Unraveling Water–Defect Coupled Degradation via Deuterium Isotope Labeling in Prussian Blue Analogue Cathodes for Long‐Life Sodium‐Ion Batteries

ABSTRACT

Sodium‐based Prussian blue analogues (NaPBA) are promising cathodes for sodium‐ion batteries owing to their high capacity and low cost. Nonetheless, rapid capacity fading during prolonged cycling remains a critical challenge, as the underlying degradation mechanisms are not fully understood. Here, we employ deuterium isotope labeling to unveil a dual‐coupled degradation pathway in NaPBA cathodes, where crystalline water‐induced lattice distortion is coupled with defect‐triggered electrolyte decomposition. Density functional theory screening identifies In(OTf)₃ as a multifunctional electrolyte additive, which suppresses both degradation processes via synergistic cation–anion effects. Molecular dynamics simulations and isotope‐ratio mass spectrometry (IRMS) reveal that In³⁺ strongly coordinates with crystalline water, suppressing its repeated insertion/extraction and preventing framework collapse, while OTf^‐− anions passivate [Fe(CN)₆]⁴⁻ vacancies, reducing solvent adsorption and inhibiting electrolyte decomposition at defect sites. Consequently, the NaPBA||Na cells incorporating In(OTf)₃ exhibit significantly improved electrochemical performance, demonstrating 80.63% capacity retention after 1000 cycles at 1C and nearly threefold higher discharge capacity at 20C compared to baseline cells, outperforming the most reported PBA‐based batteries. This cation–anion modulation strategy provides a general design principle for electrolyte engineering in PBA‐based energy‐storage systems.

Through deuterium isotope labeling, we demonstrate a dual‐coupled degradation mechanism in Sodium‐based Prussian blue analogs (NaPBA) cathodes involving “crystalline water‐induced distortion accumulation coupled with defect‐triggered electrolyte decomposition”. The In(OTf)₃ additive exhibits synergistic cation/anion effects that simultaneously immobilize crystalline water and passivate [Fe(CN)₆]⁴⁻ vacancy defects. This dual action mitigates structural distortion while suppressing electrolyte decomposition side reactions. Consequently, the In(OTf)₃‐containing NaPBA||Na cell achieves 80.63% capacity retention after 1000 cycles.

1. Introduction

Sodium‐ion batteries (SIBs) are an attractive alternative to lithium‐ion systems for large‐scale energy storage owing to the abundant reserves, widespread distribution, and low cost of sodium resources [1, 2, 3]. As a key battery component, cathode materials fundamentally determine the electrochemical performance of batteries [4, 5, 6]. Among various cathode materials, Prussian blue analogues (PBAs) stand out owing to their open framework for rapid Na⁺ diffusion and energy‐efficient low‐temperature synthesis [7, 8, 9, 10, 11]. Specifically, low‐cost Fe‐based PBAs offer high theoretical capacity (170 mAh g⁻¹) and avoid the Jahn‐Teller distortions associated with Mn‐based systems, making them highly attractive for commercialization [8, 12, 13].

Despite their promise, PBA cathodes synthesized via aqueous co‐precipitation inevitably contain crystalline water and [Fe(CN)₆]⁴⁻ vacancy defects, leading to severe structural degradation and capacity decay [14, 15, 16]. Specifically, crystalline water predominantly resides at defect sites as coordinated water and within interstitial spaces as interstitial water. While coordinated water stabilizes the framework, interstitial water undergoes repeated deintercalation during cycling, driving cumulative lattice distortion [17, 18]. Unfortunately, direct experimental investigation of the above intrinsic mechanism remains challenging due to the low crystalline water content in PBAs and their susceptibility to ambient moisture, which complicates reliable characterization [19]. Concurrently, vacancy defects act as catalytic sites for parasitic electrolyte decomposition, generating byproducts that obstruct Na⁺ transport [20].

Conventional strategies to mitigate these issues primarily focus on synthesis optimization, such as hydrothermal methods or post‐synthesis thermal treatments [21, 22, 23]. However, hydrothermal processes risk generating toxic byproducts (e.g., NaCN), while thermal treatments often struggle to selectively remove interstitial water without disrupting the stabilizing coordinated water [24]. Moreover, approaches to reduce vacancy defects (e.g., chelating agents [25] or reaction temperature adjustments [18, 26]) show limited efficacy. Fundamentally, the synergistic degradation mechanism between crystalline water and vacancy defects remains poorly understood, and a strategy that selectively targets harmful interstitial water without disrupting beneficial coordinated water is still lacking. Therefore, developing simple and effective approaches to simultaneously suppress both degradation pathways is critical for advancing sodium‐based PBAs (NaPBA) cathodes.

Compared to energy‐intensive solid‐state cathode modifications, electrolyte engineering offers a more flexible, scalable and manufacturing‐compatible solution. Well‐designed additives can modulate interfacial chemistry without altering the intrinsic cathode structure, enabling simultaneous mitigation of multiple degradation modes. However, rational additive design requires molecular‐level understanding of interactions between additive species, crystalline water and vacancy defects, which remains insufficient in PBA systems.

Herein, we elucidate a dual‐coupled degradation mechanism in NaPBA cathodes wherein crystalline water‐induced lattice distortion is intimately intertwined with defect‐triggered electrolyte decomposition. Through deuterium isotopic labeling and density functional theory (DFT) calculations, we reveal that crystalline water undergoes repeated insertion/extraction in tandem with Na⁺ cycling, driving cumulative structural distortions and eventual framework collapse. Concurrently, [Fe(CN)₆]⁴⁻ vacancy defects catalyze parasitic decomposition of electrolyte solvents, generating surface byproducts that impede ion transport kinetics. Guided by these mechanistic insights, we establish a rational additive selection criterion based on strong water affinity and minimal interaction with fluoroethylene carbonate (FEC). Theoretical screening identified indium trifluoromethanesulfonate (In(OTf)₃) as an optimal multifunctional additive that operates via synergistic cation/anion mechanisms: In³⁺ strongly coordinates with water molecules to confine them in the electrolyte bulk, preventing their deleterious shuttling; OTf⁻ preferentially adsorbs at vacancy defects to passivate catalytic sites. This cation/anion‐synergistic suppression of dual‐coupled degradation is validated through comprehensive theoretical and experimental investigations. As a result, NaPBA||Na cells containing In(OTf)₃ retain 80.63% capacity after 1000 cycles (approximately one‐third of the baseline cell) and deliver nearly triple the discharge capacity of control cells at 20 C after three high‐rate cycles. We further develop an In(OTf)₃‐containing NaPBA||Na pouch cell that lights LED panels, confirming its practical viability. These mechanistic insights can provide effective guidance for developing advanced electrolytes toward long‐cycle PBA‐based energy storage systems.

