The chemical evolution of solid electrolyte interface in sodium metal batteries

Abstract

The solid electrolyte interface (SEI) formed on the anode is one of the key factors that determine the life span of sodium metal batteries (SMBs). However, the continuous evolution of SEI during charging/discharging processes complicates the fundamental understanding of its chemistry and structure. In this work, we studied the underlying mechanisms of the protection effect offered by the SEI derived from sodium difluoro(oxalato)borate (NaDFOB). In situ nuclear magnetic resonance (NMR) shows that the prior reduction of DFOB anion contributes to the SEI formation, and it suppresses the decomposition of carbonate solvents. Depth-profiling x-ray photoelectron spectroscopy and high-resolution solid-state NMR reveal that the DFOB anion is gradually turned into borate and fluoride-rich SEI with cycling. The protection effect of SEI reaches the optimum at 50 cycles, which triples the life span of SMB. The detailed investigations provide valuable guidelines for the SEI engineering.

The chemistry and protection effect of solid electrolyte interfaces evolve remarkably as battery cycling proceeds.

INTRODUCTION

Low-cost and high–energy density rechargeable stationary batteries are essential components of electric vehicles and the integrated electrical power grid based on renewable energies (1–3). Sodium metal batteries (SMBs) are attractive candidates because of the abundance of sodium resources along with the high theoretical specific capacity (1165 mAh g⁻¹) and the low voltage (−2.71 V versus standard hydrogen electrode) of sodium metal (4–7). Unfortunately, the utilization of sodium metal anode encounters several challenges, including dendrite growth, parasitic reactions at the anode/electrolyte interface, and large volume change, which result in battery failure and safety issues (8–11).

Forming a desirable solid electrolyte interface (SEI) protective layer is an efficient way to stabilize Na metal and to improve the battery performance and cycle life (11–14). SEI arises from the chemical and electrochemical reactions between the electrolyte and the highly reactive sodium anode (15, 16). A favorable SEI can prevent the excessive degradation of electrolyte, limit the dendrite growth, and minimize the volume change but still allow the fast transport of sodium ions (17). Various strategies can be applied to tune the chemistry, morphology, and thickness of SEI, such as engineering solvation structure (18, 19), adopting superconcentrated electrolytes (20–22), using additives (23–26), and constructing an artificial SEI (27–31). However, the strategies for controlling SEI are generally empirical as the chemical compositions of SEI are highly complex and difficult to characterize (32). On the other hand, the structure and composition of SEI continuously evolve with cycles of battery charging/discharging. The chemistry of SEI cannot be understood as a simple and stationary picture. The formation processes of SEI and its intrinsic correlation to battery performance are vital knowledge not only for SMBs but also for other battery systems (33). Routine characterization techniques for studying SEI are mostly limited to scanning electron microscopy (SEM) for probing morphology and x-ray photoelectron spectroscopy (XPS) for examining surface chemical composition (34–36). Recently, solid-state and in situ nuclear magnetic resonance (NMR) have been proven to be powerful techniques that can extract holistic chemical information of SEI and unveil the electrochemical processes in operating batteries (37–43).

In this study, we try to monitor the formation process of SEI and to understand its protection effect in SMBs using electrochemical measurements, SEM, XPS, and advanced NMR techniques. Our study is focused on the electrolyte composed of sodium difluoro(oxalato)borate (NaDFOB) and carbonate-ester solvents, which exhibits excellent electrochemical performance in SMBs and has been shown to form stable SEI on sodium anode surface (44, 45). It is demonstrated that DFOB anion has a low lowest unoccupied molecular orbital and reduces before common organic solvents (46). The SEI formation mechanism of NaDFOB is fundamentally different from those of conventional electrolytes or organic additives. Here, we investigate how the protective SEI extends the battery life, how the chemical structures evolve under different cycling stages, and why certain SEI formation is more favorable for battery performance. We apply in situ ¹H, ¹⁹F, and ¹¹B NMR to track the degradation of electrolyte components in operating cells. In addition, we reveal the chemical changes in SEI composition under different cycling stages with solid-state NMR and XPS depth profiling. The characterizations demonstrate that the DFOB anion is gradually reduced into borate- and fluoride-rich SEI, which prevents the decomposition of carbonate solvents. The SEI formed with 50 cycles offers the best protection as its chemical composition reaches the optimum. Our study offers an in-depth understanding of the SEI growing processes in SMBs as well as the intrinsic correlations with electrochemical performance.

