Fig. 18). Notably, the OH^- ligand enables the lowest Zn²⁺/Zn potential (−1.28 V) thanks to its very high complex formation constant (logβ₄ = 15.5, Supplementary Table 11), which effectively reduces the free Zn²⁺ concentration through the strong complexation effect. We have analyzed the complex formation and dissociation equilibrium in the [Zn(OH)₄]²⁻ electrolyte and obtained a low Zn²⁺ concentration of ~1.62 × 10⁻¹⁶ M at the equilibrium state, which translates to a theoretical Zn²⁺/Zn potential of −1.26 V vs. SHE based on the Nernst equation (Supplementary Fig. 19). This value is very close to the measured potential (−1.28 V). However, Zn batteries get short-circuited after ~120 h, due to the well-known Zn corrosion in strong alkaline conditions (pH > 14). Comparatively, NH₃ is a strong complex agent (logβ₄ = 9.46) and a weak base (pK_b = 4.75; K_b: base hydrolysis constant)⁴³, achieving a good balance in redox potential (−1.08 V) and cycling stability (~200 h, Fig. 5a and Supplementary Fig. 20). Similarly, we have calculated the free Zn²⁺ concentration based on the zinc-ammonia complex formation and dissociation equilibrium, which is 1.43 × 10⁻¹⁰ M and corresponds to a theoretical Zn²⁺/Zn potential of −1.05 V (Supplementary Fig. 21). This value is close to that of standard [Zn(NH₃)₄]²⁺/Zn couple (−1.04 V, Supplementary Table 12) and consistent with the experimental potential (−1.08 V), thereby validating the complex formation approach in regulating the metal redox potential.

Fig. 5. Characterizations of anolyte electrolytes and Zn plating performance.

a GCD curves of symmetrical Zn‖Zn cells in different anolytes (0.5 mA cm⁻² for 0.5 mAh cm⁻²). b, c SEM images of Zn deposits in different anolytes. d The plating efficiency comparison (1 mA cm⁻² for 0.5 mAh cm⁻²). e, f Solvation structures and RDF plots of 0.5 m [Zn(NH₃)₄]²⁺ and 0.5 m [Zn(NH₃)₄]²⁺ + 5 m LiCl. In this figure (b, c), the Zn plating was conducted using a current density of 1 mA cm⁻² with a capacity of 0.5 mAh cm⁻². The battery was disassembled and the Zn electrode was retrieved after the first plating process. The testing temperature is 25 ± 1 °C. In this figure (e, f), the white, red, magenta, green, yellow, and blue colored spheres represent H, O, Zn, Cl, Li, and nitrogen atoms, respectively.

Considering the presence of LiCl in the catholyte, we also introduced various amounts of LiCl (2.5 and 5 m) to the zinc-ammonia anolyte, which could facilitate the Cl⁻ anion migration and the full cell operation. Interestingly, 2.5 and 5 m LiCl salts further reduce the potential to −1.0 and −1.12 V, respectively (Fig. 5a and Supplementary Fig. 22), likely due to the increased Zn²⁺···Cl⁻ coordination and the suppressed Zn²⁺ free concentration. Again, this observation indicates the role of supporting electrolytes in regulating the Zn²⁺/Zn potential. Besides the suppressed potential, 5 m LiCl prolongs the cycling life to 450 h (Supplementary Fig. 23), owing to the much-improved and well-defined Zn plating morphology (Fig. 5b, c and Supplementary Fig. 24). We reason that the electrostatic shielding effect plays a role in the LiCl-added electrolyte (Supplementary Fig. 25), where inert Li⁺ cations accumulate around Zn tips, regulate Zn²⁺ migration flux, and thus improve the plating behavior and morphology^54–56. Indeed, the LiCl-added electrolyte exhibits a higher plating efficiency (98.7%) than the pristine one (92.5%, Fig. 5d and Supplementary Fig. 26).

To gain insight into the solvation structure, we conducted MD and RDF simulations on the pristine and LiCl-added zinc-ammonia electrolytes. As shown in Fig. 5e, the Zn²⁺-N(NH₃) CN is 4.0 in the zinc-ammonia solution, but the Zn-O CN significantly decreases from 5.47 (1 m ZnCl₂) to 0.281, indicating the near-complete replacement of water molecules in the primary solvation shell by NH₃ ligands. This transformation signifies a fundamental change in the solvation structure, from fully hydrated [Zn(H₂O)₆]²⁺ species to ligand-coordinated ZnL_x complexes (L = NH₃). This complexion effect well explains the observed potential drop from −0.76 to −1.08 V (Fig. 5a), as the formation of stable Zn-NH₃ complexes effectively reduces the population of free and dissociated Zn²⁺ ions. Adding 5 m LiCl does not change the Zn-N coordination, since NH₃ is a stronger complex agent than Cl⁻ (Fig. 5f); however, it slightly increases the Zn²⁺-Cl coordination from 1.825 to 1.908, suggesting the formation of more complex species and contributing to the reduced potential. Overall, the introduction of NH₃ and Cl⁻ ligands can significantly transform the coordination chemistry and solvation structures of Zn²⁺ ions (Supplementary Fig. 27), leading to the reduced concentration of dissociated Zn²⁺ ions. Based on the Nernst equation, the suppressed Zn²⁺ concentration directly translates to the decreased redox potential. More importantly, there is a good agreement between the calculated redox potential and measured potential (Supplementary Figs. 18, 19, and 21), since the potential difference is merely 20–40 mV. Therefore, the complex formation, solvation structure change, and Nernst equation provide a fundamental and compelling explanation for the reduced Zn²⁺/Zn potential in anolyte electrolytes. We identify 0.5 m [Zn(NH₃)₄]²⁺ + 5 m LiCl as the optimal anolyte owing to its low potential, long cycling, and enhanced plating efficiency.

