Reinvented sodium anode by creating a metal-bulk storage matrix with an expanded 3D plating/stripping mechanism

Abstract

Direct metal anodes are plating/stripping processes without a supporting framework and bulk ion conductivity; they are the electrodes susceptible to collapse and limiting the electrochemical reaction to the two-dimensional surface. The focus of this era is mostly on building a solid electrolyte interface (SEI). However, simply building protective layers cannot address essential issues; a thorough transformation of the metal electrode bulk is critical. We propose a reconstructed sodium metal anode (RSMA) by implanting an activatable ion-conductive network to the bulk. NaPF₆ will be activated with an electrolyte to conduct ions and form an anion-derived SEI. Conductive polymers become the supporting skeleton; thus, the RSMA has a metal-bulk storage matrix and an expanded three-dimensional plating/stripping mechanism and permits the homogeneous deposition/dissolution of Na⁺ in high dimensions. Last, RSMA symmetric cells were stably cycled for 2700 hours and achieved a 100% depth of discharge. RSMA||PB cells can achieve 10-coulomb cycling and a proof-of-concept pouch cell energy density of 367 watt-hours per kilogram.

A sodium anode with 3D plating/stripping and ion-conductive networks forms protective layers, boosting battery performance.

INTRODUCTION

Environmental and resource sustainability is critical for the development of next-generation battery technologies, and the development of high-specific-energy battery technologies is imminent for meeting the expanding markets of renewable energy storage in smart grids and automotive electrification. The metal anode, with its high specific capacity and independence from resource conditions, is favorable for the development of highly reversible, high-specific-energy metal batteries (1, 2). Unfortunately, the formation of dendrites triggered by inhomogeneous deposition behavior poses a serious threat to the safety of batteries and limits the further development and application of metal batteries. These active metal anodes (e.g., lithium or sodium) are severely hampered by the electrochemical kinetic hysteresis and the chemical instability of the unevenly thick and thin solid electrolyte interface (SEI), which do not allow for the rapid migration of cations at the electrodes (3–6). In addition, the metal anode inherently lacks an effective support framework and is intrinsically soft and viscous, with volume expansion in plating/stripping prone to structural collapse (7, 8). They are only electrically conductive, with insufficient ionic conductivity, limited to two-dimensional surfaces for deposition, and prone to dendrite growth and local volume changes, leading to short-circuiting and safety issues (9–12).

In recent years, the mainstream direction of constructing stable metal anodes is still focused on electrolyte modifications (e.g., high-concentration electrolytes and local high-concentration electrolytes) or the fabrication of artificial SEI protective layers (13–18). It is worth affirming that these modifications are important strategies to passivate metal anodes to prevent continuous parasitic reactions with liquid electrolytes. They focus on the formation of a protective layer on the metal surface to alleviate some of the problems of active metals under mild charging and discharging conditions (19). Unfortunately, these means of surface engineering alone cannot completely overcome the defects of metal electrodes. These modifications are still challenging to attempt stable plating/stripping at higher current densities and even under high depth-of-discharge (DOD) conditions. After all, under such demanding conditions, large volume changes and low ionic conductivity of the metal bulk phase often led to rapid SEI failure, internal disorder of the bulk phase, and rapid structural collapse (20). In addition to surface modification, current studies have used three-dimensional (3D) collector matrices (e.g., porous carbon, 3D metal-organic frameworks, and alloys) with high surface area and electronic conductivity to host active metals and maintain structural stability during cycling (21–26). However, in addition to adding many inactive components, the central problem with metal electrodes still lies in their intrinsic structural properties, especially how to achieve 3D ionic and electronic conductivity in the interior of the metal electrode. An ideal substrate should ensure fast and uniform cation flux, excellent plating/stripping kinetics, and strong self-supporting capability for stable operation under harsh conditions, both on the surface and inside. This requires not only a rational design of the surface SEI but also a complete transformation of the entire electrode bulk phase. To date, such multiple transformations still lack rational and simple design. Furthermore, the metal anode needs to be reinvented to deal with the inherent electrochemical instability of metal batteries.

Therefore, we propose the concept of a reconstructed metal anode (RMA) to achieve a radical transformation of metal anodes. As an example, a reconfigured sodium metal anode (RSMA) with a 3D self-supporting ionic/electronic conducting skeleton was formed by preimplantation of an activatable ion-conductive network and polymer skeleton and describes its effects on electrolyte and SEI formation and electrochemical properties. A simple mechanical synthesis was used to prepare the mixed metal salt skeleton of RSMA (as shown in fig. S1 and the experimental methodology), thereby altering the structural stability and mode of action of the sodium metal. Sodium hexafluorophosphate (NaPF₆) was activated and enriched at the electrode/electrolyte interface by the electrolyte to form a confined dissolved state, leading to a 3D reactive zone, where there is no longer a two-dimensional plane, allowing Na⁺ to enter the interior uniformly for plating/stripping in high dimensions, resulting in rapid Na⁺ penetration and electron transport. The retained NaPF₆ and PPy act as a 3D salt skeleton with a specific crystalline structure in the sodium metal bulk, supporting the entire electrode. Meanwhile, on the basis of the confined high salt concentration, the decomposition of PF₆⁻ becomes the main source of SEI components (including NaF, Na₃PO₄, and Na_xPO_yF_z) without the need for cumbersome adjustments of the solvation structure. Last, the symmetric cells with RSMA composite electrodes were stably cycled in carbonate electrolytes at 1 mA cm⁻² with 1 mAh cm⁻², 5 mA cm⁻² with 5 mAh cm⁻², and 10 mA cm⁻² with 10 mAh cm⁻² for 2700, 270, and 140 hours, respectively. The thin RSMA electrodes loaded on copper also show the advantages of a 3D metal-salt-polymer skeleton, achieving 100% DOD cycling. The RSMA||PB full cells achieve excellent long-term cycling (>9-month cycle life with an 82.1% capacity retention), rate performance (100 C), variable temperature operation (−40° to 60°C) and long-term cycling under high current density at 10 C. Actual sodium metal batteries (SMBs) exhibit stable operation [anode/cathode capacity (N/P) ratios of 4, 2.5, and 1] at 4.5 V. On the basis of anode and cathode active materials, the proof-of-concept pouch cell (N/P ratio of 2.5) achieves an energy density of 367 Wh kg⁻¹ and a capacity retention of 85.6% after 200 cycles, a record high for pouch batteries. In addition, the excellent performance with other sodium salts [sodium bis(trifluoromethylsulfonyl) imide (NaTFSI) and sodium bis(fluorosulfonyl)imide (NaFSI)] and on lithium proves the broad application prospects of the RMA concept and strategy.

