Kosmotropic aqueous processing solution for green lithium battery cathode manufacturing

Abstract

The global pursuit of carbon neutrality is driving efforts toward environmentally friendly aqueous electrode manufacturing. However, the inherent chemical reactivity of water with cathode materials remains a challenge to achieving this goal. Here, we design a class of aqueous processing solutions based on the kosmotropic effect. Ion hydration shells in the kosmotropic solutions are restructured to form an ordered state of anion–water clusters and to stabilize local hydration structure adjacent to cathode materials. Consequently, interfacial side reactions and structural degradation of Ni-rich cathode materials are mitigated. The kosmotropic solution-processed LiNi_0.8Co_0.1Mn_0.1O₂ cathode achieve high specific and areal capacities (≥ 205 mAh g^–1 and ≥ 3.7 mAh cm^–2) together with stable cyclability, which are comparable to those of commercial N-methyl-2-pyrrolidone (NMP)-processed cathodes. Techno-economic analysis demonstrates that this kosmotropic solution approach reduces energy consumption in battery manufacturing by 46% compared to the NMP-based process, highlighting its practical and sustainable viability.

Current lithium-ion battery cathode production relies on toxic solvents, raising another environmental concern. Here, authors present a kosmotropic aqueous processing solution, enabling sustainable, cost-effective cathode production while preserving commercial electrochemical performance.

Introduction

Advancements in clean energy storage are crucial for accelerating the transition to a sustainable future. Lithium-ion batteries (LIBs) are widely favored for their high energy density, long cycle life, and balanced electrochemical properties¹. To continuously meet the ever-increasing demand for high energy density, LIBs favor the utilization of transition metal (TM) oxide-based cathode materials (Li_xM_yO_z: M = Ni, Co, and Mn), which are optimized for high-capacity and high-voltage operations. However, the current manufacturing process for energy-dense battery cathodes uses N-methyl-2-pyrrolidone (NMP). This teratogenic and toxic processing solvent is classified as a Substance of Very High Concern (SVHC) under the European Union Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) regulation². Environmental and safety concerns surrounding the NMP solvent have prompted efforts to reduce and ultimately eliminate its use. This issue was further highlighted by the European battery regulation (2023/1542), which was announced in June 2023 with the common goal of minimizing the harmful effects of batteries on the environment. A dry-coating process, which does not utilize processing solvents, has been proposed as a promising approach to address this issue. However, its lack of compatibility with current commercial wet-based electrode production lines hinders its practical implementation in the near future³. Therefore, an urgent transition to scalable cathode manufacturing using environmentally benign processing solvents, which should also be compatible with existing electrode fabrication processes, is essential to comply with environmental regulations and ensure sustainable battery production.

Water, the greenest substance on Earth (Supplementary Fig. 1), offers a promising solution to the challenges above and exhibits advantages for cathode production, including its innate ubiquity, environmental friendliness, and potential for ease of recycling due to the use of water-soluble electrode binders (compared to organic solvent-soluble fluorine binders)⁴. Nonetheless, certain challenges are associated with the aqueous cathode manufacturing process stemming from the interfacial side reactions between water molecules and cathode materials. The relevant chemical reaction is described in the equation⁵ below (1).

$${\mathrm{Li}}_x{M}_y{\mathrm{O}}_z+p{\mathrm{H}}_2\mathrm{O}\rightleftharpoons {\mathrm{H}}_p{\mathrm{Li}}_{x-p}{\mathrm{M}}_y{\mathrm{O}}_z+p{\mathrm{Li}}^{+}+\left(p\right){\mathrm{OH}}^{-}$$ 1

As indicated in this equation, H₂O dissociates into H⁺ and OH⁻, and H⁺ tends to spontaneously exchange with Li⁺ at the cathode material surfaces, resulting in undesired Li⁺ leaching and structural degradation (H_pLi_x−pM_yO_z) of the cathode materials⁶ (Fig. 1a, left). In addition, the simultaneous generation of OH⁻ increases the pH of the slurries, inducing the surface passivation of the cathode materials and corrosion of Al current collectors⁷. The adverse effects of water molecules are particularly significant in high-capacity and high-voltage cathode materials⁸.

Fig. 1. Cathode manufacturing using aqueous processing solution based on kosmotropic effect.

a Interfacial side reaction between water and cathode materials during aqueous electrode processing: water (left) vs. kosmotropic aqueous solution (right). b Schematic of the cathode manufacturing process using kosmotropic aqueous processing solution and a photograph of the obtained cathode.

Previous works implemented to resolve these problems have focused on modifying the surface of cathode materials⁹, adding mild organic acids for in situ surface coating¹⁰, increasing Li⁺ concentration in water to reduce Li⁺ leaching (Le Chatelier’s principle)^5,11, and applying coatings on Al current collectors⁷. Despite these efforts, the underlying challenge of water’s inherent chemical reactivity with cathode materials remains unresolved. To address this limitation and enable the use of water as a processing solvent for cathode manufacturing, it is crucial to understand and manipulate the local hydration environment around the cathode materials.

Herein, we introduce a class of aqueous processing solutions based on the kosmotropic effect, a phenomenon common in biochemistry^12,13, but rarely applied to batteries. Incorporating the kosmotropic anions resulted in the restructuring of the ion hydration shell to form a highly ordered state of anion−water clusters and the regulation of the local hydration structure adjacent to cathode materials (Fig. 1a, right). This reorganization of the ion hydration shell influences water molecules’ thermodynamics and physicochemical properties, thus mitigating the undesired interfacial reactions at the cathode material surfaces during cathode fabrication.

