Introduction

Lithium–sulfur batteries (LSBs), known for their high theoretical specific capacity and environmental friendliness, face practical problems such as polysulfide shuttling and lithium dendrite formation1,2,3. In contrast, magnesium–sulfur batteries (MSBs) have emerged as a prospective alternative, which mitigates the dendrite growth issues and offers a large theoretical volumetric capacity (3833 mAh cm−3)4,5,6,7,8,9. Despite these advantages, the practical development of MSB is hindered by low sulfur conversion efficiency and severe capacity decay10,11. The strong coulombic interactions between Mg2+ and polysulfides that significantly retard redox kinetics. Additionally, the poor electronic/ionic conductivity of discharge products, magnesium sulfide (MgS), leads to pore blockage and premature capacity termination12. To enhance reaction kinetics and optimize battery performance, these critical challenges must be effectively addressed.

The high energy density of MSB originates from multi–electron redox reactions involving complex phase transitions from solid sulfur to MgS13,14. During the transition processes, polysulfides (PS) solvated in the electrolyte serve as fundamental intermediates and directly govern the performance of the batteries15. Prior work has established that strong Mg2+–PS coordination dominates MgPS behavior16,17, but the specific solvation structure of MgPS and its profound influence on the fundamental conversion chemistry itself have remained elusive. Early attempts based on conventional magnesium electrolytes (such as the Grignard reagent/aluminum chloride system) were made, and the instability issue in the presence of polysulfides was pointed out18. Breakthroughs in the design of electrolyte systems for MSB, such as boron-based and phenolate-based systems with better compatibility, have also been achieved19,20. Besides, recent strategies demonstrate that introducing additives21,22,23 or catalyst24 to the electrolyte can modify the interphase to suppress polysulfide dissolution and diffusion. Nevertheless, the fundamental role of solvent–directed MgPS solvation in steering conversion kinetic and its critical impact on battery performance remain largely unexplored.

In this work, we reveal that the solvation structure of MgPS is a previously overlooked regulator of the conversion chemistry. To demonstrate this, a series of common ether solvents were selected to evaluate their ability of solvation modulation for MgPSs: 1,2-dimethoxyethane (DME), diethylene glycol dimethyl ether (G2), triethylene glycol dimethyl ether (G3) and tetraethylene glycol dimethyl ether (G4). The lone pair electrons on oxygen atoms of ether solvents enable strong coordination capability, which can stabilize the MgPS complexes. Although these solvents show analogous molecular structures, they demonstrate different electrochemical performance when applied in MSB. To systematically explore the role of solvent, we investigate the solvation–modulated property of polysulfide speciation and explore the relationship between the solvation environment of MgPS and battery performance. We found that solvent shields the interaction between Mg2+ and PS and governs MgPS solvation. Is/p, calculated by the Mayer bond order ratio of Mg-solvent (Is) to Mg-PS (Ip), was defined as a parameter to evaluate the shielding ability: solvation environment with high Is/p value exhibits higher shielding ability and exerts stronger effect on the Mg-S interaction, facilitating efficient growth of MgS with the uniform three-dimensional (3D) morphology (Fig. 1). Based on the simulation results, solvents affect the intrachain bonding of polysulfides, which concurrently changes the energy barriers during sulfur conversion. Electrochemical characterization and spectroscopy analysis prove that this solvation effect on MgPS manifest distinctly in voltage profiles, polysulfide conversion, and electrode thermodynamics. Meanwhile, solution–mediated nucleation and growth of MgS also show strong correlation with solvation effect. Among the prepared electrolytes, the G2–based electrolyte, which facilitates a favorable solvation environment with high shielding ability for MgPS, promotes the polysulfide speciation and conversion. High sulfur utilization with 3D morphology deposition of MgS was observed in G2–based electrolyte. Moreover, the G2–based electrolyte demonstrates stable cycling for 120 cycles at 0.1 C in coin cells and exhibits high capacity in pouch cells (940.80 mAh g−1 in the 1st cycle; > 600 mAh g−1 after 18 cycles), underscoring its suitability for MSBs. These results not only shed light on attainable reaction mechanisms of MgPS speciation and MgS electrodeposition, but also give guidance for solvent selection criteria in designing practical electrolytes for MSB.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Conceptual diagram illustrating the role of solvation modulation in MSB(the pink and blue items represent for the different solvents; the green and orange atoms represent for Mg atoms and sulfur atoms, respectively).

