Introduction

The ever-growing demands for energy storage drive the development of low-cost and high energy density devices1,2,3. The SOCl2-based primary batteries hold ultra-high energy density of up to 710 Wh kg−1 based on the redox of 2SOCl2 + 4e- → S+ SO2 + 4Cl-, φθ/V = 3.6 V, long shelf storage life, and feasible application under extreme conditions. Thereby, the successful conversion of rechargeable Na-Cl2 systems from SOCl2-based primary batteries demonstrates greater promise for cutting-edge battery based energy storage systems possessing higher energy density, longer shelf life, and superior environmental adaptability compared with current Li-ion/Li-S/Li-air batteries4,5,6. In a typical rechargeable Na-Cl2 battery, the SOCl2-based electrolyte is firstly decomposed and combined with Na-ions from the Na-anode to form NaCl particles deposited on the surface of the porous cathode. Over cycling, the deposited NaCl particles are reversibly reacted (2NaCl ↔ 2Na+ + Cl2 + 2e-) to complete the charge storage7. Nevertheless, the reaction reversibility between NaCl and Cl2 is far from satisfactory, leading to very short cycle life of Na-Cl2 batteries and significantly blocking their practical use.

The kinetics of charge-transfer and mass-transport across multiphase interfaces, involving gas (Cl2), liquid (Na+ and Cl- in electrolytes) and solid (NaCl and porous carbon host) in the cathode, are the key factors to improve the electrochemical performance8,9. Typically, the microporous carbon nanospheres offer large surface area and pore volume for multi-phase conversion to accommodate Cl2 and NaCl4. Considerable efforts have been devoted to designing cathode materials, such as CO2-activated graphite and NH2-functionalized MOFs, for accelerating the electrochemical conversion process10,11. Recently, a bicontinuous structured carbon cubosome with heteroatomic doping allows efficient ion and electron transport and promotes Cl2 adsorption and conversion, delivering a maximum rate ability and cycle life of 16 A g−1 and 250 cycles, respectively12. Although these strategies have improved the reaction kinetics of the Li/Na-Cl2 battery by providing physical or chemical confinement on Cl2 species, the cycling stability has been a big problem, with the life time mostly less than 500 cycles10,11,12,13,14. Obviously, the deeper understanding of the kinetics of the three-phase reversible reaction of Cl2/NaCl is quite limited so far.

In this work, we demonstrate that the substantially increased voltage polarization is critically important in the battery failure along its charge/discharge cycling. The passivation behavior caused by the deposition of the insulated NaCl on the porous cathode material plays a crucial role in the three-phase chlorine (NaCl/Na+, Cl-/Cl2) conversion kinetics. More interestingly, a self-depassivating electrode with I- as a mediator is reported, in which the existence of I- facilitates the NaCl deposition with low-crystallinity and small-particle size due to the strong coordination between I- and Na+, affording much lower polarization for chlorine conversion. Consequently, the Na-Cl2 battery with this self-depassivated electrode exhibits an increased discharge and decreased charge voltage platforms of 3.14 to 3.43 V and 4.10 to 3.89 V, respectively, corresponding to a substantially improved energy efficiency of 76.6 to 88.2%. Such a cell can provide a record-long cycling performance over 2000 cycles with minor increase of polarization, which is four times longer even compared to the longest cycle life reported previously. Moreover, this Na-Cl2 battery also exhibits excellent low-temperature performance with over 800 stable cycles at −20 oC, demonstrating its great promise for practical applications.

Results and discussion

Analysis of the passivation failure mechanisms

When looking back at the primary SOCl2 batteries, their failure mechanism is attributed to the overloaded voltage polarization, in which the excessive insoluble discharge products (metal chloride) are deposited at the cathode-electrolyte interface, passivating the electrodes15,16. Therefore, a deeper view of the fast fading mechanism of rechargeable Na-Cl2 batteries considering the solid NaCl and exploring the corresponding strategy to reduce the polarization upon repeated cycling is a prerequisite for achieving ultra-long cycle Na-Cl2 batteries.

As illustrated in Fig. 1a, the Na-Cl2 battery was assembled with metal Na as the anode and reduced graphene oxide (rGO) as cathode support (Supplementary Fig. 1, Supplementary Fig. 2, and Supplementary Text 1). Normally, the Na-Cl2 battery works from the first irreversible discharge process of SOCl2 (eq1, 2) to the subsequent reversible charging/discharging process of NaCl and Cl2 (eq3, Supplementary Text 2 and Supplementary Fig. 3). The produced NaCl as the active chlorine species launches the reversible NaCl/Cl2 reaction, and the AlCl3 reacts with NaCl that generated upon the first discharging process of SOCl2 in anode or cathode to totally convert into Na[AlCl4] as an ionic conductor during reversible cycling4.

Fig. 1: Failure mechanisms of Na-Cl2 battery.
figure 1

a Schematic of the rechargeable Na-Cl2 battery, composed of a Na metal anode, a reduced graphene oxide (rGO) cathode, and electrolyte. b Time-voltage curve and c charge-discharge curves at different cycle number of assembled Na-Cl2 batteries. d Ex-situ SEM image of rGO cathodes after discharged. e The Na-Cl2 battery failure process upon cycling with the original electrolyte (blue) and cycling performance after reassembling (red) with the refreshed SOCl2 electrolyte, separator, Na metal anode, etc., except for the cycled cathode. f The discharge-charge curves of Na-Cl2 battery at 5, 100, and 220th cycles (blue) and the discharge-charge curves of the Na-Cl2 battery at 5, 25, and 55th cycles after reassembling (red).