RESULTS

Preparation and characterization of the RSMA

The RSMA was achieved by a simple calendering and folding process, which resulted in the formation of a composite structure consisting of stable and activatable NaPF₆ crystals densely packed in a PPy skeleton with an electrochemically active sodium metal matrix. To verify the stability of NaPF₆ and PPy, we analyzed the composition of RMSA using x-ray diffraction (XRD) (Fig. 1A). Peaks at 5° to 20° are present with characteristic peaks associated with the Capton membrane and PPy (27). The diffraction peaks after RSMA electrode loading are consistent with the PDF standard cards of metallic Na (PDF no. 22-0948) and NaPF₆ (PDF no. 07-0292). No diffraction peaks of the decomposition products of NaPF₆ and PPy (e.g., NaF) are detected, suggesting that they are still stored in their original form in the sodium electrode, which not only forms a stable polymer-salt skeleton but also provides preparation for the subsequent formation of a confined high concentration of the NaPF₆ system at the interface (28, 29). The morphology was further analyzed by scanning electron microscopy (SEM), elemental mapping spectroscopy, and optical photographs (Fig. 1, B to E, and fig. S2). Compared with the bare Na surface, the NaPF₆ powders filled the uneven voids inside the sodium metal, and the overall RSMA was flatter and smoother. Meanwhile, the F, C, P, and N elements were densely distributed on the RSMA, and the electrode components were completely fused, resulting in the preparation of RSMA electrodes with more moderate hardness and viscosity. We further performed energy-dispersive x-ray spectroscopy analysis and thickness analysis on different regions of RSMA formed by the same batch of roll pressing to verify the uniformity of Na, PPy, and NaPF₆ and the thickness of roll pressing (figs. S3 and S4 and table S1). From energy-dispersive x-ray spectroscopy, it is seen that the contents of Na, F, P, C, and N in these regions are basically close to each other with the ratio of basically 13.6:3.83:1:0.23:0.065, which is close to the theoretical value. In addition, the cross-sectional thicknesses are all around 77 nm. Such composite electrodes with effective thickness control imply higher DOD and sodium metal utilization (30). The surface morphology of RSMA electrodes was further examined using atomic force microscopy (AFM) (Fig. 1F). The average surface roughness of RSMA was 112 nm over a size area of 5 by 5 μm, showing low undulation and a flat surface morphology, confirming that the electrodes prepared by mechanical mixing do not have an additional effect on the surface roughness. Also, many particles can be seen on the surface, further indicating that NaPF₆ is uniformly distributed on the sodium electrode. The composition of the RSMA electrodes was further analyzed using x-ray photoelectron spectroscopy (XPS) (Fig. 1, G and H). The presence of C peaks (C─C/C═C and C─N peaks) corresponding to PPy on the RSMA and the high distribution of P and F elements at 0 and 50 nm further confirmed the ability to form high concentrations of NaPF₆ on the RSMA surface. The electronic conductivity of RSMA was tested using a four-probe test system. The electronic conductivity of RSMA was slightly lower than that of bare Na, but all were within 10⁵ S cm⁻¹. Although the RSMA contains 30 wt % NaPF₆, which has an impact on the overall electronic conductivity, the results are on the same order of magnitude as bare Na (tables S2 and S3). To understand the RSMA from a more microscopic viewpoint, we further observed the fragments of RSMA using transmission electron microscopy (TEM) (Fig. 1I). By combining the mapping images of the corresponding elements, a uniform distribution of Na, P, F, and N elements can be clearly seen. The O elements are essentially microgrid substrates from TEM, with no signals from RSMA-related shapes. As a result, the large amount of uniformly distributed NaPF₆ close to the electrolyte will be rapidly activated and form a confined-domain high-concentration dissolved state, which becomes the reactant for the subsequent anionic decomposition and provides the “seed” for the formation of the F-rich inorganic SEI layer. The subsequently reacted RSMA (including the formed stable SEI, internal NaPF₆, and PPy) not only provides a reliable support for the sodium metal but also provides a stable bridge for rapid ion/electron transport.

Fig. 1. Characterization of the RSMA.

(A) XRD diffraction pattern. a.u., arbitrary units. (B) SEM and (C) corresponding mapping images. (D and E) Optical image and electrode thickness. (F) AFM surface topography. (G) C 1s XPS spectra and (H) atomic content of RSMA. (I) TEM images and elemental mapping of RSMA.

