Introduction

The chlorine (Cl2) redox reaction is utilized for gas‒liquid phase-conversion reactions and exhibits a reaction potential of 1.36 V versus SHE (standard hydrogen electrode at 25 °C) and a theoretical gravimetric capacity of 756 mAh g−1 based on a one-electron Cl-based conversion reaction1,2. Looking into the ever efforts on developing Cl2-based batteries, the earliest prototype was first proposed in 1884 as a ZnǁCl2 battery system and was further utilized with a 500 kW h battery run by Energy Development Associates in the 1980s3. However, research on Cl-redox electrodes was largely limited within the past four decades due to the leakage of gaseous Cl2, causing safety concerns. Although the Cl-redox feature was recently exploited by confining the oxidized Cl-0.25 inside a graphite interlayer based on intercalation chemistry4, the most common approach to fix the oxidized Cl2-based cathode is to apply adsorption-type host materials, such as activated carbon (AC) and graphite5,6,7,8, in which interactions between Cl2 and the carbon-based host materials involve physical adsorption. As a result, the diffusion of gaseous toxic Cl2 could not be fully eliminated. Another possible strategy is to store gaseous Cl2 in low-temperature water (below 9.6 °C) to form a solid-state chlorine hydrate as Cl2•xH2O (x ≈ 5.9)9,10, in which the electrochemically active Cl2 gas is released by heating once the ZnǁCl2 battery is discharged. However, the main drawback is the necessity of a complex thermal and gas management system to run the gaseous Cl2 electrode loop9,10.

Irreversible Cl-redox reactions in battery electrodes generally occur in aqueous electrolytes not only because of the Cl2 fixing issue at the positive electrode side but also because of the competing oxygen evolution reaction (OER as 1.23 V versus SHE at 25 °C), which exhibits an onset potential 0.13 V lower than that needed to trigger the Cl2 evolution reaction (ClER) (similar to the competing relationship between the OER and ClER in the seawater splitting process11). In addition, the tendency of Cl2 to solubilize in water aggravates the instability of the Cl2 electrode12. To avoid the OER, the aqueous electrolytes were replaced with molten metal chlorides that operate at high temperature (for example >130 °C in ref. 7); however, the flowing of a gaseous Cl2 electrode was still necessary in the LiǁCl2 and AlǁCl2 cell configurations7,13,14. Regarding the metalǁCl2 configurations, NaǁCl2 and LiǁCl2 batteries have been recently developed by confining the oxidized Cl2 inside highly microporous carbon15. However, stabilizing reversible Cl-based redox reactions (ClRRs) in aqueous electrolytes is challenging because of the drawbacks associated with the electrodes and the electrolyte, i.e., fixing the oxidized Cl0 without producing gaseous Cl2 and simultaneously ensuring that the oxidation reaction potential of Cl- is below the OER.

Based on this state-of-the-art knowledge, we further explore feasible strategies to exploit ClRR for battery applications. The selection principles are utilized to fix agents at the electrodes and electrolyte. Regarding fixing agents for oxidized Cl0 at the electrode side, the ideal goal is to store Cl0 in the solid state; the strategy of fixing Cl2 by H2O molecules in solid-state chlorine hydrate (Cl2•xH2O)1,16 sheds light on the feasibility of exploiting other coordinating agents to fix oxidized Cl0. Theoretically, there is one basic requirement for the coordinating agents, namely, they should form strong chemical bonds with oxidized Cl0. Given the Lewis acid nature of Cl0, i.e., its strong electrophilicity, Cl0 spontaneously produces gaseous Cl2 molecules in the energy stable form; thus, it should be more energetically favourable for coordinating agents to bind with Cl0 and prevent the production of Cl2 molecules. Based on this consideration, we examined I atoms with high electronegativity and larger atomic weights17, showing the possibility of using I atoms as fixing agents for Cl0 to store the final products in terms of the solid state. On the other hand, regarding the electrolytes, the main goal is to regulate the H2O activity of the applied electrolytes to lower levels, retarding the onset potential of the OER and inhibiting the dissolving tendency of the final products of Cl0.

In this research work, we apply I atoms as the coordinating agents to fix the oxidized Cl0 based on interhalogen coordinating chemistry in a highly concentrated ZnCl2 electrolyte, i.e., 30 m ZnCl2 (where m is molality as mol kg–1 calculated as moles of solute divided by the mass of the solvent) and a Zn metal negative electrode. Reversible ClRR can be obtained and deliver an average discharge voltage of 1.88 V and a discharge capacity of 210 mAh gI2−1 at 0.5 A gI2−1 and 25 °C (based on the mass of fixing agents of iodine unless otherwise specified, Supplementary Note 1). Two halogen-based redox centres based on Cl and I as the I-based positive electrode were examined, and the electrode accommodates cascade interhalogen reactions with three-electron transfer processes, i.e., two electrons from I and one electron from Cl. This interhalogen electrode in a hybrid Zn metal battery system enables a discharge capacity of 612.5 mAh gI2−1 at 0.5 A gI2−1 and 25 °C and an average discharge voltage of 1.48 V, which translates to a calculated specific energy of 905 Wh k gI2−1.

