Introduction

The rapid expansion of industries like metal plating, mining, and chemical manufacturing has significantly increased the release of heavy metal-contaminated wastewater, especially in developing nations. Remarkably, heavy metal ions are usually not biodegradable, and most importantly, most heavy metal ions such as Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺, are highly toxic even at low concentrations. Therefore, effectively removing heavy metal ions from wastewater is a critical concern for global health1,2 In past decades, scientists have developed various methods to address this issue with chemical precipitation being popular due to its simplicity and ease of operation3. However, while it is more effective for wastewater with high initial concentrations of heavy metals, it is less efficient for lower concentrations. Along these lines, the ion exchange process is used for selective heavy metal recovery but has limitations such as resin durability, vulnerability to high temperatures and oxidation, and high operational costs4,5 Moreover, the adsorbent adsorption method in which a porous solid is utilized, has also been developed as an alternative method6,7 Meanwhile, membrane separation technologies such as ultrafiltration (UF), reverse osmosis (RO), and nanofiltration (NF) have gained considerable attention for their high efficiency in removing heavy metal ions from water8,9.

The advent of graphene and its derivatives has shown great promise for desalination and wastewater treatment due to their outstanding properties, including high quality, low weight, excellent electrical conductivity, and thermal stability10,11 Previous research has successfully demonstrated the separation of heavy metal ions using graphene based membranes. For example, Amal Kanta Giri et al. investigated stacked graphene membranes with varying channel widths for heavy metal ion separation, finding that wider channels offered higher water permeability while narrower channels provided better ion rejection12. Li et al. compared different functionalized graphene membranes and discovered that boron-doped graphene showed superior water permeability and ion rejection relative to other functionalized groups13. Jafar Azamat et al. effectively incorporated fluorine and hydrogen atoms into boron nitride nanosheet (BNNS) membranes, enhancing ion separation from water14. These studies strongly suggest the superiority of nanomaterials in ion separation applications.

Metal-organic framework (MOF) is a type of inherently porous nanomaterials, which have attracted numerous attentions from the heavy metal ion removal area15,16,17,18,19 as well as other membrane separation areas, e.g., gas separation20,21,22. Due to the inherent porosity, MOF can allow the spontaneous trap and adsorption of heavy metal ions via electrostatic attraction forces14,18. For example, Zhong and co-workers experimentally constructed a MOF material based on ethylenediaminetetraacetic acid, which featured a broad-spectrum heavy metal ion trap for up to 22 types of ions with removal efficiencies of > 99% for single-component adsorption, multi-component adsorption, or in breakthrough processes16. Boix et al. demonstrated that the composite materials based on MOFs and inorganic nanoparticles can effectively and simultaneously remove multiple toxic heavy metals from water23. Using a theoretical approach, Gu and colleagues demonstrated that the heavy metal cations can be tightly trapped by the negatively charged open sites inside a MOF membrane, leading to the efficient removal of heavy metal ions24. In addition, ultrathin 2D MOFs25 have been extensively investigated for heavy metal ion removal. Qu and co-workers synthesized an ultrathin 2D water stable MOF nanosheet (Zn(Bim)(OAc) MOF, with average thickness of 7.05 nm), which shows high efficiency and selectivity for removing Pb2+ and Cu2+ ions in water solution26. Chen and colleagues experimentally realized the few-layered CoCNSP MOF membrane from a facile-operating method, which showed high distribution coefficients, fast capture dynamics, and ultrahigh uptake capacities for some heavy metal ions27. Wang et al. showed that the single-layered sulfur-rich two-dimensional (2D) MOF nanosheets can be served as an ideal adsorbent materials for the capture of Hg2+ from aqueous solution, featuring considerable capacity, efficiency, effectivity, and recycling ability28. Therefore, previous investigations showed the good performance of 2D MOF for heavy metal ion capture and separation29. However, very limited effort has been devoted to investigate the capacity of porous 2D MOF as a sieving membrane for heavy metal ions separation. In this study, we employed MD simulation approaches to explore the heavy metal ion removal capacity of one typical 2D MOF, Cu-THQ. The Cu-THQ is a 2D porous MOF, featuring nanopores that display a uniform size with dimensions slightly smaller than the hydrated ion. In addition, to the best of our knowledge, there is no report investigating the ion removal capacity of such MOF. Thus, we choose this MOF to explore its heavy metal ion removal performance. Our MD simulation resuls show that the single-layered Cu-THQ MOF membrane has excellent performance for heavy metal ion removal.