Full size image

First irreversible discharging process:

$${4{{\rm{Na}}}}^{+}+{4{{\rm{AlCl}}}}_{3}+{2{{\rm{SOCl}}}}_{2}+{4{{\rm{e}}}}^{-}\to 4{{\rm{Na}}}[{{{\rm{AlCl}}}}_{4}]+{{\rm{S}}}+{{{\rm{SO}}}}_{2}$$
(1)
$${4{{\rm{Na}}}}^{+}+{2{{\rm{SOCl}}}}_{2}+{4{{\rm{e}}}}^{-}\to 4{{\rm{NaCl}}}+{{\rm{S}}}+{{{\rm{SO}}}}_{2}$$
(2)

Reversible charging/discharging process:

$$2{{\rm{NaCl}}}\leftrightarrow {2{{\rm{Na}}}}^{+}+{{{\rm{Cl}}}}_{2}+{2{{\rm{e}}}}^{-}$$
(3)

The battery operated on two redox reactions of Na/Na+ at the anode and Cl-/Cl2 at the cathode in the reversible process. Three phases are involved in the electrochemical reduction of NaCl/Cl2 conversion in the cathode, including solid NaCl and rGO support, liquid Na+, Cl- in the electrolyte, and the gaseous Cl2. NaCl is formed during discharging (2Na+ + Cl2 + 2e- ↔ 2NaCl)4. We found that the electrode polarization tends to become more severe with the increase of the cycle number but the specific capacity is relatively stable upon battery operation (Fig. 1b, c). When the charge voltage exceeded the protection range of test (5 V), the battery operation suffers from a capacity dive. Then, we reinjected the fresh electrolyte but the battery was not reactivated, which means the electrolyte depletion was not the main cause for the polarization increase. Ex-situ SEM of the 2nd discharged battery and the failed battery revealed that the cathode was gradually covered by thick layer of NaCl as the cycle progressed (Fig. 1d and Supplementary Fig. 4a). Given the insulating property of NaCl, it is reasonable to conclude that gradual deposition of NaCl passivated the electrode surface. To verify it, we reassembled this battery with the refreshed electrolyte, separator, Na metal anode, etc., except for the passivated cathode. As a result, this battery still has a fast capacity decay (Fig. 1e, f). By showing the charging and discharging curves of failure and after reassembled, it is clear to see that the polarization of the charge process was still very severe. This confirms the cathode completely loses the possibility of resurrection with the main cause being NaCl passivation, not the active chlorine loss. The main failure mechanism includes the following: (1) NaCl easily accumulates and forms large particles arising from the mismatch between nucleation and growth kinetics. The large NaCl particles with poor conductivity are difficult to convert to Cl2 in the subsequent charging process, leading to an increased charging polarization17,18. (2) The dense micron-level NaCl particles into the surface holes of the electrode tend to block the diffusion channels of Na+ and Cl2, hindering the nucleation of NaCl in subsequent cycles and further increase the polarization. Moreover, as the cycle repeats, this situation will become more severe. (3) The large NaCl particles also make it difficult to fully utilize the active component, thus accelerating the irreversible consumption of SOCl2, resulting in the capacity rollover fade of Na-Cl2 batteries (Supplementary Fig. 4b, c)19. Therefore, we propose that the mode of NaCl deposition plays a crucial role in dictating the cycle life of Na-Cl2 batteries, where the deposited large-particle NaCl film severely passivates the porous cathode during cycling, causing capacity diving.

The self-depassivation process

Fundamentally solving the deposition issue of NaCl is the critical step to overcome the above issues in long-cycle stable Na-Cl2 batteries. We introduced iodine into the electrolyte as an additive aimed to modulate the NaCl deposition mode and achieve NaCl self-depassivation in the cathode (Fig. 2a). First, when I2 was added upon the first discharging process, the I2 could enhance the reducing activity of SOCl2 by forming SOCl2·I2 complex for the first discharge reaction (Eqs. 4, 5). At the ending of the first discharging, the I- altered the deposition of NaCl, mainly through a co-deposition of NaCl and NaI (Eqs. 6, 7), facilitating the uniform formation of higher-reactivity NaCl with low-crystallinity and fine particles than that of the normally formed NaCl in Na-Cl2 batteries (Fig. 2b and c). Second, in the reversbile charging/discharging process with NaCl/Cl2 as the redox species, the deposited NaI would dissvole into the electrolyte to produce I- and discharged into NaI as the co-deposition with NaCl again. This dissolution-deposition process of I- is reversible. As a result, by tuning the NaCl deposition and oxidation activity using iodine, the cathode passivation by NaCl is significantly suppressed, which ensures efficient charge-mass transfer and prevents voltage polarization during charging (Fig. 2d, e).

Fig. 2: Schematic of self-depassivation process by iodine species.
figure 2

a Mechanistic diagram of I2 in Na-Cl2 battery. Comparison of the main reversible discharge process of Na-Cl2 battery with and without I- mediator. b In the original electrolyte system, a thick and uneven film of NaCl was induced by Cl2 reduction during discharge. The rGO cathode in this condition is called the original cathode, abbreviated OC. c The high activation voltage and large charge voltage polarization of OC Na-Cl2 battery after 100 cycles. d When I- is introduced in the electrolyte as a mediator, it could be strongly coordinated with Na+, facilitating its incorporation into the NaCl lattice during discharge. This results in a disarrangement of the NaCl and forms sparse, low-crystallinity small NaCl particles, substantially improving the reactivity of NaCl and enabling a self-passivated cathode. Simultaneously, the generated NaI on the electrode is dissolved in the electrolyte, producing I- to participate in the next reversible cycle with Na+ and Cl2. The rGO cathode in this condition is called the self-depassivated cathode, abbreviated SDC. e The low activation voltage and steady charge curves of SDC Na-Cl2 battery after 100 cycles.