Results and discussion

Electrochemical investigation on the redox activity of iodine-based electrodes in Cl-ion aqueous electrolyte solutions

We first studied ClRR as an independent reaction. Therefore, the working potential range was confined within 1.7 V to 2 V to capture the electrochemical features of the ClRR. The electrolyte applied here was 30 m ZnCl2 at 25 °C in a two-electrode pouch cell configuration. To verify the role of I in fixing ClRR, a blank control experiment was conducted in which the corresponding cyclic voltammetry (CV) and galvanostatic charge‒discharge (GCD) profiles were collected from the I-containing and I-absent samples, respectively. There was one noticeable cathodic peak with the I-containing electrode, and the corresponding peak current density was 21 times larger than that of the I-absent electrode, i.e., 3.48 mA cm−2 versus 0.165 mA cm−2. Such pronounced contrast indicates the promotion of ClRR from I-containing electrodes (Fig. 1a).

Fig. 1: Electrochemical characterizations of iodine activity towards ClRR.
figure 1

a CV profiles at 1 mV s−1 and b GCD profiles of the I-containing electrode cycled at 0.5 A gI2−1, and I-absent electrode cycled at 0.5 A gAC−1 based on AC mass, both in 30 m ZnCl2 electrolytes and in two-electrode pouch cells at 25 °C. c Operando gaseous Cl2 detection of Cl2 evolution during the charging process of I-absent and I-containing electrodes. Note that the scales of the abscissas are different by 10-fold.

Full size image

The capacity of the cell with the I-absent electrode does not show an apparent discharging plateau from the Cl-redox reactions, which is assigned to the dominant surface charge storage capacity and slight Cl0-redox capacity; together, these capacities contribute to the output capacity of the AC host in the positive electrode, as shown in Supplementary Fig. 1. On the other hand, the corresponding capacity of the I-containing electrode is 215 mAh gI2 −1 at 0.5 A gI2−1 and 25 °C, and the charge storage mechanisms of the iodine (ClRR) and AC (capacitive) jointy contribute (Fig. 1b). The average discharging voltage is 1.88 V, which is different from the voltage plateau resulting from the iodine-based redox as I0 ↔ I− (~1.25 V) and I0+ ↔ I−(~1.60 V). Notably, the valence state of iodine was maintained at +1 in the specified voltage window (1.7 V to 2 V), and the corresponding redox behaviours of the two I-based electron-transfer reactions transitioning from initial I− to I0 and to I+ are shown in Supplementary Fig. 2, in which GCD and in situ Raman measurements (in the 0.6 V to 1.75 V cell voltage range) and analyses are reported18,19.

The cathodic peak in the CV curve shown in Fig. 1a correlates to one electron-transfer reaction, which can be assigned to the Cl0/Cl− reduction reaction or extraction behaviour of the anionic ions as [ZnClx]2-x in the carbon-based host at the positive electrode20. To confirm the mechanism, in situ measurements of the Cl2 gas were carried out to monitor the Cl2 evolution at both the positive and negative electrodes (Supplementary Fig. 3). For the I-containing electrode, no pronounced signals of Cl2 gas were detected at different charge potentials from 1.7 V to 1.95 V, and the electrode released a small amount of Cl2 from 1.95 V to 2.0 V (5.1 ppm at 2.0 V). In stark contrast, for the I-absent electrodes, gaseous Cl2 was produced after charging to approximately 1.85 V and continuously increased during charging to 2.0 V (49.6 ppm at 2.0 V state) (Fig. 1c). The role of I as a fixing agent for oxidized Cl0 is schematically illustrated in Supplementary Fig. 4.

Effect of the electrolyte concentration on the Cl-based redox reaction

After confirming the feasibility of applying I at the positive electrode of a Zn | |AC cell to fix the ClRR, we further optimized electrolytes to support the cycling stability of the ClRR. As mentioned above, one main goal is to suppress the competing OER to obtain a highly reversible ClRR21,22. To optimize the electrolyte, linear sweep voltammetry (LSV) was measured for electrolytes with different ZnCl2 concentrations from diluted 1 m to concentrated 30 m with the Ti-foil current collector as the positive electrode23; this was achieved by coupling with the Zn anode in a two-electrode cell configuration to identify the corresponding onset potentials of the OER and ClER (Supplementary Fig. 5a). With increasing electrolyte concentrations, the OER onset potential shifted to higher values, and conversely, the Cl− oxidation reaction shifted to lower working potentials (Supplementary Fig. 5b). The decrease in the Cl− oxidation potential can be explained by the Nernst equation associated with the increase in Cl− concentrations (Supplementary Note 2). In contrast to the 1 m ZnCl2 electrolyte, the AC electrode tested in 30 m ZnCl2 electrolyte produced a cathodic peak at 1.83 V in the CV profiles, further verifying the decrease in Cl− oxidation potential (Supplementary Fig. 5c). Therefore, the 30 m ZnCl2 electrolyte is verified to decrease the Cl− oxidation potential below the OER potential, resulting in a suppressed OER.

We further studied the influence of electrolyte concentrations on the electrochemical behaviours of the I-containing electrode. Specifically, the corresponding CV curves in a two-electrode cell configuration show that the ClRR behaviour became relevant when the electrolyte concentration increased from 1 m to 30 m, i.e., 1, 2, 5, 10, 20, and 30 m (Fig. 2a). No reduction peaks are observed when the concentration is less than 10 m, which could be attributed to the competing OER that occurs before Cl2 evolution. In contrast, reduction peaks appeared when the electrolyte concentrations increased to above 10 m, in which the ClRR dominated rather than the OER. This indicates that I as a coordinating agent can promote Cl2 fixation well only in high-concentration electrolytes. On the other hand, the b-value for different scanning rates from 0.1 to 1 mV s−1 approaches 0.5 of the reduction peak, corresponding to the nature of conversion reactions (detailed discussions on the b-value are provided in Supplementary Fig. 6).