2. Results and Discussion

To investigate crystalline water migration in PBAs during electrochemical cycling while eliminating ambient moisture interference, deuterium‐labeled sodium‐based PBA (NaPBA‐D₂O) was synthesized via co‐precipitation in D₂O. Conventional sodium‐based PBA (NaPBA‐H₂O) was also prepared in aqueous solution for comparative analysis. Scanning electron microscopy (SEM) images (Figure 1a,b) revealed that NaPBA‐D₂O particles were cubic, approximately 1∼2 µm in size, with morphology and particle dimensions nearly identical to those of NaPBA‐H₂O. Energy‐dispersive x‐ray spectroscopy (EDS) further confirmed homogeneous distribution of Na, Fe, C, N, and O in both samples (Figures S1 and S2). X‐ray diffraction (XRD) patterns (Figure S3) revealed that both materials adopted a monoclinic phase, confirming that D₂O‐based synthesis does not affect the crystalline structure of PBAs.

FIGURE 1.

SEM image of (a) NaPBA‐H₂O and (b) NaPBA‐D₂O. (c) Fourier transform infrared spectra of NaPBA‐H₂O and NaPBA‐D₂O. (d) Schematic illustration of D₂O co‐extracted with Na⁺ from the NaPBA framework during charging. (e) Comparison of deuterium isotope content in NaPBA‐D₂O electrodes before and after charging. (f) Schematic illustration of D₂O co‐intercalated with Na⁺ into the NaPBA framework during discharging. (g) Comparison of deuterium isotope content in NaPBA‐H₂O electrodes before and after discharging. Optimized configurations of (h) initial NaPBA, (i) NaPBA with H₂O co‐intercalated Na⁺ ions, and (j) NaPBA with Na⁺ intercalation.

Given the high sensitivity of vibrational spectroscopy in distinguishing chemical bonds, Fourier transform infrared spectra (FTIR) were acquired for both samples. As shown in Figure 1c, the FTIR spectrum of NaPBA‐D₂O exhibited three new vibrational bands within the 2580∼2700 cm⁻¹ range, corresponding to D─O stretching vibrations of D₂O [27]. Notably, complete exclusion of environmental H₂O during synthesis was extremely difficult: both precursors and centrifugal washing solvents contained hydrogen atoms, promoting H/D exchange with D₂O to form H₂O. Since H₂O has a lower molecular weight and faster reaction kinetics than D₂O, residual H₂O persisted in NaPBA‐D₂O, explaining the broad H₂O stretching band near 3400 cm⁻¹ in its FTIR spectrum [28]. H₂O in NaPBA exists in forms analogous to D₂O, including crystal water confined within the NaPBA framework (interstitial water and coordination water) and a small amount of surface adsorbed water [17].Solid‐state deuterium nuclear magnetic resonance (SSNMR) spectroscopy further confirmed this finding, showing a distinct deuterium signal in NaPBA‐D₂O that is absent in NaPBA‐H₂O (Figure S4). Collectively, these spectroscopic results confirm the successful synthesis of D₂O‐labeled NaPBA‐D₂O. Thermogravimetric analysis (TGA) was conducted on both NaPBA‐D₂O and NaPBA‐H₂O samples (Figure S5) to quantify their water (D₂O and H₂O) content. The water contents of NaPBA‐D₂O and NaPBA‐H₂O were 8∼9 wt%, which was similar to previously reported values [29]. We further evaluated the basic electrochemical performance of both samples. Under the identical charge/discharge protocol, NaPBA‐D₂O and NaPBA‐H₂O exhibited nearly similar electrochemical profiles and specific capacities (∼103 mAh g⁻¹ at 100 mA g⁻¹; Figure S6). This confirms that D₂O isotopic labeling does not perturb the Na⁺ storage mechanism of PBAs, which validates the use of NaPBA‐D₂O for investigating crystalline water migration within the PBA framework. Compared to conventional spectroscopic methods, stable isotope ratio mass spectrometry (IRMS) offers substantially higher sensitivity. It can detect ultra‐low deuterium isotope concentrations, such as the natural abundance of only 0.015% in water, enabling precise analysis of trace species [30]. Therefore, IRMS delivers robust experimental support for probing the migration dynamics of trace crystalline water within PBAs. To track the migration of crystalline water during electrochemical processes, we fabricated pristine NaPBA‐D₂O cathodes (NaPBA·nD₂O) and assembled NaPBA·nD₂O||Na cells. As illustrated in Figure 1d, D₂O molecule alongside Na⁺ ions is expected to migrate from the PBA cathode framework into the electrolyte during charging, which means a reduction in the D₂O content within the PBA. IRMS results in Figure 1e confirmed this phenomenon. Compared to the pristine NaPBA·nD₂O electrode, the deuterium abundance in the charged electrode (NaPBA·(n−x)D₂O) decreased by approximately 50%. Subsequently, to verify whether electrolyte D₂O re‐enters the PBA framework during discharge (accompanied by Na⁺ intercalation), we extracted the electrolyte‐soaked separator from fully charged NaPBA·nD₂O||Na cells. This separator was then reassembled with an isotopically unlabeled, charged‐state PBA cathode (NaPBA·(n−x)H₂O) and a Na metal anode to form the NaPBA·(n−x)H₂O||Na cell. Upon discharge completion (Figure 1f), D₂O alongside Na⁺ insertion migrated back into the PBA framework. IRMS results were consistent with this process, showing increased deuterium abundance in discharged electrodes (NaPBA·(n−x)H₂O·xD₂O) compared to the pre‐discharge NaPBA·(n−x)H₂O electrode (Figure 1g).