Stable RSMA with interface chemistry

To evaluate the most suitable ratio of the electrodes, different electrodes were prepared using different mass fractions of NaPF₆ powder, and it was found that the use of 30 wt % NaPF₆ exhibited the best cell performance (fig. S5 and note 1). Electrochemical impedance spectroscopy (EIS) tests of cells after being rested for different periods of time also showed differences between the RSMA and bare Na. In addition, Raman spectroscopy and ¹⁹F nuclear magnetic resonance also demonstrated the effect of RSMA on the electrolyte. The amount of NaPF₆ dissolved in the RSMA was calculated to be about 1 M by immersion experiments (figs. S6 and S7 and note 2) (31, 32). Therefore, the enrichment of NaPF₆ at the electrode/electrolyte interface would contribute to the formation of a more stable SEI. Afterward, TEM was used to characterize the morphology and thickness of SEI. The SEI derived from bare Na was thick and inhomogeneous. However, the RSMA-derived SEI was thinner and more homogeneous (~13 nm) and exhibited excellent interfacial properties (Fig. 2, A and B). The different electrodes after cycles were further analyzed by mapping and elemental line scanning, and both the electrode bulk and SEI were different (fig. S8 and note 3). High-resolution TEM of the SEI formed by different electrodes also showed that the SEI formed by bare Na is amorphous without any lattice stripes and diffraction rings. The RSMA is not only obvious with many lattice stripes, but also the diffraction rings of the inorganic component in the SEI are clearly visible (fig. S9 and note 4). In addition, we also analyzed the morphology and elements of the different electrodes after cycling (figs. S10 and S11 and note 5). The bare Na particles after cycling contain large amounts of C and O elements, and their organic SEI leads to a rapid decrease in cell performance. In contrast, RSMA particles formed by the rolling process have a regular structure with fewer C and O elements, forming an excellent SEI. The interfacial decomposition products of the different electrodes were investigated using deep sputtering XPS (Fig. 2, C and D). The C 1s spectra of both electrodes contain C─C (284.8 eV), C═O (287.5 eV), and C─F (289.5 eV), which are species derived from the decomposition of carbonate solvents [e.g., fluoroethylene carbonate (FEC)]. The C─N peak at 25 nm (286.2 eV) in the RSMA is from PPy, further suggesting that the electrode forms a thin SEI (33, 34). In addition, bare Na also has C─O (286.8 eV) and CO₃²⁻ (289.5 eV) peaks from the massive decomposition of the solvent. The peaks at 688 and 684.5 eV in the F 1s spectrum are C─F and NaF, respectively, indicating the electrochemical reaction of FEC. The crystalline Na₂CO₃ and C─F formed by the solvent are unstable and tend to decompose when in contact with the electrolyte and Na, increasing the porosity of SEI and leading to electrode surface roughness and corrosion (35–37). In contrast, the RSMA showed a considerable proportion of NaF and Na_xPO_yF_z (686.5 eV), demonstrating the involvement of NaPF₆ in SEI formation (38). The solid-phase NaPF₆ is rapidly activated when the RSMA encounters the electrolyte, forming a confined-domain high-concentration dissolved state, which in turn changes the composition of the SEI (28). In the O 1s and P 2p spectra, their peaks also correspond to the C 1s and F 1s spectra (fig. S12 and note 6) (39). With increasing sputtering depth, the signals of bare Na inorganics (e.g., NaF) decrease, and more organics are present in the inner layer. On the basis of the XPS results, the F element content of SEI in the RSMA is higher than that of bare Na, close to the Na element, and the relative content within the dense layer is close to 35%, which is the main component of SEI (Fig. 2, E and F). Combining the F 1s and P 2p spectra, the preferential decomposition of PF₆⁻ at the electrode interface is due to the participation of more PF₆⁻ in the Na⁺ solvation structure and the enrichment of NaPF₆ at the RSMA interface to form a 3D reactive zone with high concentration, resulting in the presence of more NaF, along with an increase in the content of P-containing substances (including Na_xPO_yF_z, OPF_x, and phosphate Na₃PO₄) (40, 41). These beneficial components occupy many of the internal sites of the SEI (Fig. 2G), exhibiting improved densification and homogeneity, which is essential for stable cycling under harsh operating conditions.

Fig. 2. SEI characterization of RSMA and mechanism of SEI formation.

TEM images of SEI thickness of (A) bare Na and (B) RSMA. XPS depth profiles of (C) C 1s and (D) F 1s for the sodium metal after 15 cycles at 1 mA cm⁻² and 1 mAh cm⁻² in bare Na and RSMA symmetric cells after Na stripping. (E and F) Corresponding atomic concentrations from the C 1s, O 1s, P 1s, F 1s, N 1s, and Na 1s spectra on the different Na electrodes for various durations of Ar⁺ sputtering. (G) Normalized ratios of different species in the SEI formed on the different Na electrodes after Na stripping. 3D rendering for the NaO⁻, CH₂O⁻, NaF⁻, and PO₃⁻ secondary ions and overlay in TOF-SIMS tested for the SEI formed in (H) bare Na and (I) the RSMA. (J) Depth profiling of different fragments for bare Na and the RSMA.

Time-of-flight secondary ion mass spectrometry (TOF-SIMS) also provided more intuitive evidence to elucidate the formation mechanism of the superior SEI. Ion fragments of SEI components were first detected on the side in contact with the electrolyte (figs. S13 and S14). NaF⁻, NaO⁻, PO₃⁻, and CH₂O⁻ were the characteristic ion fragments of NaF, Na₂O, and Na─P─O components and the organic components, respectively. The intensity of NaF⁻, NaO⁻, and PO₃⁻ was very low in bare Na, with the SEI being mostly occupied by these organics, which is consistent with the XPS and TEM results. The high intensity of NaF⁻, POF₂⁻, and PF₆⁻ in the RSMA implies preferential decomposition of NaPF₆. Further depth profiling of the resolved 3D structure for the SEI and the spatial distribution of the components also indicate a very high content of inorganic fragments such as NaF⁻ and PO₃⁻ in the RSMA SEI, forming a thin and dense SEI (Fig. 2, H to J, and fig. S15). It is noteworthy that the increasing intensity of NaPF₆, from the massive decomposition of the SEI to the internal aggregation, affects the overall RSMA. From these results, the electrolyte has more PF₆⁻ involved in the Na⁺ solvation structure, leading to a different reaction pathway from the dehydrogenation hydrolysis reaction in the ethyl methyl carbonate (EMC) electrolyte, which avoids corrosive substance formation but promotes the generation of P─O substances (note 7) (39, 42). The whole electrode ensures fast internal electron transport and prevents electron tunneling through the inorganic-rich SEI, which guides Na⁺ to achieve homogeneous deposition and improves the reversibility of sodium plating/stripping.