Owing to these advantageous effects, the kosmotropic aqueous processing solution enabled the scalable slurry-cast fabrication of various Ni-rich cathodes (LiNi_0.6Co_0.2Mn_0.2O₂ (NCM622), LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811), and LiNi_0.9Co_0.05Mn_0.05O₂ (NCM9½½)) with stable cycle life comparable to those of conventional NMP-processed cathodes (Fig. 1b). In addition, the kosmotropic solution-processed Ni-rich cathodes exhibited high specific and areal capacities (≥ 195 mAh g^–1 and 3.6 mAh cm^–2, respectively) compared to previously reported aqueous solution-processed cathodes. In addition to the electrochemical benefits, the material cost of the kosmotropic solution is 96% lower than that of the NMP processing solvent. From a comprehensive techno-economic analysis, we propose that the use of the kosmotropic aqueous processing solution can reduce the operational (~ 23%), and capital (~ 95%) costs associated with cathode manufacturing by eliminating the conventional solvent recovery process. Furthermore, this approach substantially reduced energy consumption in electrode manufacturing by 46%, demonstrating its viability as a platform technology for environmentally friendly and cost-competitive electrode manufacturing applicable to various high-capacity electrode materials.

Results

Reorganization of ion hydration shells via Hofmeister effect

Water, comprised of two hydrogen atoms and one oxygen atom, is one of the most elementary molecules on Earth¹⁴. Despite its simple molecular structure, water forms intricate and dynamic configurations via extensive hydrogen bond (H-bond) networks, adopting an open tetrahedral arrangement, unlike other liquids with non-directional van der Waals interactions¹⁵.

Ions, as categorized by the Hofmeister series, can influence the molecular structure and dynamics of water by altering the strength of its H-bond network¹⁶. The kosmotropic ion exhibits a high charge density and can strengthen the H-bond structure¹⁷. On the other side of the series, the chaotropic ion has a large and charge-delocalized molecular structure¹⁸, exhibiting weaker intermolecular interactions with water and disordering the H-bond networks¹⁷. Owing to its molecular polarity, anions exert a more pronounced influence on the structure of water than cations¹⁹. Based on the kosmotropicity of anions, we investigated the effects of four representative Li salts: Li bis(trifluoromethanesulfonyl)imide (LiTFSI), Li trifluoromethanesulfonate (LiOTf), Li acetate (LiOAc), and Li sulfate (Li₂SO₄). Among various cation candidates, Li⁺ was selected to effectively suppress the Li⁺/H⁺ exchange reaction in Eq. (1) described above.

To gain insight into the hydration states of these anions, we calculated their electrostatic potential (ESP) values using density functional theory (DFT) (Fig. 2a). The dipole moment in water molecules causes the preferential coordination of H atoms (H-bond donors) with the highly electronegative sites of anions (e.g., O atoms as H-bond acceptors)^20,21. The DFT calculations revealed that the ESP values of the hydration shells gravitated towards more negative values with an increase in the kosmotropicity of anions. It is expected that the fluoroalkyl group of TFSI⁻ or OTf⁻ can act as an electron-withdrawing group, reducing the electron density of the O atoms. In contrast, the methyl group of OAc⁻ is a weak electron-donating group, resulting in a relatively higher electron density around the O atoms. In SO₄^2–, the electronegativity difference between sulfur (2.58) and oxygen (3.44) causes electron density towards the oxygen atoms, resulting in a higher electron density around the O atoms. Therefore, the ability of H-bond acceptors is strengthened in the order: TFSI⁻< OTf⁻< OAc⁻< SO₄²⁻.

Fig. 2. Reorganization of ion hydration shell by Hofmeister effect.

a, b Electrostatic potential maps (a) and hydration structures (b) of different anions (TFSI⁻, OTf⁻, OAc⁻, and SO₄²⁻) with varying chaotropicity and kosmotropicity. Gray, blue, white, red, yellow, and cyan spheres represent carbon, nitrogen, hydrogen, oxygen, sulfur, and fluorine, respectively. c H-bond number of anion-water clusters and interaction energy between anion and water molecules calculated from the molecular dynamics (MD) trajectories. d Fourier transform infrared (FT-IR) spectra representing O–H stretching mode of aqueous solutions and absorbance difference $\left(\mathrm{\Delta }A=A_x-A_{{\mathrm{H}}_2\mathrm{O}}\right)$ under equimolal Li⁺ concentration at 1.0 m. v_CH indicates C–H stretching vibration of OAc⁻. e Inversion-recovery plots of aqueous solutions obtained from ¹H NMR spectra. f, g 2D ¹H-¹H NOESY spectra of chaotropic solution (1.0 m LiTFSI in H₂O) (f) and kosmotropic solution (0.5 m Li₂SO₄ in H₂O) (g) using a mixing time (T_mix) of 1.33 s at 298 K. Positive phase peaks are depicted in red, and negative peaks are shown in blue with identical contour levels.

The structural rearrangement of anion−water clusters via H-bonds was elucidated by monitoring the molecular dynamics (MD) trajectories (Fig. 2b). The average number of H-bonds and the interaction energies with water molecules increased with the kosmotropicity of the anions (Fig. 2c). Particularly, the H-bond number and interaction energy of SO₄²⁻ with water molecules were 2.5 and 13.7 times higher than those of the chaotropic TFSI⁻.

