Manipulating diffusion energy barrier in graphene oxide subnanochannels for precise radionuclide separation

Abstract

Nuclear energy plays a crucial role as a clean energy source in modern society. The use of nuclear energy will result in the generation of a large amount of radioactive nuclear wastewater. Separation of nuclides from radioactive wastewater is crucial for the safe disposal of nuclear wastes and the sustainable development of resources. However, it remains a great challenge to achieve precise separation between different radionuclide ions due to their similar properties. Herein, we constructed a radiation-resistant graphene-based membrane via ethylenediaminetetraacetic acid (EDTA) functionalization with highly stable and aligned two-dimensional subnanochannels, which exhibits adjustable ion diffusion energy barrier and ultrahigh radionuclide ion selectivity. The functional groups within the GO-EDTA channel exhibit strong affinitive binding interactions with Sr²⁺ and La³⁺. The mono/multivalent metal-ion selectivity up to 485 and 1300 for Cs⁺/Sr²⁺ and Cs⁺/La³⁺, respectively, outperforms other reported membranes. Besides, the channel can still maintain stable separation performance under irradiation conditions. Furthermore, using quartz crystal microbalance, we break down the contributions of partitioning at the pore mouth and intrapore diffusion to the overall energy barrier for salt transport, indicating that the precise separation of ions is achieved by regulating the diffusion energy barrier. This work provides a mechanism for the design of membranes with high ion-ion selectivity and demonstrates the application potential of nuclear resource recycling.

Graphical abstract

Keywords

Radionuclide separation / Ion separation / Subnanochannels / Energy barrier regulation / Confined ion transport

Highlight

	● EDTA-functionalized GO nanochannels with radiation resistance were prepared.
	● The subnanochannels exhibit high Cs+/La3+ selectivity of 1300.
	● A pure CsCl product with a purity of 99.5% was obtained via electrodialysis.
	● Revealing the ED is the main energy barrier for ion transmembrane transport.

1 Introduction

As a pivotal energy source in modern society aimed at reducing carbon emissions, nuclear energy inevitably generates substantial volumes of radioactive nuclear wastewater due to its operational characteristics. Radioactive wastewater (Xiao et al., 2017; Jin et al., 2025) is a persistent byproduct of nuclear power generation and post-accident remediation, whose management has become a major global environmental and technical challenge (Smith et al., 2023; Lin et al., 2024). Effective treatment is not only crucial for current environmental protection but also vital for the long-term sustainable development of nuclear energy (Xiao et al., 2017; Sun et al., 2024; Kumar et al., 2025). Among the diverse radionuclides requiring stringent control, cesium-137 (¹³⁷Cs⁺) is particularly challenging (Chen et al., 2020; Tang et al., 2022). Its high fission yield, long half-life (~30.2 years), high environmental mobility due to its solubility, and intense gamma emission pose significant long-term risks to ecosystem integrity and human health (Manos et al., 2006; İnan et al., 2026). Consequently, the selective removal and recovery of Cs⁺ from complex wastewater streams is imperative (Rauwel and Rauwel, 2019; Fuller et al., 2022). This objective aligns with the evolving paradigm of the circular nuclear economy, under which recovered ¹³⁷Cs⁺ can be utilized in high-value industrial applications such as radiotherapy sources and industrial gauges, thereby transforming hazardous waste into a resource and minimizing ultimate disposal volumes (Yan et al., 2023; Tang et al., 2024). Traditional separation techniques, including chemical precipitation, solvent extraction, and ion exchange, have been applied to cesium remediation (Wang and Zhuang, 2019; Chen et al., 2020; İnan et al., 2026). However, these methods often face limitations when applied to high-volume, low-concentration wastewater containing complex saline matrices. Challenges include insufficient selectivity toward competing ions, high operational costs, and the generation of secondary solid waste fractions.

Membrane-based separation technologies (e.g., nanofiltration, NF; reverse osmosis, RO) excel in operational continuity, energy efficiency, and scalability, with separation predominantly governed by size exclusion and charge effects (Shen et al., 2022a, 2022b; He et al., 2024; Xiao et al., 2024; Dai et al., 2025). Nevertheless, for target substances with large and polydispersed sizes, charge effects are generally negligible. This limitation significantly affects the selective recovery specificity of such membranes for monovalent Cs⁺, particularly in the presence of multivalent cations (e.g., Sr²⁺, La³⁺), which have similar hydration radii or exhibit analogous membrane surface interaction behaviors (Wen et al., 2016; Liu et al., 2019; Peng and Zhao, 2021). Meanwhile, the existing membrane materials have limited radiation resistance, which leads to damage to the membrane structure during the treatment of radioactive nuclear wastewater, thereby preventing them from exerting their intended separation functions. These inherent limitations have driven the exploitation of func-tionalized membranes, which integrate selective binding sites within sub-nanoscale transport channels (Esfandiar et al., 2017; Lu et al., 2020; Xu et al., 2023, 2024). Two-dimensional materials such as graphene oxide (GO) have emerged as highly promising platforms due to their tunable interlayer spacing and rich surface chemistry, enabling the construction of structurally well-defined layered nanochannels (Zhang et al., 2015; Chen et al., 2017; Morelos-Gomez et al., 2017; Guan et al., 2023; Wang et al., 2023; Li et al., 2024). Nonetheless, pristine GO membranes are typically limited by interlayer swelling in aqueous solutions and a lack of inherent selectivity for target ions such as Cs⁺, thus requiring targeted functionali-zation to simultaneously confer stability and specificity (Liu et al., 2015; Alemayehu et al., 2022; Lin et al., 2023; Wang and Liang, 2024).