Stability of RSMA

For insight into the improvement of sodium deposition behavior by the RSMA, the deposition morphology of sodium was observed using SEM for different cycles at a current density of 1 mA cm⁻² and a capacity of 1 mAh cm⁻² (Fig. 3, A to F). During the cycling process, the sodium deposits on the bare Na surface showed a loose and porous morphology and even obvious cracks, failing to form a dense Na deposition layer. This obviously fails to block the penetration of the electrolyte and the occurrence of side reactions and is accompanied by the growth of sodium dendrites, which is unfavorable to the subsequent deposition. The dissolution of rooted sodium is more likely to occur during the stripping process, resulting in the deposition layer detaching from the substrate and the formation of dead sodium, which cannot meet the safety and performance requirements of the cells (12, 18). The RSMA not only offers excellent SEI protection but also provides better deposition uniformity. It acts as a hybrid ion/electron–conductive network that enables 3D deposition of Na⁺ with smooth and flat overall deposition, effectively reducing dendritic growth and “dead sodium.” This means less electrolyte and active sodium consumption, which can be maintained over long cycles. The deposition cross section and higher Na plating capacities of RSMA showed lower volume fluctuations and dense and dendrite-free deposition (figs. S16 and S17). In addition, the RSMA showed some porosity (P₀ = 15.4%) and better electrolyte wetting capacity (22°), demonstrating its potential for 3D Na plating/stripping (figs. S18 and S19 and notes 8 and 9), further increasing the capacity to explore the evolution of Na plating/stripping morphology. The stripping of RSMA at high capacity revealed the 3D skeleton characteristic of metal-salt-polymer. During the plating after stripping, the RSMA was redeposited uniformly and the overall morphology was smooth and dense. The side morphology of RSMA is also flat and dense without volume expansion at high capacity (fig. S20 and note 10). AFM provides a more intuitive view of the flatness of the deposited sodium metal (Fig. 3, G and H). In an area of 5 by 5 μm, the average surface roughness values of bare Na and RSMA are 174 and 48 nm, respectively. The smaller degree of undulation of RSMA contrasts impressively with the “hills and valleys” of the bare Na surface (43). In addition, the Cu electrode of the Na||Cu cells after cycling also showed a uniform and flat deposition pattern of RSMA, maintaining a low deposition thickness (39.6 μm versus 68 μm) (fig. S21). The dynamic process of sodium deposition was observed in a transparent electrolytic cell with the aid of an optical microscope (movies S1 and S2 and fig. S21). Within the initial 10 min, a small amount of flocculent sodium deposits quickly appeared on the surface of bare Na. This inhomogeneous flocculent deposit leads to the rupture of the SEI, making Na⁺ tend to enrich at the crackout, which will further enlarge the inhomogeneity of the Na⁺ flow and intensify the dendrite growth. During the last 80 to 100 min, Na⁺ subsequently migrating to the surface continues to be deposited in the flocculated region, resulting in dendrites becoming larger and gradually wrapping around the entire electrode (Fig. 3I). As for the RSMA, it not only offers good SEI protection, but the electrode with a skeleton serves as an excellent ion/electron transport bridge that can be consistently and uniformly densely deposited for up to 100 min without dendritic growth, which is consistent with the SEM and AFM characterization results (Fig. 3J and note 11) (44).

Fig. 3. Characterization of RSMA plating/stripping morphology.

SEM top views of sodium deposition morphology for (A to C) bare Na and (D to F) the RSMA under different cycles at 1 mA cm⁻² and 1 mAh cm⁻². AFM 3D topography analysis of the Na cycle in (G) bare Na and (H) the RSMA at 1 mA cm⁻² and 1 mAh cm⁻². In situ dynamic observations of (I) bare Na electrode and (J) RSMA electrode deposited continuously for 100 min at a current density of 0.5 mA cm⁻².

Compatibility of the RSMA

In addition, to further verify the kinetics of the sodium plating/stripping reaction of the electrode, cyclic voltammetry (CV) scan and Tafel polarization test for the symmetric cells after cycling were performed (Fig. 4A and fig. S23). The CV peak current response and peak area of RSMA were larger, while the exchange current density was 10 times that of bare Na. This suggests that the RSMA has faster ion transport kinetics and more active sites for sodium deposition, facilitating highly reversible sodium plating/stripping (45). Na⁺ needs to undergo two processes, desolvation at the SEI interface and diffusion in the SEI layer before reaching the anode surface to exchange electrons, which together determine the speed of Na⁺ transport at the electrode interface (46). The electrochemical impedance spectra of temperature variation were measured for R_ct and R_SEI at different temperatures (45, 46). The interfacial resistance of the RSMA is smaller than that of the bare Na electrode at different temperatures, indicating that it has lower deposition resistance (fig. S24). The R_ct and R_SEI values fitted by the equivalent circuit are listed in table S4. Combined with the Arrhenius equation, we can quantify the activation energy required for Na⁺ desolvation and passage through the SEI (denoted as E_a1 and E_a2, respectively) and thus characterize the kinetic rapidity of Na⁺ deposition on the electrode surface (Fig. 4, B and C). The smaller E_a1 and E_a2 for the RSMA suggest faster Na⁺ transfer kinetics.

Fig. 4. Electrochemical properties of RSMA anodes.

(A) CV scan curve of bare Na and RSMA symmetric cells after 10 cycles of reactivation at 1 mA cm⁻² and 1 mAh cm⁻². (B and C) Arrhenius behavior and activation energy of charge transfer and Na⁺ diffusion through the SEI film in bare Na and the RSMA. CE test in Na||Cu cells (D) using different electrodes at a current density of 1 mA cm⁻² and a deposition capacity of 1 mAh cm⁻² and (E) under different current densities and a deposition capacity of 1 mAh cm⁻². (F) Variation of the polarization voltage with the number of cycles during the deposition of the symmetric cells. Voltage profiles of Na||Na symmetric cells at (G) 1 mA cm⁻² and a capacity of 1 mAh cm⁻², (H) 5 mA cm⁻² and a capacity of 5 mAh cm⁻², and (I) 10 mA cm⁻² and a capacity of 10 mAh cm⁻². h, hours. (J) Cycle time comparison between the recently reported works and this work. (K) Rate performance of the symmetric cells.