The strength of H-bond networks was assessed by probing the O−H stretching vibrational energies via Fourier transform infrared (FT-IR) spectroscopy (Fig. 2d). The FT-IR analysis revealed an increase in strong H-bond (~ 3200 cm⁻¹) and a corresponding decrease in weak H-bonds (~ 3600 cm⁻¹) with an increase in the kosmotropicity of anions²². This trend was quantitatively confirmed using the Hildebrand solubility parameter, which captures the cohesive interactions among solution components²³ (Supplementary Fig. 2 and Supplementary Note 1). These findings confirm that the kosmotropic anions restructured the ion hydration shells into an ordered state.

The dynamic anion−water interaction in different H-bond networks was investigated by measuring the longitudinal relaxation time (T₁)²⁴ obtained from the inversion-recovery plots of ¹H nuclear magnetic resonance (¹H NMR) spectra (Fig. 2e). Equimolal aqueous solutions of Li⁺ were employed to eliminate the influence of paramagnetic ⁷Li nuclei, which can alter the T₁ values^25,26. The SO₄²⁻ exhibited the lowest T₁ value, indicating the limited mobility of the water molecules in the presence of a highly kosmotropic anion²⁴.

Two-dimensional ¹H-¹H nuclear Overhauser effect spectroscopy (NOESY) analysis was conducted to investigate the stereochemical relationship of protons in the aqueous solutions (Fig. 2f, g). The ¹H NMR chemical shifts indicated the electronic environments of protons influenced by different H-bond networks. In the kosmotropic solution (0.5 m Li₂SO₄ in H₂O), a downfield shift was observed compared to the chaotropic solution (1.0 m LiTFSI in H₂O), reflecting a reorganization of H-bond networks from free water to more immobilized and bound water states. Notably, the appearance of NOEs in the kosmotropic solution (blue lines in Fig. 2g) provides insight into the spatial proximity (within approximately 5 Å) of protons in the structured H-bond networks. These NOE cross-peaks in the kosmotropic solution indicated the tight intermolecular associations of protons within the anion-water clusters, demonstrating stereochemical rearrangement of the H-bond networks by the kosmotropic anion.

Local hydration structures of kosmotropic/chaotropic aqueous solutions adjacent to cathode materials

To assess the extent of detrimental Li⁺/H⁺ exchange reactions at the interface between cathode materials and water, we investigated the local hydration structures of the aqueous solutions utilizing MD simulations. The anion−water clusters of kosmotropic aqueous solution and chaotropic aqueous solution were theoretically investigated within the distance of 10 Å from the surface of cathode materials (NCM811) (Supplementary Figs. 3, 4, Supplementary Table 1, Supplementary Note 2, and Supplementary Datas 1–16).

Analysis of the number density distribution function (n(r)) of water, anion, and Li⁺ along the distance (r) from the NCM811 indicated that the distribution of water and Li⁺ remained nearly consistent, irrespective of the anion species (Supplementary Figs. 5 and 6). The sulfur atom of SO₄^2– in the kosmotropic solution was notably concentrated near the surface (r < 5.75 Å), whereas the central nitrogen atom of TFSI^– in the chaotropic solution was located in the bulk region (r > 5.75 Å) (Fig. 3a, top)²⁷. The preferential adsorption of SO₄^2– on the cathode materials can be attributed to the high charge density of kosmotropic anions^28–30. Furthermore, the charge distribution (e(r)) of the anions indicates that the oxygen atoms (strong H-bond acceptors) of SO₄^2– are predominantly located in the surface region (Fig. 3a, bottom). Higher concentrations of SO₄^2– near the cathode surface can interact with reactive H⁺ at the NCM811−water interface, consistent with the previously reported result for LiCoO₂ in salt-in-water electrolytes³⁰.

Fig. 3. Local hydration structures of kosmotropic/chaotropic aqueous solutions adjacent to cathode materials.

a Number density distribution function (n(r)) (top) and charge density distribution function (e(r)) (bottom) of anions in aqueous solutions at NCM811surface obtained from MD simulations. b Comparison of water stabilization energy (ΔE_s) in aqueous solutions upon addition of Li salts: LiTFSI vs. Li₂SO₄. c Residence time correlation function (C_i(t)) of water in aqueous solutions and corresponding residence time (τ_water) in the first coordination shell of anions. d Displacement of anions from the surface of NCM811 for 5 ps. e Schematic of the local ion hydration structure at the NCM811 surface.

To further understand the impact of anions, the stability of water molecules was assessed by analyzing the shift in stabilization energy (ΔE_s)³¹ (Fig. 3b and Supplementary Note 3). Compared to the chaotropic LiTFSI with a ΔE_s value of 0.25 kJ mol⁻¹, the kosmotropic Li₂SO₄ displayed a larger ΔE_s value of 0.80 kJ mol⁻¹, highlighting the role of order-making ions on the stabilization of water molecules.

The dynamics of water molecules in the surface region of NCM811 were investigated using the residence time correlation function (C_i(t)) from the MD trajectories (Fig. 3c). The C_i(t) of water in the kosmotropic solution decayed gradually, suggesting that the kosmotropic SO₄^2– can establish long-lasting hydration clusters³². The formation of stable hydration clusters indicates a marked restriction in the free motion of water molecules. The restricted mobility was quantitatively validated by investigating water molecules’ residence time (τ_water). The interaction of water with kosmotropic anions prolonged the residence time by a factor of 2.9 compared to the interaction with chaotropic anions.