Herein, a radiation-resistant EDTA-functionalized graphene oxide (GO-EDTA) membrane with highly stable poly-carboxylate modification and well-ordered sub-nanoscale channels was constructed, which exhibits exceptional selectivity for monovalent over multivalent radioactive ions, surpassing the performance of previously reported membranes. Subsequently, employing experimental methods (quartz crystal microbalance-QCM and energy barriers test) in conjunction with molecular simulations, we demon-strated that this monovalent/multivalent ion selectivity stems from differences in EDTA’s binding energies for ions of different valence states, which leads to a significant increase in the transmembrane diffusion barrier for multivalent ions, as revealed through the precise decoupling of partition and diffusion barriers. Furthermore, the practical efficacy of the GO-EDTA membrane has been validated by its successful application in recovering cesium from simulated nuclear wastewater, yielding a CsCl product with a purity exceeding 99.5% via electrodialysis. This study not only introduces a high-performance separation membrane material but also provides an in-depth elucidation of the fundamental mechanisms of ion transport in functionalized subnanochannels, offering a valuable framework for the design of advanced membranes for sustainable nuclear wastewater treatment and critical resource recovery.

2 Experimental section

2.1 Quartz crystal microbalance (QCM) testing

The partitioning of salt ions within the GO-EDTA layers was investigated using a QCM (Biolin Scientific, Sweden). First, the active layer was isolated from PES membrane coupons and transferred onto 5-MHz gold sensors (QSX 301). In brief, the composite membrane was placed in a DMF solution to dissolve the PES substrate. The resulting free-standing GO-EDTA film was then picked up using the gold sensor and dried in a fume hood for 24 h to eliminate residual organic solvents prior to testing.

Two parameters of the GO-EDTA layer coated on the QCM sensor were determined using a QCM: the areal mass of the layer itself (m_AL, ng/cm²), and the amount of partitioned salt partitioned into the layer (m_salt, ng/cm²) during exposure to salt solutions. For each sensor, the m_AL value was obtained as the frequency-derived mass difference between two air-phase QCM readings—taken before and after isolating the active layer. The frequency change (Δf) was converted to mass change (Δm) using the Sauerbrey Eq. (1):

(1)\mathrm{\Delta }m=-C\frac{\mathrm{\Delta }f}{n},

where Δm represents the areal mass change of the active layers (ng/cm²), Δf denotes the resonant frequency shift, C signifies the mass sensitivity constant (17.7 ng/(cm² ·Hz) at f = 5 MHz), and n is the harmonic number (n = 5). All measurements were conducted at a crossflow rate of 0.03 mL/min and a system temperature of 25 °C.

2.2 Determination of the partition coefficient

The partition coefficient (K) is defined by Eq. (2):

(2)K=\frac{c_m}{c_b},

where c_m and c_b represent the salt concentrations in the membrane and bulk solution, respectively. Given c_b, once c_m is obtained, K can be determined. The membrane concentration c_m is derived from the measured mass of salt partitioned into the membrane (m_s,p) using Eq. (3):

(3)c_m=\frac{m_{s,p}}{M{W}_s{V}_p},

where MW_s represents the molecular weight of the salt, V_p denotes the membrane volume-calculated from the membrane’s area and the thickness. The partitioned salt mass, m_s,p, was measured using QCM. Moreover, by conducting these partition tests at various temperatures, the ion partition energy barrier was extracted using the Arrhenius equation.

2.3 Transmembrane energy barrier test

The transmembrane energy barrier for various membranes and ions was evaluated by examining the temperature dependence of flux using the Arrhenius equation.

(4)F=F_0{\mathrm{e}}^{-\frac{E_a}{RT}},

where F denotes the flux, E_a represents the activation energy, R denotes the gas constant, T signifies the temperature. Based on transition state theory (TST) and the Eyring equation, the activation enthalpy (ΔH*) and entropy (ΔS*) were further derived.