After analyzing the structural composition, deposition morphology, and kinetic processes of the SEI layers formed by the different electrodes, the electrochemical reversibility and stability of anodes were evaluated using various electrochemical performance tests to verify the superiority of RSMA. The Na||Cu cells were first assembled for analysis (Fig. 4D), and the reversibility of coulombic efficiency (CE) on the Cu foil at a current density of 1 mA cm⁻² and a capacity of 1 mAh cm⁻² was about 87% in bare Na. However, the SEI produced by bare Na in the presence of a conventional carbonate electrolyte was fragile and inhomogeneous and poorly tolerated high currents (47). Impressively, the RSMA achieved stable Na⁺ plating/stripping over 200 cycles with an average CE of 97.2%. In addition, the RSMA had minimal voltage hysteresis (160 mV) (fig. S25, A and B). In the CE measurements at different current densities, the RSMA showed little variation in CE, excellent ion diffusion, and good current suitability (Fig. 4E). The corresponding charge/discharge curves (fig. S25, C and D) show that the overpotential of the RSMA system is low at high current densities. It is worth noting that because of the unique properties of carbonate electrolytes, there is a voltage drop at the end of the discharge phase of the Na||Cu cells, which is due to the polarization of the Na plating process at a later stage (note 12) (48–52). Nucleation and growth overpotentials for Na deposition at different current densities as seen from the first cycle discharge curves (fig. S26) reveal the remarkable effect of RSMA in conferring highly reversible plating/stripping of Na at different current densities (47). After that, we performed postanalyses such as CV, EIS, XPS, and SEM to validate the role of RSMA in Cu electrodes and to explain the improved performance of Na||Cu cells (figs. S27 to S30 and notes 13 to 15). Compared with previous work on Na||Cu cells (fig. S31 and table S5), our work demonstrates advantages in both the cycle life and CE (31, 46, 53–62). The electrochemical performance of RSMA over a long period of time was evaluated using symmetric cells (Fig. 4, F and G, and fig. S32). At a current density of 1 mA cm⁻² and a capacity of 1 mAh cm⁻², the RSMA showed improved performance compared to bare Na, displaying smaller plating/stripping overpotential, stable cycling profiles, and an exceptionally long cycling life (2700 hours), whereas bare Na shorted out after 230 hours of cycling. The average polarization voltage within the cycling process of the RSMA symmetric cells is about 150 to 180 mV, which ensures a much smaller overpotential with a long cycling time. The low R_int of RSMA in the electrochemical impedance curve (fig. S33) indicates a beneficial SEI layer to facilitate the kinetics of Na⁺ migration. In addition, the RSMA can be maintained stably by cycling for more than 1000 hours when the electrolyte without FEC is used, which also proves the role of RSMA laterally (fig. S34). In addition, by changing the plating/stripping capacity to 3 mA cm⁻² with 3 mAh cm⁻², 3 mA cm⁻² with 1 mAh cm⁻², 1 mA cm⁻² with 5 mAh cm⁻², 5 mA cm⁻² with 5 mAh cm⁻², and 10 mA cm⁻² with 10 mAh cm⁻² (Fig. 4, H and I, figs. S35 to S39), the cells with the RSMA still exhibited more than 350-, 450-, 1100-, 250-, and 140-hour cycling capability, respectively. Figure 4J and table S6 summarize the cycling performance of Na||Na cells, which outperformed most recent carbonate-based electrolyte modification strategies in the literature (5, 41, 54, 60, 63–72). The sodium plating/stripping was fixed at 1 mAh cm⁻², cycled at different current densities, and then returned to a long cycle of 1 mA cm⁻². The overpotential of the plating/stripping of RSMA is smaller than that of bare Na, and the stability of plating/stripping remains excellent after going back to low current cycling, which shows excellent multiplicity performance (Fig. 4K, fig. S40, and note 16). In addition, the RSMA also showed advantages over the NaPF₆ liquid-phase high-concentration system (fig. S41 and note 17). To extend the concept and strategy of reconstituted metal anodes, symmetric cells were assembled with different sodium salts (NaTFSI and NaFSI), all of which exhibited activatable RSMA with excellent cell performance. In addition, the same was true for the Li metal, where reconstructed Li metal anodes with activatable lithium hexafluorophosphate (LiPF₆) showed even more enhancement in both symmetric and full cells (figs. S42 and S43 and note 18). Thus, the strategy has good prospects for application in other metal anodes and salts as well.