Performance of Zn‖Zn concentration batteries

To assemble Zn‖Zn concentration batteries, it is crucial to separate the catholyte and anolyte electrolytes. Herein, we propose a polymer gel electrolyte as an alternative approach for demonstration purposes, although more advanced membranes can be developed in future. This gel electrolyte comprises ion-conducting KCl salts in the agar polymer matrix and provides commutable Cl^- anions for charge balance (Supplementary Fig. 28). Supplementary Fig. 29 illustrates the working mechanism of concentration Zn‖Zn batteries. Upon charging, Zn cathode loses electrons, gets oxidized, and becomes Zn²⁺ ions (E₊ = −0.42 V vs. SHE), whereas Zn²⁺ ions in the anolyte receives electrons, get reduced, and yield Zn metal (E₋ = −1.12 V vs. SHE). The discharge is a reversed process. To balance the charge, Cl^- anions serve as charge carriers and shuttle between the catholyte and anolyte.

As shown in Fig. 6a, the Zn‖Zn concentration battery exhibits an average voltage of ~0.70 V, agreeing well with the cathode/anode potential difference (−0.42 V vs. −1.12 V). The Zn cathode delivers a specific capacity of 750 mAh g⁻¹, approaching its theoretical capacity (820 mAh g⁻¹). The concentration battery exhibits promising rate (Supplementary Fig. 30) and cycling performance. After 750 cycles (1500 h, ~2.1 months), the discharge voltage decreases from 0.70 to 0.63 V (Fig. 6b and Supplementary Fig. 31), corresponding to a high retention of ~90% (0.63/0.7 = 90% over 750 cycles; 0.013% voltage decay per cycle). We also conducted GCD tests at higher current density (0.125 mA cm⁻²) and capacity (0.5 mAh cm⁻). As shown in Fig. 6c, the concentration battery maintains well-overlapping GCD curves, indicating its reaction reversibility under more demanding conditions. The charge/discharge plateau appears at ~0.76/0.65 V, leading to an average voltage of ~0.70 V. After 1000 h, this battery experiences no short circuits, validating its cycling stability (Fig. 6d).

Fig. 6. The concentration Zn‖Zn battery performance.

a GCD curves at 12.5 μA cm⁻², the inset is the charge curve of Zn cathode at 20 mA g⁻¹. b Battery voltages during cycling. c, d GCD curves and cycling performance (0.125 mA cm⁻² for 0.5 mAh cm⁻²). e The voltage, capacity, and specific energy properties of various Zn-ion cathodes. A star symbol denotes Zn metal in this figure (e), which has a well-defined capacity and voltage. Colored ellipses denote PBAs, Mn based, V based, and organic cathodes; the extent of each ellipse reflects the range of reported capacities and voltages across the literature.

We highlight that by leveraging the concentration battery mechanism, Zn metal could work as a high-capacity cathode, thus having the potential to open emerging opportunities for Zn batteries. Considering the reasonable voltage (0.7 V) and high capacity (750 mAh g⁻¹), Zn cathode achieves a competitive specific energy of ~525 Wh kg⁻¹, which rivals many Zn-ion cathodes^57,58 (Fig. 6e and Supplementary Table 13), including Prussian blue analogues (PBAs), manganese/vanadium-based oxides, and organic compounds.

Performance of Zn‖cathode concentration batteries

Beyond the Zn‖Zn system, the concentration mechanism readily applies to various Zn‖Cathode batteries, significantly increasing their working voltages and specific energies. Note that battery voltage results from the potential difference between cathode and anode, both of which vary with electrolyte properties. For conventional batteries (Fig. 7a), the cathode-anode pair usually shares one type of charge carriers in one electrolyte. When electrolyte properties (concentrations or solvation structures) are altered, cathode and anode will likely experience an equivalent or comparable potential shift, as determined by the Nernst equation (Fig. 7b and Supplementary Fig. 32). Consequently, the battery voltage tends to remain constant or similar, because the potential shift will be canceled out (Fig. 7c). This explains why many M‖cathode batteries exhibit similar voltages^31,59, despite distinct electrolytes are used (Supplementary Table 14). For instance, the Li‖LiCoO₂ battery always delivers a voltage of 3.8 V (Supplementary Fig. 33), no matter in diluted (0.16 M), regular (1 M), or concentrated electrolytes (>3 M).

Fig. 7. Analysis and characterization of regular and concentration M‖cathode batteries.

a, b Scheme and Nernst equation analysis of the conventional M‖cathode Galvanic battery, where the potential shift will be cancelled out for the metal (M) anode and the host (H) cathode. c Scheme of the potential shift of cathode/anode in various electrolytes. d, e Scheme and Nernst equation analysis of the proposed M‖cathode concentration battery, where the potential shift will be harnessed for the metal (M) anode and the host (H) cathode. f GCD curves of the traditional and concentration Zn‖Zn₃[Fe(CN)₆]₂ battery. g–i GCD curves of the concentration Zn‖LiMn₂O₄, Zn‖MnO₂, and Zn‖S batteries.