2.4 Electrodialysis method for treating nuclear wastewater

In an electrodialysis transport experiment using simulated nuclear wastewater, 100 mL of a mixed solution containing 0.1 mol/L Cs⁺ and 0.01 mol/L each of Sr²⁺ and La³⁺ was injected into the feed tank (facing the selective layer). The concentrated compartment was filled with 100 mL of CsCl solution, which also served as the electrolyte. A constant current of 0.018 A was applied. Permeate concentrations in the concentrated chamber were was determined using ICP–OES. The solute diffusion permeability was then derived using Eq. (5):

(5)J=\frac{\mathrm{\Delta }C}{S\times \mathrm{\Delta }T},

where C is the change in solute concentration in the concentrated chamber, S represents the effective membrane area, and T refers to the test duration.

The selectivity S_ij of cation i over cation j was then calculated as the ratio of their fluxes:

(6)S_{ij}=\frac{J_i}{J_j},

where J_i is the flux of cation i.

2.5 Density functional theory calculation

ORCA 6.1.1 package was used to calculate the binding energy between EDTA anion and the metal ion hydrated cluster. The hydration cluster conformational search employed the GFN-xTB2-based explicit solvation algorithm in ORCA 6.1.1. Geometry optimization and frequency analysis were performed using the B97-3c composite method. No imaginary frequencies were present in the optimized geometry. The single-point energy was calculated by ωB97M-V functional with the minimal argumented def2-TZVP basis set and def2-TZVP/J auxiliary basis set. The solvation effect of water was considered by the Solvation Model Based on Density (SMD). The binding energy (BE) was calculated Eq. (7):

(7)\begin{array}{rl}{E}_{\text{binding}}& =E_{\text{complex}}-(E_{\text{metal hydratecluster}}\\ & +E_{\text{EDTA Anion}}\pm N\times E_{\text{water}}),\end{array}

where E_complex denotes the single point energy of the complex, E_{metal hydrate cluster} refers to the single point energy of metal ion hydrated cluster, E_{EDTA Anion} indicates the single point energy of EDTA anion, and E_water corresponds to the single point energy of single water molecule. N is the number of water molecules involved in the binding process. A positive sign is used when water molecules are released during binding, while the negative sign is employed if water molecules are consumed. The calculated binding energy is expressed in kcal/mol.

2.6 Molecular dynamics simulation

Molecular dynamics (MD) simulations were performed using the GROMACS 2019 software package under constant number of atoms, volume, and temperature (NVT) conditions (T = 300 K). Water molecules were simulated using the TIP3P model, while EDTA molecules, La³⁺, Cl^–, and Cs⁺ were described using the OPLSAA force field. The cutoff radii for both the Lennard-Jones (LJ) interactions and electrostatic interactions were set at 10 Å, and long-range electrostatic interactions were handled using the particle-grid Ewald (PME) method. The initial simulation box size was 4 nm × 10 nm × 2 nm, with a channel approximately 2 nm long and 1.2 nm wide (calibrated based on the distance of OH groups in GO), and two EDTA molecules were inserted into the channel. Periodic boundary conditions were applied. The oxidation ratio of GO was controlled at O/C = 10%. The temperature was controlled by a Nosé–Hoover thermostat. The Newtonian equations of motion were integrated using the velocity Verlet algorithm with a time step of 2 fs. Before the ions were introduced, the system underwent a 25 ns equilibration period. After the ions were added, each system was simulated for an additional 50 ns. To avoid additional osmotic pressure caused by differences in ion concentration, Cs⁺ ions with a concentration of approximately 2.5 mol/L were added to the inlet side of the channel, La³⁺ ions at 2.5 mol/L, and Cl^– ions to maintain the charge neutrality of the system. Plotting and analyses were performed using the VMD software.

3 Results and discussion

3.1 Preparation and characterization of EDTA-functionalized GO channels

A suspension of graphene oxide (GO) nanosheets was prepared by dissolving GO powder in water, followed by ultrasonic treatment and centrifugation. The resulting nanosheets exhibit an ultra-thin thickness of approximately 1 nm, confirming the single layer of GO nanosheet (Zhang et al., 2023; Liu et al., 2024) (Fig. S1). With a flat, defect-free structure and a flake size of approximately 1 μm, such nanosheets are well-suited for the fabrication of uniformly aligned laminar membrane (Fig. S2).

