Step-by-step desolvation enables high-rate and ultra-stable sodium storage in hard carbon anodes

Significance

Hard carbons are considered to be the most promising anode of sodium-ion batteries in terms of low cost, easy synthesis, and sustainability. However, hard carbon suffers from poor rate capability, limited life span, and plateau capacity decay with rate and cycling. Herein, we report a step-by-step desolvation strategy to reduce the high activation energy for one-step desolvation on hard carbons. It comprehensively promotes the overall improvement of performance, achieving the longest life span and minimum capacity decay rate at all evaluated current densities without compromising plateau capacity.

Abstract

Hard carbon is regarded as the most promising anode material for sodium-ion (Na-ion) batteries, owing to its advantages of high abundance, low cost, and low operating potential. However, the rate capability and cycle life span of hard carbon anodes are far from satisfactory, severely hindering its industrial applications. Here, we demonstrate that the desolvation process defines the Na-ion diffusion kinetics and the formation of a solid electrolyte interface (SEI). The 3A zeolite molecular sieve film on the hard carbon is proposed to develop a step-by-step desolvation pathway that effectively reduces the high activation energy of the direct desolvation process. Moreover, step-by-step desolvation yields a thin and inorganic-dominated SEI with a lower activation energy for Na⁺ transport. As a result, it contributes to greatly improved power density and cycling stability for both ester and ether electrolytes. When the above insights are applied, the hard carbon anode achieves the longest life span and minimum capacity fading rate at all evaluated current densities. Moreover, with the increase in current densities, an improved plateau capacity ratio is observed. This step-by-step desolvation strategy comprehensively enhances various properties of hard carbon anodes, which provides the possibility of building practical Na-ion batteries with high power density, high energy density, and durability.

Lithium-ion (Li-ion) batteries have achieved great success in portable and mobile devices in the past decades (1–4). However, Li reserves in the Earth’s crust are very limited and unevenly distributed, which has failed to meet increasing demands for Li-ion batteries. By contrast, the well-distributed sodium (Na) element has much higher abundance and lower cost (5–7). Moreover, the compatibility of Na-ion battery technology with cheap and lightweight Al current collectors makes it more sustainable than Li-ion batteries (8, 9). However, significant deficiencies in cycling stability, power density, and energy density remain unresolved for the practical application of Na-ion batteries.

Crucial steps for the commercialization of Li-ion batteries are the introduction of graphite anodes and the successful handling of interfacial stability based on in-depth analysis of the electrolyte desolvation process (10–12). Similar to the Li-ion battery configuration, hard carbon is expected to be the most promising anode material for Na-ion batteries due to its superiorities in high capacity, low operating potential, high abundance, and low cost (13, 14). Nonetheless, its electrochemical behaviors, such as rate performance, cycling life span, and initial Coulombic efficiency (ICE), are far inferior to those of the graphite anodes in Li-ion batteries (15–17). Major efforts have been dedicated to extending capacity and promoting rate capability by material modifications, such as heteroatom doping (18–22), increasing specific surface areas (23–25), introducing defects (26, 27), and pore mouth tightening (28). However, most of these strategies come at the expense of losing plateau capacity at low voltage and reducing ICE. When the defect, porosity, and specific surface areas in hard carbons are reduced, the ICE can increase, while it is unfavorable for maintaining a high reversible capacity (29, 30). Manipulation of the property of hard carbon anodes has generally failed to gain comprehensive improvement in electrochemical performance.

Incorporating a state-of-the-art hard carbon anode and an optimized ether electrolyte to regulate interfacial properties can simultaneously boost rate performance, life span, and ICE (31–33). For instance, Hou et al. (34) developed a flexible hard carbon paper anode, showing a capacity of 170.0 mAh g⁻¹ at 2 A g⁻¹ as well as good cycling stability in 1 M NaOTf-Diglyme (G2) electrolytes. Dong et al. (35) reported a xylose-derived hard carbon sphere that exhibited high ICE (84.93% at 1 A g⁻¹) and excellent rate capability in 1 M NaClO₄-G2 electrolytes. The superiority of ether electrolytes can be mainly ascribed to the following two aspects. First, the ether electrolyte possesses a higher LUMO (lowest unoccupied molecular orbital) than commonly used carbonate electrolytes, which can effectively resist excessive decomposition of electrolytes (36–38). Second, the ether electrolyte yields a high-quality solid electrolyte interphase (SEI) and guarantees the smooth cointercalation of solvated Na⁺ in hard carbons, thereby providing faster charge transport kinetics than in ester electrolytes (35, 39, 40). However, even when modified ether electrolytes are combined with optimized hard carbons, their electrochemical behaviors are still inferior to those of graphite anodes for Li-ion batteries and incapable of meeting the criteria for practical application of Na-ion batteries. For example, current progress in rate performance cannot satisfy the standard of fast charging in 15 to 10 min corresponding to 4 to 6 C (the goal of the U.S. Department of Energy) (41). Furthermore, as the galvanostatic current density increases, the plateau capacity (below 0.1 V versus Na⁺/Na) decays rapidly, as it is controlled by diffusion and suffers from slower kinetics (42). Therefore, increasing the current density inevitably leads to a rapid decay of plateau capacity. Unfortunately, there is no feasible access to further improve the rate performance while maintaining the plateau capacity of hard carbon anodes.