RESULTS

Study the protection effectiveness of SEI

Here, we adopt a disassemble-reassemble strategy to demonstrate the protection effectiveness of NaDFOB-derived SEI (Fig. 1A). First, symmetric Na/Na cells were fabricated with NaDFOB-based electrolyte, i.e., 1.0 M NaDFOB in ethylene carbonate (EC):dimethyl carbonate (DMC) (volume ratio = 1:1). The Na metal plates were passivated by galvanostatic cycling at a current density of 1.5 mA cm⁻² and a capacity of 0.1 mAh cm⁻². After certain cycles, we disassembled the symmetric cell and extracted the SEI-coated sodium plates. Last, the SEI-coated sodium plates were reassembled as electrodes in new cells, either Na/Na symmetric cells or Na₃V₂O₂(PO₄)₂F (NVOPF)/Na cells, using commercial 1.0 M NaPF₆ or 1.0 M NaClO₄ electrolytes. The reassembled Na/Na symmetric cells were cycled at a current density of 1.5 mA cm⁻², while reassembled NVOPF/Na cells were cycled at 10 C (1.3 A g⁻¹). The long-term cycling stability in the reassembled cells is compared.

Fig. 1. Protection effectiveness of SEI.

(A) Schematic diagram of the disassemble-reassemble strategy. (B) Time to failure of reassembled Na/Na symmetric cells with protected Na electrodes. NaPF₆ (1.0 M) in EC:DMC was used as the electrolyte, and the cells were cycled at a current density of 1.5 mA cm⁻². (C) Long-term cycling performance of reassembled NVOPF/Na cells with protected Na anodes. NaClO₄ (1.0 M) in EC:PC was used as the electrolyte, and cells were cycled at 10 C (1.3 A g⁻¹).

In the reassembled Na/Na symmetric cells, time to failure is used as a metric to compare the effectiveness of SEI protection. The time to failure is determined at the time when the fluctuation of peak-to-peak voltage exceeds ±10% relative to the steady-state value. The voltage-time profiles of Na/Na cells are provided in fig. S1. In these battery tests, most SEI-coated Na anodes demonstrate an increased time to failure than that of the bare Na electrode, confirming the protection effect of NaDFOB-derived SEI (Fig. 1B). Even after a single pre-cycle of SEI coating, the time to failure increases notably. No sudden voltage drop due to short circuit was observed in the cells with SEI-coated electrodes, suggesting that NaDFOB-derived SEI effectively suppresses the dendrite growth. The time to failure maximizes at about 80 hours (triple the life span than that of the bare electrode) for the electrode with 50 pre-cycles of SEI coating, and it decreases for the electrodes with longer pre-cycles of SEI coating. For the electrode with 200 pre-cycles of SEI coating, the time to failure decreases to about the same as that of bare Na.

In the reassembled NVOPF/Na cells, the battery life span is determined when the capacity degrades abruptly while they are cycled at a rate of 10 C. The capacity degradation of NVOPF/Na cells mainly comes from the sodium metal anode as NVOPF is a stable high-voltage cathode material with an average working potential of 3.8 V (figs. S2 and S3) (47). The SEI-coated Na anodes demonstrate an improved life span than that of the bare Na anode (Fig. 1C and fig. S4). In general, the effectiveness of SEI protection in NVOPF/Na cells follows the similar trend as that in Na/Na symmetric cells. The anode with 50 pre-cycles of SEI-coating achieves the longest life span, retaining 94% of the initial capacity after 2500 cycles (90 mAh g⁻¹) without notable degradation, while the life span of the anode with 200 pre-cycles of SEI-coating only improves by about 15% compared to that of the bare Na anode.

We further performed the electrochemical impedance spectra (EIS) of Na/Na symmetric cells with 1.0 M NaDFOB/EC:DMC electrolyte under various cycling stages (fig. S5). The interfacial resistance can be obtained from the semicircle at higher frequencies. The interfacial resistance decreases with cycling proceeds until the 50th cycle and then reaches a steady value of ~500 ohms. It implies the continuous growth of SEI when the cycling number is lower than 50 and that a relatively stable interface structure is established when the cycling number is over 50. These findings raise important questions about the fundamental chemistry of the SEI, i.e., how it is formed and protects the sodium electrode in SMBs and what the critical chemical components that contribute to its protection effect are.