To gain insight into the diffusion behavior of ion hydration clusters, the displacement of anions was monitored for 5 ps (Fig. 3d)³³. Despite the low molecular weight and small radius of the geometry of SO₄^2– (Supplementary Fig. 7), its hydration cluster exhibited a smaller displacement than the TFSI^–-centered cluster. These observations confirm that the mobility of both anions and water molecules within the hydration shells was reduced in the kosmotropic solution. The ordered, long-lasting kosmotropic hydration shell is primarily located near the surface of NCM811. This modification in the local hydration structure influences the dynamics of water molecules, potentially mitigating undesired interfacial chemical reactions on the NCM811 surface (Fig. 3e).

Chemical reactivity of kosmotropic/chaotropic aqueous processing solutions with cathode materials

To understand the influence of the hydration structure of kosmotropic and chaotropic aqueous processing solutions on their chemical reactivity with cathode materials, NCM811 particles were immersed in the respective solutions and subsequently collected for analysis (Supplementary Fig. 8). First, we quantified Li⁺ leaching from the NCM811 by measuring the change in the Li⁺ concentration of the immersion solution using inductively coupled plasma-optical emission spectroscopy (ICP-OES) (Fig. 4a, top). The change in the Li⁺ concentration of the kosmotropic solution was significantly lower than that of water or the chaotropic solution, indicating the suppression of Li⁺ leaching from the NCM811.

Fig. 4. Chemical reactivity of kosmotropic/chaotropic aqueous solutions with cathode materials.

a Change in Li⁺ concentration (top) and pH (bottom) of aqueous solutions (from the model study presented in Supplementary Fig. 8). The Li⁺ concentration of the solution was measured using ICP-OES. b Corroded area of the Al current collectors after exposure to various aqueous electrode slurries. The area was estimated from binary SEM images in Supplementary Fig. 9. c Intensity ratios of I₀₀₃/I₁₀₄ and (I₁₀₆ + I₁₀₂)/I₁₀₁ of the NCM811 obtained via X-ray diffraction (XRD) analysis. d–f Cross-sectional scanning electron microscopy (SEM) images and corresponding binary images of aqueous solution-processed cathodes: water (d), chaotropic solution (e), kosmotropic solution (f). Red arrows indicate intergranular cracks of NCM811 secondary particles.

To further assess Li⁺/H⁺ exchange reactions, the pH change of the immersion solutions was measured. The insertion of H⁺ into NCM811 would increase the pH values of these solutions. Notably, the kosmotropic solution exhibited the smallest pH increase compared to water and the chaotropic solution (Fig. 4a, bottom). This pH increase results in the corrosion of Al current collectors, thus increasing electrochemical impedance. The surface alterations to the Al current collectors were characterized after casting electrode slurries prepared from different aqueous processing solutions (Supplementary Fig. 9). Scanning electron microscopy (SEM) images revealed numerous defects and pinholes on the current collectors exposed to the water- and chaotropic solution-based electrode slurries, which can be attributed to high OH⁻ concentrations. In contrast, those exposed to the kosmotropic solution-based slurry exhibited substantially less corrosion (Fig. 4b).

The microstructural degradation of NCM811 was investigated using SEM (Supplementary Fig. 10). After immersion in the water or chaotropic solutions, the secondary NCM811 particles disintegrated into primary particles. In contrast, particles immersed in the kosmotropic solution remained relatively intact. This observation was quantitatively confirmed by analyzing particle size distribution using Zetasizer (Supplementary Fig. 11). The particle size tends to decrease in the order of NMP > kosmotropic solution > chaotropic solution > water. This finding demonstrates that the microstructural degradation of secondary NCM811 particles is ascribed to the undesired interfacial side reactions with water along the crystal grain boundaries³⁴.

Change in crystallinity of the rhombohedral NCM811 particles with R-3m space group was investigated using X-ray diffraction (XRD) analysis (Fig. 4c and Supplementary Fig. 12). The intensity ratio of I₀₀₃/I₁₀₄, which is inversely proportional to the degree of cation disorder, was compared^35,36. The NCM811 particles immersed in the water and chaotropic solution exhibited low I₀₀₃/I₁₀₄ intensity ratios. In contrast, the particles immersed in the kosmotropic solution displayed a high I₀₀₃/I₁₀₄ intensity ratio of 1.43, similar to that of pristine NCM811 (1.46). The hexagonal ordering of the particles was investigated by comparing the (I₀₀₆ + I₁₀₂)/I₁₀₁ ratios^37,38. The NCM811 particles in the kosmotropic solution exhibited reduced (I₀₀₆ + I₁₀₂)/I₁₀₁ values compared to those in water and the chaotropic solutions, indicating the stabilization of hexagonal ordering within the crystal lattice. These results demonstrate the effectiveness of the kosmotropic solution in preserving the structural stability of the NCM811 particles.

Electrochemical performance of cathodes manufactured with aqueous processing solutions

To test the aqueous processed cathodes, we prepared cells comprising of Li metal anodes and NMC811 cathodes with areal mass loading and electrode density densities of 18 mg_NCM811 cm⁻² and 3.0 g cc⁻¹, respectively (See Methods and Supplementary Table 2 for details). The aqueous cathode slurries exhibited dispersion stability similar to those formulated in NMP, suggesting that the incorporation of kosmotropic or chaotropic salts had a minimal effect on the dispersion of electrode slurries (Supplementary Fig. 13). After calendaring, the structures of the electrodes were analyzed using cross-sectional SEM and corresponding electron probe microanalyzer (EPMA) images (Supplementary Fig. 14). Intergranular cracks were observed on the NCM811 particles in the water- and chaotropic solution-processed cathodes (red arrows, Fig. 4d, e). In contrast, the particles in the kosmotropic solution-processed cathodes remained comparatively unimpaired (Fig. 4f). These observations imply that external pressures can amplify the microstructural degradation of cathode materials after the aqueous processing.