In contrast, the concentration battery mechanism can strategically harness the potential shift and enable high-voltage full cells. Specifically, the cathode resides in a concentrated catholyte, which elevates its redox potential, whereas the anode is immersed in a metal-complex anolyte, which decreases the Mⁿ⁺/M redox potential (Fig. 7d, e and Supplementary Fig. 34). Rather than canceling each other, these potential shifts act in concert, leading to enhanced voltages (Fig. 7c) and boosted specific energies. Notably, the electrolyte stability window expands to ~2.8 V (Supplementary Fig. 35), which supports the operation of various cathodes, including PBAs, metal oxides, and elements.

Zn₃[Fe(CN)₆]₂ (zinc ferricyanide) is a typical PBA material for Zn-ion storage (Supplementary Fig. 36)^60,61, which delivers a reasonable voltage of ~1.54 V in the conventional battery configuration (Supplementary Fig. 37). When applied in the concentration M‖Cathode design, this cathode exhibits similar GCD curves but demonstrates an improved voltage of 2.24 V (Fig. 7f), raising its voltage by +0.7 V. Consequently, the specific energy increases from 105 to 148 Wh kg⁻¹, marking a 41% improvement. This 2.24 V concentration battery exhibits a stable cycling performance of 500 cycles (Supplementary Fig. 38). Similarly, the Zn‖LiMn₂O₄ (LiMn₂O₄: lithium manganese oxide) concentration battery exploits the ~0.7 voltage enhancement and achieves a high voltage of ~2.45 V (Fig. 7g and Supplementary Figs. 39 and 40). The specific energy increases from 180 to 245 Wh kg⁻¹, corresponding to 36% improvement. This 2.45 V concentration battery shows promising cycling life (Supplementary Fig. 41).

Besides, the concentration mechanism can be utilized to fabricate high-voltage primary batteries, including Zn‖MnO₂ and Zn‖S batteries (MnO₂: manganese dioxide; S: sulfur). As shown in Fig. 7h, the MnO₂ cathode (Supplementary Fig. 42) delivers a high voltage of 2.27 V and a high capacity of ~276 mAh g⁻¹, corresponding to a competitive specific energy of 626 Wh kg⁻¹. This cell voltage is higher than other primary Zn batteries, such as zinc-carbon (1.5 V), alkaline zinc (1.5 V), zinc-silver oxide (1.55 V), and zinc-mercury oxide batteries (1.35 V, Supplementary Table 15). Additionally, the Zn‖S primary battery exhibits a high capacity of 972 mAh g⁻¹ and a reinforced voltage of ~0.95 V (Fig. 7i), raising its specific energy from 286 to 939 Wh kg⁻¹, corresponding to 228% improvement (Supplementary Fig. 43).

Extension to other metal-based concentration batteries

We vision that the concentration battery holds great potential to extend to various metal batteries, including monovalent (silver), divalent (manganese, cobalt, nickel, copper, cadmium), and trivalent metals (aluminum, indium, antimony). Following the same principle, the catholyte and anolyte are concentrated solutions and metal-complex solutions, respectively. For readers’ convenience, we list some soluble salts that could work as concentrated catholytes in Supplementary Table 16. On the anolyte side, numerous complex agents can be explored and optimized, such as ammonia, citrate, tartrate, ethylenediaminetetraacetic acid, and ethylenediamine, etc.

To demonstrate the efficacy of our approach, we conducted preliminary studies on Cu‖Cu concentration batteries. The catholyte is a concentrated copper nitrate (Cu(NO₃)₂) solution (6.66 m), which increases the Cu²⁺/Cu potential from +0.34 to +0.48 V vs. SHE (Fig. 8a). Meanwhile, the anolyte is 0.5 m [Cu(NH₃)₄]²⁺ complex solution (Supplementary Fig. 44), which decreases the Cu²⁺/Cu potential to −0.10 V (Fig. 8a). When combined into a concentration Cu‖Cu battery, we obtain an average voltage of ~0.58 V (Fig. 8b), aligning well with the potential difference. Besides, it supports stable cycling for 575 h (Supplementary Fig. 45), indicating its reaction reversibility. With the emergency of aqueous Cu batteries, our data could facilitate the demonstration of high-voltage Cu full cells. Overall, this result consolidates our approach in developing efficient concentration batteries for energy storage.

Fig. 8. Extension to concentration Cu‖Cu batteries.

a GCD curves of symmetrical Cu‖Cu cells in different electrolytes (0.2 mA cm⁻² for 0.2 mAh cm⁻²). b GCD curves of concentration Cu‖Cu batteries (10 μA cm⁻² for 10 μAh cm⁻²).

Advantages and limitations of concentration batteries

In this section, we highlight several unique advantages of concentration batteries, and we also discuss their current limitations and propose alternative strategies toward applications.

Firstly, concentration batteries represent an appealing yet underexplored version of Galvanic batteries, which enables the use of multivalent metals as cathodes. In traditional aqueous batteries, metals are almost exclusively employed as anodes, whereas cathodes are typically intercalation or conversion-type materials with limited capacity. By contrast, the concentration battery elevates the Mⁿ⁺/M redox potential through the solvation effects, thereby facilitating metals to function as cathodes. It is worth noting that metals such as Mn, Fe, Zn, Cu, and Sb possess high theoretical capacities (660–976 mAh g⁻¹), high density (> 6.6 g cm⁻³), low cost (0.424–6.0 USD kg⁻¹), and high electric conductivity (10⁵–10⁷ S cm⁻¹), making them attractive cathode alternative to common cathodes like vanadium pentoxide (V₂O₅) and MnO₂ (Fig. 9a, b, and Supplementary Table 17)^2,3,9,10. Therefore, concentration batteries hold the promise to open alternative opportunities for dual-metal plating chemistry for energy storage.