Method

The single layer 2D Cu-THQ MOF, which is comprised of oxygen, copper and carbon atoms, is displayed in Fig. 1a; this 2D MOF has been experimentally synthesized and characterized as a promising CO2 catalyst 30,31 Different number of Cu-THQ MOF layers were utilized to construct sieving membranes. The simulation was mainly consisted of four components, a piston (consisting of a graphene nanosheet), water reservoir containing 40 heavy metal ions (wastewater including Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺ ions), the sieving membrane composed of different numbers of Cu-THQ MOF layers and freshwater. Seven simulation systems were separately built, i.e., four systems each with 40 ions of either Cd²⁺, Cu²⁺, Hg²⁺, or Pb²⁺, one system with mixed ions (containing 10 of each heavy metal ion) and two systems with 3 and 5 layered Cu-THQ MOF sieving the Cd2+ ion. To keep the integrity of membrane, the box sizes (i.e., x and y axis) along the MOF surface were set to 5.36 and 4.65 nm, making the MOFs fitly bond at the boundary. Pressures ranging from 20 MPa to 100 MPa with increments of 20 MPa were exerted on the piston to accelerate the transport of waters through the membrane.

Fig. 1
figure 1

(a) Model of the Cu-THQ membrane. Red, blue and cyan spheres indicate the oxygen, copper and carbon atoms. (b) The simulated system is composed of a water reservoir containing heavy metal ions (wastewater) flanked by a Cu-THQ membrane on one side and a graphene piston on the other side. The heavy metal ions are shown as light blue spheres, the graphene piston is displayed by gray spheres. The box boundary is also illustrated. The configuration is drawn by VMD software package.

Full size image

The GROMACS software package32 was employed to conduct the simulations, and the VMD software33 was utilized during the analysis and visualization. The CHARMM36 force field34,35 was used to describe the heavy metal ions, while the SPC/E water36 model represented water molecules. The force field parameters for the graphene piston were adopted from a previous work37, and those for the ions were also derived from a previous study38. The force field parameters of Cu-THQ were built similarly to the method of previous studies,39,40 where the Leonard Jones (LJ) parameters (used to calculate vdW energy) were derived from the Universal force field (UFF)41, whereas the atomic charges (used to calculate Coulomb energy) were calculated by the DDEC method42,43,44,45. Such method (i.e., using the UFF force field to describe the LJ parameters and the DDEC charges for all the MOF atoms) has been usually employed to describe MOF force field parameters in molecular modeling46,47,48,49,50. Moreover, we have conducted DFT calculations of the solvent (water) and one typical solute (Pb2+) binding to the open metal in MOF (Fig. S1). We originally positioned these two molecules above the open metal in MOF. After the optimization calculations, the water molecule keeps its binding to the open metal in MOF with a corresponding adsorption energy of -0.182 eV. In contrast, the optimization shifts the Pb2+ ion from the top position of the open metal to the top position above the benzene ring, resulting in a large adsorption energy (-3.123 eV). The ion displacement indicates that the benzene ring in MOF has more robust attraction to the heavy metal ions, due to the cation-π interaction. Therefore, the ion interaction at the top of the open metal in the MOF structure is not as stabilizing as the position on the top of the benzene ring. The high adsorption capacity of the MOF to heavy metal ions can aid in the ion rejection of the MOF. In addition, although obtaining precise force field values of the open metal in MOF remains challenging, the UFF is currently one of the best methods to qualitatively describe the interactions between MOF and other molecules, which has been widely applied in previous studies.

In all the simulations, the position of the membrane was restrained, and for each simulation, three parallel trajectories were performed. The temperatures during the simulations remained constant at 300 K using the V-rescale thermostat51. Periodic boundary conditions were implemented in all directions (x, y, and z)52. The van der Waals (vdW) interactions were calculated with a cut-off distance of 1.2 nm. The SETTLE algorithm53 was utilized to restrain the geometry of the water molecules. Throughout all the simulations, a time step of 2.0 fs was employed, and each simulation lasted for 100 ns.