Solvation structures can be effectively regulated by materials with nanopores, such as 3A zeolite molecular sieves (43) and Metal Organic Frameworks (44), due to the size effect and can exhibit unique interfacial transport behaviors. The low-cost 3A zeolite molecular sieve with a well-defined pore size was used as model material in our follow-up studies. According to the International Zeolite Association, 3A zeolite falls under the Linde Type A framework type (45). It is composed of silicon-oxygen and aluminum-oxygen tetrahedra with a linear formula of K_nNa_12-n[(AlO₂)₁₂(SiO₂)₁₂] xH₂O, which is widely used in gas separation, purification, and drying, etc. (46).

In this work, we report a step-by-step desolvation strategy to improve the rate capability and cycling stability of hard carbon anodes in both ester and ether electrolytes. With the introduction of a 3A zeolite molecular sieve film, the direct desolvation on hard carbon anodes evolves into a step-by-step desolvation with largely reduced activation energy. Furthermore, the step-by-step desolvation enables a beyond concentrated electrolyte configuration. It causes the formation of a thin and inorganic-dominated SEI with largely reduced activation energy of Na⁺ transport through the SEI (Fig. 1). As a result, the synergistic improvement on fast Na⁺ transport kinetics then endows rarely reported rate performance. When the step-by-step strategy is applied, the hard carbon anode shows the longest life span and minimum capacity decay rate among all reported literature at all evaluated current densities of 0.2, 0.5, and 1 A g⁻¹. Furthermore, with the increase in current densities, an extensively improved capacity proportion at the plateau region was obtained. During long-term cycling, a plateau capacity ratio of up to 73% could be maintained at a current density of 0.5 A g⁻¹. Insights from this desolvation study will shed light on the practical application of Na-ion batteries, which is also expected to yield promising results in other types of rechargeable batteries.

Fig. 1.

(A and B) Comparison of the kinetics of desolvation and Na⁺ transport through the SEI in pristine ether electrolytes (A) and predesolvated electrolytes (B). (C) Schematic illustration of the step-by-step desolvation process.

Results and Discussion

Issues of rate performance and cycling stability also existed during early research into graphite anodes in Li-ion batteries (47, 48). However, these problems were properly resolved by electrolyte modification based on in-depth understanding of the desolvation process (49–51). In addition, long-standing distinctions regarding the electric double layer and pseudocapacitive energy storage mechanisms of porous and layered materials are well unified with the introduction of the concept of confinement microenvironment during the desolvation process (28, 52). Given the importance of the desolvation process for energy storage, we considered applying it to address these challenges in hard carbon anodes. For practical operation, we focused on regulation of the desolvation process by size effect, as it is directly related to the nanoconfined microenvironment, which was also verified in our previous works (53, 54).