Full size image

First irreversible discharging process with I2:

$${4{{\rm{Na}}}}^{+}+{4{{\rm{AlCl}}}}_{3}+{2{{\rm{SOCl}}}}_{2}+{4{{\rm{e}}}}^{-}\to {4{{\rm{NaAlCl}}}}_{4}+{{\rm{S}}}+{{{\rm{SO}}}}_{2}$$
(4)
$${8{{\rm{Na}}}}^{+}+{2{{\rm{SOCl}}}}_{2}\cdot{{\rm{\cdot }}}{{{\rm{I}}}}_{2}+{8{{\rm{e}}}}^{-}\to 4{{\rm{NaCl}}}+4{{\rm{NaI}}}+{{\rm{S}}}+{{{\rm{SO}}}}_{2}$$
(5)

Reversible charging/discharging process with I2:

$$2{{\rm{NaCl}}}\leftrightarrow {2{{\rm{Na}}}}^{+}+{{{\rm{Cl}}}}_{2}+{2{{\rm{e}}}}^{-}({{\rm{electrochemical\; process}}})$$
(6)
$${{\rm{NaI}}}\leftrightarrow {{\rm{Na\cdot \cdot \cdot I}}}({{\rm{self}}}-{{\rm{dissolving}}})$$
(7)

Effect of iodine on the first discharge

To fully clarify the discharge mechanism, we first need to identify the existential state of the iodine additive in the electrolyte. Raman spectra of original electrolytes and electrolytes with iodine additive were examined to gain more insights into the coordination environment in the electrolytes. As shown in Fig. 3a, different characteristic peaks of various stretching vibrations were observed in the Raman shift range from 100 to 1200 cm−1. The vibration peaks at 194/487, and 285/344 cm-1 were detected for the pure SOCl2, corresponding respectively to the SCl2 and SOCl16,20. When AlCl3 was added, new peaks appeared around 216 and 526 cm−1, corresponding to Al-Cl16,21. In addition, the addition of NaFSI as well as NaTFSI did not lead to a obvious change in the displayed rang due to the trace amount. After dissolving the iodine into the mixed solvent, the peaks of S-Cl (194/487 cm−1) in SOCl2 gradually weakened and the I-I peak (210 cm−1) gradually appeared with the increase of I2 content (Supplementary Fig. 5), which confirms that I2 cooperated with SOCl2 to form a SOCl2·I2 complex and remained stable (Supplementary Fig. 6 and Supplementary Text 3)13.

Fig. 3: Spectroscopic, computational and electrochemical investigations of electrolyte with iodine addition.
figure 3

a Raman spectra of different SOCl2 electrolytes, including SOCl2, SOCl2-AlCl3, SOCl2-AlCl3-NaFSI-NaTFSI (abbreviated as SANN), SANN with different amount of I2 (0.02, 0.2, 1.0 wt%, abbreviated as SANN + 0.02/0.2/1.0 wt% I2). b Formation energy of the structures in different times. c Comparison of the calculated lowest occupied molecular orbital (LUMO) energy levels and highest occupied molecular orbital (HOMO) energy levels of SOCl2·I2 complex at different time. d Gibbs free energy for the possible decomposition processes of SOCl2 and SOCl2·I2 complex components. e CV curves of the SANN, and SANN + 0.2 wt% I2 electrolytes in three-electrode systems with a scan rate of 2 mV s−1. f The first-cycle discharge curves of SDC battery and OC battery.

Full size image

Molecular dynamics simulations were carried to verified the formation of the complex intermediate. The starting (t = 0 ps), intermediate (t = 0.4, 0.6, and 0.8 ps), and final (t = 1 ps) configurations were shown in Supplementary Fig. 7a. The I2 to SOCl2 distances decreased from 5.56 Å (0 ps) to 2.70 Å (0.8 ps), which is closed to the I-I bonding length (2.86 Å), strengthening the nonbonded electrostatic interaction. Thus, this further validates that I2 interacts with SOCl2 to form a SOCl2·I2 complex13. Further calculations of binding energies and LUMO and HOMO energy levels were carried out based on the above structures (Fig. 3b, c). The binding energies of the various structures decrease compared to its pristine state, indicating a tendency for the SOCl2·I2 structure to stabilize in close to I2. The LUMO and HOMO energy level of complexations between SOCl2 and I2 at different stages were calculated based on density functional theory (DFT). The LUMO diagram shows orbital lobes that give a qualitative representation of the reactive sites22,23. At first, the LUMO of the complex is located on the I2, and as time evolved, the orbital lobes on SOCl2 gradually become visible and interact significantly with I2, which shows an increase in the reductive activity on SOCl2 under the influence of I2. Moreover, we have derived the formation energies as well as the LUMO and HOMO energy levels of the different SOCl2·I2 complex structures (including SOClI, SOCl2I, and SOI2 as neutral) by DFT calculation. The band gap of SOCl2·I2 complex gradually narrows with increasing iodine content, representing its higher conductivity, and the gradually decreased LUMO energy level represents its superior reduction properties (Supplementary Fig. 7b and c)24. The Gibbs free energy of the reaction involving SOCl2 and the SOCl2·I2 complex was calculated. The distributed Gibbs free energies of the SOCl2·I2 complexes were found to be consistently lower than that of SOCl2 (Fig. 3d and Supplementary Table 1), indicating that the introduction of I2 effectively reduces the reaction barrier of SOCl2 and enhances its reduction potential25.