Fig. 2: Physicochemical and electrochemical characterizations of ZnCl2-based aqueous electrolyte solutions at various concentrations.
figure 2

a CV curves of the I-containing electrode in ZnCl2 electrolytes with different concentrations at a scanning rate of 1 mV s–1. b Cycling stability of ClRR in 10, 20, 30 m ZnCl2 electrolytes at 0.5 A gI2−1, respectively, and the specific capacity was calculated based on the mass of I2 in ZnǁCl-I pouch cells at 25 °C. c Ex situ Raman profiles revealing the different hydrogen-bond networks of water in different electrolytes after 100 cycles corresponding to b. d Raman results of uncycled electrolytes with different concentrations, in which the deconvolution peaks of strong hydrogen bonds (H-bonds), weak H-bonds and non-H-bonds are centred at 3228 cm−1, 3432 cm−1, and 3594 cm−1, respectively, for the 1 m ZnCl2 electrolyte. e Rate performance of Zn | |Cl-I cells from 0.5 to 0.5 A gI2-1 and the corresponding GCD curves f. g Cycling stability of the Cl-redox reaction after 300 cycles at 0.5 A gI2-1.

Full size image

We further studied the cycling performance of the ClRR in different electrolytes (10, 20, 30 m ZnCl2) (Fig. 2b), in which the electrochemical stability was enhanced as the concentration increased. The low CE of the first cycle occurs due to the activation of I2 in the positive electrode (Supplementary Fig. 7). Throughout 100 cycles for the Cl-redox reaction, the electrode capacity in the 30 m electrolyte remains at 167 mAh gI2−1, showing better performance compared to the 20 m case of 127 mAh gI2−1 and the 10 m case of 114 mAh gI2−1 (Fig. 2b). To trace the origin of these performance differences, cycled electrolytes were ex situ characterized via Raman measurements. The results showed that I3− ions are present in lower concentration electrolytes (10 m, 20 m) but not in the 30 m case (Fig. 2c). This implies that the I species serving as Cl0 coordinating agents dissolve in the low-concentration electrolytes and consequently result in losses of capacity during battery cycling. The activity of water determines the solubility of halide species, in which polar halide species, such as I3− and I+ ions, easily dissolve into polar H2O molecular networks according to the like-dissolves-like rule. Via analysis of the ex situ Raman measurements, it was verified that the deconvolution proportion of strong hydrogen bonds centred at 3228 cm−1 decreases as the salt concentration increases; the tendency was increased for the nonhydrogen bonds (Fig. 2d and Supplementary Fig. 8)24,25, indicating that the water activity is suppressed with reduced hydrogen bonds from 1 m to 30 m electrolytes. Based on these analyses, 30 m electrolyte was selected as the suitable electrolyte for the ClRR.

To further explore the variations in different electrolyte systems, the solvation structures of the electrolytes, i.e., 1 m, 10 m, 20 m, and 30 m ZnCl2 electrolytes, were analysed by molecular dynamics (MD) simulation. Specifically, snapshots of simulation boxes of various concentrations as well as zoomed-in images are displayed for the corresponding major solvation structure of Zn2+ and their proportions in the system (Supplementary Fig. 9). In addition, the radial distribution functions (RDFs, g(r)) and coordination number (n(r)) of Zn-Cl and Zn-O were plotted to further determine the coordination structure of Zn2+ ions (Supplementary Fig. 10). As \({C}_{Z{{nCl}}_{2}}\)increases, the coordination number of H2O decreases from approximately 5.6 to 1.6, while that of Cl− increases from approximately 0.4 to 3.8. Therefore, Cl− competes with H2O to coordinate Zn2+ and gradually replaces the highly active coordinated H2O at high concentrations, which suppresses the reactivity of water and helps alleviate water-induced parasitic reactions. The ratio distributions of the corresponding solvation structures in various systems are calculated by setting the radius of the first solvation shell (i.e., the distance of the first valley in the RDF plot) as the cut-off distance.

At low concentrations, such as 1 m ZnCl2 electrolyte, Zn(H2O)62+ is the major solvation structure, which takes up 64.44% Zn2+ ions in the system, indicating the high reactivity of water towards Zn2+. As the concentration increased, the Cl− ions gradually dominated the first solvation shell, with Zn(H2O)3Cl3− and Zn(H2O)4Cl2 becoming dominant at 10 m and 20 m. However, when the concentration increased to 30 m, the proportion of Zn(H2O)3Cl3− and Zn(H2O)4Cl2 decreased, while Zn(H2O)2Cl42- grew to occupy 32.07% of the Zn2+ ions in the system (Supplementary Fig. 11). Notably, four-coordinated ZnCl42- occurs when the concentration increases to 20 m and becomes even more abundant at 30 m, which implies stronger binding between Zn2 and Cl−.