Fig. 9. Advantages and limitations of concentration batteries.

a Gravimetric capacities and densities of metals and common cathodes (V₂O₅ and MnO₂). b The price of metals and common cathodes (V₂O₅ and MnO₂). c The proposed anolyte electrolyte with higher osmotic pressures. d The scheme of concentration batteries based on anion-exchange membranes. e GCD curves of AEM-based Zn‖Zn concentration batteries (10 μA cm⁻² for 10 μAh cm⁻²). f GCD curves of AEM-based Zn‖ZnFe-PBA concentration batteries at 50 mA g⁻¹. In this figure (a), colored spheres denote Fe, Mn, Zn, Cu, and Sb metals, each of which exhibits well-defined capacity and density values. In contrast, colored ellipses denote MnO₂ and V₂O₅ materials, whose reported capacities vary across the literature and are therefore represented as ranges. In this figure (d), AEM denotes an anion-exchange membrane.

Secondly, concentration batteries efficiently harness electrode potential shifts induced by two distinct electrolytes, providing a different route to fabricate high-voltage aqueous batteries. Unlike conventional batteries that rely on developing high-potential cathodes, concentration batteries leverage thermodynamic potential differences arising from solvation structure and activity changes in separated electrolytes. This advantage is substantiated by our concentration M‖cathode batteries (Fig. 7f–i), where various cathodes (ZnFe-PBA, LiMn₂O₄, MnO₂, and S/C) exhibit significantly enhanced voltage and specific energy compared with conventional aqueous batteries.

Thirdly, concentration batteries offer a potentially cost-effective and scalable architecture. Particularly, the ability to operate as a symmetrical M‖M system, where both electrodes utilize the same metal as active material, eliminates the need for expensive or specially engineered cathodes. This symmetry simplifies the material development and manufacturing costs, and it allows the use of identical electrode fabrication processes on both sides. This configuration is advantageous for large-scale manufacturing, as it minimizes the process complexity and ensures consistency across cells.

Despite these attractive features, concentration batteries also face several notable limitations that require careful optimization and engineering. One primary challenge arises from the use of two distinct electrolytes, which inevitably introduces a difference in electrolyte composition and osmotic pressure. This difference promotes water migration and gradual mixing between two compartments. In the present study, a salt bridge is employed as a tentative strategy to decouple two electrolytes and retard water migration. Nevertheless, over extended cycling, partial water exchange can still occur, leading to electrolyte concentration changes and the gradual cell voltage decline (Supplementary Fig. 46).

Indeed, when two diluted electrolytes with comparable osmotic pressure (e.g., 0.5 m zinc sulfate and 0.5 m copper sulfate) are employed, the Zn‖Cu battery exhibits much improved cycling stability with suppressed voltage decay (Supplementary Fig. 47), suggesting the role of osmotic imbalance in driving electrolyte mixing. To further substantiate this effect, we disassembled the Zn‖Zn concentration batteries after cycling and tested the density of cycled electrolytes. We found that the catholyte density decreased from 2.26 to 2.03 g cm⁻³, whereas the anolyte density increased from 1.10 to 1.11 g cm⁻³ (Supplementary Fig. 48). We also evaluated the Zn²⁺/Zn potential in the cycled electrolytes, where the Zn²⁺/Zn cathode potential decreases from −0.42 to −0.479 V, while the Zn²⁺/Zn anode potential raises from −1.12 to −1.089 V (Supplementary Fig. 49). These observations provide direct evidence for water exchange driven by osmotic pressure differences.

To mitigate this challenge, we propose introducing redox-inactive additives to increase the overall concentration and osmotic pressure of the anolyte without significantly altering the Zn²⁺/Zn anode potential. Sugars represent a large class of carbohydrate compounds with high and tunable solubility in water (Supplementary Table 18). Owing to their strong hydrogen-bonding interactions with H₂O molecules, sugars can effectively increase the electrolyte osmotic pressure while remaining electrochemically inert. As a proof of concept, we investigated the solubility of sucrose in the 0.5 m [Zn(NH₃)₄]²⁺ + 5 m LiCl anolyte. The maximal solubility reaches 5.445 m sucrose (Fig. 9c), and the resulting anolyte exhibits an osmotic pressure comparable to that of the 30 m ZnCl₂ + 5 m LiCl catholyte, as evidenced by the unchanged solution heights during storage (Supplementary Fig. 46). Notably, the introduction of sucrose only slightly increases the Zn²⁺/Zn anode potential from −1.12 to −1.0 V (Supplementary Fig. 50), which is within an acceptable range and likely due to the reduction in water activity.

A second limitation of concentration batteries is that electrolytes constitute part of the active mass, meaning that relatively large electrolyte volumes are required to maintain charge and mass balance. In this regard, concentration batteries share similarities with redox-flow batteries rather than conventional rechargeable batteries. At the current stage of development, this characteristic makes it challenging to directly make concentration batteries using coin-cell or pouch-cell formats. To address this issue and move toward more integrated configurations, we propose anion-exchange membranes (AEMs) as a more feasible alternative to salt bridges (Fig. 9d and Supplementary Fig. 51). AEMs are commercially available, relatively inexpensive, thinner, and lighter than salt bridges (Supplementary Fig. 52), and they have been widely used in aqueous batteries to decouple distinct electrolytes and redox reactions^62–70, as summarized in Supplementary Table 19. We assembled Zn‖Zn concentration batteries based on two AEMs, a 30 m ZnCl₂ + 5 m LiCl catholyte, a 0.5 m [Zn(NH₃)₄]²⁺ + 5 m LiCl + 5.445 m sucrose anolyte, and a supporting electrolyte of 5 m LiCl + 5.445 m sucrose (Supplementary Fig. 53). As shown in Fig. 9e, this AEM-based concentration battery delivers a voltage of ~0.57 V and maintains stable cycling performance. Furthermore, common cathodes like ZnFe-PBA and LiMn₂O₄ can be readily integrated with AEMs to construct high-voltage aqueous concentration batteries (Fig. 9f and Supplementary Figs. 54 and 55), thereby validating the feasibility of transitioning from salt bridge model systems to more practical cell configurations.