The cyclic voltammetry (CV) curves of the SANN, and SANN + 0.2 wt% I2 electrolytes were measured using a three-electrode cell with a glassy-carbon electrode as the working electrode (Supplementary Fig. 8), Na metal as the counter electrode, and carbon rod as the reference electrode in the glove box. As shown in Fig. 3e, the SOCl2 reduction potential in SANN + 0.2 wt% I2 electrolyte (-0.63 V vs. Ref. C) was higher than that in the SANN electrolyte (-0.73 V vs. Cref). Notably, the onset oxidation potential of SOCl2 in SANN + 0.2 wt% I2 electrolyte occurred at 0.24 V, which was earlier than SANN electrolyte (0.64 V). The results validated that the subtle interactions between I2 and SOCl2 decreases the energy barrier of SOCl2 reaction, which is consistent with the phenomenon of two-electrode CV (Supplementary Fig. 9)26. Figure 3f and Supplementary Fig. 10 presented the first discharge curves of SDC battery and OC battery. Two discrete plateaus present at 3.27 and 2.92 V in OC battery were associated with Eq. 1 and Eq. 2. In the SDC battery, both discharge platforms (3.40 and 3.06 V) were higher than the OC battery, which was attributed to the fact that the cooperation of I2 with SOCl2 greatly enhanced the reduction activity of SOCl2. These findings are consistent with the results obtained from three-electrode CV and computational calculations.

The effect of iodine on the crystallization of NaCl

Furthermore, the I- formed by first-cycle discharges affects the solvation structure and interfacial composition in the electrolyte, which will influence the interfacial transfer process of Na+, mainly involving the desolvation behavior of Na+ and diffusion across the interface27. The molecular dynamic (MD) calculations were conducted to explore the Na+ coordination based on the electrolyte environment after the first discharge ([AlCl4]-, SOCl2, I-). A MD simulation was conducted to obtain the radial distribution function (RDF) of Na+ ions and the specific atom of the solvents based on the periodic box of the designed electrolyte structure. RDFs of Na+ and different atoms in the SOCl2 solvents were further calculated to distinguish the component that was affected mainly by I-. As shown in Supplementary Fig. 11, the coordination number (CN) of Na+−ClSOCl2 based on the 0.5 nm dropped from 3.94 for the SANN electrolyte to 3.38 for the SANN-I2 electrolyte, suggesting that the coordinated SOCl2 solvent in the solvation shell was reduced. In addition, the CN of Na+ and I- in the additive electrolyte indicates that the coordination between Na+ and I- was 0.36. The above coordination states suggested that the I- additive significantly affects the solvated structure of Na+. The schematic illustration of molecular interaction between the solvent (SOCl2), the solute of [AlCl4]-, and I- additives is shown in Fig. 4a. The frontier molecular orbital energy levels can be adopted to evaluate the redox stability of the materials. The HOMO and LUMO energy levels of the components in the electrolytes were calculated by DFT method. As illustrated in Fig. 4b, the coordination structure with I- has a higher HOMO ( − 6.66 eV) and smaller energy gaps than the original coordination structure, indicating the enhancement of the redox activity of the coordination structure with I-, which may be mainly attributed to the effect of I- on SOCl2.

Fig. 4: Evolution of the NaCl crystal on the cathode tuned by iodine.
figure 4

a Schematic of the molecular interaction of the original SOCl2 electrolyte and SOCl2 with iodine additive. b Comparison of the solvated complexes’ calculated LUMO energy levels in the SOCl2 electrolytes with and without iodine additive. c De-coordination energy of different Na+ coordination structure. d In-situ XRD spectra of rGO cathode in OC battery and SDC battery. e Ex-situ TEM image and f schematic of NaCl crystals of the OC after cycling, the inset image is the selected area electron diffraction (SAED) of NaCl crystals. The light blue balls represent sodium ions, the soft green color represents chloride ions, and the purple color represents iodide ions. g Ex-situ TEM image and h schematic of co-deposited crystals of NaCl and NaI after cycled. i XRD pattern of NaCl and NaI co-deposition. j k High-angle annualar dark field scanning transmission electron microscopy (HADDF-STEM) image and EDS mapping of Cl, Na, I and S on the SDC. l High-resolution I 3d XPS spectra change of SDC upon discharge/charge.

Full size image

Typically, the solvated structure of the Na+ affects the transport and reaction kinetics of Na+, which is mainly reflected by the desolvation energy (Fig. 4c). Surprisingly, the Na+ can more easily desolvate SOCl2 as well as [AlCl4]-, etc., but with a strong binding energy to I-. It is postulated that species exhibiting strong coordinating abilities are more likely to infiltrate the inner layers28. Consequently, during subsequent charging and discharging processes, I- is possibly prone to accompany Na+ into the lattice of NaCl29,30, which opens up the possibility of iodine co-deposition with NaCl and result in irregular crystallization of NaCl and impeding further growth of NaCl crystals. As a consequence, the resulting NaCl exhibits enhanced oxidizing activity.