As shown in Figure S7, the intensity of the ─OH absorption peak decreases upon charging and recovers after discharging compared with the initial NaPBA‐D₂O electrode. This phenomenon qualitatively verifies that water molecules co‐intercalate and co‐deintercalate with Na⁺ ions during cycling. FTIR results merely provide rough qualitative evidence. By contrast, stable isotope mass spectrometry exhibits unique superiorities in quantitative analysis and accurate tracing of hydrogen isotope species. Notably, the migration of crystalline water accompanied by Na⁺ intercalation and deintercalation is not completely reversible, and water participates in parasitic reactions on both electrodes. On the cathode side, crystalline water can promote the decomposition of electrolyte. On the sodium metal anode side, a small amount of crystal water that shuttles across the separator is reduced to release H₂ (Figure S8). Collectively, by employing deuterium isotope labeling combined with IRMS, we effectively eliminated interference from environmental water molecules and demonstrated that crystalline water within the PBA framework undergoes reversible migration, synchronously with Na⁺ deintercalation and intercalation during electrochemical cycling.

DFT calculations were performed to examine whether this reversible migration of crystalline water influences the crystal structure of PBAs. We selected the PBA crystal structure at the late discharge stage as the initial state, since PBAs remain stable in the monoclinic phase during this period, facilitating direct comparison of unit cell volume changes before and after Na⁺ intercalation. As shown in Figure S9, upon co‐intercalation of H₂O with Na⁺, the PBA unit cell volume expands by 1.78%. In contrast, with Na⁺ intercalation alone, the volume change is reduced to 0.99%. Detailed analysis of lattice parameter changes showed that H₂O/Na⁺ co‐intercalation induced more significant expansion along the a‐, b‐, and c‐axes compared to Na⁺‐only intercalation (Figure 1h–j). This suggests that reversible H₂O migration during charge/discharge cycles results in more significant volumetric deformation of the PBA framework. In addition to the widely recognized phase‐transition stress associated with Na⁺ (de)intercalation [14] and the intrinsic stress caused by [Fe(CN)₆]⁴⁻ vacancies [15], our results indicate that the insertion and migration of crystalline water also act as a crucial source of lattice stress. The dynamic ingress and egress of H₂O molecules disrupt hydrogen bonding and alter the local coordination environment within the Fe─N≡C─Fe network, leading to anisotropic lattice fluctuations and transient mechanical perturbation. Such water‐derived stress, when coupled with phase‐transition and vacancy‐induced strains, could further intensify the cumulative lattice distortion of PBAs upon prolonged cycling, potentially culminating in structural degradation and framework collapse (Figure S10).

Beyond bulk‐phase optimization, electrolyte engineering offers a promising alternative to mitigate structural degradation. Given that water molecules possess lone electron pairs on their oxygen atoms, we propose that electrophilic additives are capable of capturing crystallization water via a coordination mechanism. During charging, crystalline water within the framework exits into the electrolyte alongside Na⁺ ions (Figure 2a). These electrophilic additives can then coordinate with the lone pair electrons of water molecules, effectively anchoring the water molecules within the bulk electrolyte phase (Figure 2b). The aforementioned interaction prevents water molecules from re‐entering the PBA framework during discharge, thereby maintaining the structural integrity of the cathode material. Lewis acids constitute an important class of electrophilic additives [31]. The strength of their coordination with H₂O requires careful optimization. Excessive Lewis acidity can induce the decomposition of FEC, a key film‐forming additive for Na metal anodes (Figure S11) [32]. Such decomposition generates HF, which directly attacks and degrades the PBA framework. To balance these competing requirements, we establish a screening criterion requiring strong H₂O adsorption but weak FEC interaction, and computationally evaluate the adsorption energies for Al‐, Ga‐, and In‐based Lewis acids with both species (Figure 2c,d). The selection of Group 13 metal ions (Al³⁺, Ga³⁺, and In³⁺) is predicated on their vacant orbitals, which readily accept lone‐pair electrons from H₂O molecules. Moreover, we also analyze the interaction of these additives with FEC or H₂O (Figures 2e, S12, and S13), where F atoms in FEC and O atoms in H₂O both tend to interact with the metal cations of the additives. As shown in Figure 2f, Al(OTf)₃ exhibits the strongest H₂O adsorption among the six evaluated additives. However, its strong affinity toward the fluorine atoms of FEC may promote massive FEC decomposition, thereby resulting in an obvious decline in the Coulombic efficiency (CE) of batteries (Figure S14). Our comprehensive analysis reveals that In(OTf)₃ achieves an optimal balance, demonstrating strong H₂O coordination while maintaining weak FEC interactions. This selective adsorption profile establishes In(OTf)₃ as the optimal candidate for precise water molecule capture.

FIGURE 2.

Schematic illustration of the mechanism for electrophilic additives capturing bound H₂O in NaPBA frameworks during (a) charge and (b) discharge processes. (c) Comparison of the adsorption energy of FEC with different additives. (d) Comparison of the adsorption energy of H₂O with different additives and NaClO₄. (e) Charge transfer between H₂O and In(OTf)₃. Cyan and yellow isosurfaces represent electronic loss and gain, respectively. (f) Comprehensive analysis of the adsorption energy of different additives for FEC and H₂O. (g–i) Molecular dynamics simulation snapshots of H₂O distribution in In(OTf)₃‐containing electrolytes under different states. (j) The variation of the distances between four water molecules and In(OTf)₃ over time. (k) Radial distribution functions and (l) coordination numbers of H₂O around In(OTf)₃ at the first 5 ps and last 5 ps.