Results

Solvation–mediated Mg2+–PS coordination chemistry

Polysulfide species serve as the solvated intermediates in the electrochemical conversion of sulfur electrodes. Their formation and evolution directly govern the energy efficiency, cycling stability, and reaction kinetics of the battery25. To screen the interactions among Mg2+, solvents, and polysulfides and gain deeper insights into the solvation behavior of MgPS, we employed molecular dynamics (MD) simulations to investigate the structures of polysulfide species in four ether solvents :1,2-dimethoxyethane (DME), diethylene glycol dimethyl ether (G2), triethylene glycol dimethyl ether (G3) and tetraethylene glycol dimethyl ether (G4). Supplementary Fig. 1 exhibits the radial distribution function (RDF) analyses and coordination numbers in four electrolytes (The atomic coordinates of the optimized structure can be found in the Supplementary Data 1). It can be found that Mg2+ not only coordinates with O from the solvents and TFSI- in the solvation shell, but also coordinates with S from polysulfides due to the strong Lewis interaction. Here, the representative reaction intermediate MgS4, which bridges the gap in the solid–liquid–solid transformation process, was selected as the focus for a detailed illustration. The solvation structures of MgS4 derived from electrolytes with different solvents are shown in Fig. 2a. The calculated values of Mayer bond orders and Mulliken charge distribution are listed in Supplementary Tables 1–4. (The solvation structures and corresponding atomic numbers are shown in the Supplementary Fig. 2). According to Supplementary Tables 1–4, it can be seen that for four solvation environments, the terminal sulfur atoms show a more negative Mulliken charge value, indicating a great charge enrichment in two ends of S4 chain. At the same time, the Mayer order of Mg-S bond exhibits a strong–weak–strong sequence along the S4 chain and the Mayer bond order of middle S–S bond is weaker than other S-S bonds, suggesting that Mg2+ mainly interacts with sulfur atoms at both ends of S4. Figure 2b and Supplementary Table 5 demonstrate the average Mayer bond order of Mg-O(Is) and Mg-S(Ip) in four coordination environments from Fig. 2a. It can be inferred that, due to the strong interaction between Mg2+ and PS, Ip is stronger than the Is for all solvents in Fig. 2b. Moreover, the values of Is and Ip changes significantly with the solvation environments, indicating solvation shields the strong electrostatic interaction between Mg2+ and PS. To assess the role of the solvent on tailoring the solvation structure, the Is/p ratio (the value of Is divided by the value of Ip) is introduced as a parameter and represents the shielding ability of solvation. Among four solvents, G2-based solvation environment shows the highest Is/p and thus has the strongest effect on MgPSs, indicating effective modulation of Mg-PS interaction. Meanwhile, the Mayer bond order of middle S–S bond in G2–based solvation environment is the weakest (Fig. 2a) and is prone to cleavage, which benefits the MgPS solvation dynamics as well as sulfur conversion kinetics. Moreover, because the high shielding ability of G2, the Mulliken charge of Mg in G2–based solvation environment is the highest (Fig. 2b), suggesting that the valence state of Mg is less influenced. The reaction pathway in the implicit solvation environment by considering the MgS8, MgS4, and MgS is built in Fig. 2c. Attributed to the solvation effect on the coordination between Mg2+ and PS, the energy barrier from solvated polysulfides to terminal solid MgS shows different values. It can be shown that G2 with high shielding ability also facilitates a lower Gibbs free energy change (ΔG), enabling a reduced conversion barrier and accelerated redox kinetics for MgPS. The simulations explicitly decipher the cooperative effects of solvent–Mg2+–PS and demonstrate the critical role of solvation mediation in Mg2+–PS coordination, providing a molecular–scale basis for MgPS solvation dynamics.

Fig. 2: Solvation effects on MgPS chemistry and battery performance.
Fig. 2: Solvation effects on MgPS chemistry and battery performance.The alternative text for this image may have been generated using AI.
Full size image

a Comparisons of the solvation shells of MgPS with the middle S–S coordination bond order given by DFT calculations. b Coordination bonds of Mg–O (Is), Mg-S (Ip) and the Mulliken charge of Mg within the solvation structures of MgPS. c Gibbs free energy changes of MgS8, MgS4 and MgS in four solvent environments(the inset figure shows the conversion of polysulfides within the implicit solvation environment: the green balls and yellow balls represent the Mg atoms and sulfur atoms, respectively; the stick model in the backgrounding represents the solvent environment). d Galvanostatic charge–discharge profiles of the electrolytes at 0.05 C in MSB (1 C (60 min) = 1675 mA g−1). e Peak current and specific capacities of the batteries.