Here, 3A zeolite molecular sieves with a well-defined nanopore (3.2 Å) were chosen to regulate desolvation and related interface behaviors. Then, the desolvation process and charge transport kinetics on the molecular sieve film–coated hard carbon anodes were analyzed in detail. To avoid direct contact of the hard carbon with the liquid electrolyte, we first attached the 3A zeolite molecular sieve film on the hard carbon anode and then dropped the electrolyte onto the 3A zeolite molecular sieve film. In addition, a 3A zeolite molecular sieve film that was larger than the hard carbon anode was used to avoid penetration of the liquid electrolyte to the hard carbon anode through the edge. These two points were crucial to ensure the effectiveness of the strategy. Note that solvated Na⁺ was not completely desolvated to remove all coordinating solvents and form naked Na⁺ when passing through the nanopores of molecular sieves but formed a solvent structure with a higher degree of aggregation, which can be seen as partial desolvation or predesolvation. The predesolvation on zeolite was well demonstrated by Raman and Fourier transform infrared spectroscopy (FTIR). Because of the size effect, the solvation structure of the solvent-separated ion pair (SSIP) in bulk electrolytes was predesolvated into a contact ion pair (CIP) or aggregate (AGG) with a smaller size under the electric field and was stored in the nanochannel of zeolites (Fig. 1C). The intuitive manifestation of this predesolvation is that the concentration of the electrolyte in the zeolite film was significantly higher than that of bulk electrolytes, which was elucidated in our previous report (43, 54, 55). In this study, the ether after the desolvation process within 3A zeolite molecular sieves was denoted as predesolvated electrolytes. By establishing the functional relationship between the area of coordinated PF₆⁻ in Raman spectra with the electrolyte concentration, the concentration of the predesolvated electrolyte could be quantitatively obtained. As shown in Fig. 2A, a beyond concentrated electrolyte with a concentration of 2.5 M was reached after predesolvation, far exceeding the bulk electrolyte (1 M). The unique predesolvation process possessed different interfacial charge transport kinetics. Therefore, the activation energy of desolvation on hard carbons was first investigated by fitting the electrochemical impedance spectroscopy (EIS) results to the Arrhenius law (32) (Fig. 2 B and C). The hard carbon used in this study was prepared by a typical polymer pyrolysis method. Detailed materials synthesis methods and characterization are presented in SI Appendix, Figs. S1 and S2. For the hard carbon anodes using predesolvated electrolytes, it delivered a desolvation activation energy (Ea₁) of 10.87 KJ mol⁻¹, which is significantly lower than that using pristine 1 M NaPF₆-G2 electrolytes (21.87 KJ mol⁻¹) (Fig. 2D). Obviously, the predesolvation process on the molecular sieve film significantly reduces the activation energy for the final desolvation on hard carbons. The one-step desolvation process in the pristine electrolytes evolves into step-by-step desolvation after the introduction of the zeolite film. As shown in Fig. 1, the predesolvation first occurred on the molecular sieve film for the bulk electrolyte, and then the predesolvated electrolyte was further desolvated on the hard carbon anode before transport in hard carbons. In addition, the activation energy of Na⁺ transport through the SEIs (Ea₂) can also see a decrease. The SEI derived from predesolvated electrolytes presents an Ea₂ of 16.36 KJ mol⁻¹, which is much smaller than the SEI formed in pristine 1 M NaPF₆-G2 electrolytes (22.76 KJ mol⁻¹). This is attributed to the formation of thin and inorganic-dominated SEIs with higher ionic conductivity in predesolvated electrolytes, which is discussed in detail later.

Fig. 2.

(A) The relationship between electrolyte concentration and coordinated PF₆⁻. (B and C) Nyquist plots of Na|hard carbon batteries in pristine 1 M NaPF₆-G2 electrolytes (B) and predesolvated electrolytes (C) at various temperatures. (D) Corresponding Arrhenius plots. (E and F) CV curves at various scan rates with pristine 1 M NaPF₆-G2 electrolytes (E) and predesolvated electrolytes (F). (G) Fitting results of b value for batteries using 1 M NaPF₆-G2 electrolytes and predesolvated electrolytes.

The improved rate capability of hard carbon in predesolvated electrolytes can also be seen from cyclic voltammetry (CV) curves. At all evaluated scan rates from 0.1 to 2 mV/s, the battery applying predesolvated electrolytes showed much larger currents than that using pristine 1 M NaPF₆-G2 electrolytes (Fig. 2 E and F). Then, the rate-determining step for the interface reaction was evaluated by introducing the equation I = av^b to fit b values, which was widely used for kinetic evaluation (34, 35). As shown in Fig. 2G and SI Appendix, Fig. S3, very good linear fit results were observed for these two electrolytes. The fitted b value for the battery using predesolvated electrolyte was 0.863, which is larger than the case using pristine 1 M NaPF₆-G2 electrolytes (0.793). The larger b value indicates a surface-controlled reaction in predesolvated electrolytes and is responsible for improved kinetics for hard carbons. An optimized electrode–electrolyte interface derived by predesolvated electrolytes facilitates faster charge transport. The unique properties enabled by step-by-step desolvation are expected to dramatically improve rate capability.