Moreover, Ab initio molecular dynamics (AIMD) simulations were conducted to elucidate the hydration dynamics of In(OTf)₃ within the electrolyte system. The initial, transition and final configurations of H₂O and electrolyte species around In(OTf)₃ (Figure 2g–i) revealed progressive migration of water molecules toward In³⁺ centers driven by electrostatic forces. Distance analysis relative to the In³⁺ cation (Figure 2j) demonstrated convergent behavior of all four monitored H₂O molecules during the 20 ps trajectory. Notably, initially distal H₂O‐1 and H₂O‐3 underwent a marked decrease in distance to the cation (distance reductions greater than 3.0 Å) by the end of the simulation. Meanwhile, the coordination distances of proximal H₂O‐2 and H₂O‐4 both slightly decreased during the first 2.5 ps and then remained stable. Figure 2k depicts the radial distribution functions (RDF) of H₂O molecules around In³⁺ ions at different time intervals. Compared to the first 5 ps, the RDF peak intensity around 2.5 Å increased markedly during the last 5 ps, reflecting improved H₂O density within the primary solvation shell of In³⁺ ions. Correspondingly, the average coordination number of H₂O molecules within 4 Å of In³⁺ ions also increased significantly during this period (Figure 2l). In contrast, the density and coordination number of H₂O molecules around Na⁺ ions decreased progressively (Figure S15). These results confirm that In(OTf)₃ exerts a strong electrostatic attraction on water molecules, surpassing that of the Na⁺ cation. Complementing these theoretical insights, isotope labeling studies with D₂O coupled with IRMS provided experimental validation. During charging (Figure S16a), D₂O co‐migration accompanied by Na⁺ deintercalation resulted in a reduction of D‐isotope abundance in the post‐charged PBA cathode (NaPBA·(n−x)D₂O). Critically, the introduction of In(OTf)₃ suppressed D‐isotope reincorporation upon discharge (Figure S16b), indicating effective confinement of D₂O within the bulk electrolyte phase and inhibition of the D₂O/Na⁺ co‐insertion pathway. Moreover, compared to the baseline cell, the additive significantly lowers the total hydrogen volume percentage from 1.08% to 0.33% (Table S1), which can be ascribed to an indium‐rich SEI that suppresses the hydrogen evolution reaction [33]. Isotopic mass spectrometry reveals characteristic D signals in both cells, verifying the anodic decomposition of D₂O. However, the nearly identical deuterium‐to‐protium atomic ratios in both systems clearly demonstrate that the decreased D₂O co‐intercalation is not due to In³⁺‐accelerated D₂O decomposition. Instead, In³⁺ cations bind D₂O molecules via coordination interactions and confine them in the bulk electrolyte. Collectively, our combined theoretical and experimental evidence demonstrates that In(OTf)₃ successfully captures crystalline water from the PBA framework, thereby preventing its recurrent co‐migration with Na⁺ ions during cycling, which is a key factor in mitigating structural degradation.

To assess the impact of In(OTf)₃ on the electrochemical performance of practical battery systems, we assembled NaPBA‐H₂O||Na cells (denoted as NaPBA cell) using baseline electrolyte (1 M NaClO₄ in PC/FEC) and electrolyte with In(OTf)₃ additive. Electrochemical impedance spectroscopy (Figure 3a) revealed that the NaPBA cell with additive exhibited a lower AC impedance compared to the baseline cell, indicative of improved kinetic performance. Charge‐discharge profiles (Figure 3b,c) at a current density of 1C (170 mA g⁻¹) showed that the voltage hysteresis for the NaPBA cell with additive was only 0.245 V, significantly lower than the 0.305 V observed for the baseline cell. This result corroborates the impedance data and suggests faster redox kinetics. The baseline NaPBA cell maintained stable discharge capacity during the first 50 cycles, but suffered a notable capacity fade thereafter, with the discharge capacity decreasing from an initial value of 99.7 to 86.9 mAh g⁻¹ after 200 cycles. This fade primarily stemmed from degradation of the low‐voltage plateau, caused by declining redox activity of high‐spin Fe coordinated to nitrogen within the PBA framework. Given that Fe─N bonds (high‐spin Fe) exhibit longer bond lengths and higher reactivity than Fe─C bonds (low‐spin Fe), they are more susceptible to rupture due to structural distortions, resulting in the loss of redox activity. In contrast, the NaPBA cell with additive retained a capacity of approximately 98.6 mAh g⁻¹ after 200 cycles, signifying a significant mitigation of capacity fade related to high‐spin Fe redox processes.

FIGURE 3.

(a) Nyquist plots of NaPBA||Na cells using baseline electrolyte and additive‐containing electrolyte. Charge/discharge curves of NaPBA||Na at different cycles using (b) baseline electrolyte and (c) additive‐containing electrolyte. (d) Long‐term cycle performance (inset: charge/discharge curves) of NaPBA||Na cells using baseline electrolyte and additive‐containing electrolyte. The initial five cycles were conducted at a low current density of 0.1 C to facilitate electrode activation. (e) Comparison of capacity retention of NaPBA||Na cells using additive‐containing electrolyte with other recently reported works. (f) Rate performance of NaPBA||Na cells using baseline electrolyte and additive‐containing electrolyte. (g) In situ XRD intensity contour map of NaPBA||Na cell using additive‐containing electrolyte (M: monoclinic phase; C: cubic phase). (h) Photograph of NaPBA||Na pouch cell lighting the LED patterns.

