Deciphering the liquid-solid interactions in dealkalization of O3 layered oxides

Abstract

O3-phase layered oxides are among the mainstream positive electrode active materials for advanced batteries due to a stable topological lattice framework and potential for tunability. However, surface residual alkali due to sensitivity hinders their large-scale application. Water washing, an industrial surface residual alkali removal method in lithium-based positive electrode materials, brings about severe issues in sodium-based materials, such as lattice collapse and extensive active alkali metal ion leaching. Here, we propose an interaction mechanism between host solid-phase positive electrode materials and guest liquid-phase solvents, which elucidates the dependent structural degradation on the molecular configuration of dispersed solvents and the alkali metal–oxygen bond covalency during dealkalization. For H₂O (H₃O⁺), self-propagating molecule intercalation into Na slabs and subsequent protonation induce the leakage of Na ions, leading to lattice destabilization. In contrast, the efficient dealkalizing agent—ethylene glycol—prevents further structural degradation due to the constraint of size effect. Based on the time-dependent deterioration of the extended scope of positive electrode materials, an adaptable analytical framework is established for stability assessment against the liquid phase. Our work provides fundamental theoretical guidance for liquid-phase engineering of O3-phase layered oxides.

High-performance battery materials often degrade during standard cleaning processes. Here, the authors identify why water washing causes structural collapse in sodium-based oxides and demonstrate that ethylene glycol can safely remove impurities while maintaining the material’s integrity.

Introduction

Driven by the rapid advancement of high-performance rechargeable battery technologies, the transition to clean energy has significantly developed. Owing to the open two-dimensional diffusion channels, layered oxides have become prominent positive electrode active materials (PEAMs) for high-power and specific energy rechargeable batteries. O3-phase transition metal oxides (TMOs) have emerged as commercially viable positive electrode materials for advanced lithium-ion batteries. In contrast to the limited availability and uneven geographical distribution of lithium resources, the progressive advancement of sodium-based batteries facilitates the development of clean energy technologies, which are not subject to regional disparities.

Nevertheless, the scaled application of O3 phase materials is hindered by challenges associated with the harsh alkaline synthesis conditions and the unstable framework. The resulting surface residual alkali (SRA), which subsequently causes cradle-to-grave issues have become a shackle in its large-scale application¹. Compared with Li-based transition metal oxides (LiTMO₂), Na-based transition metal oxides (Na_xTMO₂) tend to be more reactive with ambient air due to the wider alkali metal (AM) layer space for the larger ionic size (1.02 Å for Na⁺ vs. 0.76 Å for Li⁺), as evidenced by higher SRAs content². In particular, the O3-type NaTMO₂ are highly susceptible to induced increased SRAs, active Na loss, and structural degradation³. This sensitivity restricts their mature applications in several ways: (1) contamination of powder surface⁴; (2) gelation of electrode slurry⁵; (3) performance deterioration⁶; and (4) electrically-induced gas generation at high voltage⁷.

Passive prevention strategies (chemical property tuning⁸ & inert coating layers⁹) and active removal strategies (dealkalization) are commonly used to address SRAs issue. Liquid-phase washing dealkalization (as shown in Fig. 1a), which is synergistically applied with secondary sintering, has become a widely used industrialization strategy of Li-based Ni-rich PEAMs¹⁰. The aqueous dealkalizing solvents, featured by high efficiency but potential destruction, include water or weakly acidic solutions such as boric acid¹¹, nitric acid¹², hydrochloric acid¹³, NH₄H₂PO₄¹⁴, LiH₂PO₄¹⁵, etc.^16,17 The organic solvents mainly include methanol¹⁸, ethanol¹⁹, N-methylpyrrolidone²⁰, etc. Unfortunately, O3-NaTMO₂ encounter challenges in dealkalization due to their sensitivity to aqueous media (Fig. 1a). Researchers noted that NaTMO₂ undergo substantial active Na leaching and layered structures degradation when treated with moisture/water, reflected in a sudden drop in specific capacity^21–25. This failure is often attributed to Na⁺/H⁺ exchange (analogous to LiTMO₂)^26,27, yet it lacks an explanation for profound structural degradation over time. The scientific issue of liquid-solid contact involved in the liquid-phase dealkalization process has impeded the further scale-up application of O3-NaTMO₂ materials.

Fig. 1. Feasibility of liquid-phase dealkalization for O3-NaTMO₂ materials.

Schematic (a), processability (b), electrochemical performance (c), and gas release behavior (d) of NFM materials (air exposed for 24 hours at 50% relative humidity, short for 50% RH) before and after dealkalization. a liquid-phase dealkalization applied on a large scale in Li-based positive electrode active materials (short for PEAMs), while still exposed to difficulties for Na-based PEAMs. Surface residual alkali is abbreviated as SRA. b Slurry mixing and coating images in scale-up electrode process. c Electrochemical performance testing of A-NFM and EG-NFM. The discharge curves for the initial and first 200 cycles are tested at 26 °C. d Charge and discharge voltage profiles of A-NFM and EG-NFM with concentration of CO₂ detected from differential electrochemical mass spectrometry (DEMS), demonstrated in model pouch images. The initial charging curve is tested at 26 °C. The pink-shaded area represents the additional capacity provided by the decomposition of Na₂CO₃.

In this work, we target the coupling liquid-solid interactions during the dealkalization of O3-NaNi_1/3Fe_1/3Mn_1/3O₂ (NFM for short, a typical O3-NaTMO₂ which has been put into small-scale industrial application in HiNa Battery, NATRIUM, etc.), employing comprehensive characterizations to comparatively investigate the dealkalization effect of two typical solvents, water and ethylene glycol. The effectiveness of the ethylene glycol is confirmed through evaluations of processability, electrochemical performance, and safety. A forward/reverse structure profiling is conducted to reveal interdependent failure behavior. Furthermore, a mechanism of liquid-solid interactions, facilitated by solvent molecular intercalation, is proposed to elucidate the coupling interaction process. Based on the established analytical framework, we convert the variable to host O3 layered oxide: extending the scope to more NaTMO₂ and Li-based Ni-rich oxides, employing water as reference solvent, to verify the generalizability of this mechanism. “AM loss” is defined to quantify the extent of structural damage, thereby assessing the stability tolerance of PEAMs against the liquid phase. This work contributes to the optimization and innovation of dealkalization technology. Moreover, it facilitates the scaled application of highly sensitive oxide PEAMs, further enhancing the potential of cheap and widely distributed sodium-based energy storage technology for future clean energy storage applications.

