The energy-free purification of trace thallium(I)-contaminated potable water using a high-selective filter paper with multi-layered Prussian blue decoration

Abstract

Thallium is a highly toxic metal, and trace amount of thallium(I) (Tl⁺) in potable water could cause a severe water crisis, which arouses the exploitation of highly-effective technology for purification of Tl⁺ contaminated water. This report proposes the multi-layered Prussian blue (PB)-decorated composite membranes (PB_x@PDA/PEI-FP) based on the aminated filter papers for Tl⁺ uptake. Extensively characterization by Fourier transform infrared spectrometer-attenuated total reflectance, scanning electron microscope, thermogravimetric analysis, X-ray photoelectron spectroscopy and X-ray diffraction were performed to confirm the in situ growth of cubic PB crystals on filter paper membrane surfaces via the aminated layers, and the successful fabrication of multi-layered PB overcoats via the increasing of aminated layers. The effect of PB layers on Tl⁺ removal by PB_x@PDA/PEI-FP from simulated drinking water was evaluated as well as the influence of different experimental conditions. A trade-off between PB decoration layer number and PB distribution sizes is existed in Tl⁺ uptake by PB_x@PDA/PEI-FP. The double-layered PB₂@PDA/PEI-FP membrane showed the maximum sorption capacity, but its Tl⁺ uptake performance was weakened by the acid, coexisting ions (K⁺ and Na⁺) and powerful operation pressure, during filtrating a large volume of low-concentrated Tl⁺-containing water. However, the negative effect of coexisting ions on the Tl⁺ uptake could be effectively eliminated in weak alkaline water, and the Tl⁺ removal was increased up to 100% without any pressure driving for PB₂@PDA/PEI-FP membrane. Most importantly, PB₂@PDA/PEI-FP displayed the high-efficiency and high-selectivity in purifying the Tl⁺-spiked Pearl River water, in which the residual Tl⁺ in filtrate was less than 2 μg·L^–1 to meet the drinking water standard of United States Environmental Protection Agency. This work provides a feasible avenue to safeguard the drinking water in remote and underdeveloped area via the energy-free operation.

Key words： membrane adsorption; Prussian blue; energy-free filtration; potable water; trace thallium(I)

1 Introduction

Heavy metal ions such as Pb²⁺, Cd²⁺, Hg²⁺ and Tl⁺ in water have become a severe threat to human health because of their high toxicity and enrichment in organs [1]. Among these heavy metal ions, Tl⁺ has the strong affinity to the K⁺-carried enzymes, which interfered with the exchanging of K⁺ at the cellular level; therefore, Tl⁺ shows the higher toxicity to human organs than mercury, lead, and arsenic [2,3]. The United States Environmental Protection Agency (USEPA) stipulates that the allowable concentration of thallium in drinking water and industrial wastewater should be lower than ~2 and ~140 μg·L^–1, respectively. The effective and handy techniques are urgently needed for treating Tl⁺-containing water or sudden Tl⁺ pollution incidents. To address it, a wide range of treatment technologies such as chemical precipitation [4,5], ion-exchange [6,7], adsorption [8,9], solvent extraction [10,11], and electrochemical deposition [12] have been exploited. Nevertheless, the technologies mentioned above have displayed the inherent limitations when treating the trace Tl⁺-containing water, requiring a large amount of chemicals, high cost, or secondary pollution.

Nowadays, membrane-based separation technology is very attractive due to its continuous, environment-friendly and scalable process as well as low-energy consumption and zero by-product production [13,14]. Among them, polymeric porous membranes and their surface modification with both functional organics and nanoparticles have obtained increasing attention in the treatment of water and wastewater containing heavy metals. Tian et al. [15] fabricated a thin film composite nanofiltration (NF) membrane with electronegativity and small pore size, which exhibited excellent heavy metal rejection of Ni²⁺, Zn²⁺ and Cu²⁺ over 98% and Cd²⁺ more than 92.7% via the Donnan effect and size exclusion. Although the clean water could be produced by the separation of NF membranes, heavy metals were remained in the contaminated water. To capture the trace heavy metals from contaminated water, many researchers have proposed methods for the preparation of adsorption membranes. Pei et al. [13] used the polyelectrolyte active layers to graft onto the regenerated cellulose membrane surface. This membrane exhibited the high-efficiency removal of multiple ions such as Cu²⁺, Pb²⁺ and Cd²⁺ via adsorption and was easy to be regenerated and reused. To construct the “state-of-the-art” adsorption membranes, some nano-adsorbents such as titanium dioxide (TiO₂), ferric oxide (Fe_xO_y), manganese oxide (Mn_xO_y), metal-organic frameworks (MOFs) have been grown in situ on membrane surfaces for capturing the trace heavy metals in water. All results reveal that nano-adsorbents promote greatly the adsorption capacity of membranes to heavy metals. Efome et al. [16] reported that the maximum adsorption capacities of Cd²⁺ and Zn²⁺ reached up to 225.05 and 287.06 mg·g^–1, respectively, for the commercial polyacrylonitrile membrane decorated with Zr-based MOF-808. These suggest adsorption membrane can act as a potential candidate for the treatment of heavy metals-polluted water.

However, most of adsorption membranes, which possess the electrostatic interaction or multiple-coordination with metal ions, are unselective in removing the various heavy metals. Obviously, facing the nontoxic K⁺ and high-toxic Tl⁺, previous adsorption membranes are incapable of accurately capturing the Tl⁺ in the actual water due to the interference of resembling K⁺. To resolve it, a few studies have integrated the Prussian blue (PB, Fe₄[Fe(CN)₆]₃) with the porous membranes for treating Tl⁺-contaminated water and wastewater. This is because the cubic lattice structures of PB and its analogs are exhibiting the greater adsorption capability and better selectivity of capturing Tl⁺ from water/wastewater over competing cations. Wang et al. [17] and Shi et al. [18] both reported that the as-fabricated PB-based membranes exhibited an outstanding Tl removal efficiency (> 90%) at various operating conditions and co-existing competing cations. Unfortunately, the above membrane supporters were originated from fluoropolymers, which cause refractory solid waste pollution after discarding. Moreover, with the aim of obtaining high flux values or accelerating treatment efficiency, previous PB-based membranes were needed to be driven by a cross flow system with high pressures (around 1 bar or higher), inevitably intensifying the energy consumption and operation cost. Thus, exploiting the low-cost adsorption membranes without sacrificing their specificity is urgent for purification of Tl-containing water, especially in rural and underdeveloped areas.