According to the umbrella sampling method54,55,56, we calculated the potential of mean force (PMF) value by pulling the typical Cd2+ and water molecule axially from the nanopores interior to the wastewater chamber. The harmonic force exerting on the molecules was according to the formula, F = K×(d-d0), where K is a constant (2000 kJ/mol/nm2), and the vertical distance (d) between nanopores and targeted molecules was limited to a reference distance, d0. The sampling interval was set to 0.1 nm. At each reference distance d0, the system was equilibrated for 2 ns, followed by MD simulations for 10 ns. The free energy curves were obtained by the weighted histogram analysis method of the g_wham tool57,58,59.

The DFT calculation systems consisted of a water molecule or a Pb2+ ion adsorbed on the top position of the open metal in MOF. Their periodic boxes had a dimension of 20 × 20 × 20 Å3 in a 2 × 1 × 1 supercell, containing 111 atoms for the H2O@MOF system and 109 atoms for the Pb2+@MOF systems. A vacuum layer of about 20 Å was used to avoid interactions between the periodic cells in the z-axis direction.

All the density functional theory (DFT) calculations were carried out with the Vienna ab initio simulation package (VASP)60. The exchange-correlation function was described by the Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation61, with dispersion corrections considered through the van der Waals correction (DFT-D3)62. The projector augmented wave (PAW) method63 was employed with a plane wave cutoff of 600 eV. Reciprocal space was sampled using a 1 × 1 × 1 k-point grid during geometry optimization until the convergences of energy and force reached 10− 5 eV and 0.01 eV/Å, respectively.

The adsorption energies (Ead) of the water molecule or Pb2+ ion to the MOF were defined as Ead = EH2O/Pb2+@MOF - EMOF - EEH2O/Pb2+, where EH2O/Pb2+@MOF, EMOF and EH2O/Pb2+ were the energies of the corresponding optimized structures, respectively.

Result

Figure 1 illustrates the MOF membrane and the corresponding simulation system. The Cu-THQ MOF is a 2D structure comprising many ordered circular nanopores with diameters of approximately 1.1 nm; each nanopore is surrounded by 12 oxygen atoms. The Cu-THQ MOF is positioned in-between the wastewater chamber and the freshwater chamber. By applying external pressures to the graphene piston, water molecules can traverse the Cu-THQ nanopores to reach the targeted freshwater chamber. Using the protocol. we examined the ability of Cu-THQ MOF to separate four typical heavy metal ions: Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺.

Figure 2 shows the water transport capacity through the Cu-THQ MOF. The time-dependent filtered water numbers under five successive pressures (20, 40, 60, 80, and 100 MPa) are depicted in Fig. 2a. These curves exhibit a distinct linear relationship, suggesting that each pressure enables continuous water transport through the Cu-THQ membrane. Higher pressures lead to steeper increases in filtered water numbers due to the greater pushing force exerted at these pressure values. We summarized the water flow rates under five pressures as shown in Fig. 2b. The water flow rate is obtained from the slopes of the linear regions in the time-dependent filtered water number curves (Fig. 2a). The water flow rates for each heavy metal ion show comparable values at the same applied pressure value, although we observed slight differences at the higher pressures (80 and 100 MPa). The water flow rates in all the five pressures demonstrate a linear relationship, indicating that the water flow rate is proportional to the external pressure. Considering the slopes of the water flow rate and the membrane surface area, we calculated the water permeability of the Cu-THQ membrane for separating Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺ ions, as shown in Fig. 2c. The water permeabilities are 16.8 ± 0.1, 16.0 ± 0.6, 17.5 ± 0.5, and 16.7 ± 0.2 L cm⁻² day⁻¹ MPa⁻¹ for Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺ ions, respectively. These results indicate that the Cu-THQ MOF membrane allows a fast water permeability for the process of heavy metal ion removal.

Fig. 2
figure 2

(a) Time-dependent filtered water number. The data are obtained from the simulations of the Cd2+ separation. (b) Water flow rates at different pressures in the separating process for the four heavy metal ions. (c) Water permeability in the separating process for the four heavy metal ions.