The capacity of the water-processed cathodes degraded significantly after 200 cycles because of the active Li⁺/H⁺ exchange^5,39 during the aqueous cathode fabrication (Supplementary Fig. 15 and Fig. 5a). The chaotropic solution-processed cathode showed improved cyclability than the water-processed cathode, however, failed to maintain the cycle retention after 300 cycles. In contrast, the kosmotropic solution-processed cathode delivered a high initial specific capacity of 205 mAh g_NCM811⁻¹ with a stable cycle retention (over 80% after 400 cycles), which is comparable to that of the NMP-processed cathode. We analyzed the cathodes after 400 cycles using electrochemical impedance spectroscopy (EIS, Supplementary Fig. 16). The charge transfer resistance (R_ct) and the film resistance (R_film) of the kosmotropic solution-processed cathodes were comparable to those of the NMP-processed cathodes. Meanwhile, stable cycling performance (85% retention after 300 cycles) was also observed at higher Li₂SO₄ concentrations of 1.0 and 2.0 m (Supplementary Fig. 17).

Fig. 5. Electrochemical performance of NCM811 electrodes prepared by aqueous processing solutions.

a Cycling retention (expressed as discharge-specific capacities of cathode materials) of the cathodes at a charge/discharge current rate of 0.2 C/0.5 C (1 C = 3.74 mA cm⁻²). The cathode material mass loadings were 18.2 mg cm⁻². b Material cost of aqueous processing solutions compared to NMP (a commercial non-aqueous processing solution). c Cycling retention (expressed as capacities normalized by material costs) of the cathodes. d Galvanostatic initial charge/discharge profiles as a function of mass loading of kosmotropic solution-processed cathodes at a current rate of 0.1 C/0.1 C. These profiles were obtained after one formation cycle (0.1 C/0.1 C). e Cycling retention of kosmotropic solution-processed NCM622, NCM811, and NCM9½½ cathodes. The cathode material mass loadings were 18.2 mg cm⁻². f Comparison of kosmotropic solution-processed cathodes and previously reported aqueous solution-processed cathodes in terms of four parameters: specific capacity of cathode materials (x-axis), areal capacity (y-axis), types of cathode materials (heatmap), and cathode material composition (X_AM) of the electrode (diameter). The gray dashed lines represent the mass loading of cathode materials. The number assigned to each circle corresponds to the serial number in Supplementary Table 7 (refs. ^{5,10,11,39,48–61}) The electrochemical tests were conducted within a voltage range of 3.00–4.25 V and at a temperature of 25 °C.

The material costs of the aqueous processing solutions and NMP were analyzed in Fig. 5b (see calculation details in Supplementary Table 3). The kosmotropic Li₂SO₄ salt (0.33 USD g^–1) is more cost-effective than the chaotropic LiTFSI salt (5.52 USD g^–1) owing to its elemental abundance and ease of synthesis: the cost of the kosmotropic solution (0.02 USD g^–1) is merely 4% of the price of NMP. Fig. 5c demonstrates that the kosmotropic solution-processed NCM811 cathode can achieve higher capacities normalized by the material costs over the entire cycles, surpassing the NMP-processed cathode (see calculation details in Supplementary Table 4). This material cost-normalized capacity interpretation highlights a significant economic and performance advantage over current NMP-based cathode manufacturing.

To address potential concerns regarding the presence of precipitated kosmotropic/chaotropic salts in the aqueous solution-processed cathodes, we estimated the specific capacities of the cathodes based on their total mass, including the mass of the precipitated salts (Supplementary Fig. 18 and Supplementary Table 5). Owing to the low molar mass of Li₂SO₄ (55.0 g mol_Li^–1), the kosmotropic solution-processed cathode achieved a specific capacity of 159 mAh g_cathode^–1, closely mirroring the NMP-processed cathode (162 mAh g_cathode^–1). Meanwhile, the chaotropic solution yielded a diminished specific capacity of 142 mAh g^–1_cathode, which can be attributed to the heavier LiTFSI salt (287.1 g mol_Li^–1).

Furthermore, the kosmotropic solution-processed cathode delivered up to 188 mAh g^–1 under a fixed charge current density of 3.0 mA cm^–2, comparable to the NMP-processed cathode (Supplementary Fig. 19). In contrast, the chaotropic solution-processed cathodes exhibited inferior performance, which can be ascribed to the pore blockage as shown in the cross-sectional SEM images (Supplementary Fig. 20) and the reduced particle size of NCM811 that could increase the tortuosity of ion conduction (Supplementary Fig. 11). In contrast, the kosmotropic solution-processed cathode maintained a porous structure. Energy-dispersive spectroscopy (EDS) mapping further confirms the dense distribution of O atoms in the through-thickness direction of the cathode, originating from the precipitated LiTFSI salt. This result is consistent with the molar volume calculations, which revealed that the volume occupied by chaotropic LiTFSI is approximately 8.7-fold higher than that of kosmotropic Li₂SO₄ (Supplementary Table 5a). The ionic resistances of the electrodes (R_ion) were evaluated using EIS in a blocking-symmetric cell (electrode | |electrode) configuration (Supplementary Fig. 21). Under the constrained amount (E/C ratio = 2 g Ah^–1 per each electrode) of liquid electrolyte, which surpasses the total volume of the electrode and separator⁴⁰ (Supplementary Table 6). R_ion values appeared to be equilibrated after 6 h of aging, confirming complete wetting of the electrodes (Supplementary Fig. 21d). The kosmotropic solution-processed cathode exhibited a low R_ion (~ 7.5 Ω cm²), closely resembling that of the water-processed cathode (~ 6.0 Ω cm²). In contrast, the chaotropic solution-processed cathode presented a significantly high R_ion at ~ 80 Ω cm², suggesting that the increased tortuosity stemming from the blocked pores and the reduced particle size of NCM811 further hinders ion transport.