In simple filtration system, gravity-driven filtration membranes (GDFMs) show relatively stable fluxes with few energy-consuming in filtrating the germs and organic contaminants-rich water. Among GDFMs, filter papers (FPs) originated from abundant natural resources, have obtained great attention in securing the potable water because of their high porosity, water permeability, dimensional stability, low cost and biodegradable [19]. However, little literature reported the application of FPs in purifying the heavy metals-polluted water. The immobilization of PBs on FP surfaces may have a potential feasibility to purify Tl-contaminated water via a dead-end flow system. Safeguard the drinking water in remote areas even though the occurrence of Tl⁺-discharge in surface water. However, the detachments of PBs from FP surfaces are more likely to be caused by the water power, water soak and changes of water level. Thus, constructing the PBs layer with good stability and high efficiency remains a key challenge since the PBs adsorption layer and FPs typically exhibit poor compatibility [20,21].

Motivated by this challenge, we used the typical mussel-inspired polydopamine (PDA)/polyethyleneimine (PEI) intermediate layers to combine FPs with PBs. The PDA/PEI intermediate layer with abundant amino groups was thinner, smoother, and more stable, which could work as a defect-free functional layer for chelating with metal ions [22,23]. By virtue of the PDA/PEI layer, PBs could be in situ grown and assembled on the FP surfaces. Drawing inspiration from this co-deposition methodology, we fabricated multilayer of PBs on the FP surfaces, and discussed their efficiency and selectivity in capturing the Tl⁺ from water with various temperature, concentration, pH value, operation pressures, and competition cations (e.g., K⁺, Na⁺, Mg²⁺ and Ca²⁺) via a simple filtration system. The double-layer PBs-based FP shows satisfactory filtration performance with high removal performance for various Tl⁺-containing water with reasonable water flux. Moreover, the double-layer PBs-based FP exhibits good stability and reusability as the bio-glue effect of the PDA/PEI layer immobilizes the PBs via strong coordination complexes between Fe²⁺/Fe³⁺ and aminol groups. In combination with this facile preparation, high Tl⁺ adsorption selectivity, and efficient Tl⁺ removal performance and low energy consume, this indicates that the prepared PBs-based FPs represent a promising strategy in high-toxic heavy-metal polluted water purification and reclamation.

2 Experimental

2.1 Materials

FPs (membrane size, d₁ = 50 mm, pore size, d₂ = 15–20 μm) were bought from Fuyang Beimu Filter Co., Ltd. (Zhejiang, China). Dopamine hydrochloride (DA·HCl), PEI (M.W. 600, 99%), and sodium ferrocyanide decahydrate (Na₄Fe(CN)₆) were purchased from Shanghai Titan Scientific Co., Ltd. (Shanghai, China). The Tl⁺ stock solution (1000 mg·L^–1) was prepared by dissolving TlNO₃ (99.9%, Aldrich, USA) in deionized water. The Tl⁺ working solutions with various concentration were prepared daily from the stock solution via dilution with the calculated amount of deionized water. The NaCl, KCl, MgCl₂, and CaCl₂ were all purchased from Aladdin Reagent Co., Ltd. (Shanghai, China), and used in interfered experiments to evaluate the effect of co-existing ions on Tl⁺ adsorption. The HCl aqueous solution (36%–38%) and NaOH, obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China), were utilized to control the pH of the Tl⁺-containing solutions.

2.2 Fabrication of the PB-based FPs

First, the DA·HCl (2.0 g·L^–1) and PEI (2.0 g·L^–1) were both dissolved in Tris-HCl solution (50 mmol·L^–1, pH = 8.5) [24,25]. Then, the circled FPs with diameter of 50 mm were immersed into the PDA/PEI solution for surface modification based on previous studies. After 24 h of deposition, the resultant FPs were washed thoroughly by deionized water [26], and marked as the PDA/PEI-FP. Subsequently, As-obtained PDA/PEI-FPs were put into the 100 mL of mineralized solution containing Na₄Fe(CN)₆ (4.0 g·L^–1) for their in situ surface decoration with PBs. After PDA/PEI-FPs wetted completely, HCl water solution (4 mL, 36%–38%) was promptly added into the mineralized solution to launch the hydrolysis reaction, and then the solution temperature was heated up to 60 °C for accelerating the in situ growth of Na₄Fe(CN)₆ crystals on PDA/PEI-FP surfaces. The whole reaction was maintained for 24 h to ensure the sufficient PBs decoration on PDA/PEI-FP surfaces, and the final obtained sample was marked as PB₁@PDA/PEI-FP. Immediately after the preparation of PB₁@PDA/PEI-FP, the PDA/PEI surface modification and PBs in situ coating were repeated twice, and the as-prepared samples were marked as PB₂@PDA/PEI-FP and PB₃@PDA/PEI-FP, respectively. Lastly, all PB_x@PDA/PEI-FP (x = 1, 2, and 3) samples were dried by an oven at 60 °C for 24 h, then weighed (Table S1, cf. Electronic Supplementary Material, ESM), packed in a Ziploc bag, and sent for characterization.

2.3 Membrane characterizations

The morphologies of all membrane samples containing FP, PDA/PEI-FP, and PB_x@PDA/PEI-FP (x = 1, 2, and 3) were examined by scanning electron microscopy (SEM, JSM-7610FPlus, JEOL, Japan), and their surface element distributions were detected by an energy dispersive spectroscopy (EDS, INCA X-Act, Oxford Instruments, UK). For all PB_x@PDA/PEI-FP (x = 1, 2, and 3) samples, the changes of chemical groups, valence states and contents of chemical elements, crystal structure of surface-immobilized PBs, and mass fraction of PBs were analyzed by attenuated total reflectance-Fourier transform infrared spectrometer (ATR-FTIR, Escalab, USA), X-ray photoelectron spectroscopy (XPS, ThermoFisher Nexsa, USA), X-ray diffraction (XRD, Bruke D8 Advance, Germany), and thermogravimetric analysis (TGA, METTLER TOLEDO TGA/DSC1 analyzer, Germany), respectively. By using the inductively coupled plasma-mass spectrometry (ICP-MS, NexION 300, PerkinElmer Inc., USA; detection limit = 0.01 µg·L^–1), the effect of concentration, pH, operation pressure, competition ions on the Tl⁺-adsorption efficiency of PB_x@PDA/PEI-FP (x = 1, 2, and 3) samples were supervised via a filtration system. Prior to metal ions analysis, the aqueous samples were prepared according to previous literature. The humic acid concentrations, chemical oxygen demands, and NH₃-N contents of actual Peral River water before and after membrane filtration were measured by a UV-vis spectrometry (Tu1810, Persee, China) and the multi-functional water quality analyzer (GNST-900S, Genesite, China), respectively.