Lastly, we emphasize that this work primarily aims to revisit the fundamental mechanism of concentration batteries and demonstrate feasible alternative pathways for their realization. Many engineering and technical challenges, such as long-term membrane stability, electrolyte management, and scalable cell design, are beyond the scope of the present study. Addressing these challenges will require interdisciplinary collaboration among electrolyte chemists, membrane scientists, and battery engineers. We hope that the mechanistic insights and design principles presented here will stimulate more research efforts toward further optimization.

Discussion

We have developed an alternative concentration battery technology that potentially redefines how cell voltage can be expanded. This approach utilizes the same redox couple (Mⁿ⁺/M) at both the cathode and anode, leveraging the electrolyte concentration and solvation structure differences to produce a meaningful full cell voltage. Using a concentrated catholyte and a metal-complex anolyte, the M‖M concentration battery achieves a feasible cell voltage of ~0.7 V, higher than the standard of 59/n millivolts. Moreover, this concentration cell configuration is highly compatible with various cathodes (PBAs, oxides, and elements), further increasing their voltages by 0.7 V. As a result, high-voltage aqueous Zn‖Cathode batteries (2.2–2.5 V) have been demonstrated. Additionally, the concentration mechanism is applicable to other metal systems (such as copper), further validating its efficacy for energy storage.

Methods

Electrolyte preparation

All aqueous electrolytes were prepared using high-purity distilled water (resistivity > 1.0 MΩ·cm at 25 °C) produced by a Barnstead Mega-Pure Still (Model MP-11A, Thermo Scientific, Waltham, MA, USA). To prepare ZnCl₂ catholyte solutions, we added 5, 25, 75, and 150 mmol ZnCl₂ salts (Sigma Aldrich, ≥98% purity) into 5 g deionized water, leading to 1, 5, 15, and 30 m ZnCl₂ electrolytes, respectively. To make the 5 m LiCl + 30 m ZnCl₂ electrolyte, we further dissolved 25 mmol LiCl salts (Sigma Aldrich, ≥99% purity) into the above 30 m ZnCl₂ solution with thorough stirring and dissolution. Other catholyte solutions (such as ZnBr₂ and ZnI₂, Sigma Aldrich, ≥98% purity) were prepared similarly.

To make the 0.5 m zinc-ammonia complex solution, we first prepare 0.8 m ZnCl₂ electrolyte and then slowly add concentrated ammonia solution (25 wt.%, Fisher Scientific) dropwise with continuous stirring. In the beginning, white precipitation occurs due to the Zn(OH)₂ formation. Subsequently, the precipitation re-dissolves because of the [Zn(NH₃)₄]²⁺ complex formation. The molar ratio of NH₃/Zn²⁺ is slightly over 6:1 (6.1:1) to ensure full dissolution and complexation. After dissolution, extra water is added to make the Zn²⁺ concentration reach 0.5 m. The LiCl salts (Sigma Aldrich, ≥99% purity) are added to prepare the 0.5 m [Zn (NH₃)₄]²⁺ + 5 m LiCl solution. To make the 0.5 m [Zn (NH₃)₄]²⁺ + 5 m LiCl + 5.445 m sucrose anolyte, we could add 1.4 g sucrose (Sigma Aldrich, ≥99.5% purity) to 1 g electrolyte of 0.5 m [Zn (NH₃)₄]²⁺ + 5 m LiCl. We also added 15 m LiCl to 1 m ZnCl₂ to make the [ZnCl₄]²⁻ complex solution. The 1 m [Zn(OH)₄]²⁻ solution was made in a similar method, where the NaOH/ZnCl₂ molar ratio is slightly over 6:1 (6.1:1) to ensure full dissolution and complexation. The anhydrous sodium hydroxide (NaOH) pellets were purchased from Sigma Aldrich with a purity of ≥99% purity. The 0.5 m [Cu(NH₃)₄]²⁺ solution was prepared in a comparable manner to the zinc-ammonia electrolyte, using the copper nitrate trihydrate as the precursor (Sigma Aldrich, puriss. p.a., 98–103%).

To compare the electrolyte osmotic pressure, we used the commercial anion-exchange membrane (fumasep®FAB-PK-130, Fuelcellstore) to decouple these two electrolytes, which could allow water molecules to diffuse during the storage process.

Material synthesis

The Zn₃[Fe(CN)₆]₂ material was prepared by a simple precipitation method. Typically, a solution of ZnSO₄ (Sigma Aldrich, 99% purity; 40 mL, 0.09 M) was added dropwise to the other solution of potassium ferricyanide (K₃[Fe(CN)₆]) (Sigma Aldrich, purity ≥99.0% purity; 40 mL, 0.06 M) under constant stirring. After reacting for 4 h, the dark-brown precipitates were washed and centrifuged multiple times, which were allowed to dry naturally in air. Each washing cycle involved centrifugation at 1000 × g for 30 min using an Adams Analytical Centrifuge (Model CT-3200, Clay-Adams, Inc., USA).