During the charging and discharging process, active sulfur species in both solid and solvated phases control the reaction thermodynamics and provide the capacity26. To figure out how the solvation–mediated intermediates influence the energy density of the battery, constant current charge–discharge and cyclic voltammetry tests were conducted using various electrolyte solvents and iodine additive. The galvanostatic charge–discharge profiles were displayed in Supplementary Fig. 3a–d and the profile of the third cycle at 0.05 C was shown in Fig. 2d. During the discharge reaction, magnesium metal is oxidized, producing Mg2+ ions that migrate toward the positive electrode through the separator and electrolyte. Simultaneously, electrons are transferred to the active sulfur materials via an external circuit to keep the positive electrode at a neutral charge state. It can be concluded that four types of electrolytes exhibited a certain capacity and featured a long plateau with a significant portion of the capacity, originating from the liquid–solid conversion step. Supplementary Fig. 3a shows that in the 1st cycle, DME–based cell shows the high discharge capacity around 1100 mAh g−1 at 0.05 C and exhibits two main voltage plateaus (~1.4 V and ~1.0 V). However, because of the irreversible MgS conversion, the high polarization leads to premature termination during charge (coulombic efficiency ~69%), leading to the gradual capacity decay in the following cycles and the cell failure. Based on the discussion in simulation results, it can be inferred that G2 provides the favorable solvation environment for polysulfides, which facilitating the reaction kinetics. Therefore, the charge–discharge profiles of G2-based cell have small polarization(~0.62 V) and reasonably high reversibility during charge and discharge. Meanwhile, the two redox peaks observed at approximately 1.9 V and 2.0 V in Supplementary Fig. 3b attribute to the redox reaction of iodine additive. For the G3–based cell, in the first discharge cycle, compared to G2, the solvation environment modulated by G3 has weaker promoting effect on polysulfide conversion, resulting in lower sulfur utilization and reduced capacity release. Notably, unexpectedly prolonged charging profiles near 1.8 V can be observed in the charge curves of G3-based cell in Supplementary Fig. 3c. During this process, the conversion of long-chain polysulfides to short-chain species exhibits slow kinetics, leading to the polysulfide shuttle effect and continuous loss of active sulfur. For G4–based cell, the high viscosity and poor wettability with the electrodes resulted in erratic curve in the first cycle. Then in the following cycles, sulfur undergoes incomplete conversion, which results in low voltage plateaus and low discharge capacity. Furthermore, because of the high dissociation energy barrier of MgS, G4-based cell suffers from a large polarization in charging process, ultimately leading to insufficient charge capacity (Supplementary Fig. 3d). These results confirm that the solvation environment critically determines polysulfide conversion efficiency, with the optimal coordination strength of G2 enabling both low polarization and stable cycling, and this is the core requirements for practical MSBs.

The key results derived from the galvanostatic charge–discharge profiles can be correlated with the CV analysis of four ether–based cells at a scan rate of 0.2 mV s−1 (Supplementary Fig. 3e–h). Two main reduction peaks at ~1.23 V and ~0.88 V were observed both in cells with DME and G2 baseline solvent, which can be attributed to the reaction from long–chain polysulfides to short–chain polysulfides and short–chain polysulfides to MgS, respectively. During charging, the two overlapped oxidation peaks belonged to the MgS oxidation in the G2–based cell start at 1.55 V. In contrast, the cell with the DME solvent shows two separate oxidation peaks at 1.60 V and 1.96 V, which demonstrates a higher energy barrier during the transformation from MgS to elemental sulfur. The small peak at ~1.47 V is ascribed to the iodide reduction in the DME– and G2–based electrolyte27. In the case of cells with the G3 and G4 solvent, although the first reduction peak appears around 1.23 V, the peak voltage originated from MgS formation in G3 and G4 is lower than DME and G2, revealing different polysulfides speciation in the electrolyte. Notably, apart from the obvious charge/discharge voltage variations and the pronounced reduction peak shifts, when comparing the value of peak current and capacities of discharge plateaus (Fig. 2e), it is recognizable that the significant numerical fluctuations exhibit a certain regularity consisted with the replacement of solvent. This variation implies that the electrochemical motive forces for the reversible conversion between elemental sulfur and MgS are drastically changed by solvation of MgPS.

Supplementary Table 6 presents the inherent properties of various solvents, such as viscosity and dielectric constant, which have been proved to have impacts on the sulfur transformation and its reaction mechanism. However, based on the above analysis, these properties cannot independently govern the electrochemical performance. Combined with the molecular simulation, we can comprehensively evaluate the role of solvent and make judgments from a more profound perspective: solvent tunes the coordination of Mg2+–PS and can provide favorable MgPS solvation environment. The feeble binding force of Mg2+ toward polysulfide species, coupled with the diminished inter-bond S-S coordination, promotes the acceleration of the reduction reaction via electron acquisition, which contributes to high capacity and reversibility of MSB during the charge/discharge process.

Solvation–mediated MgPS phase transition and nucleation

Redox–active MgPSs play an important role in phase transition at the electrode–electrolyte interface. The solubility of MgPS directly relates to the reaction pathways and kinetic behavior of sulfur, causing following impact on shuttle effect, battery efficiency, and cycling stability for enhancing overall battery performance28,29. Polysulfides dissolved in different solvents with/without iodine additive were analyzed using UV−vis spectroscopy (Fig. 3a and Supplementary Fig. 4). It can be learned that soluble S8 shows signals at 260 and 280 nm30. Owing to the difference in Donor number (Supplementary Table 6), the intensity of S8 peak in DME and G2 is higher than G3 and G4. The UV peak at 315 cm−1 attributed to S62–/S42– and the absorbance distributed in 355 and 409 nm belong to S62− and S42−, respectively31. Notably, overlapped with the sulfur peak, the UV peak at 234, 294 and 366 nm originates from iodide ions32,33. The stability of these main intermediates determines the reaction rates of the disproportionation/dissociation/recombination of polysulfides and thereby affects the phase transition of the MSB.