Results

Dealkalization efficacy of ethylene glycol

In this study, the fresh NFM PEAMs were exposed to air (25 °C and 50% relative humidity) for 24 hours²⁸. These treated materials, referred to as A-NFM, were subsequently utilized for implementing liquid-phase dealkalization. To achieve efficient and minimally invasive dealkalization, the selection of solvents necessitates two key criteria: (1) effectiveness in dissolving SRAs, (2) ability to minimize the structural damage of the PEAMs. In response to this, we evaluated several potential solvents based on solubility and viscosity properties (Supplementary Fig. 1). Polyhydric alcohols, rich in hydroxyl groups, generally enhance SRA solubility. We favor those without long alkyl chains to avoid steric hindrance and maintain high polarity. Glycerol exhibits relatively better solubility for Na₂CO₃, while high viscosity renders it impractical for application. Ethylene glycol (EG), commonly used as an automotive coolant, combines favorable fluidity and considerable solubility. In this work, EG is selected as a candidate solvent for comparison analysis with water. The samples dealkalized using EG were designated as EG-NFM.

A comprehensive evaluation was conducted on the processability, electrochemical properties, and gas release of NFM materials both before and after dealkalization, as shown in Fig. 1. The results of the rheology tests on the electrode slurry demonstrated a significant reduction in viscosity after dealkalization with EG (2.76 Pa·s → 2.22 Pa·s at a shear rate of ~50 s⁻¹). Both the storage modulus (G′) and loss modulus (G″) as functions of shear stress exhibit a leftward shift of the thixotropic point, indicating enhanced fluidity of the slurry^29,30. Simulation tests of slurry calendering further illustrate the enhanced processability of EG-NFM (Supplementary Fig. 2). Scale-up validation of the electrode manufacturing process indicated that materials with a high alkalinity surface encounter conundrums due to severe slurry gelation. The jelly-like slurry poses challenges for coating the collector, while the dealkalized positive electrode slurry achieves a uniform coating (Fig. 1b). The scanning electron microscopy (SEM) images of the electrode cross-section also illustrate that the binder failure in the high alkalinity slurry leads to warping and peeling, while the dealkalized sample exhibits strong adhesion (Supplementary Fig. 3).

The electrochemical performance of the EG-NFM was evaluated using half-cells with sodium metal as the reference and counter electrode, with air-exposed material employed as a control sample. Figure 1c shows the discharge curves of the NFM positive electrode before and after dealkalization within the voltage range of 2.0–4.0 V at 0.1 C and 1 C (1 C = 150 mA/g). The initial discharge capacity of A-NFM and EG-NFM was measured to be 98.1 and 115.6 mAh/g, respectively, demonstrating the effective capacity recovery through EG dealkalization. Additionally, capacity retention improved from 67.1% to 73.1% after 200 cycles at 1 C. The comparison of rate capabilities shows a similar trend in cycling stability (Supplementary Fig. 4). Na ion diffusion kinetics were analyzed separately by galvanostatic intermittent titration technique (GITT) and cyclic voltammetry (CV) at different scan rates. Benefiting from the efficient removal of SRAs, EG-NFM exhibits a higher Na⁺ diffusion coefficient (Supplementary Note 2 and Figs. 5, 6). Water dealkalized samples (W-NFM) were similarly investigated for further comparison. The drastically declining capacity (62.5 mAh/g), combined with the poor Na⁺ diffusivity, is attributed to the reduction of active Na⁺ available for reversible extraction/insertion and severe structural degradation. The above results demonstrated the effectiveness of EG for dealkalization in sensitive PEAM while excluding water, given its detrimental effects. It is noteworthy that the specific capacity of dealkalized NFM material remains lower than that of fresh material due to aging treatment. Following actual industrial production, we conducted secondary heat treatment on EG-NFM and performed electrochemical testing (see Supplementary Table 1 and Fig. 7 for details). The results indicate that after liquid-phase dealkalization, sodium replenishment, and subsequent re-sintering, the performance of A-NFM is restored to industrial application standards. Furthermore, the dealkalizing agent can be recovered and reused under vacuum fractionation, which confirms the practicality of liquid-phase dealkalization.

To evaluate the thermal safety of batteries, differential electrochemical mass spectrometry (DEMS) measurements were conducted on the A-NFM and EG-NFM electrodes to monitor the evolution of carbon dioxide, which reflects the dealkalization efficacy on SRAs^31,32. The A-NFM electrode exhibited a sustained CO₂ signal at ~4.0 V and a significant release >4.25 V (decomposition potential for Na₂CO₃). In comparison, the EG-NFM electrode exhibit weak CO₂ signals throughout, suggesting the mitigated gas production in batteries, as demonstrated in the model pouch without swelling in Fig. 1d. Interestingly, due to the sodium ions replenished by the electrochemical decomposition of Na₂CO₃, A-NFM exhibits higher capacity during the charging (shown as the additional charging platform in Fig. 1d and Supplementary Fig. 8). This corroborates the rapid gas production observed within this potential range. Obviously, efficient dealkalization is essential for the entire manufacturing process of the PEAMs as well as the safe operation of batteries.

Surface species and structures of dealkalized NFM materials

The variability of dealkalization solvents results in distinct surface conditions and bulk phase structures of the positive electrode particles. SEM images exhibit air-exposed NFM particles concomitant with nanoscale impurity substances (shown in Fig. 2a). Initially, surface residual sodium compounds, primarily Na₂CO₃ and NaOH, were quantitatively analyzed using acid-base titration methods (Supplementary Note 3)²³. As shown in Fig. 2b, the surface residual Na content of air-exposed samples was up to 43106 ppm, predominantly consisting of Na₂CO₃ (15.23 wt.%, versus 1.75% for NaOH). Conversely, the dealkalized samples demonstrated a substantial residual Na reduction (10776 ppm for W-NFM, 7902 ppm for EG-NFM), indicating the highly effective SRAs removal capability of both solvents. Similarly, efficient dealkalization is evident in the X-ray photoelectron spectroscopy (XPS) results (detailed in Supplementary Fig. 9).