The commercial cathode of LiMn₂O₄ was purchased from Landt Battery Company.

The β-phase MnO₂ material was synthesized by a hydrothermal method. Firstly, 0.25 g of polyvinyl pyrrolidone (PVP, K-30, MW = 40,000, Sigma Aldrich) was dissolved in 20 mL H₂O. Then 0.13 g potassium permanganate (KMnO₄, Fisher Scientific, ≥99% purity) was dissolved in 10 mL H₂O, which was added dropwise to the PVP solution under continuous stirring. In a separate flask, 0.05 g ammonium sulfate ((NH₄)₂SO₄, Thermo Scientific, 99% purity) was dissolved in 10 mL H₂O, which was added dropwise to the above solution. After vigorous stirring for 15 min, the resulting mixture solution was transferred to a 100 mL Teflon lined stainless steel autoclave. The hydrothermal reaction was maintained at 130 °C for 10 h. After the reaction cooled down to room temperature, the obtained precipitate was centrifuged and washed with distilled water three times. The centrifuge condition is same as the ZnFe-PBA. The material was dried at 70 °C for six hours, which was further calcinated in a muffle furnace for 3 h (target temperature: 450 °C, ramp rate: 5 °C min⁻¹).

The sulfur/carbon composite (60 wt.% sulfur) cathode was prepared by a well-established melt-diffusion method. Typically, 0.3 g sulfur (Sigma Aldrich, 99.5 ~ 100.5% purity) and 0.2 g Ketjen black carbon (EC-300, AkzoNobel, USA) were grinded in a mortar for 30 min, and then the mixture was transferred to a planetary ball-milling jar and subjected to ball-milling at 400 rpm for 5 h. Then S/C mixture was made into a pellet under a hydraulic press at 3 tons, which was then transferred to an autoclave for melt-infusion reactions. The reaction temperature is 155 °C, and the reaction time is 6 h. After the reaction was finished, the pellet was ground into fine S/C powders for use.

Electrode preparation

The working electrode was prepared by mixing the active material (Zn₃[Fe(CN)₆]₂, MnO₂, LiMn₂O₄, or S/C), Ketjen black carbon, and polyvinylidene fluoride (MTI corporation, ≥99.5% purity, MW = 600000) binder with N-methyl-2-pyrrolidone (Sigma Aldrich, 99.5% purity) the solvent into a homogenous slurry solution (solvent-to-solid ratio of 2.5:1 by weight). The resulting slurry was manually coated (hand-cast) onto the carbon fiber current collector (Fuel Cell Store, brand: AvCarb MGL370, thickness: 0.37 mm; diameter: 1.0 cm) using a stainless-steel spatula. The carbon fiber current collector was used as received without further etching treatment. The carbon fiber papers were cut by the precision disc cutter (MTI corporation) into circular shapes with a diameter of 1.0 cm (area: 0.785 cm²). The electrodes were fully dried in an air-forced oven at 60 °C for 12 h. The mass ratio between active mass, carbon, and binder is 7:2:1. To ensure consistent mass loading, the coating thickness was carefully controlled, and the active mass loading is in the range of 1–1.5 mg cm⁻².

The zinc foil (99.95 % purity, 0.1 mm thickness, Kolamoon) was cut by the precision disc cutter into round-shape electrodes for use in symmetrical Zn‖Zn cells, whose area is 1.26 cm² (diameter: 1/2 inches). The copper foil (99.9% purity, 0.05 mm thickness, Tynulox) was also cut by the precision disc cutter into the round-shape electrodes (diameter: 1/2 inches, area: 1.26 cm²) for symmetrical Cu‖Cu cells. Both metal foils were used as received without mechanical polishing or chemical etching. To make concentration Zn‖Zn and Cu‖Cu batteries, we used the

Battery assembly and testing

To monitor the Zn²⁺/Zn redox potential, three-electrode Zn‖Zn symmetrical cells were assembled in Swagelok cells, where the working, counter, and reference electrode is Zn foil, Zn foil, and Ag/AgCl electrode (saturated, +0.20 V vs. SHE), respectively. The Swagelok cells contain PTFE (Teflon) cases, titanium rod plungers, and two plastic O-springs to ensure mechanical pressure (spring constant is not available). Glass fiber papers (Whatman, Grade GF/A, 12.7 mm diameter, 0.26 mm thickness, ~90% porosity, ~1.6 μm average pore size, separator material: high-purity borosilicate glass microfibers without organic binders) were used as separators, and the electrolyte amount is ~500 μL in Swagelok cells.

To evaluate the Zn plating efficiency, we assembled two-electrode Swagelok cells, where Zn foil is the counter/reference electrode, and the Ti foil (Futt brand, 0.03 mm thickness, purity: N/A) or Cu foil (MTI corporation, 9 μm thickness, purity: N/A) is the working electrode. For Zn plating in catholytes, Cu foil was used as the substrate, due to the intermediate Zn_xCu alloy formation and the facilitated Zn plating process. For Zn plating in anolytes, Ti foil was used as the substrate, due to its higher inhibition against hydrogen evolution side reactions. Both Ti and Cu foils are used directly for Zn plating without any pre-treatment. The Zn plating Coulombic efficiency is defined as stripping capacity (charge capacity) divided by the plating capacity (discharge capacity). The Zn plating batteries are not associated with a precycling procedure; instead, they directly start with a plating process, followed by a stripping process, and then repeat the plating/stripping cycles. The plating condition is 1 mA cm⁻² for 0.5 mAh cm⁻², and the charge cut-off voltage is 0.5 V.