Electrochemical analysis was extended to high-areal-loading NCM811 cathodes (Fig. 5d and Supplementary Fig. 22). The kosmotropic solution-processed cathodes stably maintained the specific capacity of NCM811 (~ 205 mAh g_NCM811^–1) up to an areal mass loading of 50 mg cm^–2 (~ 10 mAh cm^–2). In addition, NCM622, NCM811, and NCM9½½ processed with the kosmotropic solution achieved capacities of 195, 205, and 215 mAh g^–1, respectively, aligning closely with their practical capacities (Supplementary Fig. 23). The cycling stabilities of these kosmotropic solution-processed NCM cathodes are displayed in Fig. 5e, demonstrating the versatile application of the kosmotropic solution.

To highlight the improved cell performance, the kosmotropic solution-processed cathodes, and previously reported aqueous solution-processed cathodes are compared in Fig. 5f, Supplementary Fig. 24, and Supplementary Table 7. Previous studies have focused on enabling the aqueous processing of cathode materials that are less sensitive to water, including low-Ni NCM (Ni < 60%), LiFePO₄, LiMn₂O₄, and Li_0.5Ni_1.5MnO₄. In contrast, the kosmotropic solution-processed cathodes exhibited high specific and areal capacities (≥ 195 mAh g^–1 and 3.6 mAh cm^–2). In addition, these cathodes contain high active material content (≥ 93%), meeting the requirements of practical Li batteries.

Economic and environmental analysis of kosmotropic aqueous cathode manufacturing process

To explore the practical implications of these scientific findings for industrial applications, we designed process flow diagrams (PFDs) for the manufacturing electrode with NMP solvent and kosmotropic aqueous processing solutions (Fig. 6a, Supplementary Fig. 25, Supplementary Table 8, and Supplementary Note 4). The NMP-processed cathodes were dried at high temperatures, followed by a recovery step of NMP vapor via condensation (emphasized within the dashed outline in Fig. 6a). The recycling step is imperative given the high cost of NMP and its environmental impacts upon atmospheric release, and accounts for 45% of the total energy consumption in electrode processing (including slurry mixing, coating, drying, solvent recovery) (Supplementary Tables 9, 10, and Supplementary Note 5).

Fig. 6. Economic and environmental analysis of the kosmotropic aqueous cathode manufacturing.

a Process flow diagram of the cathode manufacturing based on the NMP solvent. b–d CO₂ emission (b), initial capital expenditure (c), and operating cost (d) of the cathode manufacturing process (NMP vs. kosmotropic solution). e Comparison of the major characteristics of the kosmotropic aqueous solution (red line) and NMP (gray line). The criteria of the hazard and green score are described in Supplementary Tables 11 and 12.

In contrast, the kosmotropic solution-based process eliminates the need for a solvent recovery step. Furthermore, the drying process of the aqueous electrode consumes less energy because of its lower vapor pressure compared to NMP. Overall, the kosmotropic aqueous cathode processing results in a 97.4% decrease in energy consumption and a 97.1% reduction in CO₂ emissions compared to the NMP-based process (Fig. 6b and Supplementary Fig. 26). Considering that the electrode processing accounts for 47% of energy consumption in battery manufacturing⁴¹, it implies that the kosmotropic aqueous electrode processing can reduce 46% of the energy consumption, demonstrating its sustainable manufacturing. Moreover, the intrinsic simplicity of kosmotropic aqueous cathode processing translates to a 95% reduction in the initial capital expenditure (Fig. 6c) and a 23% decrease in the operating costs (Fig. 6d).

The economic, environmental, and electrochemical advantages of the kosmotropic solution (0.5 m Li₂SO₄ in H₂O) over the NMP solvent are summarized in Fig. 6e. The kosmotropic solution exhibits reduced health risks, as evidenced by its favorable material hazard index of the solution (Supplementary Table 11). Moreover, the commitment to environmental compatibility is evident from the commendable green score in Supplementary Table 12. These benefits collectively contribute to a significant reduction in CO₂ emissions associated with electrode manufacturing. Utilizing the fluorine-free binder is expected to further enhance the sustainable value chain of the batteries (Supplementary Table 12b).

Discussion

The widespread use of the hazardous chemical NMP in LIB cathode manufacturing has raised significant environmental concerns. To address this imperative issue, we designed a kosmotropic aqueous processing solution to mitigate the adverse interfacial chemical reactions between water and cathode materials by forming highly ordered ion hydration shells. The kosmotropic solution-processed cathodes exhibited a stable cycle life comparable to that of the NMP-processed cathode under practical conditions and achieved their theoretical specific capacity at high areal loadings. By conducting the techno-economic analysis, we demonstrated the advantages of the kosmotropic aqueous processing solution over NMP in electrode manufacturing, achieving significant reductions in energy consumption, CO₂ emissions, and cost savings for sustainable cathode production. We anticipate that this kosmotropic solution-processed cathode approach could be extended to other electrode-active materials and battery chemistries. In addition, incorporating this kosmotropic principle into other electrode components, such as binders and conductive additives may be an effective way to mitigate water activity. This strategy provides insight into environmentally sustainable aqueous processing methodologies and encourages a path toward green and carbon-neutral battery manufacturing.