In-situ XRD was employed to investigate the potential dependence of NaCl crystalline and evolutionary process. As shown in Fig. 4d, three characteristic peaks of NaCl on OC became progressively more pronounced as the degree of discharge increases. The NaCl was gradually oxidized during the charging process until the characteristic peak disappears. We observed that the addition of iodine to the electrolyte system resulted in a decrease in peak intensity and an increase in the half-height width of NaCl characteristic peaks on SDC during discharge, as compared to the OC (Fig. 4d and Supplementary Fig. 12). This indicates reduced crystallinity of NaCl due to the presence of iodine. The characteristic peaks of OC did not completely disappear during the charging process, indicating the limited oxidation activity of highly crystalline NaCl. The progressive accumulation of irreversible NaCl resulted in the passivation of the rGO cathode and an increase in electrochemical polarization.

The ex-situ TEM and HRTEM results align with the in-situ XRD results, showing the NaCl crystal structure after cycling. Supplementary Fig. 13a shows a thick NaCl coating on the OC. The distinct lattice fringes of 2.81 Å ascribed to (200) plane in Fig. 4e, f confirmed the high crystallinity of the NaCl film. Furthermore, the SAED pattern inset of Fig. 4e showed the typical NaCl crystal planes of (200) and (220). In contrast, the NaCl deposited on the SDC appears as small particles and is well distributed (Supplementary Fig. 13b). As can be seen from Fig. 4g and h, the crystalline ring boundaries of NaCl at SDC shown in the SAED pattern become blurred and more inclined to the amorphous state, which verifies the low-crystallinity state of NaCl. Unsurprisingly, the Na-Cl2 battery with the SDC exhibited a small electrochemical polarization during cycling (Supplementary Fig. 14). This process allows for adjustment of the battery’s output voltage up to above 3.4 V from 3.1 V. It is not difficult to conclude that the reversible reaction of the NaCl in the SDC is faster than OC, resulting in an excellent electrochemical kinetic performance31,32.

As for the NaI in the deposits upon complete discharging, due to the low dosage of I2 (0.2 wt%) in this work, the iodine species is difficult to be clearly detected by XRD or TEM characterizations. Thus, ex-situ XRD and EDS of rGO with 10 wt% iodine additive under discharge to 2 V was conducted. In this condition, obvious NaI characterized peaks of (200) and (220) could be observed, and the (111) peak of NaCl undergoes a left shift, representing that the co-deposition of NaI widens the interlayer spacing at the (111) crystalline plane and prevents the growth of crystalline NaCl, especially the almost disappeared (111) crystalline plane (Fig. 4i and Supplementary Fig. 15). Figure 4j and k show the existence of iodine elements, distributing similar to that of Cl, Na elements. Notably, Supplementary Fig. 16 and Video S1-3 showed that the NaI has a much high solubility in the SOCl2 while the NaCl is almost insoluble. The high solubility of NaI in SOCl2 leads to its dissolution from the SDC. In Fig. 4l, the XPS peak of I 3d has a characteristic signal of I2, but with the discharging, this signal gradually disappeared, and not appeared in the charging process. The disappearance of I2 and dissolution of NaI upon battery operation may ascribe to the I- dissolved in electrolyte difficultly acquire electron to be oxidized to I2. Therefore, it is plausible that due to the strong coordination of I- and Na+, the I- become the fixed iodine species to undergo reversible co-deposition and dissolution process to enable low-crystallinity NaCl deposition and achieve repeated self-passivation of porous cathode.

Effect of iodine on passivation of electrode surfaces

Given the above results, we speculated that the I- can act as mediator to tune NaCl deposition and achieve self-depassivation of the electrode surface and further enhance the interfacial mass-charge transport. Ex-situ SEM tests were carried out to confirm the cathode-electrolyte interface in the Na-Cl2 battery with different electrolyte systems. Figure 5a shows the OC in the SANN electrolyte at different discharge states. When the battery was discharged to 700 mAh g−1carbon (capacity calculated based on mass of rGO materials), the NaCl on the surface of the OC was not uniform in size, and large NaCl microcrystals are present. When fully discharged, the large NaCl microcrystals formed a thick film, and almost all the rGO were covered and passivated by the NaCl microcrystals. Surprisingly, there were no excessively large NaCl microcrystals on the SDC in either discharge to 700 mAh g-1carbon or the fully discharged, and the NaCl also showed a sparse and more homogeneous state (Fig. 5b). This is due to the fact that the I- mediator leads to the destruction of intact crystals of NaCl, thereby inhibiting further growth. Additionally, the resulting membrane formed by smaller NaCl crystals tends to be thinner and more permeable to substances, facilitating rapid transport of Na+.

Fig. 5: The self-depassivation mechanism and analysis of the cathode-electrolyte interface.
figure 5

Ex-situ SEM images of the a OC and b SDC under different discharge states. In-situ Raman spectroscopy of c OC and d SDC in different states. SIMS mapping of Na e OC and f SDC. Three-dimensional depth distributions of Na3Cl2, NaCl2 and Na2Cl3 secondary ion fragments in the fully discharged g OC and h SDC. i GITT curves of the OC battery and SDC battery. Ex-situ EIS under different discharge states of j OC battery and k SDC battery.

Full size image

In-situ Raman spectroscopy of cathodes in different discharging/charging states also confirms the above conclusion (Fig. 5c and Supplementary Fig. 17). During the discharging process, the G band of the OC in the Na-Cl2 battery with the original electrolyte decreased with the decreasing potential applied, suggesting the production of NaCl and gradual coverage of the OC surface, and such peaks gradually disappeared at fully discharged due to the thick NaCl11. The G band presented as NaCl decomposes during the charging process, but the maximal peak intensities were lower than the initial status. For comparison, the changing of the SDC was also examined using in-situ Raman spectroscopy (Fig. 5d). During the charging/discharging process, the stronger peak intensity of the G band corresponds to the incomplete coverage and sparse NaCl film. Besides, the intensity of the G band has a complete restore as the initial state. Hence, the striking differences manifest the unique advantages of the I- mediator in boosting reversible reduction and oxidation of NaCl.