Long‐term cycling performances further highlight the benefits of the In(OTf)₃ additive. As depicted in Figure 3d, the specific capacity of the baseline NaPBA cell diminished sharply to 43 mAh g⁻¹ after 1000 cycles, with a capacity retention of only 43.17%. In sharp contrast, the In(OTf)₃‐modified cell retained 89.79% of its initial capacity after 500 cycles and 80.63% (82.9 mAh g⁻¹) after 1000 cycles. The capacity decay rate decreased from approximately 0.057% per cycles in the baseline cell to 0.019% per cycles with the In(OTf)₃ additive. Additionally, the baseline cell exhibited sporadic CE fluctuations (e.g., sudden drop to 1.66%, Figure 3d inset), accompanied by charging overcapacity indicative of electrolyte decomposition. This overcharge phenomenon appeared to occur randomly, suggesting that electrolyte decomposition can be associated with the exposure of highly catalytically active vacancy defects. The initially formed cathode solid‐electrolyte interphase (CEI) may possess the ability to passivate defect sites, thereby mitigating severe side reactions related to electrolyte decomposition. However, the repeated insertion and extraction of crystalline water during cycling can lead to the accumulation of PBA structural distortions, resulting in the generation of a significant number of catalytically active defect sites, which subsequently retrigger electrolyte decomposition. Remarkably, the NaPBA cell with In(OTf)₃ maintained a relatively stable CE profile over the long term, whereas the baseline NaPBA cell shows frequent and conspicuous fluctuations in CE after more than 500 cycles. This improved stability likely arises from the additive's ability to inhibit the repeated (de)intercalation of water molecules, as well as its capacity to suppress electrolyte decomposition.

Furthermore, we investigated the effect of additive concentration on the cycling stability of NaPBA||Na cells (Figures 3d and S17). Increasing the additive content from 1 wt% to 3 wt% significantly improves the capacity retention from 78.95% to 89.79% after 500 cycles, owing to the enhanced crystal water trapping and defect passivation effects. However, excessive additive concentration (5 wt%) leads to severe electrolyte decomposition induced by the strong Lewis acidity of In³⁺, resulting in a sharp drop in capacity retention to 57.18%. Therefore, the concentration gradient experiments demonstrate that 3 wt% is the optimal dosage of In(OTf)₃ additive. Additionally, the optimal electrolyte formulation is closely related to the cathode water content and areal loading. As shown in Figures S5, S18, and S19, under high loading (2.3–2.4 mg cm⁻²) and low electrolyte dosage (75 µL) conditions, high‐water‐content cathodes (15–16 wt%) exhibit inferior cycling stability compared to low‐water‐content ones (8–9 wt%) due to insufficient additive supply. This issue can be effectively addressed by either increasing the electrolyte dosage to 150 µL or reducing the cathode loading to ∼1.17 mg cm⁻², both of which can improve the capacity retention to above 96% after 100 cycles at 170 mA g⁻¹. Notably, the mass ratio of In(OTf)₃ to cathode crystal water should be maintained above 18:1 to ensure satisfactory cycling performance, while avoiding excessive electrolyte usage or low cathode loading that compromises the energy density of practical batteries. Besides, we also evaluated the effect of In(OTf)₃ on ionic transport in the electrolyte. As shown in Figure S20a, compared with the baseline electrolyte (5.16 mS cm⁻¹), the ionic conductivity of the additive‐containing electrolyte is slightly decreased to 5.08 mS cm⁻¹, which may be related to the increased viscosity of electrolyte caused by the introduction of In(OTf)₃. Nevertheless, the sodium ion transference number is slightly improved from 0.221 to 0.232 (Figure S20b,c), which is ascribed to the adsorption interaction between In³⁺ cations and ClO₄ ⁻ anions. Overall, considering the low addition ratio of In(OTf)₃, its modulation effect on the electrolyte ionic conductivity and transference number is relatively weak.

In comparison with recent literature, the NaPBA cell with additive demonstrated outstanding cycle life and capacity retention (Figure 3e), highlighting its competitive performance [8, 16, 20, 22, 26, 34, 35, 36, 37, 38]. Further evaluation of rate capability is presented in Figure 3f. Initially, the additive‐containing cell delivered a higher capacity of 61.3 mAh g⁻¹ at 20 C, compared to 52.5 mAh g⁻¹ for the baseline cell. Following three consecutive rate tests, the additive‐containing cell retained a specific capacity of 59.6 mAh g⁻¹ at 20 C, which was nearly threefold higher than that of the baseline cell (20.9 mAh g⁻¹). The improved kinetics and stability observed in additive‐containing cells can likely be attributed to the transition from hydrated Na⁺ to bare Na⁺ (de)intercalation. To examine whether In(OTf)₃ affects phase evolution during charge/discharge, in situ XRD measurements were conducted. Both the additive‐containing cell and the baseline cell underwent a similar monoclinic‐cubic‐monoclinic phase transition, confirming that the additive did not alter the fundamental phase behavior (Figures 3g, S21, and S22). Their differential capacity (dQ/dV) curves also displayed analogous processes, with two redox peak pairs corresponding to low‐spin and high‐spin Fe reactions (Figure S23). Furthermore, the NaPBA||Na pouch cell employing additive‐containing electrolyte was assembled, which effectively powered an LED array as a power source (Figure 3h), demonstrating the practical applicability of the additive‐containing NaPBA pouch cell.

To elucidate the mechanism by which In(OTf)₃ additive improves the cycling stability of NaPBA cathodes, we investigated the structural evolution of PBA during cycling through a multi‐scale characterization approach. Cross‐sectional polishing coupled with scanning electron microscopy (CP‐SEM) enabled direct observation of internal morphology within PBA particles, particularly crack propagation behavior, via cross‐sectional preparation and SEM imaging of NaPBA electrodes. As illustrated in Figures 4a–c and S24, fine microcracks developed within the particles of the NaPBA electrode using the baseline electrolyte after 100 cycles, in contrast to the uncycled initial NaPBA electrode. With extended cycling to 500 cycles, these internal cracks progressively expanded, eventually penetrating entire particles and inducing fragmentation. These cracks serve as a definitive indicator of mechanical failure that may result from the repeated intercalation and deintercalation of crystalline water within the PBA framework. Notably, the NaPBA electrode cycled in the In(OTf)₃‐containing electrolyte exhibited fewer and narrower microcracks than those in the baseline electrolyte, indicating that the In(OTf)₃ additive significantly mitigates structural damage from distortion accumulation in NaPBA particles.