In situ NMR of electrolytes in cycling cells

SEI in SMBs mainly derives from the chemical and electrochemical reactions between the electrolyte components and the sodium metal. Therefore, it is informative to track the chemical changes in the electrolyte solution during battery cycling. In situ NMR offers a great advantage to monitor chemical processes in running batteries via a nondestructive manner. Here, we fabricated a cylindrical Swagelok-type cell that is placed in a solenoidal NMR coil and connected to an external potentiostatic tester. Figure 2A schematically illustrates the design of our in situ NMR setup. This design enables us to detect the signals of electrolyte solution without disassembling the cell, thereby individually quantifying the consumption of the salt and solvents during cycling. Na/Na symmetric cells with 1.0 M NaDFOB or 1.0 M NaPF₆ electrolytes (in EC:DMC = 1:1 solvent) are investigated in our study. The cells are cycled at a current density of 1.5 mA cm⁻² and a capacity of 0.1 mAh cm⁻². ¹H, ¹¹B, and ¹⁹F signals are collected when the cells are cycled to various stages (figs. S6 and S7). Here, we focus on the sharp peaks that arise from the liquid-like electrolyte solution. Those broad signals of solid SEI or background are not examined. Clearly, the ¹¹B and ¹⁹F peaks come from the DFOB or PF₆ anion, while the ¹H peaks come from the organic solvents. The intensities of ¹⁹F, ¹¹B, and ¹H signals offer a quantitative determination of electrolyte salt and solvent consumption in operating cells (Fig. 2, B and C, and fig. S8).

Fig. 2. In situ NMR.

(A) Schematic of the in situ NMR setup. In situ NMR measurements on a Na/Na cylindrical cell using (B) 1.0 M NaDFOB/EC:DMC and (C) 1.0 NaPF₆/EC:DMC electrolytes.

For the reference 1.0 M NaPF₆ electrolyte, both ¹H and ¹⁹F signal intensities continue to decline as cycling number increases, indicating the continuous decomposition of both PF₆ anion and EC:DMC solvent. The amount of PF₆⁻ and EC:DMC decreases by about 23% when the NaPF₆-based cell is cycled to the 100th cycle and short-circuited. In contrast, for the 1.0 M NaDFOB electrolyte, only the ¹⁹F and ¹¹B signal intensities decline, indicating the continuous consumption of DFOB anion, while the ¹H signal intensity remains almost unchanged over the course of cycling, suggesting that the EC:DMC solvent is mostly intact in the NaDFOB-based cell. The NaDFOB-based cell also endures a longer cycling life in our in situ setup, which remains functional even after 150 cycles. This in situ NMR study demonstrates that the SEI formed by the NaDFOB electrolyte mainly consists of the chemical derivatives of the DFOB anion, and the SEI can effectively prevent the decomposition of organic solvents. In addition, the continuous consumption of DFOB anion also suggests that the by-products will accumulate continuously during cycling, which may result in less effective protection under longer cycling stages.

Chemical evolution of SEI on Na electrode

First, we used SEM to examine the interfacial morphology of cycled Na metal anodes that were extracted from symmetric cells under different cycling stages. The sample preparation and transfer were carried out under the protection of Ar atmosphere to avoid air contact. As shown in the SEM images (figs. S9 and S10), the NaDFOB electrolyte leads to flat and uniform deposition of Na. No obvious dendrites or mossy structures are observed. The SEI with 50 cycles appears to be smoother than the rest. Because of the low contrast for the electron density of SEI species, the thickness of SEI is difficult to identify in the cross-sectional images of SEM.

The chemical composition is also critical for the properties and effectiveness of SEI. We characterize the chemical compositions formed at different cycling stages using XPS (Fig. 3, A to D, and fig. S11). The figures show the surface XPS spectra corresponding to binding energies of C 1s, O 1s, B 1s, and F 1s for the SEIs with various cycles. In C 1s spectra, the signals of C─C (284.5 eV), C─OR (286.5 eV), C═O (288.5 eV), and Na₂CO₃ (289.6 eV) can be clearly detected. The organic species of C─C and C─OR could derive from EC:DMC solvent reduction, while Na₂CO₃ should come from the decomposition of DFOB anion. In the O 1s spectra, the broad signal can be deconvoluted into B─O (532.6 eV) and C─O (532.1 eV). In B 1s spectra, three signals corresponding to B─F, B₂O₃, and Na─B─O can be observed. F 1s spectra show the presence of NaF. These B- and F-containing inorganic compounds mainly arise from the decomposition of DFOB anion. On the basis of the XPS results, chemical compositions of DFOB-derived SEI with different pre-cycles are largely the same.

Fig. 3. XPS analysis of the cycled Na electrodes.

(A to D) C 1s, O 1s, B 1s, and F 1s surface XPS spectra of Na electrode cycled with NaDFOB/EC:DMC electrolyte for 50 cycles. (E) Atomic percentage of carbon (C), boron (B), fluorine (F), and sodium (Na) in electrodes of various deposition cycles (5th, 50th, and 200th). The results were obtained from XPS depth profiling with different Ar ion etching times. B.E. is binding energy.