Fig. 2. Surface species and structures of dealkalized NFM materials.

a Scanning electron microscopy (SEM) image of air-exposed NFM particles. b Concentration of the residual Na species on the A-NFM (24 h air exposure), W-NFM, and EG-NFM positive electrode materials measured by acid-base titration. c Time-of-flight secondary ion mass spectroscopy (TOF-SIMS) three-dimensional depth profiling of NaO₂^–, NaCO₃^–, HCO₃^– fragments in A-NFM and EG-NFM particles. SEM images, electron probe microanalysis (EPMA) line scanning and mapping results, HRTEM images of d EG-NFM and e W-NFM, with solvent molecular structure of H₂O and EG.

To further elucidate the characteristics of EG dealkalization, time-of-flight secondary ion mass spectroscopy (TOF-SIMS) was performed to precisely characterize the species near the surface of A-NFM and EG-NFM samples. The depth profiling curves of selected fragments showed that the signals of SRA-related products were significantly weakened after EG dealkalization (Supplementary Figs. 10 and 11). Intensity of NaO₂^– fragments, initially decreased followed by an increase, is influenced by the bulk particle (Fig. 2c). The abnormal increase in HCO₃^– fragments intensity for dealkalized samples is further corroborated by Fourier Transform Infrared (FT-IR) spectroscopy results (Supplementary Fig. 12). Based on the results above, we have summarized the mechanism of EG dealkalization as a synergistic process: (1) the dissolution and liquid phase scouring of SRAs; (2) the tuning of particle surface alkalinity by the weak proton liquid phase environment. The intramolecular hydrogen bond formed between adjacent hydroxyl groups in EG stabilizes the deprotonated conjugate base, thereby enhancing its acidity (pKa: 14.2). This enables the tuning of alkalinity on the oxide particles' surface. This process can be described as an enhanced dealkalization process, as illustrated in Eq. (1). The simulated heat of reaction and quantitative analysis of the product water corroborate this process (shown in Supplementary Fig. 13, Tables 3-4 and Supplementary Data 12-15).

$$\text{NaOH}+{\left(\text{C}{\text{H}}_2\text{OH}\right)}_2\to \text{C}{\text{H}}_2\text{OH}-\text{C}{\text{H}}_2\text{ONa}+{\text{H}}_2\text{O}$$ 1

The morphology and bulk structure of dealkalized NFM particles are illustrated in Fig. 2d, e. The particles exhibit smooth surfaces, indicative of effective SRAs removal by both solvents. However, notable cracks are observed to be parallel to (001) plane in the W-NFM particles (Fig. 2e–i and Supplementary Fig. 14). Cross section polisher (CP) and electron probe microanalysis (EPMA) were employed in tandem to analyze the structural integrity and elemental composition within the positive electrode particles along different orientations³³. The EG-NFM sample exhibits an intact internal structure, as evidenced by the elemental mapping and line scanning scatter plots of Na in both cross-sectional directions, which displayed a uniform Na distribution. In contrast, the W-NFM sample exhibits severe cracks and a significant reduction in Na concentration in both directions (Fig. 2d, e-ii,iii and Supplementary Fig. 15, 16). The results above suggest that liquid-solid mass transfer in W-NFM occurs continuously and follows a specific path. Transmission electron microscopy (TEM) was further conducted for crystal structure analysis. Fast Fourier transform (FFT) patterns of high-resolution transmission electron microscopy (HR-TEM) images confirm the intact layered phase of the EG-NFM sample, even for the surface region (Fig. 2d-iv and Supplementary Fig. 18)²¹. Conversely, W-NFM displays dislocations and stacking faults along (003) planes in the surface region²⁴ (regions Ⅰ and Ⅱ in Fig. 2e-iv). The decreased (003) spacing (5.24 Å→ 5.10 Å) reflects the contraction of Na slabs due to AM ions loss. Based on the results in this section, disparities in the physical and chemical properties of liquid solvents dictate distinctly different mechanisms for the liquid-solid interactions that take place between them and the highly sensitive O3-phase oxides.

Reverse analysis of solvent invasion–induced structural degradation

In-depth structural profiling of the dealkalized material is crucial for investigating the mechanism of liquid-solid interactions. The limitations of X-ray diffraction (XRD) in the structural analysis of water-sensitive O3-phase layered oxides have been shown in numerous studies^{23–25,34–36}. In line with this, the XRD patterns in Supplementary Fig. 19 just display the presence of hydration peaks and partially derived phases without significant features indicative of structural transitions. Conventional characterization techniques encounter difficulties in revealing the occupation of protons within the AM layer. Therefore, time-of-flight neutron powder diffraction (NPD) data were collected on the NFM materials before and after dealkalization to achieve precise crystal structure resolution (Fig. 3a–c). According to the Rietveld refinement results (Supplementary Tables 5–8), both A-NFM and EG-NFM samples exhibit the main O3 structure and minor rock-salt NiO phase caused by the weak protonic solvent (Fig. 3a, c). The uncalibrated peaks at high d-spacing (>2.25 Å) in A-NFM is attributed to the effect of SRAs that are not observed in EG-NFM. In contrast, the degradation of the primary peak of the O3 phase (2.17 Å) in W-NFM sample, accompanied by the emergence of a derived peak (2.20 Å), typically signify the undesirable phase transition (Fig. 3b). Refinement analysis matches the derived peaks with those of the monoclinic O'3 phase, which is calculated to constitute 41.68%, while the O3 phase reduced to 54.03% (Supplementary Table 8). Indeed, the O'3 phase originates from in-plane distortions caused by Na extraction and alterations in the valence states of TM ions^37,38. The aqueous phase–induced sodium extraction elucidates the frequently observed O'3 phase in studies focused on water stability. Limited by the delocalization of intruding substance (weak non-covalent interactions), the monoclinic phase ratios obtained from the refinement can serve as an indirect descriptor to assess the degradation. In addition, the lattice parameter c of W-NFM exhibits a significant increase, suggesting the enhanced repulsive forces between the oxygen layers attributable to sodium depletion.

Fig. 3. Structure characterization and reverse analysis of NFM materials treated with dealkalization solvents.