Then, we further evaluated the universality of the step-by-step desolvation strategy in both ester and ether electrolytes. In commonly used 1 M NaPF₆-ethylene carbonate (EC)/diethyl carbonate (DEC) ester electrolytes, the battery with predesolvated electrolytes showed better rate capability at all current densities. At current densities of 0.05, 0.1, and 0.2 A g⁻¹, it delivered capacities of 314.1, 288.5, and 210.5 mAh g⁻¹, respectively, which were significantly higher than those in pristine 1 M NaPF₆-EC/DEC electrolytes (Fig. 3 A and B). Furthermore, its plateau capacity was well maintained with the increase in current densities (SI Appendix, Fig. S4). In contrast with pristine 1 M NaPF₆-G2 ester electrolytes, hard carbon showed enhanced capacity retention in pristine ether electrolytes (Fig. 3 A and C). Upon introduction of the step-by-step strategy, the rate capability further improved (Fig. 3C). As the current density increased from 0.05 to 2 A g⁻¹, negligible capacity decay was observed, and a high capacity retention of 82.3% was preserved at 2 A g⁻¹. Even at a high current density of 5 A g⁻¹, a specific capacity of 224.0 mAh g⁻¹ was still maintained, corresponding to a capacity retention of 70.3% (Fig. 3D). In contrast, in the pristine 1 M NaPF₆-G2 electrolyte, only capacities of 202.5 and 146.6 mAh g⁻¹ were retained at current densities of 2 A g⁻¹ and 5 A g⁻¹, respectively.

Fig. 3.

(A) Rate capability at various current densities from 0.02 to 2 A g⁻¹ in pristine 1 M NaPF₆-EC/DEC electrolytes and predesolvated electrolytes. (B) Charge-discharge curves at varied current densities in these two electrolytes. (C) Rate capability at various current densities from 0.05 to 10 A g⁻¹ in pristine 1 M NaPF₆-G2 and predesolvated electrolytes. (D) Charge-discharge curves at varied current densities in predesolvated electrolytes. (E) Statistical results of plateau and slope capacity at different current densities in predesolvated ether electrolytes. (F) Capacity retention of plateau and slope at varied current densities.

It is generally believed that the adsorption or intercalation of Na⁺ occurs in the slope region (above 0.1 V versus Na⁺/Na), which endows faster kinetics and better rate capability (56). The plateau region (below 0.1 V versus Na⁺/Na) is controlled by diffusion and suffers from slower kinetics to restrain the rate performance of hard carbons (42). Therefore, almost all previous reports have focused on improving capacity capability from the perspective of maintaining the slope capacity, which largely reduces the operating voltage in full cells, especially under high current densities. Nevertheless, in contrast to previous reports and in typical 1 M NaPF₆-G2 electrolytes, the proportion of plateau capacity gradually increased in predesolvated electrolytes with the increase of current densities (Fig. 3E). When the current density exceeded 0.5 A g⁻¹, the plateau capacity accounted for more than 70% of the whole capacity. At a current density of 2 A g⁻¹, the plateau capacity ratio extended to 72.0% (Fig. 3E). This anomalous phenomenon broke the traditional perception that rate capability mainly depends on the capacity contribution of the slope region. When the capacity retention rates in the plateau and slope regions were compared, the advantage of the predesolvated electrolyte in maintaining plateau capacity at high current densities was more pronounced. Capacity retention in the plateau region was significantly higher than that of the slope region. Plateau capacity retention rates were 93.4% and 89.3% at current densities of 1 and 2 A g⁻¹, respectively (Fig. 3F). In contrast, only 68.9% and 62.9% plateau capacity retention was maintained in pristine 1 M NaPF₆-G2 electrolytes. This unique step-by-step desolvation strategy provides the potential to improve rate capability from the perspective of preserving plateau capacity.

In addition to the significantly reduced activation energy for desolvation on hard carbons, the activation energy of Na⁺ transport through SEIs can also see a decrease in predesolvated electrolytes, which signals the distinct properties of their formed SEI films. As shown in Fig. 4A, a layer of amorphous SEI can be identified by high-resolution transmission electron microscopy (TEM) for the hard carbon cycled in pristine 1 M NaPF₆-G2 electrolytes (after 100 cycles). The thickness of the SEI was measured to be ∼6.3 nm (Fig. 4B). In contrast, the SEI formed in predesolvated electrolytes was much thinner (about 3.1 nm), (Fig. 4 C and D and SI Appendix, Fig. S5), which is consistent with the CV test results (SI Appendix, Fig. S6). The thinner SEI favors faster Na⁺ diffusion. Subsequently, X-ray photoelectron spectroscopy (XPS) with depth profiling was used to evaluate the composition of SEIs formed in pristine 1 M NaPF₆-G2 electrolytes and predesolvated electrolytes. The Na content of the hard carbon cycled in the predesolvated electrolyte was lower than that in 1 M NaPF₆-G2 electrolytes (Fig. 4E). However, the content of the typical inorganic component of F in the SEI formed in the predesolvated electrolytes was twice that formed in pristine 1 M NaPF₆-G2 electrolytes. Correspondingly, the signal of NaF always has a higher intensity for the SEI derived from predesolvated electrolytes, which substantiates the high abundance of NaF species in SEI derived from predesolvated electrolytes (Fig. 4 F and G). The high content of NaF in the SEI would promote stability in predesolvated electrolytes because of its high modulus and wide band gap.