Plating/stripping reversibility of RSMA under extreme conditions

Deep discharge can further assess the ability to use the active metal of the electrode, demonstrating the level of ion/electron transport and electrode stability. Figure S44 illustrates a schematic of RSMA cells under deep cycling conditions in a 1 M NaPF₆-diglyme electrolyte. Deep charge/discharge cycling is defined as the condition of 100% DOD, i.e., complete stripping of the electrodes followed by complete plating of the electrodes (19). Very little research has been conducted in the literature regarding the cycling of Na metal anodes under the harsh conditions of 100% DOD. Because thick Na foil is usually used for such tasks, only 5 to 10% or less of the metal is plating/stripping per cycle. Therefore, 100% DOD is the most drastic test protocol for anode volume change and, of course, a direct way to test electrode excellence (e.g., electrode structural stability, electrode ion/electron transport capacity, etc.); in other words, such an RSMA can achieve the so-called anode-free properties (30, 73). Figure 5A shows voltage-time plots of Cu-RSMA||Cu-RSMA cells cycled under different DODs at 1 mA cm⁻². Na is first deposited from one side of the Cu-RSMA, and then the plating/stripping of Na is repeated. Cycling at 15 mAh cm⁻² for 100% DOD starts at 20%, then 40%, 60%, and 80%, and lastly, 100%. For the RSMA, DOD experiments were also performed in a carbonate-based electrolyte (fig. S45A). A DOD of 90% was achieved in 1 M NaPF₆-EMC/FEC. In contrast, bare Na only achieved 85% DOD in a 1 M NaPF₆-diglyme electrolyte and short circuited during cycling in ester electrolytes (fig. S45, B and C). Figure 5 (B and C) shows the CE, voltage profile, and average plating overpotential of the RSMA cycled at 100% DOD at 1 mA cm⁻² and 10 mAh cm⁻² capacity. It can be observed that the RSMA cells remain stable during cycling at 100% DOD, showing an average CE of 99.2% at 1 mA cm⁻². Such voltage stability and relatively high CE values have rarely been reported for Na metal anodes that are subjected to complete plating/stripping with each cycle. In addition, we completely stripped the Na metal from the Cu-RSMA and performed SEM to analyze the structure and composition of the remaining electrode (fig. S46). The remaining electrode structure is NaPF₆ particles uniformly dotted in the network-distributed PPy skeleton, which can also be seen from the corresponding mapping images. We further examined the elemental composition and distribution of Cu-RSMA on the surface and at depth using XPS under different discharge depth conditions to verify the 100% DOD of the electrode (Fig. 5F). As can be seen from the Na 1s spectrum, the signal peak of the sodium metal (~1070.1 eV) can still be detected after 150 s of sputtering at 50% DOD (74). It is noteworthy that at 100% DOD, the signal peak of Na⁰ disappears, leaving signal peaks of NaPF₆ particles and a small amount of NaF. This not only indicates the formation of a thin inorganic SEI but also achieves the complete stripping of the sodium metal. XRD also exhibits the same results as SEM, with no diffraction peaks associated with the Na metal (Fig. 5G). These results indicate that the stability of the self-supported RSMA structure enhanced the ionic/electronic conductivity, the robust SEI formed by the confined-region high-concentration NaPF₆ dissociation state allows it to still sustain cycling under such harsh conditions, and its unique electrode reconstruction leads to excellent electrochemical performance.

Fig. 5. Deep discharge testing of RSMA anodes.

Electrochemical performance and electrode characterization of RSMA cells under deep charge/discharge cycling conditions. (A) RSMA cells cycled under different DODs at 1 mA cm⁻². (B) CE, (C and D) voltage profiles, and (E) average plating overpotentials for RSMA cycled under 100% DOD at 1 mA cm⁻² and 10 mAh cm⁻². The RSMA was rolling to the Cu foil, and then Na was repeatedly plated/stripped. (F) Depth XPS spectra of Na 1s under different DOD conditions. (G) XRD of the electrode after complete stripping of Na. h, hours.

SMBs under various conditions

On the basis of the excellent performance of RSMA, we combined it with a Prussian blue (PB) cathode to assemble a complete cell for various performance evaluations at 4.2 V. The rate performance of the Na||PB cell demonstrated the high-rate reaction kinetics of RSMA, which still had a capacity of 68.4 mAh g⁻¹ at 100 C, exhibiting much lower charge/discharge overpotential (Fig. 6A and fig. S47). Notably, the RSMA also ensures more robust battery cycling, exhibiting an initial capacity of 131.4 mAh g⁻¹ with high CE and an 82.1% capacity retention at 5600 cycles for more than 9 months (Fig. 6B and fig. S48). Cycling for such a long period of time maximally validated the toughness of the electrode. Considering the wide temperature application of this carbonate electrolyte, the stability of RSMA was tested at changing temperatures from −40° to 60°C. At all temperatures, the RMSA exhibited higher capacity and lower charge/discharge overpotential with excellent temperature adaptation (Fig. 6C and fig. S49). The full cells after 200 cycles were characterized using SEM, TEM, XPS, and EIS to help explain the excellent electrochemical performance of RSMA (figs. S50 to S54 and notes 19 and 20) (75–79). Taking advantage of the high-voltage tolerance of the carbonate electrolyte, the Na||PB cells were cycled at 1.5 C, 4.5 V, and 2 mAh cm⁻² under PB loading conditions (Fig. 6D and fig. S55). The initial discharge capacities of the full cells using bare Na and the RSMA were similar at 136.2 and 138.3 mAh g⁻¹, respectively. However, the discharge capacity of bare Na was only 86.4 mAh g⁻¹ after 300 cycles, which was attributed to the formation of brittle SEI in bare Na and the marked change of the electrode structure. In addition, the RSMA can also be discharged at a higher rate at 4.5 V (fig. S56). The full cells of RSMA still showed a capacity of 106.6 mAh g⁻¹ at 20 C and exhibited narrower voltage gaps at different rates, confirming the high-rate reaction kinetics of RSMA.

Fig. 6. RSMA on full cells.

(A) Rate performance and (B) cycle performance of Na||PB cells using different electrodes with a cutoff voltage of 4.2 V. (C) Electrochemical performance of Na||PB cells using different electrodes with temperatures from −40° to 60°C at 0.5 C. Cycle performance of Na||PB cells using different electrodes with a cutoff voltage of 4.5 V at (D) 1.5 C and (E) 10 C. (F) Cycle performance comparison between the recently reported works and this work. (G and H) Electrochemical performance of Na||PB full cells with an N/P ratio of 4:1. (I) Cycling performance of Na metal pouch cells at 4.2 V and 1 C. (J) Energy density comparison between the recently reported works and this work.