Full size image

In addition to water permeability, the heavy metal ion rejection is another important factor in assessing the membrane performance. As shown in Fig. 3, the heavy metal ion rejections by the Cu-THQ MOF membrane are nearly 100%, with the lowest ion rejection value at 98.3%±1.4%. This suggests that the 2D Cu-THQ MOF can effectively reject the heavy metal ions investigated here (i.e., Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺). The time-dependent concentration of Cd2+ in the wastewater chamber was showed in Figure S2, wherein the heavy metal ion concentration in this chamber increases successively due to the compression by the piston and the concomitant reduction of the wastewater chamber’s volume. Combined with the high-water permeability, we conclude that the single-layered 2D Cu-THQ MOF membrane is an excellent candidate for heavy metal ion removal.

Fig. 3
figure 3

Heavy metal ion Rejection rates for the four heavy metal ions, Cd2+(gray), Cu2+(orange), Hg2+(blue), and Pb2+(red) under five different pressures (20, 40, 60, 80 and 100 MPa).

Full size image

We also evaluated how the concentration of heavy metal ion in the wastewater chamber affects the performance of Cu-THQ MOF membrane. As shown in Fig. S3, the higher concentration of heavy metal ion elicits a continuously decrease in the water permeability, which may be attributed to an increase in the possibility of such ions to obstruct the nanopores in the membrane, resulting in a deceleration in the water transport through the membrane. It is commonly suggested that, a higher ion concentration leads to a reduction in the water transport through porous membranes,64,65 which is mainly linked to blockage caused by the ions. Specifically, the ions, e.g., heavy metal ions in this work, exist as a hydrated ion clusters in water, due to the strong electrostatic attraction between both molecular entities. Due to the high pressures applied in the wastewater chamber as well as the concentration difference between both the wastewater and freshwater chambers, ions often tend to transport through the membrane. When they reach the membrane pores, the hydrated ions might block the pores, resulting in the decrease in the number of waters that are transported through the membrane. Therefore, higher concentration of ions can increase the possibility of such blockage, leading to a reduction in the water permeability. Although the decrease in water permeability is observed following the increase in ion concentration, the heavy metal ion rejections remain close to 100% for the applied pressures ranging from 20 to 100 MPa, indicating the high heavy metal ion removal capacity of the single layer 2D Cu-THQ MOF system. Usually, the presence of ions hamper the water permeation across the nanoporous membranes, due to the blockage of the hydrated heavy metal ions to the nanopores, and thus a higher concentration of ions will reduce the water permeability through the membrane while increasing the ion passage64. However, the Cu-THQ membrane achieves the high ion rejection up to 100%, due to the small size of nanopores in Cu-THQ membrane, further supporting the superior performance of such membrane for heavy metal ion removal.

To reveal the reason behind the outstanding performance of Cu-THQ MOF for heavy metal ion removal from the free energy viewpoint, we performed potential of mean force (PMF) calculations based on the umbrella sampling method. During the PMF calculations, we pulled one Cd²⁺ heavy metal ion (as a representative heavy metal ion) and one water molecule from the nanopore interior (z-axis of zero nanometers) of the Cu-THQ MOF to the wastewater chamber (z-axis larger than zero nanometers). As shown in Fig. 4, the free energy barriers for transporting a Cd²⁺ ion and water molecule through the Cu-THQ MOF nanopore are 12.3 and 7.0 kJ mol⁻¹, respectively. This demonstrates that Cd²⁺ ions require more energy compared to water when transporting through the Cu-THQ membrane. Due to the strong electrostatic interactions with water molecules, the ions (e.g., heavy metal ions in this work) exist as a hydrated ion cluster in aqueous environment. Therefore, a hydrated ion has larger dimensions than a single water molecule. When hydrated ion passes through the membrane pore, the hydrated ion should lose its hydration shell to make itself small enough to fit in the dimension of the pore in membrane. The dehydration process should consume energy, which contributes to the higher free energy barrier of the heavy metal ion compared with the water molecule. Therefore, higher free energy barrier of heavy metal ion indicates the more difficult transport of heavy metal ion through the membrane, which explains the high heavy metal ion rejection. Therefore, PMF profiles indicate that water molecules are energetically more favorable to permeate through the MOF membrane compared to heavy metal ions, supporting the above findings.