EDTA is a typical negatively-charged metal polycarboxylate chelating agent that can be easily self-assembled with GO nanosheets, subsequently forming a laminar membrane through a simple filtration process (Gouaux and MacKinnon, 2005; Peng et al., 2018; Xu et al., 2023). Results from AFM, FTIR and zeta potential measurements indicate that EDTA molecules have been successfully grafted onto the surface of GO nanosheets (Figs. 1(a), S3 and S4), and the SEM−EDX analysis of the membrane surface also shows that EDTA molecules are uniformly distributed on the membrane (Fig. 1(b)). Notably, the intercalation of EDTA molecules does not disrupt the crystal structure and surface morphology of GO nanosheets (Figs. S5 and S6), and GO-EDTA exhibits a highly aligned laminar structure (Yang et al., 2022) (Fig. 1(c)). Due to the-COO– groups on EDTA, the surface potential of GO-EDTA measured by Kelvin Probe Force Microscopy (KPFM) decreased to –923 mV (Figs. 1(d) and 1(e)), and its hydrophilicity was enhanced (Fig. S7), which facilitates ion penetration and selective separation. Furthermore, the cross-linking mechanism between EDTA molecules and GO nanosheets was investigated. XPS analyses revealed that the primary form of interaction between EDTA molecules and GO nanosheets is hydrogen bonding between EDTA molecules and oxygen atoms on the surface of GO nanosheets (Xu et al., 2024) (Fig. S8). The strong interaction between EDTA molecules and nanosheets confer excellent swelling resistance on GO-EDTA. When GO-EDTA-1.5 is transferred from a dry environment to an aqueous environment, its free spacing increases only from 5.1 Å (Figs. 1(f) and S9) to 5.5 Å, whereas that of GO increases from 4.7 to 6.0 Å (Fig. 1(f)). Consequently, GO-EDTA exhibits a highly stable and well-ordered subnanochannel structure.

3.2 Ion separation performance of GO-EDTA membranes

To verify the membrane’s separation capability, a lab-made U-shaped diffusion cell was used to measure its ion separation performance, which comprises two chambers: the feed side contains a single-salt or mixed-salt solution (0.1 mol/L), while the permeate side contains deionized water (Fig. 2(a)). First, the permeation rate of cations (Cs⁺, Sr²⁺ and La³⁺) through the GO-EDTA membrane were measured at different EDTA concentrations (Fig. 2(b)). In the unmodified GO membrane, the permeation rate of monovalent ions is slightly higher than that of divalent ions, attributable to the steric hindrance of multivalent cations. Specifically, the two-dimensional channel dimension (6.0 Å, Fig. 1(f)) is smaller than the hydration diameter of Sr²⁺ (8.2 Å) and La³⁺ (9.0 Å, Table S1). With the addition of EDTA and an increase in concentration to 1.5 mg/mL, the permeation rates of coordinated Sr²⁺ and La³⁺ decreased significantly, resulting from strong binding interactions between EDTA and the multivalent ions Sr²⁺ and La³⁺, whose transport rates were 2–3 orders of magnitude lower than that of Cs⁺. This significant difference in ion permeation rates endows GO-EDTA with extremely high selectivity for separating monovalent and multivalent metal ions, with selectivities as high as 485 and 1300 for Cs⁺/Sr²⁺ and Cs⁺/La³⁺, surpassing GO by 379-fold and 714-fold, respectively (Fig. 2(c)). However, further increases in EDTA concentration compromise separation selectivity, as while maintaining a high trans-membrane coordination capacity for multivalent ions, it also enhances that of monovalent ions. Meanwhile, the GO loading was also optimized (Fig. 2(d)). Ultimately, a GO loading of 0.6 g/m² and EDTA concentration of 1.5 mg/mL were selected as the optimal membrane fabrication parameters, followed by characterization of the resulting membranes (Figs. S10–S15). By testing the permeation rate of various ions for the GO-EDTA-1.5 membrane, it exhibits a distinct cut-off effect for monovalent and multivalent ions (Fig. 2(e)). The primary reason for achieving precise separation is the difference in binding energy between EDTA molecules and monovalent or multivalent ions (Peng et al., 2018) (Figs. 2(f), S16 and S17).

3.3 Mechanism of ion separation for GO-EDTA membranes

To elucidate the separation mechanism of the GO-EDTA membrane for mono/multivalent ions, the transmembrane energy barriers (E_a) of various ions for GO and GO-EDTA-1.5 membranes (Zhou et al., 2020) were measured (Figs. 3(a) and S18). When crossing the GO membrane, the transmembrane energy barriers for Cs⁺ (6.6 kJ/mol), Sr²⁺ (9.8 kJ/mol) and La³⁺ (13.9 kJ/mol) are relatively close, resulting in poor selectivity (Fig. S18). However, in the GO-EDTA-1.5 membrane, the transmembrane energy barrier for Cs⁺ (9.9 kJ/mol) is much lower than that for Sr²⁺ (31.5 kJ/mol) and La³⁺ (62.6 kJ/mol), resulting in extremely high selectivity for mono/multivalent ions (Fig. 3(a)). Meanwhile, by testing the relationship between different CsCl concentrations and conductance, the contribution of the charge effect to selectivity was verified (Fig. S19). Furthermore, the entropy barrier (ΔS*, primarily related to steric hindrance) and enthalpy barrier (ΔH*, mainly related to interactions between ions and 2D channels) for Cs⁺ and La³⁺ on GO and GO-EDTA-1.5 membranes were calculated (Shefer et al., 2022) (Fig. S20). Upon the introduction of EDTA molecular, ΔS* did not show significant changes, but ΔH* increased by 186% and 716% for Cs⁺ and La³⁺, respectively. This confirms that the introduction of EDTA primarily enhances the interaction between multivalent ions and the nanochannels, thereby enabling precise separation. To gain insights into the role of EDTA molecules in ion separation, the partition energy barriers (EP) of ions was measured using QCM (Figs. 3(b), S21 and S22), along with the changes in EP and the diffusion energy barrier (ED) following the addition of EDTA molecules (Guo et al., 2025; Liu et al., 2025) (Fig. 3(c)). After the introduction of EDTA molecules, the EP remained essentially unchanged, but the ED values for CsCl and LaCl₃ increased by 108% and 494%, respectively. This indicates that, due to the strong binding energy between EDTA and multivalent ions, the introduction of EDTA can significantly increase the ED of multivalent ions, thereby enabling the precise separation of mono/multivalent ions.