2.4 Membrane separation tests

All membrane samples with an effective diameter of 3.8 cm were conducted by a typical filtration system. The Tl⁺-adsorption kinetic process of membrane samples was tested by using the Tl⁺-bearing solution with various volume (500 and 100 mL), concentration (0.1, 0.5, and 1.0 ppm), pH value (3, 7, and 10), competition ions (Na⁺, K⁺, Ca²⁺, and Mg²⁺) as the feed solution in the filtration system, respectively. Particularly, the operation pressure from 0 to –0.1 MPa was adjusted by a vacuum pump. In each experiment, the membrane fluxes per 10 min were calculated based on Eq. (1). The Tl⁺ removals and adsorption capabilities of various membranes were assessed by Eqs. (2) and (3), respectively. According to pseudo-first-order model (Eq. (4)), the Tl⁺ adsorption kinetics of PB_x@PDA/PEI-FP (x = 1, 2, and 3) were discussed in detail. The filtration system coupling with an ICP-MS (NexION 300, PerkinElmer Inc., USA; detection limit = 0.01 µg·L^–1) was schematic in Fig.1. Each test with our as-prepared membranes in removing Tl⁺ was repeated for five times.

Fig.1 Schematic diagram of the membrane filtration experimental setup for the removal of Tl⁺.

Full size|PPT slide

(1)J=\mathrm{\Delta }V/(S\times t),

Where J is the permeate flux (L·m^–2·min^–1), ΔV is the volume (L) of filtrate, the value of S is fixed at 0.00113 m² in the filtration system, and t is the filtration time (min).

(2)R=\left(1-\frac{C_{\mathrm{p}}}{C_{\mathrm{o}}}\right)\times 100\mathrm{\%},

Where R is the Tl⁺ removal of the as-prepared membranes, C_o and C_p are the concentration (mg·L^–1) of Tl⁺ in the feed and filtrate, respectively.

(3)Q_{\mathrm{t}}=\left(\frac{C_{\mathrm{o}}-C_{\mathrm{p}}}{m}\right)\times \mathrm{\Delta }V,

Where Q_t is the Tl⁺ adsorption capability (mg·g^–1) of membranes, ΔV is the volume (L) of filtrate, m is the weight (g) of membranes, C_o and C_p are the concentration (mg·L^–1) of Tl⁺ in the feed and filtrate, respectively.

(4)\mathrm{Ln}\left(Q_{\mathrm{e}}-Q_{\mathrm{t}}\right)=\mathrm{Ln}{Q}_{\mathrm{e}}-k_{1}t,

Where Q_t is the real-time adsorption capacity (mg·g^–1) of membrane samples, Q_e is the equilibrium adsorption capacity (mg·g^–1) of membrane samples, k₁ (mg·g^–1·min^–1) is the adsorption constant.

3 Results and discussion

3.1 Surface morphologies and chemical properties

The preparation processes and characterization of the PB_x@PDA/PEI-FP (x = 1, 2, and 3) membranes were shown in Fig.2. By surface amination and in situ mineralization, the membranes are manufactured in Fig.2(a). The microstructure of pristine FP membrane surface is a loose reticular structure consisting of interconnected, continuous, and ribbon-like fibers (Fig.2(b)). The surface morphology of ribbon-like fibers was enlarged in Fig.2(b₁), which showed a smooth surface. After PDA/PEI modification, the surface morphology (Fig.2(c) and Fig.2(c₁)) of PDA/PEI-FP membrane exhibited no obvious change in comparison with pristine FP. However, as shown in Fig.2(c₂), we observed that the amount of N element was distinctly increased on PDA/PEI-FP membrane. This result indicated that a uniform and smooth co-deposition coating was formed on the FP substrate by the Michael addition or Shiff base reaction between PDA and PEI (Fig.2(a)) [27]. By virtue of the PDA/PEI coating, PDA/PEI-FP membrane surface could provide large quantities of functional sites (e.g., –NH₂ and –OH) to coordinate with metal ions. Thus, as exhibited in Fig.2(d) and Fig.2(d₁), a few cubic PB nanoparticles were in situ grown on the PDA/PEI-FP surface after immersion in Na₄Fe(CN)₆ aqueous solution. Meanwhile, the EDS elemental maps confirmed that the N and Fe elements were highly dispersed in the resultant PBs with the cubic outlines (Fig.2(d₂) and Fig.2(d₃)). In stark contrast, PB₂@PDA/PEI-FP membrane surface exhibited much more distributed cubic PBs after redecoration with PDA/PEI and incubation in Na₄Fe(CN)₆ aqueous solution again (Fig.2(a)), and the PBs particles were more uniform and thicker than the PB₁@PDA/PEI-FP, which almost shielded the typically ribbon-like fibers of membrane substrate (Fig.2(e) and Fig.2(e₁)). Correspondingly, as shown in Fig.2(e₂) and Fig.2(e₃), the elemental maps of N and Fe on PB₂@PDA/PEI-FP membrane surface became denser, blurring the outlines of cubic PBs. These observations indicate that the surface of PB₂@PDA/PEI-FP membrane covers more PBs than that of PB₁@PDA/PEI-FP. Furthermore, with the aid of the third PDA/PEI layer, PB₃@PDA/PEI-FP membrane surface was entirely prevailed with cubic PBs (Fig.2(f)), and the crystal size and elemental mapping (N and Fe) densities of surface-decorated PBs were visibly enlarged (Fig.2(f₁–f₃)). As shown in Fig. S1 (cf. ESM), FP and PDA/PEI-FP membranes without the decoration of PBs lost almost all weight when the pyrolysis temperature increased from 50 to 700 °C, but the PB_x@PDA/PEI-FP (x = 1, 2, and 3) membranes showed residual weight after pyrolysis. The residual weight of PB_x@PDA/PEI-FP (x = 1, 2 and 3) was ~4.5, 7.2, and 10.8 wt %, respectively, which could be contributed to the PB species. All the results indicated that the multilayer PB particles were successfully coated on the FP membrane substrates via PDA/PEI intermediate layers.

Fig.2 Preparation and characterization of the PB_x@PDA/PEI-FP membranes. (a) Schematic illustration of the membrane fabrication process via surface amination and in situ mineralization. SEM images of (b, b₁) the pristine FP membrane, (c, c₁) the PDA/PEI-FP membrane, (d, d₁) the PB₁@PDA/PEI-FP membrane, (e, e₁) the PB₂@PDA/PEI-FP membrane, and (f, f₁) the PB₃@PDA/PEI-FP membrane. EDX mapping of element N and Fe on (b₂, b₃) the pristine FP membrane, (c₂, c₃) the PDA/PEI-FP membrane, (d₂, d₃) the PB₁@PDA/PEI-FP membrane, (e₂, e₃) the PB₂@PDA/PEI-FP membrane, and (f₂, f₃) the PB₃@PDA/PEI-FP membrane.