In addition, we performed time-of-flight secondary ion mass spectrometry (TOF-SIMS) to explore the NaCl distribution on the cathode surface. As a result, the sparse NaCl distribution on the SDC was observed with dispersed high-intensity yellow color, which was verified by the low intensity even at the starting of sputtering (Fig. 5e and f). In contrast, almost the whole surface of OC shows high-intensity yellow color with the sputtering deepening, strongly proves the large NaCl deposition on OC (Fig. 5g, h).

As this sparse and homogeneous NaCl exposes more active sites, it demonstrates higher reactivity and facilitates the electrochemical kinetics of the charging process. The galvanostatic intermittent titration technique (GITT) was applied to test the discharge equilibrium potential and IR drop (Fig. 5i). The Na-Cl2 battery with SDC shows a higher equilibrium potential (3.60 V) than the OC (3.55 V). The discharging potential is higher than 3.6 V and will result in the high energy efficiency. Correspondingly, compared with the relatively more significant IR drop (0.42 V) of the Na-Cl2 battery with OC in the original electrolyte system, the Na-Cl2 battery with SDC delivers a 0.15 V IR drop, which can be concluded that self-depassivating effects produced by I- mediator has a significant positive effect on the fast kinetics33. The electrochemical impedance spectroscopy (EIS) of the Na-Cl2 battery at different discharging/charging capacities validates the above conclusion. The impedance plot includes an incomplete semicircle in the high-frequency region and a tail in the low-frequency region, in which the semicircle in the high-frequency region corresponds to the charger-transfer impedance and the tail in the low-frequency region corresponds to the mass-transfer control17,34. As displayed in Fig. 5j, k, a noticeable increase in the semicircles at the high-frequency regions could be observed for the discharged samples, and this behavior can be related to the accumulation of NaCl at the rGO cathode. When charging starts, the semicircle is sharply reduced, corresponding to a rapid impedance decrease due to the removal of NaCl in the coating layer on the rGO cathode4. It is worth noting that the Na-Cl2 battery with SDC exhibits lower charge-transfer resistance and better mass-transfer rates (Supplementary Table S2).

Electrochemical performance of the SDC battery

To support the importance of this reaction mechanism for electrochemical performance enhancement, Na-Cl2 batteries were assembled to test electrochemical performance systematically. The rate performance of the OC battery and SDC battery with a cut-off capacity of 700 mAh g-1carbon was further tested. As shown in Fig. 6a and Supplementary Fig. 18a, with gradually increasing current density from 150 mA g-1 to 1500 mA g-1, the SDC battery delivered a very well-matched charge/discharge specific capacity of 647/700 mAh g-1carbon while the OC battery showed a fast discharge capacity fade at high currents (488/700 mAh g-1carbon). The rate performance of various cut-off specific capacity was tested, and the SDC battery exhibited a high discharge capacity of 1000 mAh g-1carbon while maintaining stable cycling for over 150 cycles (Supplementary Fig. 18b and c). Additionally, the high areal current density performance was assessed that a reversible areal capacity of 3.85 mAh cm−2 for 30 cycles at a current density of 2.3 mA cm−2 (based on the electrode sheet area) can be delivered (Supplementary Fig. 18d). As shown in Fig. 6b, the charge polarization voltages of SDC battery were slightly increased from 3.96 V to 4.11 V as the current density increased from 150 mA g-1 to 1500 mA g-1, while OC battery increased form 3.97 V to 4.51 V under the same conditions (Fig. 6c). The charge polarization of SDC battery (4.11 V) is much lower than that of OC battery (4.51 V), demonstrating the enhanced reactivity of thin and poorly crystallized NaCl enabled by co-deposition and dissolution of NaI. The above results highlighted that the iodine-induced self-depassivation electrode accelerates the conversion of NaCl by effectively avoiding electrode surface passivation and ensuring rapid electron transfer. The enhancement of kinetic was further validated by CV analysis (Supplementary Fig. 19 and Supplementary Text 4).

Fig. 6: Electrochemical performance of SDC batteries.
figure 6

a Rate performance of Na-Cl2 battery at 150-1500 mA g−1. The charge-discharge curves of b SDC battery and c the OC battery at a cut-off capacity of 700 mAh g−1carbon. d Cycling performance of SDC battery when the charging capacity was 500 mAh g−1carbon at 1000 mA g−1carbon. e Cycle life comparisons among SDC battery and other Na metal anode batteries reported in literature. f Cycling performance of SDC battery and OC battery at the charge capacity of 500 mAh g−1carbon at 700 mA g−1carbon under -20 oC. g The polarization change of SDC battery with the cycle proceeding. h The polarization change of OC battery with the cycle proceeding. i The polarization comparison of OC and SDC batteries with the cycle proceeding. j The electrochemical advantages of SDC batteries over other battery systems based on the Na metal anode. k In-situ differential electrochemical mass spectrometry measurements of SDC and OC cells.