The experimental and Rietveld refined neutron powder diffraction (NPD) data of the NFM positive electrode materials, including a A-NFM, b W-NFM, and c EG-NFM. d In situ variable-temperature X-ray diffraction (XRD) patterns of W-NFM and EG-NFM. e Structural evolution schematic according to the in situ XRD patterns.

To reveal the mass transfer in NFM material during liquid phase treatment, thermogravimetric analysis (TGA) was conducted on all samples (Supplementary Fig. 20). Initially, desorption of adsorbed water occurs within 30–150 °C (Point 1 before 100 °C). The delayed inflection point of W-NFM (Point 2) is attributed to the higher thermal energy required for the decomposition of intercalated water molecules^23,39. Weight loss between 150–365 °C is primarily due to the removal of proton (H), which distinguishes W-NFM samples with substantial proton occupancy⁴⁰. Decomposition of bicarbonate species also occurs at these two stages^4,23. Decomposition of Na₂CO₃ occurs between 365–540 °C^23,39. Only A-NFM samples continue to lose weight as the temperature further rises, resulting from more SRAs. In contrast, EG-NFM samples present a minimal weight loss (2.49%), suggesting limited mass transfer with the dealkalizing solvent.

For in-depth analysis of structural changes, the reverse structural analysis of dealkalized samples was conducted by the in situ variable-temperature XRD technique. The TGA curves were refined according to the evolution of the main peaks in the XRD patterns and divided into five stages, as labeled in Fig. 3d. For Stage Ⅰ (50–130 °C), the irregular peak can be attributed to the hydrated phase (consistent with Supplementary Fig. 20 and ref. ²⁴). In Stage Ⅱ (130–245 °C), the derived peak to the right of the (003)_L peak indicates the formation of the proton phase in the transition state, given that the main peak is not shifted (detailed in Supplementary Fig. 21). The weight loss at this stage measures 1.82%, corresponding to the partial removal of protons. For Stage III (245–364 °C), the peaks associated with the layered phase exhibit a shift, signifying the complete O3 → O'3 phase transition⁴¹. The structure at this stage is analyzed as a mixture of proton phase and O'3 phase for continuous deprotonation (1.68% weight loss). In Stage Ⅳ (>364 °C), complete deprotonation leads to a more pronounced shift of characteristic peaks associated with the formation of the P3 phase (Na-deficient state). Additionally, cleavage of the main peak at ~520 °C indicates the initial formation of P'3 phase. In contrast, the in situ XRD results of EG-NFM exhibit a consistent phase structure, which suggests that the material maintains its stability during EG liquid-phase treatment and avoids damaging interactions. A subtle shift of the (003)_L peak was observed at >200 °C, consistent with the temperature interval of deprotonation in the TGA curve. This laterally corroborates the slight EG proton attack.

The structure evolution from Stage Ⅰ to Stage Ⅳ is depicted in Fig. 3e. It is important to note that the vacancies caused by internal Na loss are occupied by H₂O molecules (and possibly H₃O⁺) and protons, thus maintaining metastable structures. The proton phase can be discerned in the reverse resolution here, thereby distinguishing it from the derived phase. As the temperature rises, substances that are embedded within the layered structure due to liquid-solid interactions are removed, consequently leading to disruption of the charge balance and AM occupancy as follows: (1) Na site vacancies emerge, inducing lattice distortion that triggers O3 → O'3 phase transition; (2) Na vacancies further accumulate with deep deprotonation, causing interlayer slip and O'3 → P3 phase transition^38,42. The reverse structure analysis presented in this section confirms the objective existence of liquid-solid interactions (including proton and water erosion, Na leaching), and also predicts potential structural evolution. These findings provide valuable insights and robust support for the subsequent mechanistic studies.

Mechanism of liquid-solid interactions

In O3-phase layered oxides, the Na layer is known to undergo Na⁺ insertion-extraction (electrochemical). This layer is also susceptible to the intrusion of foreign substances under external driving forces (aqueous phase), which can lead to its expansion/contraction. Existing research frequently interprets the interaction as an exchange of AM ions with protons (e.g., Li⁺/H⁺ exchange)^10,43,44. This perspective is deficient in the intrinsic driving forces and the pathways through which “exchange” transpires. Specifically, in dealkalization applications, the mass transfer between liquid and solid phases constitutes a complex, time-dependent, and coupled process, distinct from exposure to air. In this section, first-principles density functional theory (DFT) calculations are performed to clarify the liquid-solid interaction mechanism.

First, the insertion formation energy of the two solvents is presented in Fig. 4a. The insertion of water molecules into the layered structure occurs spontaneously (−1.72 eV), corresponding to the expansion of the layer spacing. Conversely, EG molecules are unable to spontaneously intercalate (0.261 eV). This is attributed to the difference in size effect of the two solvent molecules: H₂O molecules have the potential to intercalate into the Na layer, for the transition state H₃O⁺ is comparable to bare Na⁺ (1.0 ± 0.1 Å versus 1.02 Å)⁴⁵. EG molecules with a larger size experience spatial site resistance, which hinders their intercalation. Furthermore, the deprotonation free energies of the embedded solvent molecules were calculated (Fig. 4b). Embedded H₂O undergoes spontaneous deprotonation with the H captured by O slabs (hydrogen bonding dominated), while deprotonation of the embedded EG is less likely to occur. This suggests that H₂O induces a time-dependent continuum of structural damage akin to a “domino effect”. In contrast, EG affects only the near-surface region of the material due to the low free proton concentration. The atomic coordinates of the optimized computational models are provided in Supplementary Data 1–5.

Fig. 4. Liquid-solid interactions mechanism of O3-type layered oxides in dealkalized solvents.

a Insert formation energy of solvent molecules with simulated NFM crystal structure before and after intercalation. b Deprotonation free energy of embedded solvent molecules with corresponding crystal structure model. c Variations in the X-ray diffraction (XRD) patterns of the NFM materials were observed with increasing soaking time in water. d The Na leaching content histogram of the NFM materials varied with growing immersion water time. e Na leaching energy simulations in NFM crystal structure with varying hydration states (0%, 6.25%, 14.5%, 25%, 35.4%, 50%, 75%). f Na–O bond strength as reflected in crystal orbital Hamilton populations (COHP) and integrated crystal orbital Hamilton population (ICOHP) for pristine NFM and 14.5%H₂O-NFM. Liquid-solid interaction mechanisms of O3-type NFM layered oxides in water (g) and ethylene glycol (h), presented schematically as microscopic crystal structures and macroscopic bulk particles.