Methods

Material preparation

Lithium sulfate (Li₂SO₄, 99.5%), lithium acetate (LiOAc, 99.95%), lithium trifluoromethanesulfonate (LiOTf, 96.0%), N-methyl-2-pyrrolidone (NMP, 99.5%), carboxymethyl cellulose (CMC, M_w = 250,000, degree of substitution = 0.9), and polyacrylic acid (PAA, M_w = 45,000, purity) were purchased by Sigma-Aldrich. Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, 99.0%) and polyvinylidene fluoride (PVDF) were bought from Samchun and Solvay, respectively. Carbon black (Super P) powders used as a conductive additive were supplied by TIMCAL. 1 M lithium hexafluorophosphate (LiPF₆) in ethylene carbonate (EC)/diethyl carbonate (DEC) (= 1/1 (v/v)), fluoroethylene carbonate (FEC, 99.9%), and vinyl carbonate (VC, 99.9%) were purchased by Enchem. A water purification system (Direct Q–5UV, Merck Millipore) was utilized to prepare deionized water for aqueous processing solutions. Polyolefin (PE) separators with a porosity of 48% were provided by LG Energy solution. Li-metal were purchased from Honzo. Al current collectors were purchased from MTI. The material cost information used in Supplementary Tables 3–5 was obtained from the suppliers described above.

Structural and physicochemical characterization

The O–H stretching vibration of solutions was measured using a Fourier transform infrared (FT-IR) spectrometer (670, Varian). ¹H chemical shifts and longitudinal relaxation times were recorded using a 600 MHz FT NMR spectroscopy (AVANCE NEO. Bruker). ¹H chemical shifts were referenced to a dimethyl sulfoxide (DMSO)-d₆ as an external standard. The coaxial sample tube was used, with an inner tube containing DMSO-d₆ and an outer tube for the sample. The longitudinal relaxation time (T₁) was calculated with Eq. (2):

$$I=I_0\left(1-{\exp }^{-t/T_1}\right)$$ 2

where I is the peak intensity at time t, I₀ is the saturation intensity, and T₁ is the longitudinal relaxation time. The total recovery time is 6.4 s, allowing the magnetization to recover sufficiently and return to its equilibrium state, thereby minimizing saturation effects^42,43. The ¹H-¹H NOESY spectrum was acquired by co-adding four transients for each of the 256 t1 free induction decays (FIDs) using noesygpphpp pulse sequence. A mixing time of 1.33 s and a recycle delay of 5 s were adopted. The total experimental time was approximately 3.7 h. All NMR experiments were performed at room temperature. The surface and cross-sectional morphologies of the electrodes were investigated using field emission secondary electron microscopy (FE-SEM, S-4800, Hitachi), energy-dispersive X-ray spectroscopy (EDS, JSM 6400, JEOL), and field emission electron probe microanalyzer (FE-EPMA, JXA-8530F, JEOL). To investigate the chemical reactivity of the aqueous solutions with the cathode active materials, NCM811 particles were dispersed in different processing solutions and then subjected to centrifugation to collect the treated NCM811 particles and processing solutions, respectively. Details of the analysis process are provided in Supplementary Fig. 8. The change in pH value and Li⁺ concentration of the obtained processing solutions were measured using a pH meter (SevenExcellence S400, METTLER TOLEDO) and an inductively coupled plasma optical emission spectrometer (ICP-OES, 5110, Agilent), respectively. The collected particles were dried at 60 °C for 12 h, and their X-ray diffraction (XRD) patterns were recorded using an X-ray diffractometer (SmartLab, Rigaku) at 45 kV and 200 mA using Cu Kα radiation. For the particle size distribution analysis, the dried NCM811 particles were redispersed in ethanol and analyzed using a Zetasizer (ELSZneoZS, Otsuka). To observe the surface of Al current collectors after the electrode coating, the electrode active layers were gently removed from the electrode using a cotton swab soaked in dimethyl carbonate (DMC) solvent. The area of Al corrosion was quantified from binary images generated by applying 34% of the threshold program (Image J) to the SEM image of the Al current collectors. Cross-sectioned samples were thinned using a focused ion beam (FIB, Helios Nano Lab, FEI) to analyze the microstructure of NCM811 particles.

Fabrication of electrodes, cell, and electrochemical characterizations

The electrode slurries were prepared with a composition of LiNi_0.8Co_0.1Mn_0.1O₂(NCM811)/carbon black/binder = 95.5/2.0/2.5 (w/w/w). Details of the electrode composition, including residual salts, are provided in Supplementary Table 2. For aqueous solution-processed electrodes, CMC and PAA were used as binders at a composition ratio of 1/1 (w/w). The salt additives were dissolved in water for 2 h and then the CMC and PAA binders were dissolved before the formulation of electrode slurries. The binder contents in the processing solutions are 3 wt.% and 10 wt.% for CMC and PAA, respectively. No additional water is added during the electrode fabrication, leaving 47 wt.% of water in the kosmotropic electrode slurry. The electrode slurries were cast onto Al current collectors and dried at 60 °C for 6 h, followed by roll-pressing at 120 °C. The electrode fabrication process was identical regardless of the mass loading and type of cathode materials. For the NMP-processed electrodes, PVDF was used as the binder, and dissolved in NMP. The concentration of the binder solution was set at 10 wt.%. The electrode slurries were cast onto Al current collectors and dried at 120 °C for 6 h and then roll-pressed at 120 °C. The density of the fabricated electrodes was set to 3.0 g cc^–1, and the electrodes were vacuum dried at 120 °C for 12 h before the cell assembly.