Full size image

Furthermore, a high current density of 1000 mA g-1 was used to charge the SDC battery to evaluate the cycle life. The SDC battery exhibited a long cycle stability of over 2000 cycles at room temperature (Fig. 6d), which is much superior than the OC battery and far ahead of other previously reported cycle life in the literature (Fig. 6e)4,9,10,11,12,35,36,37. Due to the limitation of charging capacity, the decomposition of the electrolyte contributes to side reactions and additional capacity, resulting in the Coulombic efficiency with charging (reversible NaCl decomposition and electrolyte side reactions)/discharging (reversible NaCl deposition) exceeding 100%. In addition, the SDC battery also exhibited excellent electrochemical performance at low temperatures, delivering a maintained CE of ~100% with a current density of 700 mA g−1 and 800 stable cycles under a capacity of 500 mAh g−1carbon at −20 oC (Fig. 6f). As comparison, the OC shows a short life of no more than 200 cycles (Supplementary Fig. 20). Moreover, SDC batteries demonstrated excellent performance under high and low temperature (Supplementary Fig. 21 and Supplementary Text 5). The results verify the feasibility of the self-depassivation effect under wide temperature range. Notably, the charging voltage of SDC battery has a minor increase with the cycle proceeding even over 1500 cycles (Fig. 6g). At the 60th, 120th, 180th, 500th, 1000th and 1500th, the charging voltage are 4.05 V, 4.07 V, 4.07 V, 4.05 V, 3.98 V and 4.13 V, respectively, while the OC battery deliver a much higher charging voltage of 4.24 V, 4.43 V and 4.97 V at 60th, 120th and 180th, respectively (Fig. 6h, i). At the 180th cycle, the charging voltage approaching 5 V, causing the battery failure. The SDC battery exhibited the highest capacity, output voltage and excellent cycle life compared with various Na metal anode batteries at room temperature (Fig. 6j and Table S3)37,38,39,40,41.

Thanks to this self-passivation effect accelerating the rate of charge-mass transport, the Cl2 can also be reduced quickly, thus preventing its random diffusion and spillage. In-situ differential electrochemical mass spectrometry (DEMS) measurements also show that the SDC cell released much less Cl2 during the first cycle than the OC cell (Fig. 6k)42. The SDC cell showed a decreased tendency for the intensity of Cl2 during charging, mainly due to the stable storage of Cl2 in the unpassivated porous electrodes.

In this work, we disclose that the deposition of the insulative NaCl on the cathode plays a crucial role in determining the cycling life of Na-Cl2 batteries. The NaCl accumulation and subsequent sluggish oxidation along with the charge/discharge cycling lead to severe cathode passivation and significant overpotential increase, which finally cause capacity diving and battery failure. Surprisingly, iodine anion can be used as a mediator to promote the formation of low-crystalline, small NaCl particles and the improvement of NaCl reoxidation activity through a NaI deposition-dissolution process, resulting in a significntly improved chlorine cathode with self-depassivation functions. The resultant Na-Cl2 battery with such a self-depassivation cathode exhibits an ultra-stable cycling performance with much-lowered overpotential, achieving a cycling life of up to 2000 times. These findings open up a new avenue for the development of high performance secondary metal-chlorine batteries for pratical applications.

Methods

Materials availability

SOCl2 (99%, Aladdin), AlCl3 (99.0%, Alfa), NaTSI (DoDoChem), NaFSI (DoDoChem), I2 (99%, Aladdin), KOH (95%, Aladdin) were used in this work. All chemical reagents for electrolyte preparation were transported in argon filled packages and stored properly in the glove boxes with water/oxygen below 0.01 ppm after dried. The NaFSI and NaTFSI were dried at 100 °C vacuum oven overnight before use. Graphene oxide hydrosol was purchased from Shanxi Institute of Coal Chemistry, Chinese Academy of Sciences. The sodium metal foils we used are standard commercial products which were purchased from Shenzhen Huaxin Materials Technology Co., Ltd. with a purity of 99.7%, and the diameter of sodium metal foil is 15.6 mm and the thickness is 0.45 mm with an aluminum foil as collector.

Synthesis of rGO

The rGO materials were synthesized by one-step carbonization/activation process using KOH. Typically, original rGO hydrosol and KOH were mixed in UP water at a mass ratio of 1:4, sonicated for 30 min and then stirred for 4 h. The precursors were obtained by freeze-drying. The obtained precursor was pyrolyzed at 600 °C for 4 h with the heating rate of 5 °C min−1 under the protection of flowing Ar atmosphere in a tube furnace. Finally, the rGO materials were obtained by washing with sufficient water and drying at 80 °C overnight in vacuum oven.

Preparation of electrolyte

The electrolyte was made inside an argon-filled glovebox with water and oxygen contents below 0.01 ppm. Firstly, 1.078 g AlCl3 with 2 wt% NaTFSI and 2 wt% NaFSI were added into a 5 mL scintillation count bottle, and then 2 mL SOCl2 was added and stirred for 20 minutes. Finally, a light-yellow solution can be obtained. The obtained supernatant was mixed with 0.2 wt% of I2 to obtain the SOCl2-I2.

Preparation of rGO cathode host

The cathode was prepared by mixing rGO, Super P and PVDF with a mass ratio of 8:1:1, which were dispersed in NMP through planetary mixer to mix for 30 minutes. The obtained slurry was dipped onto stainless steel sheet collector (diameter of 15.5 mm) and dried at 80 °C overnight drying in a vacuum oven. The mass loading of rGO was 1.0 ~ 1.4 mg cm–2.

Battery assembly

All the batteries were assembled inside an argon-filled glovebox with water and oxygen contents below 0.01 ppm. To prepare coin batteries (2032 type with 316 stainless steel), glass fiber filter (GF/D, Whatman) with a diameter of 19 mm, 100 ul of electrolyte was added to the prepared rGO cathode and the glass fiber filter in two additions, sodium foil was used as anode.