To elucidate the mechanisms of ion separation at the microscopic level, molecular dynamics simulations were used to model the process of ion transport across the membrane and the associated energy changes (Fig. S23). By analyzing the radial distribution functions of different ions in the bulk phase and the channels, the hydration numbers of ions in the bulk phase and the channels were obtained (Figs. S24 and S25). The results show that Cs⁺ requires the removal of only 1.8 water molecules to enter the channel from the bulk phase, while La³⁺ requires the removal of 5 water molecules (Fig. 3(d)). This indicates that La³⁺ needs more energy to remove its additional hydration water molecules in order to enter the 2D channel, resulting in a lower flux. Furthermore, the time required for Cs⁺ and La³⁺ to enter the channel and traverse the channel was calculated (Fig. 3(e)). The results indicate that both ions require a relatively long time to traverse the channel, suggesting that ions need more energy to pass through the channel (diffusion process). This is consistent with previous experimental findings. Finally, a statistical analysis was performed on the change in the number of ions on the permeate side over time (Fig. 3(f)). The results show that after a 3 ns permeation process, the Cs⁺ concentration on the permeate side was three times that of La³⁺, which confirms the selectivity of this channel for ions at the microscopic level. In summary, the selective behavior of GO-EDTA sub-nanometer channels toward monovalent and multivalent ions, as well as their separation mechanisms, have been elucidated through both experimental and computational simulations. The sub-nanometer channel size causes multi-valent ions to release more bound water upon entering the channel, thereby requiring more energy and reducing the flux of multi-valent ions. Furthermore, EDTA’s strong binding affinity for multivalent ions requires these ions to overcome a higher energy barrier when passing through the channel. Meanwhile, differences in dehydration and electrostatic repulsion among mono/multiple ions within the channel can also cause changes in the transmembrane energy barrier for ions. Moreover, compared to channel size, interactions between the channel and ions are the primary factor driving ion separation (Fig. 3(g)).

3.4 Practical application of Cs recovery from radioactive nuclear wastewater

Since the composition of actual radioactive nuclear wastewater is extremely complex, organic matter and other interfering ions can be removed through pretreatment methods prior to membrane treatment. Moreover, the ion concentration composition of actual radioactive wastewater is not entirely consistent (Li et al., 2024; Wang and Liang, 2024). To verify the treatment capability of GO-EDTA for actual wastewater, the membrane’s selectivity toward ions was tested under different concentration profiles (Figs. 4(a) and 4(b)). During mixed-salt testing, membrane selectivity increases as the proportion of Cs⁺ rises. This is primarily because higher Cs⁺ concentrations generate a greater driving force, thereby increasing the permeation rate of Cs⁺. Furthermore, due to ion competition effect, the ion selectivity under mixed-salt conditions is higher than that in single-salt tests. Moreover, when multivalent ions occupy the binding sites of EDTA, monovalent ions can pass through the membrane more easily, leading to enhanced ionic selectivity in mixed-salt tests. In addition, the long-term stability of the membrane was evaluated (Fig. 4(c)). During a 7-d operation period, the selectivity of membrane remained stable, confirming its long-term stability. Meanwhile, the membrane can still remain stable in the presence of ultrasound and strong acidic solutions (Fig. S26). The membrane’s performance in practical wastewater treatment outperformed that of membrane materials reported in previous studies (Table S2). Furthermore, to verify the membrane’s potential for practical applications, its separation capabilities for monovalent and multivalent ions, as well as its CsCl recovery efficiency, were tested using electrodialysis (Fig. 4(d)). The results show that the GO-EDTA membrane still achieves a Cs⁺/Sr²⁺ selectivity exceeding 700 and a Cs⁺/La³⁺ selectivity exceeding 1000 (Fig. S27). Moreover, CsCl crystals with a purity of 99.5% can be obtained by evaporating and crystallizing the solution in the concentration chamber. XRD analysis also confirmed that the recovered CsCl was of high purity (Fig. 4(e)). However, when this process was performed using a commercially available NF90 membrane, the ion selectivity was only in the single digits, and the purity of the resulting CsCl product was only 54.2% (Figs. S28 and S29). Furthermore, compared to traditional solvent extraction and precipitation methods, electrodialysis does not require the use of additional organic solvents, thereby reducing secondary environmental hazards. Moreover, the energy consumption of the electrodialysis method (1.76–4.8 kWh/m³) is only a fraction of that of the traditional solvent extraction method (50–100 kWh/m³) (Deng et al., 2015; Schlumpberger et al., 2015; Alkhadra et al., 2020). To further verify the treatment capacity and radiation resistance of the membrane when treating radioactive nuclear wastewater, 60Co was used as the γ-ray radiation source to treat the prepared membrane materials. The results indicate that irradiation treatment did not cause significant changes in the morphology or chemical structure of the membrane material (Figs. S30–S32), while membrane performance remained stable after irradiation (Figs. 4(f) and S33), confirming the membrane’s potential for application under irradiation conditions.