Fig. 3: Spectroscopic characterization of magnesium polysulfide solvation and conversion.
Fig. 3: Spectroscopic characterization of magnesium polysulfide solvation and conversion.The alternative text for this image may have been generated using AI.
Full size image

a UV−vis absorption spectra of 0.3 mM of MgS4 in various solvents. b Schematic diagram of the operando Raman spectroscopy. Operando Raman spectra of in MSB with c DME and d G2 solvent and corresponding contour map e, f with voltage profile during discharge.

As illustrated in Fig. 3b, the configuration of Operando Raman spectra was constructed to monitor the MgPS speciation in DME– and G2–based electrolyte during discharge process (Fig. 3c, d). The Raman spectra at open–circuit voltage was obtained, revealing peaks at 155, 218, and 470 cm–1 in DME– and G2–based cell, which can be assigned to the signals of elemental sulfur34,35,36. For DME–based cell, as the voltage decreases, the MgSx(x = 4–8) band was observed from 279–325 cm–1, which implies that bulk sulfur is initially reduced, triggering the opening of S8 rings and the formation of liquid–phase chain polysulfides. The band of S62– species appears at a high potential of 1.43 V and gradually diminishes with Sx2–(x = 4–8) as the voltage decreases, while a weak S8 peak can still be detected. In contrast, in the G2 electrolyte, liquid–phase polysulfides emerge at 1.39 V, and their intensity increases gradually during discharge, with almost no detectable S8 peak. This is attributed to the faster conversion of solid–phase S8 to long–chain polysulfides and the subsequent shortening of long–chain polysulfides in the G2 environment, generating the absence of the short plateau region observed around 1.4 V in DME. During the conversion from S to MgS, the step involving the transformation of S42– to MgS2 and MgS contributes to the most capacity37. It can be seen that the band of S22– and MgS are observed at 492 and 273 cm–1, respectively. In DME, the S22– species appear at a higher potential around 1.12 V and convers rapidly to MgS. However, in the G2 electrolyte, S22⁻ species can still be detected at low potentials (0.5 V) and gradually converts to MgS, resulting a longer plateau with more capacity contribution compared to DME. In addition, the solid species S22– spans in different voltage range in DME and G2, indicating S nucleation and deposition may relate to solvation.

To further explore the thermodynamic and kinetic properties of MSB during discharge process, galvanostatic intermittent titration technique (GITT) at a current density of 0.05 C was adopted in four ether–based coin cell (Fig. 4a). Each step of GITT was set to be 15 min following 2 h relaxation step in order to ensure the full recovery to thermodynamic equilibrium state(see Supplementary Figs. 5 and 6: Time-dependent potential plots). The corresponding polarizations of the electrochemical reaction in four ether–based cells (Fig. 4b and Supplementary Table 7) were obtained by comparing the equilibrium potential (grey) with the transient potential in Fig. 4a. Obviously, four curves show staging features and the equilibrium potential at long plateau follow a trend of DME (1.125 V) > G2 (1.116 V) > G3 (1.098 V) > G4 (1.088 V). Besides, the polarization status can be divided into three typical stages: S8–MgSx, MgSx–MgS2 and MgS2–MgS38. Firstly, active sulfur materials receive electrons through the external electrical circuit, which triggers the successive reduction from elemental sulfur to soluble polysulfides MgSx (stage 1). Then the soluble polysulfides cleave and convert to solid MgS2 (stage 2). Solid–state transition from MgS2 to MgS occurs at the end of the discharge process (stage 3). Stage 1 corresponds to the solid-to-liquid conversion of sulfur to polysulfides in MSB, specifically the dissolution of solid sulfur in the electrolyte environment. A short plateau is observed during this stage, corresponding to the emergence of polysulfides. In the case of G3 and G4, high viscosity leads to wettability issues, which reduces the rate of electrochemical reactions and results in low active material utilization. Consequently, the first plateau exhibits a lower voltage with a distinct sloping profile. In contrast, as seen in the G2 solvent environment, better wettability and higher ionic conductivity allow the active sites on the positive electrode to function effectively and sulfur can readily transform to polysulfides without rapid diffusion away from the reaction interface, thereby resulting in a well-defined plateau region. Meanwhile, DME-based cell suffers from passivation by irreversible residues formed in previous cycles, which hinders the accessibility of active sulfur. This directly leads to the shortage of the plateau and results in incomplete sulfur conversion. In stage 2, long potential plateau and negligible hysteresis were observed in all curves, indicating that the reaction at this stage is relatively stable and proceeds continuously with lower energy barrier for conversion. Since soluble polysulfides can easily diffuse to the electron available site and acts as redox mediators to transfer electron, the reaction kinetics is faster in stages 1 and 2 than stage 3. For stage 3, the dissolved polysulfides are gradually consumed and high polarization of the solid–state reaction leads to premature termination of the discharge. Moreover, comparing the GITT curves and overpotentials of four ether–based cells, it can be concluded that the stability of the reactions and the ability to overcome the conversion barrier in each stage are relevant to the solvation of the electrolyte. Among the cells, G2–based cell has the smallest variation in polarization and show a more effectively accelerate redox kinetics of the positive electrode. This not only proves the effect of solvation on polysulfides during sulfur reduction, but also illustrates that the charge transfer process and ion diffusion rely on the solvation of the electrolyte39.