FIGURE 4.

SEM cross‐sectional images of (a) initial NaPBA electrode and NaPBA electrode cycled after 500 cycles using (b) additive‐containing electrolyte and (c) baseline electrolyte. (d) FTIR spectra, (e) high‐resolution XPS spectra, (f) N‐K edge sXAS spectra, (g) Fe‐L edge sXAS spectra of initial NaPBA electrode and cycled NaPBA electrode using baseline electrolyte and additive‐containing electrolyte. (h,i) Fe‐K edge XANES spectra of cycled NaPBA electrode using additive‐containing electrolyte and baseline electrolyte. (j) Cryo‐STEM image of initial NaPBA. Cryo‐STEM image and corresponding energy dispersive x‐ray spectroscopy images of cycled NaPBA using (k) additive‐containing electrolyte and (l) baseline electrolyte.

Beyond morphological assessment via SEM, we employed a range of spectroscopic methods to probe the evolution of valence bond structures in NaPBA electrodes. FTIR spectra (Figure 4d) revealed that, relative to the initial NaPBA cathode, the cycled baseline electrode showed a markedly weakened infrared absorption peak near 2063 cm⁻¹, corresponding to the C≡N stretching vibration in [Fe(CN)₆]⁴⁻ [39]. This attenuation indicates degradation of the NaPBA framework. Additionally, a new broad absorption band emerged around 1000 cm⁻¹, which is likely attributable to C─O vibrational modes associated with byproducts from electrolyte decomposition [40]. Conversely, the additive‐containing NaPBA cathode retained a relatively strong ν(CN) peak after cycling, implying that the In(OTf)₃ additive stabilizes the [Fe(CN)₆]⁴⁻ structure. FTIR spectroscopy primarily probes surface chemical compositions owing to its shallow detection depth. In comparison, Raman spectroscopy exhibits deeper penetration, facilitating the characterization of bulk structural changes. The Raman spectra (Figure S25) related to the ν(CN) functional group were consistent with the FTIR spectral findings described above. Considering that Fe is a core element within the [Fe(CN)₆]⁴⁻ framework, we further examined its valence state and signal intensity during cycling via x‐ray photoelectron spectroscopy (XPS). The high‐resolution Fe 2p spectrum of the initial PBA cathode showed a characteristic Fe²⁺ peak at 709.2 eV (Figure 4e) [14]. After cycling, the Fe²⁺ signal of the cycled cathode remained detectable in the In(OTf)₃‐containing cell, whereas it was significantly attenuated in the baseline cell, suggesting iron loss due to structural degradation of the framework. Given XPS primarily probes the top ∼5 nm of the surface, signal loss may also arise from masking effect by surface side products. To obtain element‐specific insights into the interfacial chemistry, synchrotron soft x‐ray spectroscopy (sXAS) was utilized, offering high sensitivity to light‐element K‐edges (e.g., N) and transition‐metal L‐edges. Figure 4f compared the N K‐edge absorption at approximately 400 eV among different cathodes [22]. The NaPBA electrode in the baseline cell exhibited a significantly weakened N peak, in stark contrast to the initial cathode and those in the additive‐containing cell. This observation indicates nitrogen loss from the degraded [Fe(CN)₆]⁴⁻ species, which is consistent with the attenuation of ν(CN) vibrational modes. The Fe L‐edge spectra (Figure 4g) exhibited a prominent peak at approximately 712 eV, which is attributed to Fe²⁺ in NaPBA [22]. This peak underwent a significant reduction in the baseline sample, confirming structural collapse of the framework. A new spectral feature emerged at approximately 706.3 eV, which is likely attributable to Fe─O bonds present in the side products [41].

To minimize surface interference from organic residues, bulk‐sensitive hard x‐ray absorption spectroscopy was conducted. The normalized Fe K‐edge XANES spectra (Figure 4h,i) display a discernible pre‐edge shift toward higher energy for the baseline electrode compared with the additive‐containing NaPBA cathode, suggesting a higher average Fe oxidation state. In the discharged state, the higher average Fe oxidation state indicates that the reduction of Fe³⁺ to Fe²⁺ during Na⁺ reinsertion is incomplete, implying partial electrochemical irreversibility. Such incomplete redox conversion suggests that a fraction of Fe sites remains oxidized, reflecting insufficient electron compensation within the framework. This effect is likely associated with sluggish Na⁺ diffusion and limited charge transfer, which may originate from structural distortion, lattice defects, or interfacial by‐products formed during cycling. The cryogenic electron microscopy (cryo‐EM) was employed to characterize the microstructural evolution and elemental distribution of NaPBA electrode at the nanoscale. As shown in Figures 4j and S26a, PBA particles from the additive‐containing cell retained intact morphology post‐cycling, similar to initial particles, with no internal cracks. Conversely, particles from the baseline electrode displayed significant through‐cracks and surrounded by fragmented debris (Figure S26b), aligning with CP‐SEM observations. EDS elemental mapping images showed uniform distribution of framework‐critical elements (C, N, and Fe) in both additive‐containing cycled NaPBA particles and initial NaPBA (Figures 4k and S27), matching cryogenic scanning transmission electron microscopy (STEM) images without aggregation or depletion. However, STEM images of the NaPBA particles in the baseline cell (Figure 4l) revealed nanovoids and localized N/Fe depletion in baseline particles, directly supporting the spectroscopic evidence of framework degradation at the nanoscale. Notably, C distribution remained homogeneous, contrasting with localized attenuation of Fe and N signals. This anomaly likely originates from carbon‐rich residues formed during electrolyte decomposition.