The structural and compositional characterizations were also employed to verify the time-dependent nature of liquid-solid interactions in aqueous solvents. Fig. 4c shows the structure evolution with the accumulating immersion time. The intensity of the characteristic peaks in the XRD pattern decreases immediately upon exposure to water, stemming from the reduced lattice integrity (Supplementary Fig. 23). It is necessary to clarify that in XRD analysis, only derived phases can be used to calibrate the structure, rather than the more precise mixed phase mentioned earlier. As the immersion time increases (>5 min), (003)_L and (104)_L peaks concurrently decline with the proton phase developing (marked with gray dotted line at 17.38°). The derived O'3/P3 phase peaks emerge sequentially. The two primary peaks completely disappeared after 2 hours, indicating a thorough degradation of the layered structure. As a crucial aspect of the interaction, the Na loss content over water immersion times was monitored using an inductively coupled plasma optical emission spectrometer (ICP-OES). As shown in Fig. 4d, the Na content in NFM material decreases rapidly upon exposure to water. This trend began to decelerate after 15 minutes, reaching 59% after 40 minutes. Indeed, sustained Na leaching drives protonation of NaTMO₂ following charge balance. Conversely, the inert nature of ethylene glycol (EG) significantly limits its impact on the NFM structure (detailed in Supplementary Fig. 24 and Table 9). Excessive immersion presented a trace amount of leached Na content (1.65% versus full occupancy), indicating a weak ion exchange, which imply its absence of a possible competitive reaction with the dealkalization.

Based on the thermodynamic predictions above, the correlation underlying the liquid-solid interaction behaviors (molecular intercalation, protonation, Na leaching) needs to be analyzed. Given the challenges in real-time monitoring of the interaction process, models of NFM in various hydration states are established to calculate the free energy of Na leaching (Supplementary Fig. 25 and Supplementary Data 6, 7). As shown in Fig. 4e, Na exhibits non-spontaneous leaching at low hydration state (0%, 6.25%). With the intercalated H₂O content increases, there is a pronounced tendency for Na to leach from the site (14.5%, 25%). This spontaneous leaching tendency terminates when the hydration state gets higher (35.4%, 50%, 75%). The result above substantiates the correlation between solvent molecular intercalation and Na loss^46,47. Aiming to further elucidate the root cause of Na leaching, we comprehend the electronic origin of the NFM structure in a specific hydration state. Crystal orbital Hamilton populations (COHP) analysis was employed to investigate the bonding nature in Fig. 4f. We conducted a comparative analysis of the Na–O bond orbitals in both pristine NFM and that with 14.5%H₂O intercalation. The reduced integrated crystal orbital hamilton population values (-ICOHP) (0.5528 eV versus 0.7388 eV) indicate that H₂O intercalation significantly weakened the strength of Na–O bonds⁴⁸. This conclusion aligns with the layer spacing simulations in Fig. 4a, given that the expansion of the Na layer weakens its O–Na–O binding. Furthermore, this rapid Na leaching may stem from the electrostatic shielding of water molecules for bonding and the rapid interfacial transfer of hydrated sodium ions^49,50.

In summary, for the highly sensitive O3-type layered NFM oxides undergoing dealkalization, we have delineated a mechanism of liquid-solid interactions facilitated by the intercalation of solvent molecules, as shown in Fig. 4g, h. The interactions transpire continuously between the layered structure and aqueous media due to the spontaneous intercalation and deprotonation of H₂O molecules, as well as the susceptibility of Na leaching in this context. Therefore, structural degradation in the aqueous phase commences on the surface and spreads to the interior of particles as duration increases (depicted by gradient blue blocks in Fig. 4g), which is consistent with earlier findings (Fig. 2e). Conversely, minimal damage to NFM particles by EG can be attributed to blocked molecular intercalation. The potential trace damage is due to the nature of the protic solvent (suggested by the dark blue surface in Fig. 4h). Obviously, for a specific host material, the significant difference in the guest solvents is the decisive factor for the marked disparity in interactions.

Mechanism adaptation—from NaNi_1/3Fe_1/3Mn_1/3O₂ to O3-type layered oxides

Liquid-solid interactions during the liquid-phase treatment of sensitive O3-type layered oxide pose a prevalent challenge for both researchers and industry practitioners. In this section, we extend the proposed analytical framework for liquid-solid interactions to a wider spectrum of O3-type oxides and demonstrate their compatibility. By integrating time-dependent structural analysis with hydration simulations, we aim to elucidate the interaction intensity between the host material and the guest solvent during the dealkalization. Here, water is selected as the guest solvent, for the failure of more positive electrode materials (Li/Na-based) in water is distinguishable and worthy of investigation.

For the host material, Li-based Ni-rich oxides (water-washed in large-scale production) and TM-substituted Na-based oxides (frequently employed to enhance water stability) are considered. We select three representative positive electrode materials, LiNi_0.8Co_0.1Mn_0.1O₂ (LNCM), NaNi_0.2Fe_0.3Mn_0.4Cu_0.1O₂ (NNFMC), and NaNi_0.33Fe_0.33Mn_0.33O₂ (NNFM) for structural/compositional analysis under water immersion, in order to elucidate the essential difference between the liquid-solid interaction process.

A time-dependent study of active AM ion loss and the evolution of XRD characteristic peaks is conducted as shown in Fig. 5a, b. The AM ion leaching content from the three materials is measured after different durations in the aqueous phase (Fig. 5a). The data were fitted as a curve to compare the effects of the host material on the interaction intensity (detailed in Supplementary Table 10). The interaction between the LNCM host and water is primarily driven by Li⁺/H⁺ exchange. This process is confined to the surface of the particles and results in only minimal Li loss, even under prolonged liquid-solid contact. Even exposed up to 24 hours, the main characteristic peaks in XRD patterns exhibit only slight intensity changes (Fig. 5b). This also explains why the washing-sintering strategy can be applied to the large-scale production of Ni-rich positive electrode powder. In contrast, sodium-based hosts undergo significant Na leaching upon liquid-solid contact, a trend that gradually moderates after ~20 min. The self-propagating degradation of the host with poor water stability accounts for this behavior. NNFM samples undergo rapid structural degradation upon immersion in water: (1) Immediate shifted peaks from O3 → O'3 phase transition; (2) Heterogeneous peaks derived after 10 min due to irreversible liquid-solid interactions; (3) Declining diffraction peak intensities from reduced crystallinity. The NNFMC samples exhibited marginally less pronounced deterioration upon water immersion: (1) Retarded decline of peak intensities; (2) Suppressed adverse phase transitions (Supplementary Note 4 and Fig. 26). Apparently, the distinction in AM and TM elements allows the O3-type layered oxides above to demonstrate disparate stability tolerance against liquid-phase, stemming from different interactions between guest solvent molecules and the host oxide lattice.