Because the beneficial performance of RSMA shows good long-term cycling stability, its cycling stability at high current densities was further investigated. The rapid charge/discharge process of the full cells at high current densities can also assess the superiority of electrodes to a large extent. Under a typical PB mass loading of about 5 mg cm⁻², the RSMA||PB cells exhibited high CE with capacities of 104.2 mAh g⁻¹ at 10 C (7.5 mA cm⁻²) for the first cycle and 85.3 mAh g⁻¹ at 10,000 cycles, with a capacity retention rate of 81.86% (Fig. 6E). In contrast, bare Na has a much lower capacity and shows obvious capacity degradation during cycling. At the 4000th cycle, the capacity was only 44.2 mAh g⁻¹ with a capacity retention rate of 49.3%. Corresponding capacity-voltage curves also show that the voltage profile of RSMA||PB cells remains stable (fig. S57). Thus, the RSMA can withstand stable and extensive cycling at such high current densities (Fig. 6F and table S7), far exceeding the recent literature in terms of cycle life, capacity retention, and charge/discharge cutoff voltage (32, 53, 80–91). Afterward, stable cycling of RSMA was similarly achieved at 150 cycles at a PB loading of 25 mg cm⁻² (fig. S58). On the basis of the beneficial electrochemical performance exhibited by the RSMA, the RSMA was also loaded on a copper foil by rolling as a Na-limited Na||PB full cell anode with a 4:1 N/P ratio to PB and cycled at a cutoff voltage of 4.5 V (Fig. 6G and fig. S59). Over 200 cycles, the RSMA achieved an average CE of 99.3%, along with a capacity retention of 81% and a narrow charge/discharge gap. This is attributed to the high dimensional response and improved interfacial properties. The rate performance of the full cells under this condition was further evaluated on the basis of the fast kinetics of RSMA (Fig. 6H and fig. S60), with capacities of 109.2 and 91.3 mAh g⁻¹ at 10 and 20 C, respectively, demonstrating excellent performance under limited RSMA. We further reduced the thickness of RSMA and lowered the N/P ratio for testing, and we can see that the RSMA can still show good electrochemical performance within a reasonable thickness range (fig. S61 and note 21). In addition, single-layer pouch cells were assembled using the RSMA to investigate the prospects for commercial application under realistic conditions such as low N/P ratio and high cathode loading. As a proof of concept, 13-cm by 13-cm RSMA||PB pouch cells were used to investigate the cycling stability of high-energy-density SMBs (N/P = 2.5). A high initial capacity of 448.2 mAh was shown at 0.1 C, while it still had a capacity of 408.1 mAh at 1 C. The capacity retention after 200 cycles was 85.6%, which is one of the highest capacity retentions of large-area pouch cells under limited Na conditions (Fig. 6I and figs. S62 and S63) (92–94). The energy density of the pouch cell was calculated from the total mass of RSMA (i.e., active Na, PPy, and NaPF₆) and PB. Encouragingly, the RSMA pouch cell achieved a weight energy density of 367.3 Wh kg⁻¹ (table S8), which is at a high level among the SMBs reported in the literature so far (Fig. 6J and table S9) (32, 91, 94–100). The RSMA pouch batteries also light up the light-emitting diode light, which stays lit for up to 5 min (fig. S64 and movie S3). In conclusion, these results demonstrate the promising application of the designed RSMA in high-energy-density full cells.

DISCUSSION

The biggest challenge of metal anodes is their continuous “disappearance” and “appearance” without any relying skeleton, and their reaction interface is limited to two-dimensional situations. It is necessary to fundamentally reinvent metal anodes rather than just focusing on surface protection. We here propose the concepts and strategies for RMAs and prepare such a sodium metal anode as an example of SMBs, which is an activatable self-supporting 3D ionic/electronic conducting “metal-polymer-salt” skeleton. The reaction zone was thus expanded to 3D. Through a series of cell configurations (Na||Cu cells, Na||Na cells, deep DOD cells, and pouch cells with a specific N/P ratio), we confirmed the feasibility of this metal reconstruction, with cell performance superior to most literature reports. Our design differs from previous studies using artificial interfaces or 3D inert primary support engineering on pristine sodium foils. Conceptually, we have reconstructed the metal anode with an expanded 3D plating/stripping mechanism and a metal-bulk storage skeleton, which not only provides other insights to overcome the challenge of poor stability of SMBs but also fundamentally overcomes the inherent structural instability of metal anodes during plating/stripping. In terms of practicality, the rolling method is technically convenient, safe, and economical. The design and beneficial performance of metal anodes, as exemplified by the RSMA, provide a beneficial concept and strategy for investigating the reversibility of plating/stripping of existing anodes, deep charging/discharging, and the long-term operation of high-voltage metal full cells at high current densities. In addition, this concept and strategy can be extended to other salt and alkali metal electrode systems, providing a beneficial theoretical basis and realization path for the design of high-efficiency, long-life energy storage systems.

MATERIALS AND METHODS

Chemicals and materials

The sodium metal was obtained from Sigma-Aldrich Corporation. The lithium metal foils (≥99.9%, Ф15.6 × 1.0 mm) were purchased from China Energy Lithium Co., Ltd. (Tianjin, China). The salts and solvents, including NaPF₆, NaTFSI, NaFSI, FEC, EMC, and LiPF₆, were purchased from Duoduo Chemical Reagent Co., Ltd., with battery-grade purity. PPy and polytetrafluoroethylene (PTFE) powder were purchased from Aladdin Co., Ltd., China, with a purity of >99%. Before preparing the electrolyte, the solvent was dried by 4-Å zeolite to ensure that the water content was below 10 parts per million (ppm). PB and LiNi_0.8Co_0.1Mn_0.1O₂ (NCM 811) cathodes were purchased from Naba Co., Ltd. Electrolytes were prepared by mixing the required amount of salt, solvent, and additive. The usage of electrolyte in a button cell was 100 μl. The preparation of electrolytes was all carried out in a glove box with O₂ and H₂O content <1 ppm.