Fig. 4: Electrochemical kinetic analysis of Mg–S batteries.
Fig. 4: Electrochemical kinetic analysis of Mg–S batteries.The alternative text for this image may have been generated using AI.
Full size image

a GITT measurements of MSB in DME and G2 under a repeating constant current pulse of 0.05 C for 15 min followed by a relaxation period of 2 h. b Corresponding overpotential of various electrolytes calculated from the GITT results. c PITT measurements of MSB in G2 (Current responses until 0.01 mA under −20 mV potential steps). d Dimensionless transient (line) in comparison with theoretical 2D and 3D models from PITT measurement of G2 based MSB. im: peak current; tm: time needed to achieve the peak current.

Based on the GITT results, it can be found that the unbearable polarization with the sluggish kinetics in stage 3 appears with the MgS nucleation and deposition. During the solid–solid conversion, the morphology and structure of the final product MgS may influence the achievable discharge capacity and the following oxidation process, which directly relates to the utilization and reversibility of active sulfur materials40. The quasi–equilibrium potential of MgPS to solid MgS was ascertained by utilizing the potentiostatic intermittent titration technique (PITT). Subsequently, the current−time transients were normalized based on the peak current (im) and the corresponding time (tm: specific time when the current reaches maximum). Then, the mechanism of MgS electrodeposition was analyzed by fitting the current-time transients to four classical electrodeposition models, including Scharifker−Hills (SH) models41 (3D instantaneous (3DI) nucleation and 3D progressive nucleation (3DP) nucleation) and Bewick−Fleischman−Thirsk (BFT) models42 (2D instantaneous (2DI) nucleation and 2D progressive nucleation (2DP) nucleation). It can be seen that the PITT curves of DME and G2 (Supplementary Fig. 7a and Fig.4c) show obvious current responses, which is ascribed to the processes of nucleation and growth in liquids/solids. By fitting and analyzing the deposition curves of MgS at the responsive current (corresponding voltage: DME at 0.94 V and 0.80 V; G2 at 0.94 and 0.84 V), the nucleation and growth of MgS in DME– and G2–based electrolytes follow the 3D growth mode, performing a transition from 3DI to 3DP as the discharge proceeds (Supplementary Fig. 8 and Fig. 4d). The 3D precipitation avoids clogging of the positive electrode and exposes more active sites, thereby improving sulfur utilization and stabilizing the electrode structure43. As for G3 and G4, there are no current signals in PITT curves (Supplementary Fig. 7b, c), suggesting that the nucleation mode is not suitable for G3– and G4–based electrolyte.

To further compare the diffusion state of polysulfides in DME and G2, we simulate the polysulfides diffusion and calculate the standard deviation by densities (DME: 0.074; G2: 0.065) (Fig. 5a, b). In DME, the polysulfides exhibit low dispersity and tend to aggregate into clusters on the positive electrode surface, which hinders the conversion and deposition. In contrast, in G2, polysulfides have higher dispersity, enabling better contact with the active sites of the electrode and facilitating more uniform deposition. Nanosized nuclei MgS particles were observed on the carbon paper (CP) via scanning electron microscopy (SEM) (Fig. 5c–f, Supplementary Fig. 9) and confirmed by the EDS (Supplementary Fig. 10). It can be observed that compared with the blank CP, the MgS particles generated in four different solvents all remain spherical shape at a size of 50–100 nm. In DME and G2, the small MgS granules deposit densely on CP and continue to precipitate on top of existing nuclei. Gradually, the particles connect together and the deposition become concentrated, which conforms to 3D nucleation mode. In contrast, in G3 and G4, a small amount of MgS particles are independently deposited on CP. Besides, reducing current density or raising sulfur concentration enhances the compactness of MgS particles (Supplementary Fig. 9b, c). The deposition of MgS on the positive electrode surface after cycling was further confirmed by XPS and TEM characterization (Supplementary Figs. 11 and 12).