4 Conclusions

In summary, a stable and radiation-resistant GO-EDTA nanochannels with extremely high selectivity for monovalent/multivalent cations was designed, demonstrating excellent cation separation efficiency for variable feed solutions with varying ion concentration ratios. The optimal GO-EDTA membrane exhibits a Cs⁺/Sr²⁺ selectivity of 485 and Cs⁺/La³⁺ selectivity of 1300. Through electrodialysis, CsCl product with a purity of up to 99.5% was obtained. Further, even after γ-ray irradiation, the GO-EDTA channel retains its structural stability and high selectivity. This performance surpasses that of traditional polymer membranes, indicating that GO-EDTA holds great potential for addressing critical separation challenges and yielding high-purity products. Meanwhile, by constructing GO-EDTA sub-nanometer channels, the transmembrane energy barrier for ions was successfully modulated, while the ion separation mechanism was elucidated, providing theoretical guidance for the design of future high-performance membrane materials. In summary, this study reveals an intriguing separation mechanism that facilitates the effective design of membranes with high ion selectivity and demonstrates their potential for application in the field of nuclear resource recovery.

References

[1] Alemayehu H G , Liu C J , Hou J J , Yang J L , Fang M N , Tang Z Y , Li L S . (2022). Highly stable membrane comprising MOF nano-sheets and graphene oxide for ultra-permeable nanofiltration. Journal of Membrane Science, 652: 120479

[2] Alkhadra M A , Conforti K M , Gao T , Tian H H , Bazant M Z . (2020). Continuous separation of radionuclides from contaminated water by shock electrodialysis. Environmental Science & Technology, 54(1): 527–536

[3] Chen L , Shi G S , Shen J , Peng B Q , Zhang B W , Wang Y Z , Bian F G , Wang J J , Li D Y , Qian Z . et al. (2017). Ion sieving in graphene oxide membranes via cationic control of interlayer spacing. Nature, 550(7676): 380–383

[4] Chen S Q , Hu J Y , Han S J , Guo Y F , Belzile N , Deng T L . (2020). A review on emerging composite materials for cesium adsorption and environmental remediation on the latest decade. Separation and Purification Technology, 251: 117340

[5] Dai Z W , Jin P R , Yuan S S , Yang J K , Agrawal K V , Wang H T . (2025). A molecularly engineered large-area nanoporous atomically thin graphene membrane for ion separation. Nature Communications, 16(1): 4626

[6] Deng D S , Aouad W , Braff W A , Schlumpberger S , Suss M E , Bazant M Z . (2015). Water purification by shock electrodialysis: deionization, filtration, separation, and disinfection. Desalination, 357: 77–83

[7] Esfandiar A , Radha B , Wang F C , Yang Q , Hu S , Garaj S , Nair R R , Geim A K , Gopinadhan K . (2017). Size effect in ion transport through angstrom-scale slits. Science, 358(6362): 511–513

[8] Fuller N , Smith J T , Takase T , Ford A T , Wada T . (2022). Radiocaesium accumulation and fluctuating asymmetry in the Japanese mitten crab, Eriocheir japonica, along a gradient of radionuclide contamination at Fukushima. Environmental Pollution, 292: 118479

[9] Gouaux E , MacKinnon R . (2005). Principles of selective ion transport in channels and pumps. Science, 310(5753): 1461–1465

[10] Guan K C , Guo Y N , Li Z , Jia Y D , Shen Q , Nakagawa K , Yoshioka T , Liu G P , Jin W Q , Matsuyama H . (2023). Deformation constraints of graphene oxide nanochannels under reverse osmosis. Nature Communications, 14(1): 1016

[11] Guo Y , He J L , Zhang J W , Sheng M , Wang Z W , Elimelech M , Wang L . (2025). Deciphering co-ion and counterion transport in polyamide desalination membranes reveals ion selectivity mechanisms. Science Advances, 11(23): eadu8302

[12] He Q , Gao J K , Chen Z Z , Ding Y J , Xia M S , Xu P T , Chen Y . (2024). Preparation of wood-based hydrogel membranes for efficient purification of complex wastewater using a reconstitution strategy. Frontiers of Environmental Science & Engineering, 18(7): 84