Full size|PPT slide

To detect the functional groups on the membrane surfaces after PDA/PEI modification and PB decoration, FTIR, XPS and XRD measurements were conducted in Fig.3. Fig.3(a₁) showed the ATR-FTIR spectra of the pristine FP. The peaks at 3457, 3025, 2979 and 1660 cm^–1 were respectively ascribed to the –OH, –CH₃, –CH₂, and –C=O stretching in acetylated glucose units, demonstrating the cellulose structure in FP [28]. In the spectrum of PDA/PEI-FP (Fig.3(a₂)), the peak of –OH (3457 cm^–1) shifted to low position (3357 cm^–1), and its intensity became weak, indicating the reduction of hydrogen bond associated peaks among the –OH groups. Based on previous literature, the –NH₂ groups often cause stronger hydrogen bond, and thus have the powerful ability in weakening the associated –OH groups. For this reason, the peak at 3357 cm^–1 (Fig.3(a₂)) was attributed to the –OH and –NH₂ in PDA/PEI-FP, in which, the –NH₂ groups were offered by the PDA/PEI layer [29]. Meanwhile, Fig.3(a₂) shows the other characteristic peak of –C=O at 1706 cm^–1 due to the existence of aromatic ring in PDA. Moreover, a mild peak at 1340 cm^–1 attributing to the ring stretching, –N–H deformation and –C=N stretching was observed in Fig.3(a₂), which implied the Michael addition or Shiff base reaction between the amine of the PEI and the catechol group of the PDA [27]. These results revealed that a crosslinked co-deposition PDA/PEI layer was formed on the PDA/PEI-FP. After PDA/PEI-FP incubation in Na₄Fe(CN)₆ aqueous solution, a distinct band at 2043 cm^–1 appeared in FTIR spectra of PB₁@PDA/PEI-FP (Fig.3(a₃)), which was ascribed to the stretching vibration mode of the –C≡N bond belonging to Fe(CN)₆^4–, indicating the decoration of PBs on membrane surface [30]. Similarly, PB₂@PDA/PEI-FP (Fig.3(a₄)) and PB₃@PDA/PEI-FP (Fig.3(a₅)) both displayed the obvious –C≡N bond, confirming the mineralization of PBs on their membrane surfaces. However, the characteristic alkyl peaks (–CH₃, –CH₂) at 3025 and 2979 cm^–1 still appeared in Fig.3(a₃), but were almost absent in Fig.3(a₄) and Fig.3(a₅), suggesting that PBs coverage on PB₂@PDA/PEI-FP and PB₃@PDA/PEI-FP surfaces was higher than that of PB₁@PDA/PEI-FP. These results are good coincidence with the above SEM results.

Fig.3 (a₁–a₅) The FTIR spectrums of FP, PDA/PEI-FP, PB₁@PDA/PEI-FP, PB₂@PDA/PEI-FP, and PB₃@PDA/PEI-FP; high-resolution XPS spectra of (b₁–b₅) Fe 2p and (c₁–c₅) N 1s for FP, PDA/PEI-FP, PB₁@PDA/PEI-FP, PB₂@PDA/PEI-FP, and PB₃@PDA/PEI-FP surfaces; (d₁–d₅) XRD patterns of FP, PDA/PEI-FP, PB₁@PDA/PEI-FP, PB₂@PDA/PEI-FP, and PB₃@PDA/PEI-FP.

Full size|PPT slide

The chemical components of all membrane samples are shown in Fig. S2 (cf. ESM). Peaks at 288 and 533 eV from FP surface are ascribed to the binding energy of C 1s and O 1s, respectively (Fig. S2(a)). For PDA/PEI-FP, a new peak at 399 eV attributed to N (1s) has occurred in the spectra curve of Fig. S2(a) [18,31]. The corresponding values of element composition are listed in Fig. S2(b). The N content of PDA/PEI-FP is ~6.7 wt %. Compared with the FP and PDA/PEI-FP membranes, all PB_x@PDA/PEI-FP (x = 1, 2, and 3) membranes exhibits new binding energy peaks at 700 eV ascribed to Fe 2p (Fig. S2(a)). The mass percentage of membrane surface-loaded Fe increased from 1.7 to 4.1 wt % along with the PB layers increasing from 1 to 3 (Fig. S2(b)). The high-resolution XPS of Fe 2p and N 1s was further adopted to analyze the changing of main chemical groups on those membrane surfaces. As displayed in Figs. S3(b₁–b₅), no peaks are observed for the FP and PDA/PEI-FP membranes, while four peaks at 720.7, 723.2 708 and 711 eV attributing to the Fe 2p_1/2 and Fe 2p_3/2 are distinctly presented on PB_x@PDA/PEI-FP (x = 1, 2, and 3) membranes, revealing the existence of Fe²⁺ and Fe³⁺ bridged by the cyano ligands (–C≡N) in those membrane surface-decorated PBs [32,33]. Furthermore, compared with the PDA/PEI-FP (Fig.3(c₂)), the peak at 398.3 eV ascribed to –NH has disappeared, but for PB_x@PDA/PEI-FP (x = 1, 2, and 3) membranes, two new peaks at 400 and 397 eV assigned to –N–C and –N-metal are found (Fig.3(c₃–c₅)), respectively. The loss of amino (–NH) functionalities demonstrated the PBs were in situ grown from the activity sites of amino [25,31]. XRD patterns (Fig.3(d₁–d₅)) were utilized to characterize the crystal type of membrane immobilized PB cubes. PB_x@PDA/PEI-FP (x = 1, 2, and 3) membranes all displayed the same characteristic peaks at 20°, 29.2°, 40.8°, 45.7°, 58.5°, 61.8° and 68.1° like those of pure PB cubic crystals, corresponding to (100), (110), (210), (211), (311), (222), and (321), respectively. This verified that the typical cubic face-centered lattice of PB crystals was located on the FP membrane surface (Fig.3(d₃–d₅)).