The conventional Zn‖Cathode batteries were assembled in two-electrode Swagelok cells, where these cathodes were separated from the Zn metal anode by glass fiber separators. The electrolyte was 30 m ZnCl₂ + 5 m LiCl. For the ZnFe-PBA and LiMn₂O₄ batteries, their charge/discharge cut-off voltage range is 0.8−1.8 and 1.5–2.0 V, respectively. We used a specific current of 100 mA g⁻¹ for ZnFe-PBA and LiMn₂O₄ batteries. For the MnO₂ battery, the discharge cut-off voltage is 0.8 V, and the specific current is 50 mA g⁻¹. For the S/C battery, the discharge cut-off voltage is 0 V, and the specific current is 100 mA g⁻¹.

The concentration Zn‖Cathode batteries were assembled in the two-electrode “breaker cell” configuration, with two different electrolytes stored in two glass vials that were further connected by the KCl/agar hydrogel gel (salt bridge; KCl, Sigma Aldrich, purity ≥99%; agar, ALDON corporation, Innovating Science Mixed Nutrient Agar, purity ≥95%). The cathode is immersed in the 30 m ZnCl₂ + 5 m LiCl catholyte, whereas Zn metal is immersed in the 0.5 m [Zn (NH₃)₄]²⁺ + 5 m LiCl anolyte. To prepare the salt bridge, we weigh 12 g of agar into a beaker and heat it in a water bath until fully dissolved. Then, 4 g KCl is added to the solution under stirring for thorough dissolution. While still hot, this KCl/agar solution is carefully transferred into a plastic tube using a dropper. The tube opening is sealed with cotton to prevent the formation of air bubbles. The solution is left to stand until the agar solidifies, which will be ready for use. To make concentration batteries based on anion-exchange membranes (AEMs), we used three cuvettes as the compartments and two layers of AEMs as separators. The AEM is much thinner and lighter than the salt bridge, and it is a more feasible approach toward applications. For the ZnFe-PBA and LiMn₂O₄ concentration batteries (salt bridge based), their charge/discharge cut-off voltage range is 2.2–2.6 and 2.2–2.65 V, respectively. We used a specific current of 100 mA g⁻¹ for the ZnFe-PBA and LiMn₂O₄ concentration batteries. For the AEM-based ZnFe-PBA concentration battery, the charge/discharge voltage range is 1.4–2.5 V, and the specific current is 50 mA g⁻¹. For the MnO₂ concentration battery, the discharge cut-off voltage is 1.5 V, and the specific current is 50 mA g⁻¹. For the S/C concentration battery, the discharge cut-off voltage is 0.75 V, and the specific current is 100 mA g⁻¹.

Galvanostatic charge/discharge curves, rate capability, and cycling performance were tested on the Landt battery cycler (CT3002AU). Electrochemical impedance spectrum (EIS) and linear scanning voltammetry (LSV) of electrolytes were tested on the Biologic SP-150 Potentiostat. LSV curves for the 30 m ZnCl₂ + 5 m LiCl and 20 m ZnBr₂ + 2 m LiBr electrolytes were tested from 0 to 1.5 V vs. Ag/AgCl with a scanning rate of 1 mV s⁻¹. EIS measurements were performed in the potentiostatic mode within a frequency range of 200 kHz to 20 mHz with a sinusoidal signal amplitude of 5 mV. All measurements were carried out at open-circuit voltage (OCV). To ensure a quasi-stationary state, the cells were rested at OCV for at least 2 h before the EIS measurements. The data were collected at a density of 6 points per decade of frequency. For quantitative analysis, the spectra were fitted using EC-Lab software.

All electrochemical tests were performed without using a climatic/environmental chamber; instead, they were conducted in an AC-controlled laboratory environment with a temperature range of 25 ± 1 °C. For the full cell testing, the Zn negative electrodes were used directly after mechanical cutting, without pre-activation or electrochemical pre-cycling procedures. For each electrochemical experiment, at least two batteries were assembled and tested to ensure reproducibility.

Physical characterization

X-ray diffraction (XRD) patterns of these cathode materials were collected on the Rigaku SuperNova equipped with a HyPix3000 X-ray detector and CuKα radiation source (λ = 1.5406 Å). The XRD patterns were recorded in the 2-theta range of 10–80° with a scan rate of 5° min⁻¹ and a step size of 0.02°. Scanning electron microscopy (SEM) images and energy dispersive X-ray spectra (EDS) mapping results were recorded at a field emission SEM (JEOL, JSM-6480LV). The SEM-EDS analysis was performed at an acceleration voltage of 20 kV.

The Raman spectra of the electrolytes were obtained using a 532 nm spectrally stabilized laser module (Innovative Photonic Solutions) in a Fiber Optic Raman Probe. Thermogravimetric analysis (TGA) was conducted on the Mettler Toledo TGA2 instrument in the nitrogen atmosphere with a ramp rate of 10 °C min⁻¹. Fourier-transform infrared spectroscopy (FT-IR) was tested by the Perkin Elmer instrument.