Fig. 4
figure 4

Potential of Mean Force (PMF) profiles for the Cd2+ ion and water molecule pulling from the Cu-THQ nanopore interior to the wastewater chamber. Zero nanometer of the z axis indicates the position of MOF membrane whereas 0 ~ 1.5 nm denotes the region of the wastewater chamber.

Full size image

Given that the above simulations used individual heavy metal ions, we then examined the performance of Cu-THQ MOF in separating wastewater containing a mixture of four heavy metal ions. As shown in Fig. S4, the water permeability is 16.4 L cm⁻² day⁻¹ MPa⁻¹, comparable to those obtained when sieving only one type of ion as shown in Fig. 2c. Meanwhile, the achieved ion rejection rates are completely 100%, indicating that the 2D single-layered Cu-THQ MOF membrane also has excellent performance in separating mixed heavy metal ions.

In practice, producing single-layered MOF membranes might be challenging. Therefore, we constructed multilayered Cu-THQ membranes (3 and 5 layers) and explored their capacity for removing heavy metal ions (using Cd²⁺ as a representative). Figure 5a shows the simulation system based on a typical multilayered Cu-THQ membrane. Figure 5b illustrates the water permeability of 1, 3, and 5-layered Cu-THQ membranes, where water permeability declines with an increase in the number of layers caused by a more long-lasting adsorption of water molecules in the Cu-THQ MOF pore interior (discussed further below). Additionally, the Cd²⁺ rejection rates (Fig. 5c) are consistently 100% for 1, 3, and 5-layered Cu-THQ membranes, indicating an excellent heavy metal ion removal capacity regardless of number of layers in the sieving system. These results suggest that fewer Cu-THQ layers offer better performance, enabling faster water permeability while maintaining similar heavy metal ion rejection rates.

Fig. 5
figure 5

(a) Simulation system of the multilayered Cu-THQ membrane system. The configuration is drawn using the VMD software package. (b) Water permeability through different layered MOF membranes, 1-layer, 3-layer, and 5-layer systems. (c) Heavy metal ion rejection rates at the same layered systems as in (b) of different layered MOF membranes.

Full size image

To understand why increased Cu-THQ layers decrease water permeability, we conducted further analyses. Figure 6 illustrates the water distributions near the membrane. Figure 6a–c show the water number density along the direction perpendicular to the membrane, indicating the highest density within the membrane. More peaks are observed with increasing Cu-THQ layer numbers, indicating more water molecule adsorption. Figure 6d–f show the 2D distribution of water molecules on the membrane, indicating more high-density regions (red regions) with an increase number of Cu-THQ layers, implying a more widespread adsorption of water molecules, supporting the observed number density profiles. The increased water molecule adsorption is attributed to stronger attraction forces as observed in the plots of water configuration inside the membrane nanopore. As shown in Fig. S5, the water configuration clearly shows an orderly arrangement of waters inside the nanopore of the Cu-THQ membrane. This specific water configuration is attributed to the strong interaction between water molecules and and the Cu-THQ nanopore boundary (see Fig. 7a), which thereby results in the density peaks observed in Fig. 6a–c as well as in the high adsorption regions in Fig. 6d–f. We calculated the interaction energy between a water molecule and the 1, 3 and 5 layered Cu-THQ MOF nanopores, wherein a water molecule is confined inside these nanopores through restraining the position of the oxygen atom of this water along the direction perpendicular to the membrane (i.e., z axis) while relaxing the movement along the other two directions (x and y axis). As shown in Fig. 7a, as the Cu-THQ layer number increases from 1 to 5, the total interaction energies between the water molecules and membrane successively increase: -4.1, -12.2, and − 16.3 kJ mol⁻¹. During this restrained trajectory to enhance the interactions, electrostatic energy (Coulomb energy) plays a dominant role. As the number of Cu-THQ layer increases, the number of negatively charged oxygen atoms in the membrane pores also increases, providing more negatively charge loci to attract water molecules, resulting in an increase in the water adsorption positions, as shown in Fig. 6. The higher adsorption energy makes it more difficult for water to leave the MOF membrane, leading to a decrease in the water flow rates (Fig. 7b). Therefore, fewer Cu-THQ layers offer better performing balance for heavy metal ion removal and water flow rates.