Fig. 4.

(A) TEM image of hard carbon after 100 cycles in pristine 1 M NaPF₆-G2 electrolytes (The pink rectangle in Fig. A corresponds to the thickness profile in Fig. B; The white dashed line in Fig. A and Fig. C is used to distinguish SEI from hard carbons). (B) Corresponding SEI thickness profile. (C) TEM image of hard carbon after 100 cycles in predesolvated electrolytes (The blue rectangle in Fig. C corresponds to the thickness profile in Fig. D). (D) Corresponding SEI thickness profile. (E) The atomic concentration percentage of the SEI composition after different Ar⁺ etching times. (F and G) NaF signal changes of SEI formed in predesolvated electrolytes (F) and pristine 1 M NaPF₆-G2 electrolytes (G) at various etching times. (H and J) C 1s depth profiles of cycled hard carbons collected from pristine 1 M NaPF₆-G2 electrolytes (H) and predesolvated electrolytes (J) (Black rectangles in Fig. H and J show organic components in SEIs). (I and K) C 1s fitting results at etching time of 50 s in pristine 1 M NaPF₆-G2 electrolytes (I) and predesolvated electrolytes (K). (L and M) Na 1s peak profiles of the cycled hard carbon in NaPF₆-G2 electrolytes (L) and predesolvated electrolytes (M). The black dashed lines in Fig. H, J, L and M show peak shifts. (N) Na 1s peak position comparison. a.u., arbitrary units.

For the C 1s spectra of the cycled hard carbons in pristine 1 M NaPF₆-G2 electrolytes, the main peak saw a clear shift toward lower binding energy, and the peak position remained unchanged after etching for 140 s (Fig. 4H). The peak shift originates from the formation of the organic SEI on hard carbons due to the decomposition of solvents. At the same time, the peak located at about 290 eV corresponds to the C=O bond of organic species in the SEI, which almost disappeared after etching for 140 s. In contrast, the hard carbon harvested from batteries using predesolvated electrolytes experienced a slight peak shift, and the SEI was completely removed after 50 s of etching (Fig. 4I), indicating a thinner SEI with less organic components. It was evident that the predesolvated electrolytes yielded a much thinner SEI than pristine 1 M NaPF₆-G2 electrolytes, which is consistent with the TEM results. Under the same etching time of 50 s, the organic species of C=O and C=O-C for the SEI generated from pristine 1 M NaPF₆-G2 electrolytes accounted for 37.1% (Fig. 4J), while the organic species only accounted for 16.8% of SEIs formed in predesolvated electrolytes (Fig. 4K), demonstrating that more organic species exist in the SEI formed in typical ether electrolytes. A positive peak shift can also be seen from the Na 1s spectra for the hard carbon cycled in 1 M NaPF₆-G2 electrolytes, demonstrating a valence state transformation toward a compound state as the etching time increased (Fig. 4 L and N). In contrast, the peak position of hard carbon harvested from predesolvated electrolytes was stable, and its valence was closer to ionic states (Fig. 4 M and N). These results also demonstrate that more ionic species of NaF/Na₂O dominated the SEI formed in predesolvated electrolytes. The thin and inorganic-dominated SEI well accounts for the reduced activation energy of Na⁺ transport through the SEIs, thus the dramatically improved rate capability. Although there were differences in the composition of the SEI formed in the two electrolytes, no additional decomposition products were detected in the predesolvated electrolyte based on the ¹H NMR test (SI Appendix, Fig. S7). The same by-product generation is attributed to the same type of solvation structures (free solvent, SSIP, CIP, and AGG), which are all involved in the formation of SEIs. The predesolvation process improved the proportion of solvation structures of CIP and AGG in the electrolyte, but did not form new species or initiate additional decomposition paths.