3.2 The adsorption performance of PB_x@PDA/PEI-FP for Tl⁺ under filtration

The dynamic membrane adsorption of Tl⁺ was evaluated in a filtration system with 500 mL Tl⁺-containing feed water. The operation pressure is controlled at –0.07 MPa. For each PB_x@PDA/PEI-FP (x = 1, 2, and 3) membrane, its Tl⁺ removal, flux, and Tl⁺ adsorption capacity per treating 100 mL of feed solution were shown in Fig.4, based on the recorded time and Eqs. (1–4). The relationship between membrane filtration performance and the size distribution of cubic PB particles is shown in Fig.4. With an increase in layers of surface-decorated PB, the Tl⁺ removal by PB-based membrane is enhanced (Fig.4(a)); however, the fluxes of PB-based membranes are declined (Fig.4(b)). The membrane fluxes are interconnected with membrane porosity and membrane pore sizes. The increase of PB layers has negative effect on the porosities and pore sizes of PB-based membranes, and thus reduces the membrane fluxes. Compared with PB₁@PDA/PEI-FP and PB₂@PDA/PEI-FP, PB₃@PDA/PEI-FP spends double time in filtrating the same volume of feed solution. The solid-liquid contact is most sufficient for the PB₃@PDA/PEI-FP when the same volume of Tl⁺-containing water transports through the PB-based membranes. Theoretically, PB₃@PDA/PEI-FP should display the largest Tl⁺ adsorption capacity (Q_e) and dynamic constant (k), but the two values (Q_e = 0.35 mg·g^–1, k = 0.05) are both lower than those (Q_e = 0.41 mg·g^–1, k = 0.1) of PB₂@PDA/PEI-FP according to the pseudo-first-order model (Fig.4(c)). More surface-decorated PB layers cannot endow the membranes with better Tl⁺-adsorption performance. As shown in Fig.4(d–f), PB coverage and PB size distribution on PB-based membranes are both rising with the increase of surface-coated PB layer. Obviously, a trade-off is occurred in Tl⁺ adsorption by as-prepared PB-based membranes, in which more PB coverage causes the improvement of Tl⁺ removal, but bigger PB particle size weakens the Tl⁺-adsorption ability [34]. Moreover, the Tl⁺ removal by FP and PDA/PEI-FP was also tested via the same filtration system. As exhibited in Fig. S3 (cf. ESM), we observe that the Tl⁺ removal is much lower than that of PB₂@PDA/PEI-FP. Obviously, the membrane matrix and its PDA/PEI mediated layer have no significant effect on Tl⁺ adsorption.

Fig.4 Membrane filtration performance: (a) Tl⁺ removal, (b) fluxes and (c) membrane adsorption capacity of Tl⁺. (d–f) The size distribution of cubic PB particles on PB₁@PDA/PEI-FP, PB₂@PDA/PEI-FP, PB₃@PDA/PEI-FP (the filtration condition: Tl⁺ concentration is 0.5 ppm, feed solution volume is 500 mL, temperature is 25 °C, vacuum pressure is –0.07 MPa).

Full size|PPT slide

3.3 Effects of the temperature, pH and coexisting ions on Tl⁺ adsorption for membranes

As discussed above, PB₂@PDA/PEI-FP displayed the best Tl⁺ adsorption performance. However, facing the complicate freshwater environmental, the Tl⁺ adsorption properties of PB₂@PDA/PEI-FP would undergo a decline, resulting from the variation of adsorption energy, adsorption sites and adsorption pores. For this reason, the effect of temperature, pH and coexisting ions on the Tl⁺ adsorption performance of PB₂@PDA/PEI-FP are revealed in Fig.5. We have observed that the increase in temperature significantly promoted the Tl⁺ removal, while shortening the filtration time, thereby boosting the k and fluxes of PB₂@PDA/PEI-FP during the filtration process (Fig.5(a) and 5(a₁)). The k increased with an increase in temperature from 25 to 45 °C, demonstrating that the adsorption of Tl⁺ by PB₂@PDA/PEI-FP is endothermic [35]. A sharp decline in Tl⁺ removal by PB₂@PDA/PEI-FP at pH 3.0 is observed in Fig.5(b) and 5(b₁). Similar to the results reported in previous literature, more hydrogen ions could lead to the protonation on the Tl⁺-adsorbent surfaces, resulting in electrostatic repulsion with Tl⁺, which was unfavorable for Tl⁺ adsorption [18,36]. On the contrary, higher pH made the surface of PB₂@PDA/PEI-FP negatively-charged and facilitated Tl⁺ adsorption. Therefore, the Tl⁺ removal of PB₂@PDA/PEI-FP was reached up to 97% at pH 10.0 (Fig.5(b)).

Fig.5 Effect of (a, a₁) temperature, (b, b₁) pH and (c, c₁) coexisting ions on Tl⁺ removal/dynamic adsorption constant (k)/fluxes, for PB₂@PDA/PEI-FP with 100 mL of Tl⁺-containing water filtrating (initial Tl⁺ concentration is 0.5 ppm, the vacuum pressure is –0.07 MPa, the concentration of each coexisting ions (e.g., Mg²⁺, Ca²⁺, Na⁺, K⁺) are 10 ppm, respectively).

Full size|PPT slide

Lastly, we investigated the effect of four common coexisting ions (e.g. Mg²⁺, Ca²⁺, Na⁺, K⁺) on Tl⁺ adsorption at both pH 7.0 and pH 10.0 (Fig.5(c) and Fig.5(c₁)). In the presence of the divalent cations (Ca²⁺ and Mg²⁺), we found that Ca²⁺ was more effective than Mg²⁺ in reducing the uptake of Tl⁺ at equivalent concentrations and pH value. Furthermore, compared with divalent cation (Ca²⁺), the monovalent cations (Na⁺ and K⁺) led to more reduction in Tl⁺ removal and k value. Especially for the K⁺, its existence severely interfered with Tl⁺ adsorption (Fig.5(c)). This is because that the surface-decorated PBs have an open zeolite-like structure with the pore size close to a hydrated Tl⁺ ion, which endows the membrane (PB₂@PDA/PEI-FP) with Tl⁺ adsorption ability [1,37,38]. Correspondingly, competitive adsorption in PBs due to obstruction of coexisting ions might be responsible for the observed decrease in Tl⁺ uptake [28]. A previous study has shown that the hydration radius of K⁺ was very close to that of Tl⁺, followed by Na⁺, Ca²⁺ and Mg²⁺, matching the disrupting power of Tl⁺ adsorption by PB₂@PDA/PEI-FP [39,40]. However, we observed that the increasing of pH (e.g. pH = 10) could effectively eliminate the effect of coexisting ions on the Tl⁺ uptake of PB₂@PDA/PEI-FP (Fig.5(c)), suggesting the high selectivity in alkaline condition.