Fig. 5: Dispersion and morphological characterization of polysulfide behavior.
Fig. 5: Dispersion and morphological characterization of polysulfide behavior.The alternative text for this image may have been generated using AI.
Full size image

Density profile with snapshots of polysulfides (yellow atoms) in a DME (pink) and b G2 (blue) by MD simulation. σ: the standard deviation of the densities; c–f SEM images of MgS deposition on carbon paper in DME, G2, G3, and G4, respectively.

Mg–S battery performance evaluation

Based on the preceding discussion and the comprehensive consideration of solvation effect, compared with DME, G3, and G4, the G2–based solvent electrolytes with iodine additive can better enable the operation of MSB. By using the three–electrode cell with Mg metal as the reference electrode, the contribution of G2–based electrolyte to the negative electrode and the positive electrode was clarified (Supplementary Fig. 13). It can be revealed that the three–electrode Mg–S cell shows similar voltage profile with two–electrode cell (Fig. 2d). The overpotential between Mg stripping and deposition at the negative electrode is only 0.4 V. Meanwhile, the positive electrode exhibits a polarization of ~0.7 V between the discharge and charge voltages. What’s more, G2–based Mg symmetric cells exhibit cycling performance of over 900 h with low polarization about 0.2 V (Supplementary Fig. 14), identifying the G2–based electrolyte exhibits chemical compatibility with electrophilic sulfur and possesses the capability for reversible magnesium deposition and dissolution. Consequently, G2 solvent is chosen to enhance the performance of MSB. By employing the G2-based electrolyte, the assembled Mg–S full cell exhibits improved performance. CV curves at different scan rates operated in G2–based electrolyte present similar shape and well–maintained redox peaks (A, B, C, D) (Fig. 6a), and there is a linear relationship between the peak response currents and square root of scan rates, proving that the highly reversible Mg storage process is driven by diffusion dynamics (Fig. 6b). Rate capability and long–term cycling performance were illustrated in Fig. 6c, d, respectively. It can be found that the G2–based MSB displays a high average specific capacity of 946.61 mAh g−1 at relatively low current density of 0.05 C, which gradually decreases to 350.20 mAh g−1 at 1.0 C, respectively. When the current density was restored to 0.5 C and subsequently to 0.2 C, the MSB presented high restoration capability. Moreover, the G2–based cell demonstrated high cyclic stability over 120 cycles at 0.1 C. The MSB pouch cell with sulfur loading of ~1 mg cm–2 was also assembled and the open–circuit voltage maintained around ~2 V (Fig. 6e). The voltage profiles of the pouch cell are displayed in Fig. 6f, showing the performance with high discharge capacity of the 1st cycle (940.80 mAh g−1) and high retention after 18 cycles. The corresponding specific energy are 31.44 Wh kg–1 in the 3rd cycle and 22.87 Wh kg–1 in the 10th cycle (Supplementary Table 8). These results suggest that G2–based electrolyte effectively facilitate kinetics and optimize the performance, demonstrating the prospects for practical MSB.

Fig. 6: Electrochemical properties of G2–based MSB.
Fig. 6: Electrochemical properties of G2–based MSB.The alternative text for this image may have been generated using AI.
Full size image

a CV curves at scanning speed from 0.2 mV s−1 to 0.5 mV s−1 and b linear fitting of the peak current (absulote value) versus the square root of the scanning rates. c Rate performance from 0.05 to 1 C(1 C (60 min) = 1675 mA g−1) and d cycling performance for 120 cycles at 0.1 C. e Cycling performance (the inset figure shows the initial open circuit voltage) and f voltage–capacity profiles of the pouch cell.

According to the above observations and analysis, we demonstrate that strategic solvent selection enables precise control over the solvation structures of MgPS, directly governing polysulfide speciation and ultimately dictating the morphology and reversibility of MgS deposition (Fig. 7). The sulfur reduction pathway initiates with the direct electrochemical conversion of solid S8 to S82− on the electrode–electrolyte interface, followed by its stepwise dissociation into short–chain species. In the solvation environment of MgPS, solvent shields the high electrostatic interaction between PS and Mg2+, which attenuates the bonding strength of Mg2+–PS and activates the cleavage steps of the polysulfides. Facilitated by the supply of external electrons, MgPS subsequently migrate to the electrode–electrolyte interface and achieve phase conversion. Benefitting from the optimized solvation environment, MgPS contact well with the active sites of the electrode by diffusion force and the formed MgS performs uniform deposition, promoting a 3D morphology nucleation. The 3D nucleation provides efficient charge transfer and improves the utilization rate of sulfur. The continuous growth of MgS is inevitably impeded because of the electron–insulating property, so the consequent deposition mainly depends on the conversion of MgPS. During this process, under the solvation effect, the MgPS with a strong affinity to the solid particles prefers to deposit on the pre–existing particles along the 3D directions, constructing an ideal interphase and enhancing the reversibility of MgS conversion.

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

The schematic illustration of solvation-modulated phase transition and nucleation.