Fig. 5. Liquid-solid interactions mechanism adaptation for O3-type layered oxides.

a Fitted curves of alkali metal (AM) ion leaking content versus liquid-solid interaction time for three typical O3-type highly sensitive oxides. b Variations in the X-ray diffraction (XRD) patterns of the three alkali metal oxide materials (ATMO₂, A=Li, Na) observed with increasing soaking time in water. Mentioned materials are LiNi_0.8Co_0.1Mn_0.1O₂ (LNCM), NaNi_0.2Fe_0.3Mn_0.4Cu_0.1O₂ (NNFMC), and NaNi_0.33Fe_0.33Mn_0.33O₂ (NNFM). c Intercalation formation free energy of H₂O molecules in simulated crystal structure of positive electrode materials. The graph is divided into regions of (non)spontaneous intercalation with boundaries at 0 eV in free energy. Variations in the layer spacing are labeled on the top, and variations in the AM–O bond energy are labeled on the right for these materials. d Statistical AM loss data of O3-LiTMO₂(Ni-rich), O3-NaTMO₂(Ni/Fe/Mn based and Cu/Ti substituted) and O3-NaTMO₂(Ni/Fe/Mn based). Detailed data sources can be referred to in the Supplementary Table 11.

Furthermore, we extend the analytical framework of the interaction mechanism to encompass multiple positive electrode categories, analyzing it from the perspective of the host material. The free energy of water intercalation in LNCM, NNFMC, and NNFM oxides is simulated in Fig. 5c, coupled with the potential stability against aqueous phase. The atomic coordinates of the optimized computational models are provided in Supplementary Data 8–11. The LNCM is located in the non-spontaneous water intercalation zone (6.56 eV), whereas the Na-based oxides are both situated in the spontaneous water intercalation zone (−0.38 eV for NNFMC and −1.72 eV for NNFM). This result underscores the critical influence of AM with the functionality of TM doping on structure evolution in liquid-solid interactions. The simulated results are also consistent with the interaction mechanism mediated by solvent molecular intercalation. First, the AM element plays a crucial role in determining the intercalation of water molecules. The narrow AM layer spacing in Li-based oxide, owing to the smaller size of Li⁺, significantly impedes the intercalation of water molecules, thus confining phase transition and Li dissolution primarily to the surface layer. Conversely, Na-based oxides, characterized by broader layer spacing, are susceptible to continuous erosion from water molecules. Furthermore, the electrostatic force of TM–O–Na can be modulated through TM substitution, which can strengthen local structure against water attack and limit active Na leaching^4,25,51. In conclusion, the variation in AM dictates the distinct performance of Li/Na-based oxides under aqueous treatment, which depends on the AM–O bonding strength. For NaTMO₂, tuning the TM–O covalency to enhance the Na–O bonding (higher Na–O bond energy) is a suitable approach.

As is evident from the preceding discussion, PEAMs exhibit diverse and complex behaviors when exposed to liquid phases. It is crucial to establish a unified standard for evaluating their stability during liquid-phase processing. In this section, considering the general damage water inflicts upon PEAMs, we define the battery reversible capacity degradation as “AM loss” (which means active AM available for electrochemical reaction in layered oxides) to quantify the extent of structural damage, thereby assessing the stability tolerance of PEAMs against liquid-phase. Through an extensive review of water treatment studies on the aforementioned positive electrode categories, we have compiled representative data, as illustrated in Fig. 5d. Water-treated Ni-rich LiTMO₂ positive electrodes experience less than 5% AM loss. In comparison, Ni/Fe/Mn-based NaTMO₂ positive electrodes experience a drastic AM loss (almost halved) that renders water-treated positive electrodes unavailable. The introduction of water-stabilizing elements (Cu, Ti et. al) can improve the resistance of NaTMO₂ to water-induced structural damage despite unavoidable AM loss. Within the context of O3-type layered oxides examined in this work, “AM loss” serves as a concrete standard for stability tolerance evaluation, which demonstrated its utility for the specific Li/NaTMO₂-H₂O pair. Obviously, the development of this criterion requires a more comprehensive database regarding the diversity of host materials and guest solvents in liquid-solid pairs.

In summary, we have developed a comprehensive evaluation system for liquid-solid interactions of active layered positive electrode materials in liquid-phase media. From the perspective of the O3-TMO host, both AM and TM collectively determine the AM–O bond covalency, which serves as a crucial determinant of lattice stability. From the perspective of the guest solvent, the characteristics of the molecule (such as molecular weight, binding force, polarity, etc.) determine the exchange tendency between the liquid-phase and TMOs. Furthermore, iterative optimization of studies on the liquid-solid interactions for more active materials and solvents can facilitate refinement of analytical models and the expansion of databases.

Discussion

In this study, we investigated the liquid-solid interaction behavior between the O3-type PEAMs and the liquid-phase solvent during the dealkalization process. We propose a mechanism facilitated by solvent molecular intercalation by integrating the interaction behavior of the solvent and the positive electrode materials. Our findings reveal that water produces serious damage to the crystal structure of O3-NFM, including the active Na loss, lattice distortion, and layer slip. In the reverse structure analysis, this degradation is attributed to the intercalation of H₂O molecules and subsequent protonation, leading to phase transformation from O3 to derivative phases. We have elucidated the critical role of H₂O molecular intercalation in facilitating continuous proton invasion and Na leaching. In contrast, EG demonstrates efficient residual alkali removal while maintaining structural integrity, owing to the size-effect-induced shielding of glycol molecules by the AM layer. The proposed liquid-solid interaction mechanism provides insights into the interdependent relationship between solvent molecular intercalation, Na-based oxide protonation, and active Na leaching. The result not only explains the markedly different structural impacts of these two dealkalizing solvents on O3-NFM materials but also correlates these effects with their distinct performance outcomes.