The difference in SEI composition inevitably leads to different mechanical properties, which would largely affect interface stability and cycling stability. Therefore, the mechanical property of the SEI formed in these two electrolytes was measured by atomic force microscopy (AFM). The typical sphere morphology of hard carbons was observed in both cases (Fig. 5 A and D) by AFM, which is consistent with the TEM and scanning electron microscopy (SEM) observations. The elastic modulus of cycled hard carbons is shown in Fig. 5 B and E. The elastic modulus of the SEI formed in predesolvated electrolytes was much higher than that in 1 M NaPF₆-G2 electrolytes as reflected in mappings. Seven sites were chosen in each mapping to study the mechanical properties, and the corresponding force-displacement curves are shown in Fig. 5 C and F. The average elastic modulus of the SEI generated from predesolvated electrolytes was 8.74 GPa, which is much higher than the SEI formed in pristine 1 M NaPF₆-G2 electrolytes (4.49 GPa). In addition, the force displacement curves during loading and unloading were not fully reversible (Fig. 5C) for the SEI formed in pristine 1 M NaPF₆-G2 electrolytes. However, under similar force loading, the predesolvated electrolyte-derived SEI exhibited elastic deformation with small displacements (Fig. 5F), further demonstrating a higher modulus than the SEI formed in pristine electrolytes. This high modulus is directly related to the high proportion of NaF in the SEI formed in predesolvated electrolytes. Incorporating the inherent wide bandgap of NaF, the inorganic-dominated SEI formed in predesolvated electrolytes is expected to provide a stable interface and greatly improved cycling stability.

Fig. 5.

AFM analysis of SEI formed in these two electrolytes. (A and D) AFM height images of hard carbons harvested from pristine 1 M NaPF₆-G2 electrolytes (A) and predesolvated electrolytes (D). (B and E) Corresponding two-dimensional maps of elastic modulus. (C and F) Representative force-displacement curves of selected sites for the SEI formed in 1 M NaPF₆-G2 electrolytes (C) and predesolvated electrolytes (F).

Here, three current densities were used to evaluate the cycling stability of hard carbon anodes in predesolvated electrolytes. At all current densities evaluated, the hard carbon anodes with predesolvated electrolytes exhibited the longest cycle life and best cycle stability reported to date. Specifically, at a current density of 0.2 A g⁻¹, the cell presents a reversible capacity of 255.1 mAh g⁻¹ in predesolvated electrolytes after 1,900 cycles, with a decay rate of 0.0068% per cycle (Fig. 6A). Slight capacity decay was observed from the detailed charge-discharge curves (Fig. 6B). In contrast, the hard carbon anode using the pristine 1 M NaPF₆-G2 electrolyte showed rapid capacity fading (SI Appendix, Fig. S8). Only 228.2 mAh g⁻¹ of capacity remained after 783 cycles. At a current density of 0.5 A g⁻¹, the hard carbon anode with predesolvated electrolyte still maintained a high capacity of 260.4 mAh g⁻¹ after 4,000 cycles, with a capacity retention rate of 94.1% (Fig. 6 C and D). When the current density was further increased to 1 A g⁻¹, the hard carbon exhibited a high capacity of 251.9 mAh g⁻¹ after 4,500 cycles, corresponding to a capacity retention of 91% (Fig. 6E). The almost overlapped charge-discharge profiles further illustrate this high stability in predesolvated electrolytes (SI Appendix, Fig. S9). In contrast, the batteries using pristine 1 M NaPF₆-G2 electrolyte not only had lower capacities at current densities of 0.5 A g⁻¹ and 1 A g⁻¹ but also had a sharp capacity decline after about 1,000 cycles. After repeated cycling, both the hard carbon anode and the 3A zeolite molecular sieve film maintained good integrity and a clean surface in predesolvated electrolytes (SI Appendix, Figs. S10–S12). However, for the hard carbon cycled with the pristine 1 M NaPF₆-G2 electrolytes, the electrode was heavily coated by by-products due to electrolyte decomposition, and the hard carbon was not well distinguished (SI Appendix, Fig. S13). In addition, when the step-by-step strategy was applied in ester electrolytes, the cycling stability also dramatically improved (SI Appendix, Figs. S14 and S15).

Fig. 6.

(A–D) Long-term cycling performance at 0.2 A g⁻¹ (A) and 0.5 A g⁻¹ (C) and corresponding charge-discharge profiles (B and D). (E) Long-term cycling performance at 1 A g⁻¹. (F and G) Variation of plateau capacity ratio with the number of cycles in predesolvated electrolyte (F) and pristine 1 M NaPF₆-G2 electrolyte (G). (H) Nyquist plots of Na|Hard carbon batteries using predesolvated electrolytes after various cycles. (I) Comparison of stability and cycle number of hard carbon anodes using predesolvated electrolytes and data reported in previous literature.