The binding energy of one water molecule with the rest of the structure was calculated to explore the evolution of dominant solvation structures (Supplementary Fig. 12 and Supplementary Note 3). Specifically, the binding strength of one water molecule of Zn(H2O)62+ in 1 m ZnCl2 of −34.77 kcal mol−1 is much higher than that for Zn(H2O)4Cl2 and Zn(H2O)2Cl42- in high concentration electrolytes (−3.09 kcal mol−1 and −2.77 kcal mol−1, respectively), indicating that water molecules remain much more stable in Zn(H2O)62+. The low water binding energy of Zn(H2O)4Cl2 and Zn(H2O)2Cl42- and the small difference of 0.32 kcal mol−1 between them imply that water molecules in these structures are easily pulled out and replaced by Cl− ions in highly concentrated electrolytes. On the other hand, the reactivity of water can also be suppressed by breaking the hydrogen bond (H-bond) network in the aqueous electrolyte as \({C}_{Z{{nCl}}_{2}}\) (the concentration) increases. The H-bonds in different simulation boxes are shown to explore the impact of \({C}_{Z{{nCl}}_{2}}\) on the H-bonds formed between water and water (Supplementary Fig. 13), from which we can intuitively observe the decline of the H-bonds number from low concentration to high concentration. Specifically, we calculated the average hydrogen bond number formed by one water molecule for each concentration (Supplementary Fig. 14). The average H-bond number of one water molecule is 1.21 in 1 m ZnCl2, while the H-bond number decreases to 0.53, 017, and 0.08 in the 10 m, 20 m, and 30 m systems, respectively. These results demonstrate that the H-bond network is broken as the concentration of ZnCl2 (\({C}_{Z{{nCl}}_{2}}\)) increases, which also contributes to the lower reactivity of water in the high \({C}_{Z{{nCl}}_{2}}\) case. This is consistent with the electrochemical stability window test in which the OER was suppressed to enable the stable realization of the Cl− ion redox reaction.

Regarding the electrochemical behaviour of the ClRR, the rate performance and corresponding GCD profiles were characterized at various operating specific currents in a ZnǁCl-I pouch cell at 25 °C (Fig. 2e, f). The ClRR enables a cell discharge capacity of 210 mAh gI2−1 with an average cell discharge voltage of 1.88 V at 0.5 A g−1 for the 20th cycle, correlating to a specific energy of 388.5 Wh k gI2−1. In addition, it could retain 75 mAh gI2−1 at a specific current of 5 A g−1, delivering a specific power of 1725 W k gI2−1. Long cycling stability was achieved with a discharge capacity of 140.8 mAh gI2−1 and 67.5% retention after 300 cycles at 0.5 A gI2−1 (Fig. 2g).

Physicochemical characterizations of the I-containing positive electrode before and after cycling

A set of spectroscopy experiments and calculations were performed to depict the interacting manners and structural evolution of I-containing positive electrodes during Cl oxidation/reduction. First, in situ Raman spectroscopy was performed to measure five charge–discharge states marked in the GCD curves (Fig. 3a and Supplementary Fig. 15). Two Raman peaks are always present at different charged/discharged states. Specifically, the first peak centred at 204 cm−1 is correlated to the presence of I+Cl- (Supplementary Fig. 2), while the peak at 291 cm−1 originates from the Cl− coordinating Zn2+ species adsorbed on the electrode as hydrated [ZnCl2+x(H2O)y] x- in the electrolyte. During the charging process, three Raman peaks appeared at 172 cm−1, 223 cm−1, and 345 cm−1, which disappeared during the reverse discharging process. The newly emerging rather than shifting Raman peaks indicate that new species are produced, which correlates to the presence of new structural configurations after Cl0 oxidation26.

Fig. 3: Spectroscopic and computational characterizations of ClRR.
figure 3

a In situ Raman characterizations of different charged and discharged states of the Cl-redox reaction. b Ex situ UV‒vis results and c ex situ XPS to detect the Cl-I electrode at different charged states. d Energy profiles of possible reaction pathways with different corresponding products. e Configurational structure and valence states (atomic Mullikan charge) of ICl3- as the final product.

Full size image

The ex situ ultraviolet/visible (UV/vis) spectroscopy of different charged states showed that the peak at 204 nm remained stationary as the peak intensity increased, and the peak at 326 nm originating from Cl-I bonding intensified; this might occur due to the oxidation from Cl− to Cl0 (Fig. 3b) since Cl0 was previously verified at a peak position of approximately 330 nm27,28. The new peaks at 472 nm, 573 nm and 669 nm might be attributed to the new configuration of the final products. Furthermore, ex situ X-ray photoelectron spectroscopy (XPS) was performed to analyse the valence states of Cl in different charged states. The binding energy at 199 eV emerged, which corresponds to Cl0 as the neutral valence state (Fig. 3c)20,29. This further verified the oxidation reaction from Cl− to Cl0 to complete the conversion reaction. Of note, even though it was posttreated and analysed in a vacuum atmosphere for XPS measurement, Cl0 was detected without any major evaporation, indicating that Cl was fixed by I. The XPS results of Cl at the initial and fully discharged states are shown in Supplementary Fig. 16. When checking the variations in the valence states of iodine as coordinating agents, the shape of the I 3d peaks remained almost unchanged with a reversible 0.2 eV shift in binding energy in charged and discharged states (Supplementary Fig. 17). The slight shift in the valence states of I can be attributed to electron redistribution due to the redox behaviour of Cl species during charging/discharging (Supplementary Note 4). Overall, these spectroscopic results verify the presence of oxidized Cl after charging.