Fig. 6
figure 6

(a–c) Water number density along the direction perpendicular to the membranes in the 1, 3 and 5 layers of Cu-THQ systems. (d–f) Water density maps near the membranes of 1, 3 and 5 Cu-THQ layers projecting onto the membrane surface.

Full size image
Fig. 7
figure 7

(a) Interaction energies, including vdW, Coulomb, and total energies, of one water molecule for the three membranes systems. (b) Water flow rates through three membranes under different pressures.

Full size image

In fact, the 2D Cu-THQ MOF has been already experimentally synthesized,30,31 however, the creation of a large-scale Cu-THQ MOF system adequate for sieving applications remains challenging to achieve. The possible solution is constructing a membrane by stacking many small-sized Cu-THQ MOF, resembling a recently devised graphene membrane66, which may compromise the separation performance somewhat due to the possible covering of the nanopores. Although we propose the good performance of the single-layered Cu-THQ MOF membrane, the multilayered MOF membrane can be realized in a realistic application more easily, because the single-layered Cu-THQ MOF may be affected by the environment more easily, resulting in its mechanically instability (structural fragility). In addition, considering that this MOF comprises a copper cation coordinated by oxo ligands, which may be susceptible to degradation in the aqueous heavy metal salt milieu, a multilayered Cu-THQ MOF membrane may be more robust in tolerating the salt environment during the separation process. Although there are some potential drawbacks of the Cu-THQ MOF membrane, a recent experimental investigation67 has showed the successful synthesis of a stable Cu-THQ MOF nanocomposite membrane. Specifically, they synthesized the Cu-THQ MOF through the chemical reaction between Cu2+ and THQ, which shows structural and chemically stability in water. By incorporating into the polyamide, the authors achieved the stable Cu-THQ nanocomposite membrane. They also demonstrated that this Cu-THQ nanocomposite membrane has good performance for NaCl desalination, featuring 98.9% NaCl rejection rate and 2.9 ± 0.3 L/m2/h/bar water permeance. Therefore, based on our results, we believe that this nanocomposite membrane can also be effective for a high efficient removal of heavy metal ions from wastewater in experimental conditions. Commonly, the MOF membrane may be not flat and instead presents different orientations in realistic application. Varied orientations of the MOF nanosheets may slightly reduce the transport of water but without compromising water permeability. Furthermore, the MOF membranes remain rigid throughout all the simulations presented here. Nonetheless, we think that the rigid MOF membrane does not have distinct difference compared with a more flexible MOF membrane in the two main sieving properties, that is heavy metal ion removal and water permeability. Since our results suggest that the performance mainly depends on the open pores in the membrane and the pore size is unlikely to change in a more flexible MOF membrane, we conclude that the rigid MOF membrane has sight differences compared with the flexible MOF membrane in its performing to remove heavy metal ions.

Conclusion

In summary, using MD simulations, we explore the potential application of Cu-THQ MOF-based membranes for heavy metal ion removal. Four typical heavy metal ions, namely Cd²⁺, Cu²⁺, Hg²⁺, and Pb²⁺, are considered. Firstly, we examine the capacity of a single-layered Cu-THQ MOF membrane for separating individual ions. The simulation results show that single-layered Cu-THQ MOF membranes possess water permeability of approximately 16–17.5 L cm⁻² day⁻¹ MPa⁻¹, accompanied by ion rejection rates of nearly 100%, indicating an excellent heavy metal ion removal performance. The PMF profiles suggest that water molecules face a lower free energy barrier compared to heavy metal ions when transporting through the Cu-THQ MOF membrane, supporting the observed performance. We also test the performance of multilayered Cu-THQ membranes, and the results present a negative correlation with more MOF layers leading to a decrease in water permeability, even though the ion rejection rates remain at around 100%. Additional analyses demonstrate that more MOF layers elicit stronger water adsorption inside the nanopores, resulting in slower water flow rates. Therefore, our findings provide a promising 2D MOF membrane for heavy metal ion removal, which should be useful for future sieving applications.