[13] İnan S , Özkan B , Kusumkar V . (2026). Synthetic inorganic layered materials for cesium and strontium removal from aqueous solution: a review. Water, Air, & Soil Pollution, 237(1): 35

[14] Jin H H , Hu Y Z , Shen Z W , Pan H , Bao H L , Yin L S , Zhao G X , Ji Z Y , Wang X K , Huang X B . (2025). Electrochemical upcycling of uranyl from radioactive organic wastewater with a self-standing covalent-organic framework electrode. Nature Communications, 16(1): 3574

[15] Kumar A , Rana S , Dhiman P , Sharma P , Lai C W , Sharma G . (2025). Recent progress in integration of photocatalysis and nano-filtration for advanced wastewater treatment. Frontiers of Environmental Science & Engineering, 19(11): 150

[16] Li C , Li Z , Guan K C , Chiao Y H , Zhang P F , Xu P , Gonzales R R , Hu M Y , Mai Z , Yoshioka T . et al. (2024). Polyethylenimine-assisted preparation of Zr-MOF/GO membranes for radionuclide separation. Desalination, 592: 118118

[17] Lin T , Chen T , Jiao C , Zhang H Y , Hou K , Jin H X , Liu Y , Zhu W K , He R . (2024). Ion pair sites for efficient electrochemical extraction of uranium in real nuclear wastewater. Nature Communications, 15(1): 4149

[18] Lin Z Y , Zhong J D , Sun R Y , Wei Y Z , Sun Z H , Li W Y , Chen L Y , Sun Y R , Zhang H B , Pang J H . et al. (2023). In situ integrated fabrication for multi-interface stabilized and highly durable polyaniline@graphene oxide/polyether ether ketone special separation membranes. Advanced Science, 10(25): 2302654

[19] Liu G P , Jin W Q , Xu N P . (2015). Graphene-based membranes. Chemical Society Reviews, 44(15): 5016–5030

[20] Liu H , Huang X X , Wang Y , Kuang B A , Li W B . (2024). Nanowire-assisted electrochemical perforation of graphene oxide nanosheets for molecular separation. Nature Communications, 15(1): 164

[21] Liu K R , Epsztein R , Lin S H , Liu L , Qu J H , Sun M . (2025). Electrostatic-driven dehydration of ions in nanoporous membranes. Science Advances, 11(35): eadv0174

[22] Liu X J , Wu J L , Hou L A , Wang J L . (2019). Removal of Co, Sr and Cs ions from simulated radioactive wastewater by forward osmosis. Chemosphere, 232: 87–95

[23] Lu J , Zhang H C , Hou J , Li X Y , Hu X Y , Hu Y X , Easton C D , Li Q Y , Sun C H , Thornton A W . et al. (2020). Efficient metal ion sieving in rectifying subnanochannels enabled by metal–organic frameworks. Nature Materials, 19(7): 767–774

[24] Manos M J , Chrissafis K , Kanatzidis M G . (2006). Unique pore selectivity for Cs⁺ and exceptionally high NH₄⁺ exchange capacity of the chalcogenide material K₆Sn[Zn₄Sn₄S₁₇]. Journal of the American Chemical Society, 128(27): 8875–8883

[25] Morelos-Gomez A , Cruz-Silva R , Muramatsu H , Ortiz-Medina J , Araki T , Fukuyo T , Tejima S , Takeuchi K , Hayashi T , Terrones M . et al. (2017). Effective NaCl and dye rejection of hybrid graphene oxide/graphene layered membranes. Nature Nanotechnology, 12(11): 1083–1088

[26] Peng H W , Zhao Q . (2021). A nano-heterogeneous membrane for efficient separation of lithium from high magnesium/lithium ratio brine. Advanced Functional Materials, 31(14): 2009430

[27] Peng Y G , Huang H L , Zhang Y X , Kang C F , Chen S M , Song L , Liu D H , Zhong C L . (2018). A versatile MOF-based trap for heavy metal ion capture and dispersion. Nature Communications, 9(1): 187

[28] Rauwel P , Rauwel E . (2019). Towards the extraction of radioactive cesium-137 from water via graphene/CNT and nanostructured Prussian blue hybrid nanocomposites: a review. Nanomaterials, 9(5): 682

[29] Schlumpberger S , Lu N B , Suss M E , Bazant M Z . (2015). Scalable and continuous water deionization by shock electrodialysis. Environmental Science & Technology Letters, 2(12): 367–372

[30] Shefer I , Lopez K , Straub A P , Epsztein R . (2022). Applying transition-state theory to explore transport and selectivity in salt-rejecting membranes: a critical review. Environmental Science & Technology, 56(12): 7467–7483

[31] Shen J , Cai Y C , Zhang C H , Wei W , Chen C L , Liu L M , Yang K W , Ma Y C , Wang Y G , Tseng C C . et al. (2022a). Fast water transport and molecular sieving through ultrathin ordered conjugated-polymer-framework membranes. Nature Materials, 21(10): 1183–1190