Preparation of RMA

Before the preparation of RSMA, premixed PPy and PTFE powders were transferred to vials and mixed using a vortex stirrer for 1 min, after which they were transferred to a ball mill for dry ball milling and stirring for 6 hours to break them into small pieces. Subsequently, the mixture was spread uniformly on a substrate plate and transferred to a hot press (which had been preheated to 150°C), applying a loaded hot press of 20 MPa for a period of time. PPy/PTFE was cooled down and then rolled repeatedly to achieve an interwoven network structure of PTFE. For the RSMA, the polished sodium block was rolled into thin sheets, an amount of NaPF₆ powder was placed between the two sodium sheets, and the sandwiched sodium sheets were rolled into thin sheets. The sodium sheets were then folded and rerolled into thin sheets. The Na/NaPF₆ complex is formed after repeating the above folding/rolling steps several times. Na/NaPF₆ and PPy/PTFE were mixed by mechanical rolling, and then the complex was cut into round sodium sheets with a 14-mm diameter using a stamping die; the electrode was named the RSMA. The mass ratio of sodium metal to NaPF₆ powder was 7:3. The PPy powder accounted for 1% of the total mass, and PTFE accounted for 0.5% of the total mass. For the preparation of RLMA, the lithium flakes were first stacked and then extruded into large discs by a press, then mixed with LiPF₆ and PPy/PTFE (same ratio as the RSMA), and tumbled in the same way as the RSMA. The entire preparation process was carried out in an argon-filled glove box (less than 0.1 ppm of water and oxygen).

Characterization

EIS (IM6, Zahner Elektrik, Germany) tests of button cells were performed in a frequency range from 0.1 to 10⁵ Hz with an amplitude voltage of 5 mV. XRD experiments were performed on a Rigaku Ultima IV x-ray diffractometer with Cu Kα radiation (λ = 1.5418 Å) at a scan angle window between 10° and 90°. The morphology of bare Na and the RSMA was obtained by field emission SEM (HITACHI S-4800) and AFM (SPM 5500; Keysight Technologies, Santa Rosa, CA). The contact angle tests were performed on Na and the RSMA in a DSA-30 Droplet Shape Analyzer device, under vacuum, with the electrolyte being the carbonate system used in this paper. High-resolution TEM and mapping images of the cathode electrolyte interface were obtained by a Talos F200 type electron microscope. All characterizations of the Na metal and SEI were carried out at −178°C and obtained by a Talos F200X S type electron microscope. To analyze the chemical composition of the SEI, characterization of electrodes after cycling was performed using XPS measurements (Thermo Fisher Scientific ESCALAB Xi⁺, US). It is accompanied by depth profiling, which is performed by Ar⁺ sputtering. The cycled cells were disassembled in the glove box filled with argon (the contents of O₂ and H₂O were kept below 0.1 ppm), and the electrodes were washed with dimethyl ether before morphology and composition characterizations. TOF-SIMS was conducted using a PHI nanoTOF II TOF-SIMS (30 keV; 2 nA; ion species: Bi³⁺⁺). The in situ dynamic Na deposition behaviors were observed at 0.5 mA cm⁻² for 100 min by a ZOOM-0850C (Shanghai Puqian Optical Instrument Co., Ltd.) using Na as the working electrode. Raman results were captured on XploRA with an excitation laser wavelength of 532 nm. ¹⁹F nuclear magnetic resonance spectra were recorded at room temperature (295 K) with a Bruker Avance III HD 850 MHz spectrometer.

Electrochemical measurements

The Na metal anodes were obtained by squeezing the Na metal foil (~1 mm) into many disks (Ф 14.0, ≥99.9%). Electrochemical cycling and rate tests were assembled by 2032-type coin batteries with glass fiber (Whatman GF-C) and tested by using a battery test system (Shenzhen NEWARE Co. Ltd., China). Na||Na symmetric cells were assembled by using two pieces of Na disks as working and counter electrodes. To perform the impedance test at different temperatures, the cells were first set aside for 2 hours at different temperatures. EIS at different temperatures was carried out within an RTP-80CT high and low temperature test chamber (Guangdong Huan Rui Test Equipment Co., Ltd., China). The electronic conductivity of RSMA is tested using the RTS-9 four-probe test system. The test method of this instrument is to test by fixing the test length, so after fixing the length, only the thickness of the sample needs to be tested. CV (Squidstat prime) tests were performed within a potential range of −0.1 to 0.1 V versus Na/Na⁺ at a scan rate of 0.1 mV s⁻¹ for Na||Na cells and −0.2 to 3.0 V versus Na/Na⁺ at a scan rate of 0.5 mV s⁻¹ for Na||Cu cells. The exchange current density of electrolytes was tested by a Tafel plot (CHI660E, China) of Na||Na symmetric cells. The charge/discharge performance of electrolytes was determined by assembled full cells using PB as the cathode, the Na metal as the anode, and a piece of GF-C membrane as the separator in an Ar-filled glove box (glove box with O₂ and H₂O content <1 ppm). The cathode was prepared by coating a slurry composed of 80 wt % PB composite, 10 wt % Super P, and 10 wt % polyvinylidene difluoride onto an Al foil substrate and drying under vacuum at 120°C for 12 hours. Cycle and rate tests were performed on the NEWARE battery test system at 25°C. All Na||PB cells were tested in a voltage range between 2.4 and 4.2 V and between 2.4 and 4.5 V versus Na/Na⁺. All full cells were activated for two cycles at 0.1 C. The Na||PB full cells were fabricated with N/P ratios of 4 (the areal capacity of the anode is 12 mAh cm⁻² by RSMA rolling in Cu foil, and the areal capacity of the cathode is 3 mAh cm⁻²), 2.5 (the areal capacity of the anode is 7.5 mAh cm⁻² by RSMA rolling in Cu foil, and the areal capacity of the cathode is 3 mAh cm⁻²), 1 (the areal capacity of the anode is 4 mAh cm⁻² by RSMA rolling in Cu foil, and the areal capacity of the cathode is 4 mAh cm⁻²), and 0.85 (the areal capacity of the anode is 3.4 mAh cm⁻² by RSMA rolling in Cu foil, and the areal capacity of the cathode is 4 mAh cm⁻²). For single pouch cells, the PB cathode is cut into 11-cm by 11-cm squares and the RSMA anode is cut into 10-cm by 10-cm squares. The pouch batteries are assembled and sealed in Al plastic film packages and are subjected to charge/discharge cycling at voltages ranging from 2.4 to 4.2 V. The pouch batteries are then sealed in Al plastic film packages.

Supplementary Materials

The PDF file includes:

Figs. S1 to S64

Notes 1 to 21

Tables S1 to S9

Legends for movies S1 to S3

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S3