Indeed, our analytical framework demonstrates broad applicability to O3-type PEAMs in liquid-solid contact scenarios. We have extended our investigation to encompass both Na-based oxides with diverse TM compositions and Li-based counterparts. The distinct spatial configurations of AM layers, arising from the significant size difference in AM (Li versus Na), dictate the differential behavior in liquid-solid interactions: the narrower Li layer spacing exposes LiTMO₂ to limited damage in the aqueous phase environment. Selective substitution of TM similarly influences the intensity of interaction by regulating the Na layer. The underlying reason for the above modulation lies in the Na–O bond covalency, which directly correlates with the strength of chemical binding to oxygen. Overall, the strength of the AM–O bond not only dominates the stability of the native structure (local anchoring of AM ions and inhibition of phase transitions) but also profoundly affects the interaction of the oxides with the guest solvent.

Building upon the investigation of liquid-solid interactions, which comprehensively considers the roles of electrode materials and solvents, we have developed a systematic framework for broader application areas. This framework enables the comparison of diverse positive electrode materials and liquid-phase solvents, facilitating the creation of a guiding concept—stability tolerance against liquid-phase, which can describe the mapping relationship between microstructure/performance and different liquid-solid contact pairs. This serves as a guide for researchers and industry in the design of SRA removal strategies for highly sensitive PEAMs. Furthermore, the mass transfer between gas/liquid/solid media, which is nextricably linked to the functionality of the material, is a fundamental process that permeates manufacturing, processing, storage, working, regeneration, and recycling of electrode materials. These include material synthesis, aqueous electrode processing⁵², long-term storage²³, liquid-phase post treatment⁵³, solvation/desolvation of active ions during charge/discharge⁵⁴, and selective extraction in materials recycling⁵⁵. Design principles derived from the mentioned mechanism can be relied upon to further plan the material-electrode-battery manufacturing process.

Methods

Dealkalization of O3-NaTMO2

The O3-NaNi_1/3Fe_1/3Mn_1/3O₂ positive electrode materials (Guangdong Canrd New Energy Technology Co.) were treated separately with dealkalizing solvents. Methods of aging the pristine material are mentioned in the text. Deionized water and ethylene glycol (98%, ShangHai HUSHI Co.) were selected as dealkalizing solvents. NFM powder and solvent (1/50, mass ratio) were placed in a closed container at room temperature at 550 rpm and stirred for 5 min. After immediately solid-liquid separation, the positive electrode powder was dried at 70°C in a vacuum oven for 12 h. The NFM powder was stored in an argon-filled glove box (Shanghai MIKROUNA Mech. Tech. Co.) throughout, except for the liquid phase treatment process. In addition, NFMC was synthesized by high-temperature sintering of precursors (Wuhan Jiana Energy Technology Co.). An excess of 2% Na₂CO₃ (stoichiometric ratio: Na₂CO₃/precursors = 1.02) was added to replenish the loss of Na due to volatilization at high temperatures. The mixture was calcined at 900°C for 16 h and cooled to 200 °C in the furnace. The dealkalization step is the same as above.

Electrolytes and electrode preparation

Electrolyte contained 1 M NaClO₄ in propylene carbonate/ethylene carbonate (1:1, by volume ratio) with 5.0 wt% fluoroethylene carbonate (DoDoChem) as an additive. The slurry comprised NFM active material, Super P, and polyvinylidene fluoride (8:1:1, by weight ratio). The slurry (N-methylpyrrolidone, NMP (purity > 99.9%) as dissolving solvent for the binder) was coated onto aluminium (Al) foil current collector using a stainless steel scraper and dried at 110 °C for 12 h in a vacuum oven. Super P, polyvinylidene fluoride (5130, molecular weight: 1.1 million), and NMP were provided by Guangdong Canrd New Energy Technology Co. Al foil purity > 99.6%, thickness 18 μm. All electrodes were fabricated in air (26 ± 3 °C, 30 ± 5 %RH). As-prepared electrode samples were stored in an Ar-filled glove box with oxygen and moisture below 1 ppm. Electrode cutting equipment (MSK-T10-S) was supplied by MTI Shenzhen Co. The loading mass of the active materials for each NFM electrode is 2.5–3 mg/cm².

In-situ DEMS measurements

The gas evolution analysis during the initial charge/discharge process was carried out by differential electrochemical mass spectrometry (DEMS, QAS100 Li) with a pouch cell model. The pouch cell model was assembled with a hard carbon negative electrode (HC, MTI Shenzhen Co.), carbonate electrolyte (1 M NaPF₆ in DMC:EC:EMC = 1:1:1 Vol% with 5% FEC, from DoDoChem), multi-layer ceramic composition separator, and O3-NaNi_1/3Fe_1/3Mn_1/3O₂ positive electrode as described above. The area weight (single side) of positive electrodes is 14.65 mg/cm². Each pouch was assembled by Z-stacking multiple pieces of 2 negative electrode sheets (coated on a single side) and 1 positive electrode sheet (coated on both sides). The clamping pressure exerted by the fixture on the battery is 0.1–0.3 MPa. During the gas detection, the cells were cycled at 0.1 C (17.58 mA/mm²) within 2.0–4.5 V, and the flow rate of carrier gas Argon was controlled at 1.2 mL/min (ion source: EI, 70 eV; detector: SEM, 1100 v). The entire test was performed under a vacuum of 4.6E−6 Pa at room temperature. The full spectrum was scanned based on the expected mass number of the gas products, and the mass number m/z = 32, 44 was selected as the analyzed fragment for oxygen and carbon dioxide, respectively.

Electrode slurry rheological properties measurements

The rheological properties of the electrode slurry (composed of NFM materials before and after dealkalization) were investigated using a rotational rheometer (Haake MARS Ⅲ, ThermoElectron). In the positive electrode slurry, the mass ratios of PEAMs: Super P: polyvinylidene fluoride were 8:1:1 with a total solid material concentration of 34.5%. The mixed slurry was placed in a thermostatic test chamber at 50 °C for 5 minutes until steady state. The apparent viscosities of the slurry decreased with the decreasing shear rate from 0.01 to 4500 s⁻¹, exhibiting typical shear-thinning. The viscosity value was selected for comparative analysis at a shear rate of 50 s⁻¹. The storage modulus (G′) and loss modulus (G″) of the slurry as a function of shear stress were collected in the range of 10⁻²–10⁴Pa. The thixotropic points were observed according to the intersection of G’ and G”, for storage modulus described the solid response of the slurry at low shear stress conditions, and the loss modulus reflected the liquid response of the slurry at high shear stress conditions.