For the electrochemical tests, CR2032 coin-type cells (Li-metal anode (thickness = 100 μm and diameter = 16 mm)|polyethylene (PE) (thickness = 16 μm and diameter = 18 mm)|Ni-rich cathode (diameter = 12 mm)) were assembled in an Ar-filled glove box. Thereafter, 1 M LiPF₆ in EC/DEC)( = 1/1 (v/v)) with 10 wt.% of FEC and 2 wt.% VC was used as a liquid electrolyte. The electrolyte was injected into the cell with an electrolyte mass/electrode capacity (E/C) ratio of 5 g Ah^–1 to explore the cell performance for practical energy-dense Li-metal cells⁴⁰, if not specified. After the cell assembly, a rest period of 12 h was allowed to ensure complete wetting of the electrodes. The cells were then subjected to one formation cycle at a current rate of 0.1 C (1 C = x mA g⁻¹_{active material} (x = 195, 205, and 215 for NCM622, NCM811, and NCM 9½½ active material, respectively). The cells were cycled at a charge/discharge current rate of 0.2 C/0.5 C and a voltage range of 3.00–4.25 V in 25 ^oC environmental chamber, if not specified. The resistances of the electrodes were measured using electrochemical impedance spectroscopy (EIS) analysis in a frequency range that varies from 10⁶ to 10⁻² Hz and an applied voltage amplitude of 10 mV using potentiostat (VSP-300, Bio-Logic). 10 points per decade of frequency were acquired. The EIS analysis of the full cells was conducted after discharge and rest at an open-circuit voltage (OCV) for 1 h. To analyze the ion resistance inside the electrodes (R_ion), a symmetric cell configuration (electrode||electrode) at a fully lithiated state was used. The charge/discharge performance of the cells was investigated using a cycle tester (PESC 05, PNE Solution).

Density functional theory calculations

All density functional theory (DFT) calculations of anions (i.e., TFSI^–, OAc^–, OTf^–, and SO₄^2–) were performed using the Gaussian 16 program. Becke, 3-parameter, Lee-Yang-Parr (B3LYP) hybrid functional^44,45 and Valence triple-zeta polarization basis set (def2-TZVP) basis sets⁴⁶ were used for all calculations with the Conductor-like Polarizable Continuum Model⁴⁷ (CPCM) for the water solvation effect. The self-consistent field (SCF) was converged to 1.0e^–6 E_h. The geometry optimization convergence threshold was set to 4.5e^–4 E_h/bohr for maximum force, 3.0e^–4 E_h/bohr for root mean square (RMS) force, 1.8e^–3 bohr for maximum displacement, and 1.2e^–3 bohr for RMS displacement. Electrostatic potential maps were generated using Gaussview with a surface isovalue of 0.0004 and an energy value range of –0.06 E_h to –0.364 E_h.

Molecular dynamics simulations

All-atom simulations were performed to confirm the kosmotropic/chaotropic anion effect on the aqueous solutions. Each bulk aqueous solution comprised 4476 water molecules and 80 LiX molecules (X: TFSI^–, OAc^–, and OTF^–) to set 1 N Li⁺ concentration. In the Li₂SO₄ bulk solution system, 4476 water molecules and 40 Li₂SO₄ molecules were used. The solvent system was composed of 4476 water molecules as a control group. LiTFSI, Li₂SO₄ bulk solutions, and solvent system were placed on the z-axis surface of the NCM811 cathode for the cathode-electrolyte interphase (CEI) system. The x and y dimensions of the CEI system were fixed at those of the NCM811 cathode shown in Supplementary Fig. 4 (47.960 and 45.284 Å, respectively). To prevent interactions between the solution and the NCM811 cathode above the periodic boundary conditions, a vacuum layer of 50 Å was added to the system along the z-axis. An 80 ns NVT production run was performed at 298 K after a 10 ns NPT run for the bulk system. For the CEI system, a 10 ns NVT run was performed at 353 K, after which 50 ns NVT production run was conducted.

To calculate the stabilization energy of water in the surface region, solutions were positioned only on the surface region of the NCM811 cathode. For the surface electrolyte system, the kosmotropic solution used for the stabilization energy calculation comprised 363 H₂O molecules and 1 Li₂SO₄ molecule, whereas the chaotropic solution was comprised of 360 H₂O molecules and 2 LiTFSI molecules. The lower number of water molecules in the chaotropic solution is because the TFSI^– ion is larger, occupying more area. The total system comprised the solutions on top of the NCM811 cathode. The NVT production run was performed at 298 K after a 10 ns NPT run for surface bulk systems. After running a 10 ns NVT run at 353 K, a 20 ns NVT production run was performed for the total system. The detailed information is described in the supplementary information.

Author contributions

J.-H.K., W.-Y.K., and S.-Y.L. designed this work. J.-H.K. and W.-Y.K. performed the experimental characterization and electrochemical tests. S.B.K. and N.J.P. performed the theoretical calculations. J.D.K. performed the cost analysis of the electrode manufacturing process. S.P.H. conducted the FT NMR analysis. K.R. and S.-J.L. assisted with the data analysis. J.H.K., W.B.L., and S.-Y.L. supervised the overall project. All authors contributed to finalizing the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data generated in this study are provided in the Supplementary Information and Source Data. All other data are available from the corresponding author upon request. Source data are provided in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-56831-9.