Discussion

MgPS as the precursors of electrochemical deposition are crucial intermediates in sulfur redox reactions, yet the high charge density of Mg2+ creates strong interactions with polysulfides, severely limiting their solvation and stabilization. In this study, we draw the correlation between solvent and the regulation of polysulfide speciation and deposition for kinetic advancement. Specifically, solvent modulates the coordination between Mg2+ and PS and solvation with high shielding ability effectively modifies the binding configuration of MgPS, which directly controls the conversion reaction of MgPS. By systematically evaluating four electrolytes, we identified that solvation effect significantly influences voltage profile, MgPS speciation, and thermodynamics process of the sulfur reaction. Spectroscopic analyses including UV−vis absorption spectra and Operando Raman spectra demonstrate that the solvent dictates the speciation and conversion kinetics of polysulfides. Furthermore, solvation–dependent kinetics of sulfur degradation and MgS electrodeposition were also elucidated. Among the prepared electrolytes, the G2–based electrolyte emerges as the most promising. Accordingly, the G2–based full cell enables long cycle life and high capacity. This investigation provides substantial advancements in the fundamental understanding of MSB, promotes an intrinsic understanding of MgPS speciation and MgS nucleation, and offers a solvent selection framework to minimize the gap between theoretical and practical capacities.

Methods

General Materials

Mg(TFSI)2 (99.5%, Duoduo Chemical Reagent Co., Ltd) was vacuum-dried at 200 °C for 24 h before using. Magnesium sulfide (MgS, 99.98%, Adamas) and iodine (99.995%, Adamas) were sourced from Adamas–beta. Sulfur/carbon composites (S/C, 99.5%) were obtained from Xiamen TOB New Energy Technology Co., Ltd. Polyvinylidene difluoride (PVDF) and Super-P were obtained from Shenzhen Kejing Star Technology Co., Ltd. Solvents including 1,2–dimethoxyethane (DME, 99.5%), diethylene glycol dimethyl ether (G2, 99.5%), triethylene glycol dimethyl ether (G3, 99.5%), tetraethylene glycol dimethyl ether (G4, 99.5%,), dimethyl sulfoxide (DMSO, 99.5%) and N–Methylpyrrolidone (NMP, 99.9%) were all purchased from Adamas–beta and dried by molecular sieves (4 Å, Adamas–beta) for over 24 hours before using. All chemicals were used as received.

Synthesis of the electrolyte and magnesium polysulfides (MgPS) solution

The 0.5 M Mg(TFSI)2 electrolytes were prepared by dissolving dried Mg(TFSI)2 in DME, G2, G3, and G4, respectively. After 24 h of standing, iodine was added to achieve a final concentration of 0.5 M Mg(TFSI)2–0.1 M iodine electrolytes. Then, they were kept at 25 °C for 3 days to ensure the components were well mixed.

Magnesium polysulfide (MgSx) was synthesized by reacting MgS and elemental sulfur in a 1:3 molar ratio in DMSO under continuous stirring at 70 °C for 6 days to ensure complete dissolution and homogeneous reaction. Subsequently, the solution was vacuum-dried under 70 °C for 48 h to obtain polysulfide powders. The powders were then dissolved in solvents of DME, G2, G3, and G4 with/without iodine additive to prepare the MgSx solution. All preparation steps described above were performed in an argon-filled glove box, with both water and oxygen concentrations maintained below 0.1 ppm.

Preparation of S electrodes

The positive electrodes were prepared using the homogeneous slurry mixed from S/C, Super P, and PVDF in a 7:2:1 weight ratio in NMP. The slurry was cast onto Cu foils (16 μm, SHENZHEN KEJING STAR TECHNOLOGY CO., LTD) by doctor blade, followed by vacuum heating of the electrode at 50 °C for 12 h. The dried electrodes were cut into discs with a diameter of 12 mm for cell assembly. The active sulfur loading was around 0.6–1.0 mgsulfur cm−2.

Electrochemical measurements

Cell assembly were all carried out in an argon-filled glove box with water and oxygen contents <0.1 ppm. Electrochemical testing was conducted at 25 °C. The ionic conductivity of the electrolytes was measured using a conductivity meter (Mettler Toledo). To make coin cells (CR2025–type) for electrochemical tests, the glass fibre separator (GF/D, thickness of 675 μm, Whatman) was cut into 19-mm-diameter discs and vacuum-dried overnight at 120 °C. A piece of S cathode (diameter of 12 mm) and Mg foil (1 × 1 cm−2, Shenzhen Kejing Star Technology Co., Ltd) were separated by the GF/D separator with 90 μl of electrolyte. The galvanostatic charge‒discharge tests were conducted at 0.05 C (0.08 A g−1) or 0.1 C (0.17 A g−1) using the NEWARE battery test system. The cut-off potential ranges employed were 0.4–2.2 V for DME, 0.4–2.0 V for G2, 0.4–1.9 V for G3, and 0.4–2.1 V for G4, respectively. Corresponding cyclic voltammetry (CV) measurements were carried out on an IVIUMnSTAT analyzer at a sweep rate of 0.2 mV s−1 with the same respective potential ranges. The three–electrode electrochemical tests were conducted in hermetically sealed Teflon Swagelok cell. For the galvanostatic intermittent titration technique (GITT) tests, the cells were activated for 1 cycle and discharged at 0.05 C for 15 min followed by relaxation at open circuit for 2 h in each step. For the PITT test, the cell was potentionstatically discharged with current responses until the current was below 0.01 mA under –20 mV potential steps. The cells for MgS nucleation were assembled using carbon paper (CP, HCP020N) as the working electrode, magnesium foil as the counter electrode and Celgard 2325 as separator. A volume of 20 μL of MgSx catholyte was applied onto the CP, while 20 μL of the corresponding MgSx free electrolyte was added to the negative electrode compartment.