Having confirmed that In(OTf)₃ stabilized the NaPBA framework through spectroscopic and electron microscopy characterizations, we next investigated its role in mitigating side reactions. Electrolyte decomposition driven by highly reactive vacancy defects within the NaPBA structure is a primary source of degradation in NaPBA batteries. These vacancies originate from two main sources: inherent defects formed during liquid‐phase synthesis due to rapid nucleation and growth, and structural deterioration caused by accumulated distortions and side reactions during extended cycling. To clarify how In(OTf)₃ suppresses such decomposition, we performed DFT calculations. Optimized adsorption configurations of FEC solvent molecules and OTf⁻ anions on [Fe(CN)₆]⁴⁻ vacancy defects were shown in Figures 5a and S28, respectively. Owing to the inherent negative charge of the [Fe(CN)₆]⁴⁻ moiety, [Fe(CN)₆]⁴⁻ vacancies exhibited a strong preference for adsorbing negatively charged OTf⁻, with an adsorption energy of −2.742 eV. In contrast, the adsorption energy for FEC was only −0.350 eV, indicating that OTf⁻ anions preferentially occupy [Fe(CN)₆]⁴⁻ vacancies over FEC molecules. After OTf⁻ occupied the vacancy, the adsorption energy of FEC near the defect decreased further to −0.257 eV (Figure 5b,c), reflecting a weakened interaction. This preferential adsorption of OTf⁻ passivated the highly catalytic low‐valent iron sites by preventing their direct exposure, thereby suppressing FEC decomposition. These computational findings were supported by morphological analyses of NaPBA electrodes during cycling (Figures S29 and S30). NaPBA cathodes typically incorporated conductive carbon black and PVDF binder to establish an effective electron‐conducting network. Accordingly, pre‐cycling electrodes exhibited NaPBA particles embedded within a conductive carbon and binder matrix. However, after cycling in the baseline cell, the NaPBA electrode surfaces became entirely covered with electrolyte decomposition byproducts, which obscured the NaPBA particles. In contrast, electrodes cycled with the In(OTf)₃ additive maintained exposed NaPBA particles, which were dispersed throughout the conductive carbon network, suggesting a substantial reduction in surface byproduct deposition compared to the control.

FIGURE 5.

(a) Optimized configuration of FEC adsorbed on NaPBA. (b) Optimized configuration of FEC adsorbed on NaPBA with the vacancy site preferentially occupied by CF₃SO₃ ⁻. (c) Comparison of the adsorption energy of FEC, CF₃SO₃ ⁻ adsorbed on NaPBA and FEC adsorbed on NaPBA with the vacancy site preferentially occupied by CF₃SO₃ ⁻. (d) F 1s XPS Spectra, (e) C 1s XPS Spectra, (f) F‐K edge sXAS spectra and (g) C‐K edge sXAS spectra of initial NaPBA electrode and cycled NaPBA electrode using baseline electrolyte and additive‐containing electrolyte. (h) Cryo‐HRTEM images and the corresponding fast Fourier transform pattern of NaPBA after 200 cycles using baseline electrolyte. ToF‐SIMS depth profile of cycled NaPBA electrode using (i) baseline electrolyte and (j) additive‐containing electrolyte. (k) ToF‐SIMS 3D reconstructed mapping images and (l) elemental mapping of cycled NaPBA electrode using baseline electrolyte and additive‐containing electrolyte.

As shown in Figure 5d, the F 1s spectrum displayed two characteristic peaks at 687.9 and 684.5 eV, corresponding to C─F and Na─F bonds, respectively [42]. In the initial PBA electrode, only the C─F signal originating from the PVDF binder was observed. After cycling, both types of PBA electrodes exhibited a new Na─F peak, which is likely due to the formation of NaF, a key inorganic component of the CEI. Notably, a distinct C─F signal remained detectable in the cycled electrode of the additive‐containing NaPBA cell, whereas the C─F signal in the baseline NaPBA cell was significantly weakened. This difference likely arose from thicker byproduct layers in the control group, which masked the underlying PVDF‐derived C─F signal. A consistent trend was observed in the C 1s spectrum (Figure 5e), where the intensity variation of the C─C bond at 284.8 eV mirrored that of the C─F bond in the F 1s spectrum [43]. Moreover, the N 1s signal, which originates from NaPBA, displayed a similar trend of variation (Figure S31). Furthermore, in the cycled baseline PBA electrode, the dominant peak in the C 1s spectrum shifted from the initially predominant C─C component to one characteristic of C─O bonds. This transition suggests the formation of organic species containing carbonyl functionalities within the CEI, which is characterized by C─O functional groups. Besides, the sXAS was performed to gain deeper insights into CEI composition on cycled electrodes. The F K‐edge absorption spectrum in Figure 5f confirmed the presence of NaF on the electrode surface after cycling, consistent with XPS data. Additionally, the C K‐edge spectrum (Figure 5g) revealed a new peak at approximately 290 eV, corresponding to the COO⁻ functional group, indicative of carbonyl‐containing organic species within the CEI [44].