Cell assembly and electrochemical measurements

For half-cell fabrication, O3-NaNi_1/3Fe_1/3Mn_1/3O₂ (Guangdong Canrd New Energy Technology Co.) served as the working electrode, sodium foil as the counter electrode, and a glass fiber membrane (Whatman GF/D) as the separator. These components were assembled into 2032-type coin cells (Guangdong Canrd New Energy Technology Co.) within an argon-filled glove box (H₂O < 0.01 ppm, O₂ < 0.01 ppm). The glass fiber separator was pre-cut into circular discs with a diameter of 14 mm, a thickness of 675 μm, porosity of ~80%, and an average pore size of 2.7 μm. Sodium foil was prepared by rolling sodium ingots (99.5%, ShangHai HUSHI Co.) to a thickness of 0.3–0.5 mm and punching them into 14 mm diameter discs for immediate use. 60 μl electrolyte was added dropwise to each side of the separator. The assembled cell was then compacted under a pressure of 50 kg/cm² for 10 s to ensure adequate sealing. Galvanostatic measurements were conducted using a Land battery test system (CT2001A, Wuhan LAND Electronic Co.) housed in an environmental chamber maintained at 26 °C. Charge/discharge cycling was performed at a current density of 1 C = 150 mA/g over a voltage window of 2.0–4.0 V. Long-term cycling was carried out at 1 C following 2 cycles at 0.1 C for activation. All other electrochemical tests were conducted at 26 ± 3 °C.

Time-of-flight secondary ion mass spectroscopy (TOF-SIMS) measurements

TOF-SIMS were performed on PHI nanoTOF II Time-of-Flight SIMS. The depth profile was conducted at the high mass resolution mode (2–1850 u) with a pulsed Bi₃⁺ ion beam (30 keV, 2 nA ion current). The sputtering phase was conducted by an Ar ion gun (3 kV, 100 nA) with the area of 400 μm ×400 μm at ~10 nm/min on SiO₂ (sputtering rate), and the analyzed areas are typically 20 μm × 20 μm with the depth of 150 μm.

Neutron powder diffraction (NPD) measurements and Rietveld refinement

NPD characterization was carried out using a deep-penetrating neutron beam with high intensity and spatial resolution for complex phases analysis. Time-of-flight (TOF) NPD data were collected at the beamline of the General Purpose Powder Diffractometer (GPPD) with a 360° rotation stage at the China Spallation Neutron Source (CSNS). The mass of each set of powder samples was not less than 15 g to ensure sufficiently high diffraction intensity. The TOF method was used to collect data over a total scanning time of at least 12 hours. All the diffraction data were refined by the Rietveld method using the Z-code software.

Ex situ and in situ variable-temperature XRD measurements

Ex situ XRD measurement was performed on a Rigaku Ultima IV diffractometer. XRD patterns of NFM powder after different solvent dealkalization treatments were collected over a 2θ range of 5–90° at a scan rate of 5°/min (Cu-Kα radiation source, λ = 1.54 Å). The XRD Rietveld analysis was performed by the FullProf suite. The structural transitions of W-NFM and EG-NFM samples at the temperature range of 25–550 °C were investigated by in situ variable-temperature XRD technique. The in situ test was performed at a ramp rate of 5 °C/min, and XRD patterns were scanned at a rate of 10°/min. Data of W-NFM were collected at nodes 50 °C, 100 °C, 125 °C, 150 °C, 175 °C, 200 °C, 225 °C, 300 °C, 400 °C, and 550 °C, respectively. Data of EG-NFM were collected at nodes 50 °C, 100 °C, 150 °C, 200 °C, 250 °C, 300 °C, 350 °C, 400 °C, 450 °C, and 550 °C, respectively.

DFT calculations

Density functional theory (DFT) computations based on first principles were carried out using the Vienna Ab initio Simulation Package (VASP)⁵⁶, incorporating the projector augmented wave (PAW) method⁵⁷. For the exchange-correlation energy, the generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE) functional was adopted⁵⁸. A plane wave basis set with an energy cutoff of 500 eV was employed, and the geometry relaxation was performed until the forces on each atom converged below 0.03 eV/Å. The Brillouin zone was sampled using a 1 × 1 × 1 k-point grid. Self-consistent iterations were continued until the energy difference between successive steps fell below 10⁻⁵ eV. To prevent interactions between periodic structures, a vacuum region of 15 Å was added along the z direction. A simplified NaNi_1/3Fe_1/3Mn_1/3O₂ model is adopted to focus on solvent molecular intercalation, protonation, and leaching of Na. The model was established by expanding the cell to 4*4 and inserting 16 water molecules and 6 ethylene glycol for the simulated W-NFM model and EG-NFM model, respectively, after removing 67% Na. In the calculation of the leaching energy, different amounts of Na are removed, and the corresponding water molecules are introduced. In the subsequent calculations of the water intercalation energy barriers for LNCM and NNFMC materials, we adaptively optimized the Li/Na atoms removed from the model based on experimental results. The computation of COHP is performed via LOBSTER software⁵⁹.

Author contributions

K.Y.X. and C.S. conceived the project. K.Y.X. supervised the project. W.J.Z., X.K.X., and C.Y.L. synthesized the samples. W.J.Z. and C.S. designed and carried out the experiments, analyzed the results, and wrote the paper. J.F.Z. and A.M.S. provided guidance for the DEMS and ICP-OES experiments analysis. X.Z., M.J.H., and C.H.Y. guided scale-up electrode manufacturing and pouch cell tests. N.L., S.P.Z, and L.D. provided recommendations regarding figure composition and optimization for visualization. Y.R.Y. supported the design and analysis of results for theoretical calculations. All authors discussed the results and co-edited the paper.

Peer review

Peer review information

Nature Communications thanks Gui-Liang Xu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that the main data supporting the findings of this study are available within the paper and its Supplementary information. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70581-2.