[32] Shen L , Cheng R H , Yi M , Hung W S , Japip S , Tian L , Zhang X , Jiang S D , Li S , Wang Y . (2022b). Polyamide-based membranes with structural homogeneity for ultrafast molecular sieving. Nature Communications, 13(1): 500

[33] Smith J , Marks N , Irwin T . (2023). The risks of radioactive wastewater release. Science, 382(6666): 31–33

[34] Sun H Y , Chen Z H , Hu B , Tang J H , Yang L , Guo Y L , Yao Y X , Feng M L , Huang X Y . (2024). Boosting selective Cs⁺ uptake through the modulation of stacking modes in layered niobate-based perovskites. Nature Communications, 15(1): 8681

[35] Tang J H , Jia S Q , Liu J T , Yang L , Sun H Y , Feng M L , Huang X Y . (2024). “Ion-imprinting” strategy towards metal sulfide scavenger enables the highly selective capture of radiocesium. Nature Communications, 15(1): 4281

[36] Tang J H , Jin J C , Li W A , Zeng X , Ma W , Li J L , Lv T T , Peng Y C , Feng M L , Huang X Y . (2022). Highly selective cesium(I) capture under acidic conditions by a layered sulfide. Nature Communications, 13(1): 658

[37] Wang J L , Zhuang S T . (2019). Removal of cesium ions from aqueous solutions using various separation technologies. Reviews in Environmental Science and Bio/Technology, 18(2): 231–269

[38] Wang S , Liang S S . (2024). Acid response nanochannels of graphene oxide membranes for fast nuclide ions separation. Separation and Purification Technology, 328: 125086

[39] Wang Z P , Huang L Q , Dong X , Wu T , Qing Q , Chen J , Lu Y X , Xu C . (2023). Ion sieving in graphene oxide membrane enables efficient actinides/lanthanides separation. Nature Communications, 14(1): 261

[40] Wen Q , Yan D X , Liu F , Wang M , Ling Y , Wang P F , Kluth P , Schauries D , Trautmann C , Apel P . et al. (2016). Highly selective ionic transport through subnanometer pores in polymer films. Advanced Functional Materials, 26(32): 5796–5803

[41] Xiao C C , Yan L Q , Gao H P , Dou Z O , Xie X , Chen Y S . (2024). Selective separation and recovery of Co(II) and Ni(II) from lithium-ion battery using Cyanex 272 adsorptive membrane. Frontiers of Environmental Science & Engineering, 18(12): 148

[42] Xiao C L , Hassanzadeh Fard Z , Sarma D , Song T B , Xu C , Kanatzidis M G . (2017). Highly efficient separation of trivalent minor actinides by a layered metal sulfide (KInSn₂S₆) from acidic radioactive waste. Journal of the American Chemical Society, 139(46): 16494–16497

[43] Xu R M , Kang Y , Zhang W M , Pan B C , Zhang X W . (2023). Two-dimensional MXene membranes with biomimetic sub-nanochannels for enhanced cation sieving. Nature Communications, 14(1): 4907

[44] Xu R M , Zhang J Y , Kang Y , Yu H , Zhang W M , Hua M , Pan B C , Zhang X W . (2024). Reversible pH-gated MXene membranes with ultrahigh mono-/divalent-ion selectivity. Environmental Science & Technology, 58(15): 6835–6842

[45] Yan J , Zhang B , Li J , Yang Y , Wang Y N , Zhang Y D , Liu X Z . (2023). Rapid and selective uptake of radioactive cesium from water by a microporous zeolitic-like sulfide. Inorganic Chemistry, 62(32): 12843–12850

[46] Yang C , Long M Y , Ding C T , Zhang R N , Zhang S Y , Yuan J Q , Zhi K D , Yin Z Y , Zheng Y , Liu Y W . et al. (2022). Antifouling graphene oxide membranes for oil-water separation via hydrophobic chain engineering. Nature Communications, 13(1): 7334

[47] Zhang Q , Gao B , Zhang L , Liu X P , Cui J X , Cao Y J , Zeng H B , Xu Q , Cui X W , Jiang L . (2023). Anomalous water molecular gating from atomic-scale graphene capillaries for precise and ultrafast molecular sieving. Nature Communications, 14(1): 6615

[48] Zhang Y , Zhang S , Chung T S . (2015). Nanometric graphene oxide framework membranes with enhanced heavy metal removal via nanofiltration. Environmental Science & Technology, 49(16): 10235–10242

[49] Zhou X C , Wang Z X , Epsztein R , Zhan C , Li W L , Fortner J D , Pham T A , Kim J H , Elimelech M . (2020). Intrapore energy barriers govern ion transport and selectivity of desalination membranes. Science Advances, 6(48): eabd9045

RIGHTS & PERMISSIONS

The Author(s) 2026.