Although the electrochemical and spectroscopic results showed that Cl0 was produced, the following uncertainties remain: how and in what configurations was Cl− fixed by coordinating with I. Therefore, we investigated the possible reaction pathways of Cl− oxidation, in which the potential energy and formula of the products were calculated along different possible oxidation reaction pathways. Starting from ICl and Cl− as reactants, two Cl- oxidation reaction compete as follows: the first produces molecular Cl2 through the blue pathway, and the second is fixed by I+ atoms through the pink pathway, as shown in Fig. 3d. Specifically, the pink reaction pathway flows as ICl first coordinates with Cl- to produce ICl2-, in which ICl2- has also been determined to be energetically favourable30,31. Then, ICl2- was further oxidized to ICl2, and it was finally stabilized by Cl- in a configuration as ICl3−. A total of 3.67 eV is needed to oxidize a single Cl− ion to Cl0 (blue curve), in which the thermodynamic energy barrier is 1.05 eV higher than that for oxidizing the ICl2− ion to ICl2 2.62 eV. The oxidized products, i.e., Cl0 and ICl2, are of great interest because after examining the possibility of these two products transforming into other forms through spontaneous reactions, it was determined that they are metastable products. Cl0 spontaneously produces molecular Cl2 by combining with another Cl0 atom, and the reaction energy needed to oxidize Cl− to Cl2 is 2.5 eV. On the other hand, ICl2 coordinates with Cl− to produce ICl3−, and the total energy needed to transform ICl into ICl3− is 0.86 eV. Therefore, ICl3− is more energetically stable than Cl2 molecules, explaining why no Cl2 gas is produced during charging (as shown by the pink pathway). In addition, we examined the possibilities of further oxidizing ICl3− by losing more electrons and forming other configurations by coordinating with Cl− ions. The thermodynamic energy barriers are 4.83 eV and 3.04 eV, respectively, largely exceeding 0.86 eV as stable final products. Specifically, the pink reaction pathway flows as ICl first coordinates with Cl- to produce ICl2-, and ICl2- has also been determined to be energetically favourable. Therefore, ICl3− is assigned as the final product, the cathode reaction formula is I+Cl− + 2Cl− ⇌ ICl3− + e−, and the full cell reaction formula is 2I− + 6Cl− + 4Zn2+ ⇌ Zn(ICl3)2 + 3Zn. The reaction equations of each transferred electron and the corresponding reaction products are elaborated in Supplementary Table 1. According to the reaction pathway, the valence states of each atom in the ICl3− configuration were assigned as I+Cl0Cl−Cl−, and the reaction equations of the full cell are presented in Supplementary Note 5. Of note, the Cl-I electrode reactions based on two redox centres are different from those based on a single redox halogen center15,19, e.g., I and Cl, in terms of the redox species, electron transfer number, final products, and electrochemical performance (Supplementary Table 2).

The valence states of chlorine atoms in the I+Cl0Cl−Cl− configuration might be unstable due to the large polarity in configurational electronegativity, and thus, the electron might redistribute of each atom in the ICl3− configuration. To confirm this hypothesis, density functional theory analyses were conducted, and the most stabilized configuration was arrayed as a tetrahedron. The valence states of chlorine are decimals of −0.64, −0.64, and −0.52, rather than integers of 0 and -1, while the valence state of iodine is +0.8 (Fig. 3e). Thus, the final stable valence states of individual I and Cl in the ICl3− configuration are I0.8Cl−0.64Cl−0.64Cl−0.52. In addition, regarding the spatial structure, the distance between the chlorine and iodine atoms is 2.56 Å for Cl−0.64•••I0.8 and 2.91 Å for Cl−0.52•••I0.8. Notably, the ICl3− configuration is the new halogen-based cathode material, in which Cl0 is stabilized by interhalogen bonding chemistry. The interacting force used to fix Cl0 is chemical bonding-based interhalogen bonding rather than physical adsorption, which may accommodate enhanced interacting forces.

Electrochemical investigations of interhalogen coordinating chemistry

After determining that ClRR can be fixed by applying I as a coordinating agent, it was determined that I can also be further exploited as a redox centre to contribute capacity when the working voltage range is widened to include the reaction potential range of I species. Generally, I can contribute to the one-electron-based capacity based on the I0/I− conversion reaction, delivering a capacity of approximately 200 mAh gI2−1 with an average voltage of approximately 1.2 V at 25 °C32,33. CV profiles were measured on I-containing and I-absent electrodes, with working voltages ranging from 0.6 to 2.0 V (Fig. 4a). There were three different pairs of noticeable redox peaks for the I-containing electrode, while only one weak cathodic peak appeared at 1.83 V for the I-absent electrode. Specifically, these three cathodic peaks from higher to lower potentials at 1.84 V, 1.59 V and 1.13 V can be assigned to different reduction reactions due to Cl0/Cl−, I+/I0, and I0/I−, respectively (Fig. 4a). The CV profiles suggest that I accommodated two functionalities as the coordinating agent to fix the oxidized Cl0 and the redox centre to deliver energy.

Fig. 4: Electrochemical investigations on the multielectrode reaction of the Cl-I positive electrode.
figure 4

a CV profiles at 0.1 mV s-1 to study the two functionalities of I in I-containing and I-absent electrodes. b GCD profiles of the Cl-I electrode operated within different potential ranges at 0.5 A gI2-1 in a ZnǁCl-I pouch cell at 25 °C. c Identification of the various stages in the electrochemical reaction via cell discharge voltage analysis, in which the energy density for the 1-electron I0/I- reaction was 233.8 Wh kgI2−1. d Schematics of the interhalogen coordinating chemistry to illustrate the electron flow during battery charging and electron redistributing, in which the valence states of I and Cl redox centres are marked.