Characterizations

X-ray diffraction (XRD) patterns were characterized by a Bruker-AXS D8 Advance) with Cu-Kα1 radiation (λ = 1.5405 Å). Morphology of the rGO and cycled cathode were investigated by using scanning electron microscope (SEM, Hitachi Regulus 8230), transmission electron microscope (TEM, Thermo Scientific Talos F200X). In-situ Raman was obtained by Raman (HORIBA HR Evolution) of 532 nm wavelength. Surface chemical bonding states of the cycled cathodes were analyzed by X-ray photoelectron spectrometry (XPS, escalab 250Xi). Energy-dispersive X-ray (EDX) spectroscopy was tested by scanning electron microscope (SEM, Hitachi Regulus 8230). TOF-SIMS data were acquired from PHI nanoTOF II Time-of-Flight SIMS. Depth profiling was conducted with a primary Bi3+ beam of 30 keV and a scanning area of 100 × 100 μm2. Brunauer–Emmett–Teller (BET) surface area and pore volume were measured by a BELSORP MAX G (Microtrac MRB). In-situ differential electrochemical mass spectrometry (DEMS) was measured by QAS 100Li (Iridium filament, 5*10-4 A/mbar) with a mass range of 1−100.

All electrochemical measurements were conducted at 25 °C in a Land CT3002A thermostatic test chamber unless otherwise stated. Low-temperature measurements were carried out in freezers (DW-40L348J), and cyclic voltammetry test was carried out using the Gamry 5000. GITT was tested by executing multiple 20-minute (pulse time, τ) pulsed currents (0.05 A g−1) followed by 120 minutes relaxation intervals across the entire range of relevant voltages. Electrochemical impedance spectroscopy (EIS) analysis was performed through an alternating current (AC) signal of 10 mV (rms) in the varying frequencies from 100000 Hz to 0.1 Hz. The three-electrode CV was tested by a CHI760E workstation with the glassy carbon as working electrode, carbon rod as reference electrode, and sodium metal as counter electrode. The test was protected by a glove box filled with Ar.

Theoretical calculations

Ab initio molecular dynamic simulations of SOCl2·I2 complex. Firstly, gaussian was used to optimize the structure of SOCl2 and I2. Molecular dynamics of an I2 with a SOCl2 was done using ORCA software, time step 2 fs, initvel 298.15 K (based on batteries test temperature), and Time 10.0 ps.

Molecular dynamics simulations of Na+ coordination based on the electrolyte environment after the first discharge ([AlCl4]-, SOCl2, I-) were carried out with Gromacs 2019.6 (SoftwareX, 2015, 1-2, 19-25). One initial simulation system was constructed by uniformly mixing 300 Na+, 300 [AlCl4]– with 1377 SOCl2 molecules. Based on the first system, the second system was established by substituting 30 [AlCl4]– anions with the same number of I–. Parameters for Na+, I–, SOCl2 were obtained from OPLS-AA force field43. Parameters for [AlCl4]– were obtained from the work of Doherty et al. 44. Electrostatic interactions are computed with PME method. Van der Walls interactions are computed with Lennard Jones potential, with parameters between different atoms decided by the Lorentz-Berthelot combination rule, and the cut-off value is set as 1.2 nm. Energy minimization was conducted to relax the initial conformations, then 50-ns production runs were carried out at 298.15 K/1 bar (based on batteries test temperature). Equations of motion are integrated by the leapfrog algorithm. 3D periodic boundary conditions were applied. Quantum chemical calculations are conducted with the Gaussian 16 program. Electronic structures were computed at the B3LYP-D3(BJ)/def2TZVP level of theory.

Spin unrestricted density functional theory (DFT) calculations were conducted with periodic slab models within the DMol3 module of the Materials Studio software45,46. The method of the generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE) functional was utilized to the exchange-correlation potential47. Grimme’s DFT-D3 corrected method was used to account for weak interactions such as long rage van der Waals interaction48. The thickness of the vacuum layer was large than 15 Å. The convergence criterion was set to 10−6 au in self-consistent field computation. Energy, force and displacement was carried out until the convergence threshold of 1 × 10−5 Ha, 2 × 10−3 Ha Å−1, and 5 × 10−3 Å. Brillouin zone was sampled with the 4 × 4 × 1 Monkhorst-Pack mesh for the geometry structures and energy calculations for the slab models. The transition states were searched by using the linear synchronous transition/quadratic synchronous transit method and confirmed by frequency calculation49.

The binding energy (BE) between two molecules was calculated by: BE = EAB − (EA + EB), where EAB is the total energy of the system (SOCl2·I2), EA and EB represent the energy of molecules SOCl2 and I2, respectively.

The formation energy of SOCl2, SOClI, SOCl2I, and SOI2 were calculated according to Ef=Etotal-EO-ES-nECl-nEI, where Etotal represents the energy of the molecule, EO, ES, ECl, and EI represent the energy of a single atom in the most stable molecule (EO = 1/2O2, ES = 1/8S8, ECl = 1/2Cl2, EI = 1/2I2), respectively, and n represents the number of atoms.

The Gibbs free energy of the reaction of molecules and related decomposition products (e.g. SOCl*, Cl*, SOClI*, I*, SO2, S) were calculated by G = EDFT + ZPE – TS, where EDFT, ZPE, and S were the DFT energy, zero-point energy, and entropy, respectively, and temperature T was adopted as 298.15 K (based on batteries test temperature).