We further probed the CEI microstructure and composition on cycled NaPBA particles using cryo‐HRTEM. In the baseline NaPBA cell, cycled PBA particles were densely coated with irregular deposits (Figure S32), forming a disordered, porous and highly inhomogeneous CEI with thicknesses reaching hundreds of nanometers. HRTEM imaging combined with fast Fourier transform (FFT) diffraction ring analysis identified NaF and Na₂CO₃ as the primary inorganic components (Figures 5h and S33). In stark contrast, cycled NaPBA particles from the additive‐containing cell exhibited clean surfaces with well‐defined boundaries and minimal accumulation of deposits. The corresponding CEI was substantially thinner (<20 nm) and more uniform, with NaF as the predominant inorganic phase, as confirmed by HRTEM and FFT analysis (Figures S34 and S35). Time‐of‐flight secondary ion mass spectrometry (ToF‐SIMS) depth profiling was employed to elucidate the gradient structure and three‐dimensional distribution of components within the CEI on cycled NaPBA particles. Figure 5i,j showed ion fragment intensity profiles as a function of sputtering time for the baseline and additive‐containing cells, respectively. In the baseline cell, dominant NaF⁻ and CHO₂ ⁻ signals, corresponding to inorganic and organic CEI components, appeared within the first 100 s of sputtering. Additionally, CF₂ ⁻ (from the PVDF binder) and FeCN⁻ (from NaPBA particles) signals were initially weak and increased gradually with sputtering time. Until the sputtering time reached 400 s, the signals of all fragments gradually stabilized, indicating that the outer layer of the NaPBA particles was coated with a relatively thick CEI. By contrast, the NaPBA electrode in the additive‐containing cell exhibited enhanced combined signal intensities of CF₂ ⁻ and FeCN⁻ relative to the CHO₂ ⁻ signal, indicative of a thinner CEI layer than that of the baseline electrode. The primary composition of this CEI in the additive‐containing cell was inorganic (especially NaF), with a lower content of organic compounds. Spatial distribution analyses through ToF‐SIMS three‐dimensional reconstructions and two‐dimensional mappings of FeCN⁻, CHO₂ ⁻, NaF⁻, and NaCO₃ ⁻ further revealed the gradient structures and compositional variations after cycling (Figures 5k,l, S36, and S37). In the baseline cell, the PBA surface was coated by an inorganic‐rich CEI mainly comprising NaF and Na₂CO₃, along with organic species containing ─CHO₂ groups, which resulted in the absence of FeCN⁻ signals at the surface. NaF was distributed throughout the entire CEI, while organic components were more concentrated in the outer regions. Conversely, the PBA surface in the additive‐containing cell showed detectable FeCN⁻ signals, indicating fewer surface by‐products. This thinner, predominantly inorganic CEI was more stable and facilitated faster ion migration compared to the thicker, organic‐rich CEI formed in the baseline electrode. Moreover, In³⁺ is deposited on the surface of sodium metal to form inorganic interfacial layers such as InF₃ (corresponding to InF₂ ⁻) (FigureS38), which is also conducive to suppressing parasitic reactions on the sodium metal anode [45]. On the cathode side, FigureS39a shows that Indium element is mainly distributed on the surface of NaPBA. TOF‐SIMS depth profiling reveals that Indium element is only enriched on the cathode surface and its content decreases rapidly with etching depth, which is distinctly different from the uniform distribution of framework elements (Na and Fe) (Figure S39b). This surface distribution characteristic confirms that In³⁺ mainly participates in the formation of CEI rather than intercalating into the NaPBA lattice. Meanwhile, theoretical calculations demonstrate that the migration energy barrier of In³⁺ in NaPBA is significantly higher than that of Na⁺, making its lattice intercalation kinetically prohibited (Figure S40). Furthermore, the differential capacitance curves show no characteristic redox peaks of In³⁺ intercalation in the working voltage window (Figure S23). Collectively, the incorporation of In(OTf)₃ significantly suppressed defect‐induced electrolyte decomposition reactions. The resulting CEI, characterized by its reduced thickness and inorganic dominance, exhibits enhanced structural stability and ion transport efficiency.

3. Conclusion

In summary, we have elucidated a dual‐coupled degradation mechanism in NaPBA cathodes wherein crystalline water‐induced lattice distortion is intimately coupled with defect‐triggered electrolyte decomposition, validated through integrated theoretical calculations and multiscale experimental characterization, particularly by leveraging deuterium isotope labeling to directly track crystalline water dynamics. Specifically, repeated insertion/extraction of crystalline water during Na⁺ cycling drives cumulative structural distortions that ultimately compromise framework integrity, while [Fe(CN)₆]⁴⁻ vacancy defects catalyze parasitic electrolyte decomposition, generating surface byproducts that impede ion transport kinetics. Guided by the rational design criterion of strong H₂O affinity coupled with minimal FEC interaction, we identified In(OTf)₃ as an optimal multifunctional additive operating via synergistic cation/anion mechanisms to mitigate battery capacity fading. In³⁺ cations strongly coordinate with crystalline water molecules, thereby confining them within the electrolyte and suppressing deleterious shuttling, while OTf⁻ anions preferentially adsorb at vacancy sites, passivating these reactive defects and inhibiting solvent decomposition. Consequently, the NaPBA||Na cell incorporating In(OTf)₃ achieves an impressive capacity retention of 80.63% after 1000 cycles at 1C, with the capacity decay rate reduced to approximately one‐third of that observed in the baseline cell. Furthermore, the In(OTf)₃‐containing cell demonstrates excellent rate capability, delivering roughly three times the discharge capacity of the control group at 20C after three cycles of high‐rate testing. Additionally, the fabricated In(OTf)₃‐containing NaPBA||Na pouch cell successfully powers LED arrays, showing its practical application potential. Overall, our work precisely reveals the migration behavior of crystalline water by isotope labeling and expands strategies for suppressing PBA capacity fading through electrolyte additive engineering. We believe that the dual‐coupled degradation mechanism and the rational design of cation/anion synergistic electrolyte additives provide valuable insights for the development of advanced PBA batteries.

Author Contributions

Fangxin Ling: writing – original draft, investigation, conceptualization. Jiefeng Diao: software. Yan‐Ru Wang: methodology. Junyi Dai methodology. Mingze Ma: data curation. Ling Li: data curation. Zhen Li: visualization. Jian Feng: methodology. Ruilin Bai: visualization. Nan Hu: methodology. Junpeng Sun: resources. Hanyu Huo: data curation. Bin Ye: data curation. Xianhong Rui: formal analysis. Graeme Henkelman: software. Yu Yao: conceptualization, writing – review and editing, validation. Yan Yu: supervision, project administration, writing – review and editing, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: anie72773‐sup‐0001‐SuppMat.docx.

Ling F., Diao J., and Wang Y.‐R., et al., Angewandte Chemie International Edition (65, no. 29 (2026): e6167442, 10.1002/anie.6167442.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.