Full size image

Subsequently, the electrochemical performance of the I-Cl electrode was investigated based on the GCD profiles of the three-electron reaction exhibiting three different discharging plateaus (Fig. 4b). When regulating the working potential ranges to include more different redox reactions, the GCD curves of narrow potential ranges almost overlapped with the curves of wider ones, indicating that each plateau was independently reversible. Therefore, we elaborated three reactions separately. First, for the one-electron transfer process as the I0/I− redox electrode working within 0.6–1.45 V, the GCD profile matches well with the generally applied iodine electrode32,33, delivering a capacity of 205 mAh g−1 at 0.5 A gI2−1 and 25 °C, and the corresponding energy was 233.8 Wh kg I2−1. Second, regarding the two-electron transfer process as I+/I0/I− redox reactions within the full working range of 0.6–1.75 V, the above-obtained I0 can be further oxidized to I+Cl− with Cl− ions as the stabilizer. The total capacity doubled, and a different discharging voltage plateau appeared at approximately 1.59 V, in which the corresponding specific energy was 528.5 Wh kgI2−1 for the I+/I0/I−-based redox electrode. Therefore, it achieved a 226% enhancement in specific energy, i.e., 528.5 Wh kg I2−1/233.8 Wh kgI2−1, compared to that of the I0/I− redox electrode. Third, when the working range was widened to 0.6–2 V, the I+/I0/I− and ClRR can all be activated, delivering a tripled capacity compared to that of the generally applied I0/I− one-electron electrode reaction. Of note, it was found that the capacities delivered by these three plateaus were almost identical, i.e., approximately 200 mAh gI2−1, correlating to the identical amounts of electrons transferred in each reaction at 0.5 A gI2−1 and 25 °C. Thus, the stoichiometric ratio is 1 for one I atom to fix one Cl atom and proceed with the one-electron redox reaction. This is consistent with the above calculational results, i.e., one electron is lost from the Cl species to stoichiometrically couple with one I atom. In addition, the cycling performance within different charge ranges was stable throughout varying working potential ranges (Supplementary Fig. 18).

Overall, the I-containing positive electrode shows that three-electron reactions can deliver a capacity of 612.5 mAh gI2−1, exhibiting a 289% capacity improvement over that of the one-electron-based I0/I− redox reaction. The improved capacity and voltage jointly contribute to boosting the specific energy so that the corresponding specific energy of the Cl-I-based cathode can reach 905 Wh kg I2−1, which is 387% higher than that of the I0/I− redox electrode generally applied in the standard ZnǁI2 battery (Fig. 4c). The Zn metal cell with the I-containing positive electrode capable of a three-electron transfer reaction demonstrates improved battery performance compared to that of similar aqueous Zn-based electrochemical energy storage systems with MnO2, MXene, and V2O5 cathodes (see Supplementary Fig. 19 and Table 3).

Consequently, schematics of the charging process of the Cl-I electrode are illustrated in Fig. 4d, demonstrating the electron flow and the corresponding products throughout the charge transfer process. The interhalogen coordinating chemistry between heterogeneous halogen species determines the whole process, which can be embodied by the products as ICl and ICl3−.

Electrochemical energy storage performance of the Zn | |Cl-I battery system

Based on the previous characterizations, we assembled and tested several Zn | |Cl-I lab-scale cells capable of exploiting the I+/I0/I− and ClRR mechanisms. The CV measurements show the evolution of different redox peaks with increasing scanning rates from 0.1 to 1 mV s−1 (Fig. 5a). The kinetics of each halogen-based reaction were analysed. The relationships between the current densities of the three cathodic peaks (I) and scanning rates (ν) can be elaborated as I = aνb (Supplementary Fig. 20), where a is a coefficient, a b value of 0.5 indicates semi-infinite diffusion behaviour, and a b value of 1 implies capacitive behaviour34. Here, all b-values approaching 0.5 could be assigned to halogen-based phase conversion reactions. The evolutions of three reduction peaks were carefully checked at different scanning rates, and the current density of peak R1 corresponding to ClRR increased at a slower rate compared to those of the R2 (I+/I0) and R3 (I0/I−) peaks (Fig. 5a). Therefore, the ratio of current intensities is 2 of IR1/IR2 at a low scanning rate (0.1 mV s−1) and became 0.85 at 1 mV s−1, which indicated the sluggish kinetic response of the R1(Cl-/Cl0) reaction, as shown in close observations at 0.1 mV s−1 and 1 mV s−1 (Supplementary Fig. 21). In addition, the charge transfer resistances at different charged states were investigated via electrochemical impedance spectroscopy (EIS) measurements. The EIS analysis highlights the increase in the ionic transfer resistance during charging (Supplementary Fig. 22 and Supplementary Table 4), which might be assigned to the accumulation of chloride ions to retard the multiple steps of ion/electron transfer during charging of the Cl-I electrode.

Fig. 5: Battery performance of various Zn | |Cl-I cells.
figure 5

a CV profiles and b rate performance of the ZnǁCl-I pouch cell in 30 m ZnCl2 at 0.5 A gI2−1 and 25 °C. c Corresponding GCD curves at different currents based on the mass of iodine. d Comparison of the average voltage, capacity and cathode materials in different full battery systems, including Li+, Na+, K+, and Zn2+ ion batteries, in which the corresponding performance data are exhibited in Supplementary Table 5 for reference. The specific capacities and specific currents were based on the active materials, and it is 0.5 A gI2−1 based on the mass of iodine. e Storage performance evaluated by resting for different hours at a 100% state of charge at 0.5 A gI2−1. f The cycling stability and CE of the ZnǁCl-I full cell at 2 A gI2−1. g Digital image and h cycling performance of the ZnǁCl-I pouch cell at 2.5 mA cm−2 and 25 °C.