3.4 The effect of operation pressure on Tl⁺ uptake of membrane

The Tl⁺ removal, flux, filtration time, and k versus the vacuum pressure are shown in Fig.6. Fig.6(a) illustrates that the Tl⁺ removal and flux of PB₂@PDA/PEI-FP are strongly dependent on the operating pressure. A greater increase in flux at operating pressure above 0.05 MPa is observed. Tl⁺ removal of ~100% is achieved without any pressure driving, while a linear decrease in Tl⁺ removal accompanying with the increase of operating pressure is occurred. These results suggest that the PB₂@PDA/PEI-FP could well act as the GDFM to purify the Tl⁺-containing water [41].

Fig.6 (a) The Tl⁺ removal and fluxes of the PB₂@PDA/PEI-FP while filtrating the 100 mL of Tl⁺-containg water (0.5 ppm, pH = 7.0) as the operation pressure arranging from 0 to –0.1 MPa; (b) the Tl⁺ removal and fluxes and (c) Q_t versus time of the PB₂@PDA/PEI-FP within 100 min while filtrating the 500 mL of Tl⁺-containg water (0.5 ppm, pH = 7.0) under gravity-driven filtration (GDF).

Full size|PPT slide

Furthermore, we exhilaratingly found that very high Tl⁺ removal over 90% was maintained during the 100 min of GDF even though the filtrating volume was added to 500 mL (Fig.6(b)). However, the fluxes of PB₂@PDA/PEI-FP underwent a decline with the feed water decreasing, resulting from the reduction of gravity-driven pressure (Fig.6(b)). The corresponding adsorption kinetics of Tl⁺ are discussed for PB₂@PDA/PEI-FP based on pseudo-first-order model due to the very high R² value (0.99). It shows the k and theoretical maximum adsorption capacitiy of Tl⁺ are 0.047 min^–1 and 1.97 mg·g^–1, indicating the better Tl⁺ uptake under gravity-driven pressure. After Tl⁺ adsorption, PB₂@PDA/PEI-FP could maintain its structure (Fig. S4, cf. ESM). Since the characteristics of PB contains channels of approximately 3.2 Å, which enables PB to be compatible with the size of hydrated Tl⁺ ions. There is a potential mechanism that is an interaction between the Tl⁺ ions and the cyano group of PB to trap the Tl⁺ ions in the lattice of PB.

3.5 The feasibility of gravity-driven membranes in purifying the actual Tl⁺-containing water

Tl can be released to the environment during the exploration and utilization of Tl-bearing mineral resources, which has led to a large number of Tl contamination incidents reported worldwide [42,43]. In fact, the release concentration of Tl⁺ from Tl-bearing mineral resources (average 0.5 mg·kg^–1) is very low in the potable water [44,45]. For this reason, the experimental Tl⁺ concentration in this section is reduced to 0.01 ppm. We observed that PB₂@PDA/PEI-FP presented the excellent Tl⁺ removal over 95% for treating the ultra-low Tl⁺-containing water during the 100 min of GDF process (Fig. S5(a), cf. ESM). The residual Tl⁺ in filtrate was lower than 0.5 μg·L^–1 after the feed Tl⁺-containing water passed through PB₂@PDA/PEI-FP, which met to the drinking water standard of USEPA. Meanwhile, according to the well-fitting pseudo-first-order model, Fig. S5(b) showed the k and Q_e of Tl⁺ were 0.018 min^–1 and 0.52 mg·g^–1, respectively.

However, the components of actual water source are complicate. For instance, as exhibited in Table S1, the actual Pearl River water, located in Guangzhou Higher Education Mega Center, contains many species of metal cations and anions. This section assumed that trace Tl⁺ was discharged into the pearl river, leading to the intractable water crisis due to an extensive range of co-existing ions. To purify it, the GDF membrane (PB₂@PDA/PEI-FP) was utilized to treat the Tl⁺-polluted pearl river water in Fig.7. The PB₂@PDA/PEI-FP with a small area of ~9.0 cm² still showed a very high performance of Tl⁺ removal more than ~80% after filtrating a large volume (500 mL) of Tl⁺-polluted pearl river water (0.01 ppm) (Fig.7(a)). Furthermore, Fig.7(b) revealed that the k and Q_e of Tl⁺ were 0.012 min^–1 and 0.38 mg·g^–1, respectively which were slightly lower than those in treating the prepared Tl⁺-contaminated water in laboratory, indicating the strong anti-interference with co-existing ions. The residual Tl⁺ in the filtrate was less than 2 μg·L^–1, which met the drinking water standards of USEPA, ensuring the safety of drinking water (Fig.7(c)). Hence, our PB₂@PDA/PEI-FP has the potential to capture Tl⁺ in the actual Tl⁺-polluted water source without energy consumption, which encourages extensive applications of membrane technologies for decentralized drinking water supply.

Fig.7 (a) The Tl⁺ removal and fluxes, and (b) the kinetic analysis of PB₂@PDA/PEI-FP in remedying the Tl⁺ contaminated actual pearl river water (0.01 ppm, pH = 7.0, room temperature) via the GDF operation; (c) schematic illustration of the removal of Tl⁺ from Tl⁺-polluted water source via the emergency measure with our as-prepared membrane separation.

Full size|PPT slide

As we mentioned in the section of “Introduction,” Wang et al. and Shi et al. both reported that the as-fabricated PB-based membranes exhibited an outstanding Tl⁺ removal efficiency (> 90%) at various operating conditions and co-existing competing cations. Specifically, Wang et al. investigated the Tl⁺ removal rate was over 90% with a permeate flux of ~100 L·m^–2·h^–1 using 1 mg·L^–1 Tl⁺ feed solution (corresponding to a Tl⁺ content of 95.8 ± 9.6 mg·g^–1 in this study). Shi et al. evaluated the removal efficiency was over 90% with a pressure of 0.1 MPa using 0.5 mg·L^–1 Tl⁺ feed solution (corresponding to theoretical maximum Tl⁺ adsorption capacity of 460.4 mg·m^–2, the Tl adsorption constant (k) of 1.8 min^–1 in this study).

However, our study investigated the Tl⁺ removal rate was nearly 100% using 0.5 mg·L^–1 Tl⁺ feed solution (the k and theoretical maximum adsorption capacitiy of Tl⁺ were 0.047 min^–1 and 1.97 mg·g^–1, respectively). In cotrast to the previous study, our experiment do not require pressure to achieve a high removal rate, which really reduces energy consumption and operation cost. Plus, our membrane supporters are originated from low-cost biodegradable FPs, unlike commercial membranes like fluoropolymers which is expensive and causing refractory solid waste pollution after discarding.