The above XPS investigations were performed on the surface layer (< 5 nm thick) of the SEI. It is of great interest to know the spatial distribution of the chemical compositions. We further conducted XPS depth profiling on the SEI using Ar ion etching. Figure 3E shows the elemental abundance of carbon, boron, fluorine, and sodium changes with etching time. The inorganic components of SEI consist of boron, fluorine, and some of sodium carbonate, while organic components consist of electrochemically degraded solvents products. Note that because the sputter yield is difficult to determine for this SEI material, the exact depth is not indicated.

Although the SEIs are formed with various pre-cycles, the carbon content decreases abruptly with etching time, suggesting that the organic components primarily exist at their surface layer, and they are almost absent in the inner layers. This again confirms that the DFOB-derived SEI can effectively suppress the permeation of electrolyte. The depth profiling also indicates that the SEIs in these symmetric cells are relatively thick as it has not reached the Na metal with an etching time of 5000 s. Such inorganic layer can provide mechanical robustness and suppress the dendrite growth and the volume expansion of sodium metal. The abundance of Na is the highest for the SEI formed with 50 pre-cycles, especially in the etched layer, implying the dense structure under this cycling stage. The high abundance of Na should be beneficial for its ionic conductivity.

Because XPS cannot access the whole body of SEI, we further performed ²³Na and ¹¹B magic angle spinning (MAS) NMR to investigate the chemical compositions of SEI as a bulk (Fig. 4). These SEI materials were collected by scratching off the surface layers of cycled Na electrodes in Na/Na symmetric cells. Because of the second-order quadrupolar coupling, the ²³Na and ¹¹B signals are broadened by anisotropic effects (48). In the previous study, we used multidimensional NMR spectroscopy that identified a variety of chemical species, e.g., Na₄B₂O₅, NaF, NaBF₄, Na₂CO₃, and NaOH in NaDFOB-derived SEI (45). On the basis of these known species, we deconvolute the ²³Na and ¹¹B NMR spectra to analyze their relative populations under electrochemical evolution.

Fig. 4. Solid-state NMR analysis of SEI.

(A) ²³Na and (B) ¹¹B solid-state NMR spectra of SEI species harvested from the Na metal surface after the 15th, 50th, 100th, and 200th cycles.

The ²³Na spectrum of the SEI formed at the initial cycles (15th) shows a broad resonance near −20 to 0–parts per million (ppm), which is assigned to Na₂C₂O₄, organic oligomers, and partially Na₂CO₃. Two distinct sodium sites exist in Na₂CO₃, and the frequency distributions of these sites are broad because of strong quadrupolar interactions (49). The rightmost shoulder extending beyond −15 ppm is attributed to NaBF₄ and minor residual NaDFOB salt. The signal in the range of 0 to 20 ppm mainly contains Na₂CO₃ and sodium borate (Na_xB_yO_z) compounds. After 50 cycles, the peak belonging to Na₂C₂O₄ decreases, while the signal corresponding to sodium borate increases notably. According to ²³Na NMR, the compositions of SEI with 100 or 200 pre-cycles are slightly different from that with 50 pre-cycles. There is a minor presence of NaOH (the shoulder above 20 ppm) in SEI with 100 or 200 pre-cycles, which has been shown to be undesirable for the battery stability (45, 50). The growth of sodium borate is further confirmed by the ¹¹B signal in the range of 10 to 20 ppm, which matches with the trigonal [BO₃] species with large quadrupolar couplings (51). In ¹¹B spectra, the peak between −2 and 5 ppm corresponds to B₂O₃ from a surface-adhered glass fiber separator. The broad shoulder that extends to 40 ppm corresponds to oligomeric borates (48).

DISCUSSION

Combining spectroscopic evidence by NMR and XPS, we show that DFOB anion is preferentially reduced on Na metal anodes and forms a protective SEI layer that limits the side reactions of sodium metal. The protection is effective against both NaPF₆- and NaClO₄-based electrolytes. The chemical composition of NaDFOB-derived SEI varies across its thickness. Generally, it can be divided into a few-nanometer-thin outer layer of organic-rich species and a relative thick inner layer of stable inorganic species as shown in Fig. 5. This favorable feature enables the NaDFOB to be used as a standalone salt or as an electrolyte additive in SMBs. A 2 weight % NaDFOB additive in 1.0 M NaClO₄/EC:propylene carbonate (PC) electrolyte brings excellent protection to the Na anode (fig. S12). We show that the chemical composition of SEI evolves as deposition cycles proceed. Some part of NaDFOB is first turned into Na₂C₂O₄ and then becomes Na₂CO₃, some part is decomposed into sodium borates and then forms oligomeric borate compounds, and the rest with fluorine is turned into NaBF₄ and NaF (46, 52–54).