In addition, the high percentage of plateau capacity was well maintained during cycling. Plateau capacity ratios of over 69% were achieved at all evaluated current densities. At a current density of 0.5 A g⁻¹, an ultra-high plateau capacity ratio of about 73% was achieved over 4,000 cycles (Fig. 6F). However, for the hard carbon anode using the pristine 1 M NaPF₆-G2 electrolyte, the proportion of plateau capacity decreased with increasing current density, and the plateau capacity ratio decreased significantly over about 1,000 cycles (Fig. 6G). The ultra-stable cycling and good plateau capacity retention capability can be partially interpreted by the robust interface generated by predesolvated electrolytes. As shown in Fig. 6H, both the resistance of Na⁺ transport through SEIs and charge transfer resistance were basically unchanged over 1,000 cycles. In addition, the almost overlapping CV curves during repeated scans also demonstrated excellent stability of the interface constructed by the predesolvated electrolyte (SI Appendix, Fig. S16). Then, we compared the recently reported data on life span and cycling stability of hard carbon anodes with that applying the step-by-step strategy (Fig. 6I). At all evaluated current densities, the life span of the hard carbon anodes with predesolvated electrolytes far exceeded the highest values reported to date, while the capacity decay rate was the minimum reported so far (20, 34, 35, 57–62). Especially at current densities of 0.5 A g⁻¹ and 1 A g⁻¹, the average capacity decay rate per cycle was only 0.0015% and 0.002%, respectively, which is far below existing works.

To verify the practical potential of the predesolvation strategy, we assembled full cells using 4V-class O3-NaCu_1/9Ni_2/9Fe_1/3Mn_1/3O₂ as the cathode. A 3A zeolite molecular sieve film was used on both the cathode and anode to ensure the high voltage stability of the electrolyte and the rate and cycling stability of the hard carbon. The excellent rate performance was still well preserved in full cells (SI Appendix, Fig. S17). At current densities of 0.05, 0.2, and 0.5 A g⁻¹, capacities of 309.2, 285.5, and 268.2 mAh g⁻¹ (based on anode mass), respectively, were delivered. In addition, good cycling stability was well maintained in full cells. At current densities of 0.2 A g⁻¹ and 0.5 A g⁻¹, high capacities of 267.6 and 248.5 mAh g⁻¹ were retained after 120 and 200 cycles, respectively. The energy density reached over 260 Wh kg⁻¹ (based on the total mass of the cathode and anode active materials) at current densities of 0.03 (272.4 Wh kg⁻¹) and 0.05 A g⁻¹ (268.0 Wh kg⁻¹) (SI Appendix, Fig. S18). At high current densities of 0.5 and 1A g⁻¹, high energy densities of 230.0 and 206.4 Wh kg⁻¹ were maintained, corresponding to power densities of 430.9 and 845.6 W kg⁻¹. This indicates that this Na-ion battery can be charged to 75% of the total capacity in 15 min. The excellent fast-charging performance in full cells further validates the unique advantages of this step-by-step desolvation strategy in constructing practical Na-ion batteries.

In summary, we have demonstrated that the desolvation process is crucial in determining Na⁺ storage kinetics of hard carbon anodes by mainly affecting the activation energy of desolvation and SEI formation. With application of these insights, a universal step-by-step desolvation strategy is proposed to improve rate capability and cycling stability in both ester and ether electrolytes. One-step desolvation with high activation energy on hard carbons is effectively dispersed and reduced by the step-by-step desolvation. In addition, the predesolvated electrolyte promotes the formation of a thin and inorganic dominated SEI with lower activation energy for the Na⁺ transport. As a result, it contributes to unprecedented plateau capacity ratio increase with current densities. Moreover, the hard carbon working with the predesolvated electrolyte provides excellent rate capability and cycling life span. Overall, the proposed predesolvated electrolyte successfully unraveled the poor Na⁺ storage kinetics and interface in hard carbons and shed light on developing more practical Na-ion batteries with high power density and life span.

Materials and Methods

Preparation of Hard Carbon Anodes.