The rate performance and cycling performance of the Mg–S coin cells were assembled using positive electrodes with 0.62 mgsulfur cm−2 and 0.91 mgsulfur cm−2, respectively. The Coulombic efficiency is calculated by the ratio of charge capacity to discharge capacity. The pouch cells (one layer) were fabricated incorporating the negative electrode (Mg foil, 3 × 4 cm2, thickness of 50 μm), positive electrode with S loading of ~1 mg cm−2 (3 × 4 cm2), 200 μL electrolyte and GF/A separator (3 × 4 cm2, thickness of 260 μm, Whatman). Mg foils were freshly polished before being applied. Ni tabs (Shenzhen Kejing Star Technology Co., Ltd) were used for assembly. The cycling was performed at 0.1 C without external pressure applied during cycling. The specific energy of pouch cell is calculated based on the information listed in Supplementary Table 8. The calculation of the specific energy (W) follows:

$$W=\frac{C\times U}{M}$$

Here, C stands for the cell capacity. \(U\) represents the average discharge voltage and M represents the total weight. The mass of the packaging and tabs was not taken into the calculations. For each electrochemical condition, a minimum of five cells were evaluated and the performance curves presented in the main figures were derived from a representative cell that best reflects the median or typical behavior observed across all replicates.

Materials characterization

Ultraviolet–visible (UV–vis) spectra of MgSx were detected on a Shimadzu UV2600 UV–vis spectrometer. Operando Raman spectroscopy monitor under a 532 nm laser was detected with a perforated stainless steel positive case and a quartz window. The cell was prepared by the stainless steel collector (300 mesh, nominal pore size 40 μm, diameter of 12 mm discs) coated with the positive electrode materials (sulfur loading: 1.0 mg cm–2), Celgard 2325 separator, negative electrode (Mg, diameter of 12 mm discs), and 90 μL electrolyte. Each spectrum was acquired for 20 s with the frequency of 10%. The cells were tested at a current rate of 0.1 C (167.5 mA g−1). The Raman signals were acquired with a Raman microscope (Renishaw, British).XPS analysis was performed after five cycles. All spectra were measured after treating the sample for 10 min by Ar sputtering to remove impurities formed from the XPS instrument chamber as well as the oxide film formed during the sample transport. X-ray photoelectron spectroscopy (XPS) investigations were carried out with ULVAC-PHI, GENESIS 900, and the binding energies were calibrated against the C 1s reference peak. Small electrode fragments were taken out after the potentiostatic discharging tests. The samples were washed with the corresponding solvent to remove additional Mg(TFSI)2. Samples for transmission electron microscope (TEM, Talos F200X G2), tests were prepared after further sonication of the positive electrode in corresponding solvents and loaded onto a Cu grid. All samples were vacuum dried overnight before transferring to the microscope chamber. The microstructures of MgS on CP were revealed using scanning electron microscopy (SEM, Apreo 2S) accompanied an energy dispersive X–ray spectrometer (EDS).

Theoretical calculation

Density functional theory (DFT) calculations for geometry optimization were conducted by Gaussian 16 software package44. Geometry optimizations and energy calculations were performed using B3LYP/6–311 G(d,p)45, and the convergence tolerance quality was the default. The solvation effect was considered with the universal solvation model based on density (SMD)46,47.

All the classic molecular dynamic simulations conducted in this work were performed using the Large–scale Atomic/Molecular Massively Parallel Simulator (LAMMPS, http://lammps.sandia.gov). OPLS–AA48 parameters with 1.14*CMIA partial atomic charges were generated by LigParGen49,50,51 for the organic solvent molecules. The OPLSAA force field developed by Jorgensen was used for Mg2+ 52. To get the electrolyte structure, isothermalisobaric (NPT) ensemble runs were first performed at 328 K for 5 ns and then 298 K/328 K for 5 ns to ensure the equilibrium salt dissociation53. Then, 5 ns long canonical (NVT) ensemble runs were conducted and the last 0.5 ns were used to obtain the structure of electrolyte.