Fig. 5. Schematic illustrations of the NaDFOB-derived SEI structure at the initial cycle and continuous cycles.

In general, the protection effectiveness of SEI may be contributed by a combination of factors such as its overall chemical composition, thickness, ion conductivity, morphology, and mechanical strength. Here, we focus mostly on the contribution of chemical composition. For the SEI with short deposition cycles (e.g., less than 15 cycles), the effectiveness is limited by its unstable chemistry and low ionic conductivity. For SEI with long deposition cycles (e.g., more than 100 cycles), the effectiveness is curbed by the emergence of NaOH and thicker interface. The best protection effect is offered by the SEI formed with 50 pre-cycles, which has the optimal chemical composition. In principle, the composition and property of SEI may vary when it is deposited with a different capacity or a different current density or when NaDFOB is used as an additive. Nevertheless, the underlying formation mechanism of NaDFOB-derived SEI is correlated, and similar investigation strategies could be used.

In summary, we have studied the protection effect of NaDFOB-derived SEI and its composition evolution under various cycling stages in SMBs. In situ quantitative NMR results demonstrate that the prior reduction of the NaDFOB salt contributes to the formation of SEI, thereby effectively preventing the decomposition of organic solvents. XPS and high-resolution MAS NMR characterizations confirm that the SEI is mainly composed of inorganic borate-rich layers, which provide robust protection on sodium metal anode. The investigations on the SEI under different cycling stages reveal the chemical basis to achieve the optimal protection. Our work establishes a direct correlation between SEI compositions and cycling stability, offering new insights for the tailoring of SEI chemistry.

MATERIALS AND METHODS

Preparation of the NVOPF electrode

Na₃V₂(PO₄)₂O₂F (NVOPF) cathode was synthesized by hydrothermal method using stoichiometric amounts of V₂O_5, NaH₂PO₄·2H₂O and NaF as precursors. The detailed synthesis procedure has been provided in our previous work (47). In a typical preparation for electrode, NVOPF powder, Super P carbon black, and sodium carboxymethyl cellulose were mixed with the mass ratio of 7:2:1 in deionized water. The resulting slurry was pasted on aluminum foil and then dried in a vacuum oven at 80°C. The typical mass loading was about 2 mg cm⁻².

Materials

NaDFOB was synthesized through the reaction of Na₂C₂O₄ (Sigma-Aldrich) and BF₃·ether (Sigma-Aldrich) in acetonitrile, which has been described in previous work (44, 45). EC, DMC, sodium hexafluorophosphate (NaPF₆) in EC and DMC electrolytes, and sodium perclorate (NaClO₄) in EC and PC electrolytes were purchased from DoDoChem (battery grade 99+%).

Electrochemical measurements

CR2032 coin-type cells were constructed for electrochemical performance evaluation, XPS, SEM, and solid-state NMR characterizations. Glass fiber (Whatman D) was used as the separator. The galvanostatic cycling measurements were carried out on a Neware battery tester. The NVOPF/Na cells were cycled in a cutoff voltage range within 2.0 to 4.3 V at 10 C (1.3 A g⁻¹). An activation process was initially used at 0.5 C for five cycles. The electrochemical impedance spectroscopy (EIS) of the Na/Na cells was measured using a CHI660E electrochemical workstation with the frequency range from 0.01 to 100 kHz.

Characterizations

Structural characterization of the NVOPF cathode was performed by a powder x-ray diffraction (Ultima IV) with Cu Kα radiation. The morphology of cycled Na anodes was investigated by field-emission SEM (Hitachi SU8010 field-emission scanning electron microscope) at 5 kV. The surface analysis was conducted by XPS (ESCALAB 250Xi, Thermo Fisher Scientific Inc., USA), using a monochromatic Al Ka x-ray source at 1486.6 eV. Solid-state ²³Na and ¹¹B NMR spectra of the SEI products were collected on a 14.1-T magnet with the Bruker Avance III spectrometer using a 3.2-mm triple channel MAS probe. In situ NMR were measured on a 9.4-T magnet with the Bruker Avance III spectrometer using a static probe with an attached battery cycler. The details of NMR measurements are provided in the Supplementary Materials.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S12