Full size image

The Zn | |Cl-I cell delivers a specific capacity of 612 mAh gI2−1 at 0.5 A gI2−1 and a delivered capacity of 455 mAh gI2−1 at 8 A gI2−1 and 25 °C (Fig. 5b). The corresponding GCD curves possessed three pairs of charging–discharging plateaus as exhibited in Fig. 5c. The average cell discharge voltage and specific capacity at 0.5 A gI2−1 are well positioned compared to that of similar aqueous electrochemical energy storage systems reported in the literature (see Fig. 5d and Supplementary Table 5). Notably, the specific capacities of the Zn | |Cl-I cells are calculated based on the iodine mass, while the corresponding capacity based on the total mass of iodine and AC is 226 mAh g(I2+AC)−1 at 0.5 A gI2−1, and the corresponding capacities at other specific currents are reported in Supplementary Table 6. In addition, the specific energy is 50.7 Wh kg I2−1 based on the total mass of the positive (i.e., the mass of iodine and AC without considering the Ti current collector) and negative (i.e., Zn metal) electrodes. Moreover, Zn | |Cl-I cells with higher specific energy could be obtained by optimizing the iodine loading in the positive electrode (see Supplementary Note 6).

Self-discharge is an important parameter to embody the stability of the reaction products. The ZnǁCl-I cell exhibited 92.9% capacity retention after resting for 24 h and 71.5% retention after 96 h (Fig. 5e), and the corresponding GCD curves after resting are exhibited in Supplementary Fig. 23. This confirms that the stable fixation of Cl0 by I species and the slight capacity losses can be assigned to the dissolution of the charged electrode materials into the electrolyte. In addition, a discharge capacity retention of 95.7% was calculated after 2000 cycles at 2 A gI2−1 at 25 °C (Fig. 5f). The Zn anode cycled in 30 m ZnCl2 electrolyte showed excellent cycling stability for 200 h in symmetric cells at a current density of 1 mA cm−2 and an areal capacity of 1 mAh cm−2 (Supplementary Fig. 24). To further demonstrate the potential of the ZnǁCl-I battery towards practical applications, pouch cells were assembled with a cathode area of 80 cm2 (10 cm × 8 cm) to demonstrate the scalability of the Cl-I electrode, as shown in Fig. 5g (Supplementary Note 7). The pouch cell delivers an initial capacity of 124.7 mAh (1.56 mAh cm−2) at 200 mA (2.5 mA cm−2) and a 73.8% capacity retention, i.e., 92.1 mAh, after 300 cycles (Fig. 5h).

In summary, by applying interhalogen coordinating chemistry, we demonstrated the possibility of exploiting iodine as a coordinating agent to fix oxidized Cl0. A 30 m ZnCl2 electrolyte was selected to suppress the OER and simultaneously inhibit I-dissolving issues, guaranteeing good cycling stability. In situ and ex situ spectroscopic and calculational results jointly verified the interhalogen coordinating chemistry between I and Cl. The final product was determined to exhibit a new configuration as ICl3− through interhalogen bonding.

After fixing ClRR, the redox nature of I was exploited as well, obtaining two redox centres as Cl and I species with three-electron transfer. As a result, the Cl-I positive electrode, tested in combination with a Zn metal negative electrode, enables a specific capacity of 612 mAh gI2−1 at 0.5 A gI2−1, a specific energy of 905 Wh kgI2−1, and a 95.7% capacity retention after 2000 cycles at 2 A gI2−1 due to the stable interhalogen fixation.

Methods

Preparation of the I-containing positive electrode

First, activated carbon (AC, large surface area of ∼1000 m2/g, purchased from Sigma‒Aldrich with the product number as 902470) was coated on titanium mesh (Ti, 50 μm thickness and 100 mesh cm−2, purchased from Kaian metal wire mesh Co., Ltd.) as the hosting materials for the halogens I and Cl, in which the Ti mesh was selected as the current collector because of its corrosion-resistant property against chloride ions. AC was mixed with acetylene black and poly(vinylidene fluoride) (both purchased from Aladdin) at a weight ratio of 8:1:1 and then coated onto Ti mesh with a loading mass of AC of approximately 4 mg cm−2. After drying in a vacuum oven at 80 °C, the AC host electrode was obtained. I2 was synthesized by electrodeposition in a three-electrode glass cell at 25 °C with an AC electrode as the working electrode, Ag/AgCl electrode as the reference electrode, and Zn foil as the counter electrode, which were immersed in a 1 m ZnI2 flooded electrolyte. A constant current density protocol was applied, and the anodic current density was 1 mA cm−2 and deposited for 0.5 h at 25 °C. Thus, the predeposited capacity of I2 is 0.5 mAh cm−2, which can be calculated as 2.36 mg cm−2 based on the theoretical capacity of I2 of 212 mAh g−1. After washing with DI water and drying in a vacuum oven at 25 °C, the I-containing electrode was obtained.