Regarding the ex-situ characterizations of the Zn deposits, we disassembled the cells after the first plating process (1 mA cm⁻² for 0.5 mAh cm⁻²) in ambient air. The electrodes were harvested and immediately rinsed with distilled water to remove residual electrolyte and salts. The Zn samples were then dried at 25 ± 1 °C under vacuum in the antechamber of the Vigor glove box (O₂/H₂O content: <1 ppm). Before transporting to the SEM measurements, the Zn electrodes were sealed and stored in glass vials inside the glovebox to maintain a high-purity Argon atmosphere, which can avoid side reactions with air and moisture. Regarding the ex-situ characterizations of the cycled electrolytes, we disassembled the Zn‖Zn concentration batteries after 750 cycles (cycling condition: 12.5 μA cm⁻² for 12.5 μAh cm⁻²) in ambient air, retrieved the cycled electrolytes using pipettes, and stored them tightly in clean glass vials. Then, we used the grade cylinder to measure the density of cycled electrolytes. We took 1 mL of the cycled electrolyte and measured its mass to calculate the electrolyte density. The test for each electrolyte was repeated 3 times to calculate the average value. To provide additional evidence for the concentration change, we also assembled three-electrode Zn‖Zn symmetrical cells using the cycled electrolytes, with the Ag/AgCl as the reference electrode.

Theoretical calculations

Computational details

Molecular dynamic (MD) simulations were conducted to investigate the solvation structures of the electrolyte solutions, including 1 m ZnCl₂, 5 m ZnCl₂, 15 m ZnCl₂, 30 m ZnCl₂, 30 m ZnCl₂ + 5 m LiCl, 0.5 m Zn(NH₃)₄Cl₂ + 1 m NH₃, and 0.5 m Zn(NH₃)₄Cl₂ + 1 m NH₃ + 5 m LiCl. The modeling details are provided in Supplementary Table 20. These systems were constructed into cubic simulation boxes using PACKMOL package⁷¹. All MD simulations were performed using GROMACS package⁷² with the all-atom optimized potentials for liquid simulations (OPLS-AA) force field⁷³. Equations of motion were integrated with a time step of 2.0 fs. We first carried out energy minimization using the steepest descent algorithm. Then, an NVT ensemble (the number of particles (N), volume (V), and temperature (T) were kept as constants) was conducted using the velocity thermostat, for 2.0 ns of equilibration. Next, an NPT ensemble (the number of particles (N), pressure (P), and temperature (T) were kept as constants) was performed at 1.0 atmosphere for 2.0 ns. Finally, an NPT production run of 20 ns was performed with a time step of 2.0 fs, using periodic boundary conditions in all directions. Long-range electrostatics were treated using the PME method, with a cutoff of 1.2 Å for Coulomb and van der Waals interactions. For the systems of 0.5 m Zn(NH₃)₄Cl₂, 0.5 m Zn(NH₃)₄Cl₂ + 5 m LiCl, the significant covalent bonding between the Zn²⁺ ion and the N atoms necessitates treating the [Zn(NH₃)₄]²⁺ group as a rigid, consolidated fragment during molecular simulations to ensure computational efficiency and model accuracy. The force field parameters are provided in the Supplementary Table 21. Note that the simulated densities of these electrolytes are close to the experimentally determined densities (Supplementary Table 22 and Supplementary Fig. 56), indicating that the adopted force field parameters can accurately simulate the potential energies and solvation structures of different electrolyte systems, thereby providing a reliable analysis for the experimental phenomenon.

The potential energy was calculated using Lennard-Jones (LJ) and Coulomb interactions with consideration of both short- and long-range contributions. To compare potential energies across systems with different atom counts (due to varying electrolyte concentrations), the absolute values of potential energies were normalized by the total number of atoms.

The RDF between Zn²⁺ and coordinating atoms (M) or solvent was calculated using:

$$g\left(r\right)=\frac{1}{4{\pi }^2{\rho }_M}⟨\frac{d{n}_M\left(r\right)}{dr}⟩$$

where ${\rho }_M$ is the number of density of species M. The CN was obtained via integration:

$$\mathrm{CN}\left(r_c\right)=4\pi {\rho }_M{\int }_0^{r_c}{r}^{2}g\left(r\right)dr$$

where r_c is the cutoff radius. In this work, r = 10 Å (1.0 nm) was set for the RDF and CN calculations, but r_c = 3 Å (0.3 nm), corresponding to the first minimum in the RDF curve was used to calculate the CN of the primary solvation shell.

Author contributions

S.Q. and X.W. conceived the idea and designed the experiments. W.H. conducted most of the material syntheses, material characterizations, electrolyte preparations, battery assembly, and electrochemical characterizations, and S.C., I.U. and X.D. provided additional assistance with these experiments. A.V.-P. and L.C. conducted the Raman analysis of electrolytes. J.L.R. and L.D.-V. performed the FT-IR and TGA tests. D.M., P.C., G.M., F.Y. and X.W. provided research resources, laboratory facilities, and funding support. Z.W. conducted all the theoretical simulations under the supervision of F.Y. The initial manuscript was prepared by W.H., S.Q. and X.W. and all authors discussed the results, participated in the final manuscript revision, and approved the final version. W.H. and Z.W. contributed equally to this work.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. [A peer review file is available.]

Funding

X.W. discloses support for the research of this work from the National Science Foundation [grant number 2434152], NASA EPSCoR [grant number 80NSSC23M0189 and grant number 80NSSC24M0107], and the Center for the Resilience to Climate Change at the University of Puerto Rico-Río Piedras funded by the U.S. Department of Education [PR/Award number P116H240025]. S.Q. discloses support for the research of this work from the NASA EPSCoR [grant number 80NSSC23M0189]. Z.W. and F.Y. disclose support for the research of this work from the Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship, a program of Schmidt Sciences, LLC. All other authors declare no relevant funding.

Data availability

The data that support the findings of this study are available in the article and its Supplementary Information. Source Data are provided with this paper. Source data have been deposited in the Figshare repository with the identifier (10.6084/m9.figshare.31416362)⁷⁴. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74015-x.