Hard carbons were prepared by high-temperature carbonization of the glucose carbon spheres. Typical hydrothermal carbonization was used to synthesize the carbon spheres. Five grams of glucose was dissolved in 45 mL deionized water, placed in a sealed autoclave reactor vessel (100 mL), and reacted for 12 h at 230 °C. Subsequently, the solid product was isolated by filtration and washed several times with water and ethanol followed by drying under vacuum at 80 °C overnight. The resulting sample was carbonized at 1,400 °C for 2 h under Argon atmospheres. The hard carbon anode was prepared by mixing the hard carbon powder with acetylene black and sodium alginate in a ratio of 8:1:1 using water as solvents, and the final sticky slurry was coated on a carbon Cu foil using a scraper. Then, the electrode was vacuum dried at 100 °C overnight. For coin-cell measurements, the electrodes were punched into a plate with a diameter of 10 mm.

Preparation of the 3A Zeolite Molecular Sieve Film.

The 3A zeolite molecular sieve used in this study was purchased from Alfa Aesar, and the bead diameter was between 0.4 and 0.8 mm. The bead was ground into powder by planetary ball milling (Planetary Mono Mill PULVERISETTE 6 classic line; Fritsch). The rotational speed was controlled at 400 rpm for 5 min followed by rest for 5 min, and the total ball-milling time was 12 h. After ball milling, the powder was heated at 200 °C for 24 h to remove residual moisture absorbed in the zeolite channel. Then, the zeolite powder was mixed with polyvinylidene fluoride with a ratio of 9:1 using N-methyl pyrrolidine as solvent. The obtained slurry was homogeneously coated onto a glass plate by scraper and dried at 80 °C until the solvent was removed. Then, the molecular sieve film was soaked in methanol for several minutes, and the zeolite film was peeled off from the glass plate. The prepared zeolite film was then dried at 100 °C for 48 h under vacuum. Before use, the film was cut into small plates (12 mm in diameter).

Cell Assembly and Electrochemical Measurements.

CR2032 coin cells were used to evaluate battery performance. Na|Hard carbon cells were assembled inside a glove box to test the function of the step-by-step strategy in improving rate and cycling performance. During the assembly process, the 3A zeolite molecular sieve film was tightly pressed on the hard carbon side, and the other procedure was the same as regular assembly. Typical Celgard 2500 was used as the separator. The 1 M NaPF₆-G2 electrolyte or 1 M NaPF₆-PC/FEC electrolytes were used as electrolytes. Then, a Neware battery test system (Neware Technology Co.) was used to evaluate battery performance. The CV and EIS were collected in a CHI 660E electrochemical workstation. The Na-ion full cell was constructed using hard carbon as the anode and O3-NaCu_1/9Ni_2/9Fe_1/3Mn_1/3O₂ as the cathode. The weight ratio of the two electrodes (anode/cathode) was 1:2.8 (negative/positive capacity ratio, N/P ratio = 0.91), which was set to compensate for irreversible Na loss due to SEI formation. The full cells were tested in a voltage range of 1 to 4 V.

Characterization.

SEM was performed on a Hitachi S4800 (Hitachi Japan) instrument to analyze the morphology of cycled hard carbon anodes. X-ray diffraction (XRD) measurements were performed on a Bruker D8 Advanced Diffractometer fitted with Cu Kα (λ = 1.5406 Å) radiation at a scan rate of 5°/min. Elemental analysis was performed on an XPS (Thermo ESCALAB 250 Xi, Al Kα radiation, hv = 1,486.6 eV, America). Before XPS tests, the cycled hard carbon was washed with dimethoxyethane (DME) three times to remove residual electrolytes and then dried by evaporation in a vacuum chamber. The hard carbon anode was pasted on the loading chamber with a vacuum transfer function. In the process of transfer, the sample was sealed in the loading chamber and always protected by Ar or vacuum atmosphere. The electrode was not exposed to any reactive atmosphere during the whole process. The Raman spectra were obtained using a LabRAM HR Evolution with 532.05 nm incident radiation and a 50× aperture. The acquisition time was 15 s with two accumulations during Raman spectrum collection. TEM observations were performed on a JEM-1400Flash. The morphology and mechanical behavior of hard carbons were recorded on AFM using tapping mode imaging with an antimony-doped silicon tip (BRUKER RTESPA-300-30). NMR spectra were recorded using a 400-MHz Ultra-Shield spectrometer (Bruker). The ¹H NMR spectra of electrolytes were recorded in D₂O (99.9 atom% D; Wako Chemicals).

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. The data have been deposited in a publicly accessible database https://figshare.com/articles/journal_contribution/Fig_4A_tif/21150658/3 (63).