4 Conclusions

In this work, the multilayer PB overcoats on FP membrane substrates were successfully fabricated via several PDA/PEI intermediate layers for Tl removal. By virtue of the simple filtration system, more surface-decorated PB layers cannot endow the PB-based membranes with better Tl⁺-adsorption performance due to the trade-off between PB decoration layer number and PB distribution sizes. The double-layered PB-based membrane (PB₂@PDA/PEI-FP) displays the maximum sorption capacity, but its Tl⁺ uptake performance is reduced by the acid, coexisting ions (K⁺ and Na⁺) and powerful operation pressure during filtrating a large volume of low-concentrated Tl⁺-containing water. However, the increasing of pH (e.g. pH = 10) could effectively avoid the negative effect of coexisting ions (e.g. Na⁺ and K⁺) on the Tl⁺ uptake by PB₂@PDA/PEI-FP. And the Tl⁺ removal was increased up to 100% as the operation pressure declined to 0, revealing the highly-efficient Tl⁺ adsorption in energy-free filtration. Most importantly, PB₂@PDA/PEI-FP showed the satisfactory Tl⁺ removal in filtrate to meet the drinking water standard of USEPA after purifying the Tl⁺-polluted Pearl River water with energy-free operation, demonstrating the efficiency and high-selectivity. This work provides a feasible avenue to guarantee the actually potable water in remote area via the energy-free operation.

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

The current study was financially supported by the National Natural Science Foundation of China (Grant Nos. 22006026, 52270001), Guangdong Basic and Applied Basic Research Foundation (Grant Nos. 2023A1515012506, 2019A1515110546), Science and Technology Program of Guangzhou (Grant No. 202102080160), Project of Young Innovative Talents in Colleges and Universities of Guangdong Province (Grant No. 2019KQNCX111), Outstanding Youth Project of Guangdong Natural Science Foundation (Grant No. 2022B1515020030), Guangzhou Science and Technology Project (Grant Nos. 202201020530, 202201020200), Research Project of Guangzhou University (Grant No. YJ2023026).

Electronic Supplementary Material

Supplementary material is available in the online version of this article at https://doi.org/10.1007/s11705-023-2379-8 and is accessible for authorized users.

References

1 López Y , Reguera E . Magnetic Prussian blue derivative like absorbent cages for an efficient thallium removal. Journal of Cleaner Production, 2021, 283: 124587

2 Cheam V . Thallium contamination of water in Canada. Water Quality Research Journal of Canada, 2001, 36(4): 851–877

3 Zitko V . Toxicity and pollution potential of thallium. Science of the Total Environment, 1975, 4(2): 185–192

4 Li H , Chen J , Long J , Li X , Jiang D , Zhang P , Qi J , Huang X , Liu J , Xu R . . Removal of thallium from aqueous solutions using Fe-Mn binary oxides. Journal of Hazardous Materials, 2017, 338: 296–305

5 Liu J , Wang J , Tsang D , Xiao T , Chen Y , Hou L . Emerging thallium pollution in China and source tracing by thallium isotopes. Environmental Science & Technology, 2018, 52(21): 11977–11979

6 Li H , Chen J , Long J , Jiang D , Liu J , Li S , Qi J , Zhang P , Wang J , Gong J . . Simultaneous removal of thallium and chloride from a highly saline industrial wastewater using modified anion exchange resins. Journal of Hazardous Materials, 2017, 333: 179–185

7 Sinyakova M , Semenova E A , Gamuletskaya O A . Ion exchange of copper(II), lanthanum(III), thallium(I), and mercury(II) on the “polysurmin” substance. Russian Journal of General Chemistry, 2014, 84(13): 2516–2520

8 Li Z , Liu C , Ma R , Yu Y , Chang Z , Zhang X , Yang C , Chen D , Yu Y , Li W . . Rapid removal of thallium from water by a new magnetic nano-composite using graphene oxide for efficient separation. International Biodeterioration & Biodegradation, 2021, 161: 105245

9 Zhao Z , Xiong Y , Cheng X , Hou X , Yang Y , Tian Y , You J , Xu L . Adsorptive removal of trace thallium(I) from wastewater: a review and new perspectives. Journal of Hazardous Materials, 2020, 393: 122378

10 EscuderoLWuilloudR GOlsinaR A. Sensitive determination of thallium species in drinking and natural water by ionic liquid-assisted ion-pairing liquid-liquid microextraction and inductively coupled plasma mass spectrometry. Journal of Hazardous Materials, 2013, 244–245: 380–386

11 Yang Y , Xiao J , Shen Y , Liu X , Li W , Wang W , Yang Y . The efficient removal of thallium from sintering flue gas desulfurization wastewater in ferrous metallurgy using emulsion liquid membrane. Environmental Science and Pollution Research International, 2017, 24(31): 24214–24222

12 Ussipbekova Y , Seilkhanova G , Jeyabharathi C , Scholz F , Kurbatov A , Nauryzbaev M , Berezovskiy A . Electrochemical deposition and dissolution of thallium from sulfate solutions. International Journal of Analytical Chemistry, 2015, 7: 357–514

13 Pei X , Gan L , Tong Z , Gao H , Meng S , Zhang W , Wang P , Chen Y . Robust cellulose-based composite adsorption membrane for heavy metal removal. Journal of Hazardous Materials, 2021, 406: 124746

14 Abdullah N , Yusof N , Lau W J , Jaafar J , Ismail A F . Recent trends of heavy metal removal from water/wastewater by membrane technologies. Journal of Industrial and Engineering Chemistry, 2019, 76: 17–38

15 Tian J , Chang H , Zhang R . How to fabricate a negatively charged NF membrane for heavy metal removal via the interfacial polymerization between PIP and TMC?. Desalination, 2020, 491: 114499

16 Efome J , Rana D , Matsuura T , Lan C Q . Insight studies on metal-organic framework nanofibrous membrane adsorption and activation for heavy metal ions removal from aqueous solution. ACS Applied Materials & Interfaces, 2018, 10(22): 18619–18629

17 Wang Z , Liu S , Zhang H , Zhang Z , Jiang J , He D , Lin S . Thallium mining from industrial wastewaters enabled by a dynamic composite membrane process. Resources, Conservation and Recycling, 2022, 186: 106577

18 Shi Y , Huang L , Mahmud S , Zhang G , Li H , Wang Y , Xiao T , Zeng Q , Liu Z , Yu H . . High-efficiently capturing trace thallium(I) from wastewater via the Prussian blue@polytetrafluoroethylene hybrid membranes. Chemical Engineering Journal, 2023, 451: 138712