Electrochemical characterizations

A 30 m ZnCl2 electrolyte was prepared through dissolving 30 mol ZnCl2 salt in 1 kg deionized water at 25 °C. The 30 m ZnCl2 configuration (100 for μL for 1 cm2 size cell) was applied for all cell configurations to investigate interhalogen redox chemistry and performance, while other concentrations, e.g., 1, 2, 5, 10, 20 m ZnCl2, were applied for comparative study as described in the main text. The ZnǁCl-I pouch cells were configured by applying an I-containing electrode sized 1 cm2 (1 cm × 1 cm) as the cathode, where Zn foil (1 cm × 1 cm, 99.99% purity, purchased from Aladdin) with a thickness of 50 μm was applied as the anode. Whatman GF/F glass microfiber discs saturated with specific electrolytes are utilized as separators between the cathode and anode. Large-sized (10 cm × 8 cm) single-side coated I-containing positive electrodes and Zn negative electrodes were utilized to assemble pouch cells in an air environment. An ethylene-ethyl acrylate sheet was applied as the encapsulation layer for the pouch cells. Five cells were tested for a single electrochemical experiment for the CV, rate and cycling performance. The blank AC electrode was applied for comparison with the AC electrode containing deposited I2, while the specific gravimetric capacities of I-containing and I-absent electrodes were calculated based on the I2 mass to directly study the capacity contributed by I2. The electrochemical tests of the batteries are carried out by a CHI 760E electrochemical working station and Land 2001A battery testing system, in which cyclic voltammetry (CV) and galvanostatic charge/discharge are conducted. Linear sweep voltammetry (LSV) was performed with Ti foil as the working electrode and Zn as the anode, and the anodic scan was initiated at the voltage of the open circuit potential at a scanning rate of 0.5 mV s−1. Electrochemical impedance spectroscopy to obtain the Nyquist plots was carried out with an AC perturbation signal of ±5 mV and a frequency range from 0.1 Hz to 100 kHz to measure the selected quasistationary potentials during the anodic CV scan. All these electrochemical tests were carried out at an environmental temperature of 25 °C, and no climatic/environmental chamber was used. Since the mass of Cl− ions involved in the electrode reactions is always changing, it cannot represent the real-time capacity to count the varying mass of Cl− anions; thus, the specific capacity and specific energy of the cells are calculated based on the mass of iodine.

Material characterizations

The Raman spectra were collected for different electrolytes using a PerkinElmer Raman 400 F Spectrometer equipped with a 532 nm NIR laser, and the measurement was conducted by focusing the laser light onto the electrolyte samples in a quartz tube. The ex situ Raman analysis of electrode evolution was carried out in an electrolyte flooded cell in two-electrode Zn | |Cl-I configurations sealed by quartz glass, and the data were collected at different states of charge and discharge at 25 °C. X-ray photoelectron spectroscopy (XPS) measurements were conducted by a PHI Model 5802, and UV‒vis spectra were collected by a Shimadzu UV 3600 UV/visible/IR spectrophotometer. The electrode samples at different states were washed with DI water and then maintained under vacuum overnight to conduct XPS and UV‒vis measurements. The gas concentration detection of gaseous Cl2 was measured by a Kallu Electronics detector based on a home-designed cell, as presented in Supplementary Fig. 3. The detailed measuring protocols are described as follows: first, before charging the cell, N2 was fully injected to guarantee that the cell head space was filled with N2. Second, during charging, the valves connected to the two tubes were closed, and the cell was charged at 0.5 A g−1 to specific voltages and maintained at the voltage point for 90 s to sufficiently obtain the oxidized products. Finally, the valve of Tube 2 was opened to extract the gas in the cell head space for Cl2 detection, and the valve of Tube 1 remained closed. Three steps are used to detect the Cl2 concentration at one specific voltage. All voltage points (correlated to the voltages in Fig. 1c in the main text) were performed accordingly for gaseous Cl2 detection.

Computational method

All MD simulations were implemented in the open-source code Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) [1-SI] with the nonbonded interatomic potentials described by the Lenard-Jones (LJ) force field. The TIP4PEW model was chosen to simulate water molecules in the system. The parameters for Zn ions and Cl ions were obtained from Merz’s works of fitting LJ parameters for aqueous +2 metal cations[2-SI] and monovalent ions[3-SI], respectively. Systems of boxes with \({C}_{Z{{nCl}}_{2}}\) of 1 m, 10 m, 20 m, and 30 m were constructed to model the aqueous electrolyte with different concentrations. The compositions of these systems are shown in Supplementary Table 7. First, these systems were heated with NVT ensembles for 5 ns with different temperatures (from 300 K to 800 K) varying with the concentrations and then relaxed with an NPT ensemble for 45 ns to ensure equilibrium. Finally, an NVE run for 2 ns was performed to collect and analyse the data. The H-bond number was calculated using the Hydrogen Bond Anaysis code[4-SI] in the MD Analysis package[5-SI] with the criteria that the distance of donor O and accepter O was <3.5 Å, and the angle of (donor)O-H…O(acceptor) was >150°. Quantum chemistry calculations were carried out through the Gaussian 09 software package[6-SI]. The B3LYP/SDD basis set was used for Zn, and the B3LYP/6-311 + G(d) basis set was used for H, O, and Cl[7,8-SI]. The visualization of the simulation process was realized by VMD software[9-SI].

The electronic structure calculations, including the geometries, energies, and frequencies of all the stationary points (the reactants, transition states (TSs), products) were performed by the GAUSSIAN 09 program[10-SI]. Density functional theory calculations (DFT) were carried out by using the B3LYP (Becke’s three-parameter nonlocal-exchange functional[11-SI] with the gradient correction of Lee, Yang, and Parr[12-SI]) method together with the aug-cc-pVTZ basis set for the Cl atom and the effective core potential (ECP) for the I element. The references listed in the Computational Method are numbered from 1 to 12 in Supplementary Reference List at the end of Supplementary Information.