19 Bhattacharjee T , Islam M , Chowdhury D , Majumdar G . In-situ generated carbon dot modified filter paper for heavy metals removal in water. Environmental Nanotechnology, Monitoring & Management, 2021, 16: 100582

20 Kim H , Wi H , Kang S , Yoon S , Bae S , Hwang Y . Prussian blue immobilized cellulosic filter for the removal of aqueous cesium. Science of the Total Environment, 2019, 670: 779–788

21 Lin H , Fang Q , Wang W , Li G , Guan J , Shen Y , Ye J , Liu F . Prussian blue/PVDF catalytic membrane with exceptional and stable Fenton oxidation performance for organic pollutants removal. Applied Catalysis B: Environmental, 2020, 273: 119047

22 Qiu W , Yang H , Xu Z . Dopamine-assisted co-deposition: an emerging and promising strategy for surface modification. Advances in Colloid and Interface Science, 2018, 256: 111–125

23 Pi J , Yang H , Wan L , Wu J , Xu Z . Polypropylene microfiltration membranes modified with TiO₂ nanoparticles for surface wettability and antifouling property. Journal of Membrane Science, 2016, 500: 8–15

24 Qu F , Cao A , Yang Y , Mahmud S , Su P , Yang J , He Z , Lai Q , Zhu L , Tu Z . . Hierarchically superhydrophilic poly(vinylidene fluoride) membrane with self-cleaning fabricated by surface mineralization for stable separation of oily wastewater. Journal of Membrane Science, 2021, 640: 119864

25 Yang Y , Lai Q , Mahmud S , Lu J , Zhang G , Huang Z , Wu Q , Zeng Q , Huang Y , Lei H . . Potocatalytic antifouling membrane with dense nano-TiO₂ coating for efficient oil-in-water emulsion separation and self-cleaning. Journal of Membrane Science, 2022, 645: 120–204

26 Lv Y , Zhang C , He A , Yang S , Wu G , Darling S , Xu Z . Photocatalytic nanofiltration membranes with self-cleaning property for wastewater treatment. Advanced Functional Materials, 2017, 27(27): 1700251

27 Lv Y , Yang S , Du Y , Xu Z . Co-deposition kinetics of polydopamine/polyethyleneimine coatings: effects of solution composition and substrate surface. Langmuir, 2018, 34(44): 13123–13131

28 Mondal S , Ganguly S , Das P , Bhawal P , Das T , Nayak L , Khastgir D , Das N . High-performance carbon nanofiber coated cellulose filter paper for electromagnetic interference shielding. Cellulose (London, England), 2017, 24(11): 5117–5131

29 Yu F , Chen S , Chen Y , Li H , Yang L , Chen Y , Yin Y . Experimental and theoretical analysis of polymerization reaction process on the polydopamine membranes and its corrosion protection properties for 304 stainless steel. Journal of Molecular Structure, 2010, 982(1): 152–161

30 Qian J , Zhou L , Yang X , Hua D , Wu N . Prussian blue analogue functionalized magnetic microgels with ionized chitosan for the cleaning of cesium-contaminated clay. Journal of Hazardous Materials, 2020, 386: 121965

31 Ederer J , Janoš P , Ecorchard P , Tolasz J , Štengl V , Beneš H , Perchacz M , Pop-Georgievski O . Determination of amino groups on functionalized graphene oxide for polyurethane nanomaterials: XPS quantitation vs. functional speciation. RSC Advances, 2017, 7(21): 12464–12473

32 Forment A A , Weitz R , Sagar A , Lee E , Konuma M , Burghard M , Kern K . Strong p-type doping of individual carbon nanotubes by Prussian blue functionalization. Small, 2008, 4(10): 1671–1675

33 Cano A , Rodríguez-Hernández J , Reguera L , Rodríguez-Castellón E , Reguera E . On the scope of XPS as sensor in coordination chemistry of transition metal hexacyanometallates. European Journal of Inorganic Chemistry, 2019, 13(13): 1724–1732

34 Yuan Z , Zhao R , Sun G , Li P , Yin S , Zhou G , He G , Jiang X . Membrane flux response technology for early warning of initial surface scaling in membrane distillation. Journal of Water Process Engineering, 2023, 55: 104104

35 Huang G , Chen J , Dou P , Yang X , Zhang L . in situ electrosynthesis of magnetic Prussian blue/ferrite composites for removal of cesium in aqueous radioactive waste. Journal of Radioanalytical and Nuclear Chemistry, 2020, 323(1): 557–565

36 Wang H , Liu J , Yao J , He Q , Ma J , Chai H , Liu C , Hu X , Chen Y , Zou Y . . Transport of Tl(I) in water-saturated porous media: role of carbonate, phosphate and macromolecular organic matter. Water Research, 2020, 186: 116325

37 Tansel B . Significance of thermodynamic and physical characteristics on permeation of ions during membrane separation: hydrated radius, hydration free energy and viscous effects. Separation and Purification Technology, 2012, 86: 119–126

38 Zhang H , Qi J , Liu F , Wang Z , Ma X , He D . One-pot synthesis of magnetic Prussian blue for the highly selective removal of thallium(I) from wastewater: mechanism and implications. Journal of Hazardous Materials, 2022, 423: 126972

39 Wick S , Baeyens B , Marques F M , Voegelin A . Thallium adsorption onto illite. Environmental Science & Technology, 2018, 52(2): 571–580

40 Vincent T , Taulemesse J , Dauvergne A , Chanut T , Testa F , Guibal E . Thallium(I) sorption using Prussian blue immobilized in alginate capsules. Carbohydrate Polymers, 2014, 99: 517–526

41 Ma X , Wang Y , Tong L , Luo J , Chen R , Wang Y , Guo X , Wang J , Zhou Z , Qi J . . Gravity-driven membrane system treating heavy metals-containing secondary effluent: improved removal of heavy metals and mechanism. Chemosphere, 2023, 339: 139590

42 Belzile N , Chen Y . Thallium in the environment: a critical review focused on natural waters, soils, sediments and airborne particles. Applied Geochemistry, 2017, 84: 218–243

43 Karbowska B . Presence of thallium in the environment: sources of contaminations, distribution and monitoring methods. Environmental Monitoring and Assessment, 2016, 188(11): 640

44 Peter A , Viraraghavan T . Thallium: a review of public health and environmental concerns. Environment International, 2005, 31(4): 493–501

45 Liu J , Luo X , Sun Y , Tsang D , Qi J , Zhang W , Li N , Yin M , Wang J , Lippold H . . Thallium pollution in China and removal technologies for waters: a review. Environment International, 2019, 126: 771–790