Materials and chemicals

Detailed information on all materials and chemicals used in this work is provided in the Supporting Information. All materials and chemicals were used without further purification, except for the commercial Reillex 425 resins, which were washed to neutral pH (6–7) with ultrapure water.

Preparation of the modified Reillex 425

Reillex 425 resins (1.0 g) were dispersed in 50 mL of ethanol, followed by the addition of iodomethane (0.667 mL), (2-Bromoethyl)trimethylammonium bromide (2-BETAB, 2.643 g), (3-Bromopropyl)trimethylammonium bromide (3-BPTAB, 2.793 g), and (5-Bromopentyl)trimethylammonium bromide (5-BPTAB, 3.093 g) to obtain the modified resins. These were labelled Reillex 425-Me, Reillex 425-C2, Reillex 425-C3, and Reillex 425-C5, respectively. The mixtures were heated to 60°C in a water bath shaker (Changzhou Guoyu Instrument Manufacturing Co., Ltd, WHY-2) in the dark for six days. The products were purified by washing five times with ultrapure water and once with ethanol. Modified Reillex 425 resins were obtained by drying under vacuum (60°C for 6 h).

Characterization of the adsorbents

Scanning electron microscopy (SEM, ZEISS, EVO 18, operating at 10 kV) was used to observe the morphologies of the resins. The specific surface areas and pore size distributions were calculated using the Brunauer–Emmett–Teller and Barrett–Joyner–Halenda methods, respectively. This was done using the N₂ adsorption–desorption isotherms measured using a gas adsorption analyzer (Quantachrome Autosorb iQ, operating at 77 K). The chemical structures of the resins were characterized by solid-state ¹³C cross-polarization magic angle spinning (CP-MAS) nuclear magnetic resonance (NMR) spectroscopy (Bruker AVANCE AV400 spectrometer), Fourier-transform infrared (FT-IR) spectroscopy (Thermo Nicolet 6700 spectrometer), and X-ray photoelectron spectroscopy (XPS, Thermo VG Scientific ESCALAB 250, equipped with a monochromatic Al K_α X-ray source at an energy of 1486.6 eV). A NanoBrook 90Plus PALS particle analyzer (Brookhaven Instruments) was used to measure the surface charges of the resins. Before measuring the zeta potential, the granular resins were ground into powders. 1.00 mg of each powder was soaked in 0.5 mL ethanol and subsequently diluted in 100 mL of water, yielding a concentration of 0.01 mg mL⁻¹.

Adsorption experiments

Batch adsorption experiments using ReO₄⁻ as an analog to radioactive TcO₄⁻ were conducted to evaluate the adsorption of different Reillex 425 resins. A single adsorption experiment was performed in 4 mL of Re solution (100 mg L⁻¹) containing 1.60 mg of adsorbent on a shaker. The solution pH was adjusted to the required value by adding negligible volumes (1 mol L⁻¹) of concentrated NaOH or HCl. The adsorbents were separated from the solution by filtration through a mixed cellulose ester membrane filter (0.22 μm) after adsorption. The following Eq. (1) was used to calculate the equilibrium adsorption amount (q_e): $q_{\text{e}}=\frac{\left(c_0-c_{\text{e}}\right)\times V}{m},$ (1) where c₀ (mg L⁻¹) and c_e (mg L⁻¹) are the initial and equilibrium concentrations of adsorbate ions in the solution, respectively, both of which were obtained using an inductively coupled plasma optical emission spectrometer (ICP-OES, PerkinElmer Optima 7300DV), m (g) is the weight of the adsorbent, and V (L) is the volume of the solution.

In selective adsorption experiments, competitive anions, including I⁻ and Sb(OH)₆⁻, were mixed with ReO₄⁻. Re and Sb concentrations were measured using ICP-OES, and those of I⁻ were determined using an inductively coupled plasma mass spectrometer (ICP-MS, Thermo-VG Scientific PlasmaQuad 3).

The reusability of the modified Reillex 425 was evaluated using adsorption–desorption cycles. The adsorbents were dispersed in 20 mL of HNO₃ (2 mol L⁻¹) and shaken for 6 h to desorb the ReO₄⁻. The desorbed adsorbents were collected by precipitation and purified by washing with ultrapure water. The adsorption–desorption cycles were repeated five times.

Before the adsorption of TcO₄⁻, 0.1 mL of NH₄TcO₄ stock solution was diluted by 1.9 mL of water, and mixed with 10 mL of Ultima Gold AB cocktail to determine the active concentration using a Packard Tri-Carb 2900 Liquid Scintillation Analyzer. The active ⁹⁹Tc concentration of the stock solution was 1250 Bq mL⁻¹ (2 mg L⁻¹). The batch adsorption experiment for Tc was carried out in a TcO₄⁻ solution obtained by diluting the stock solution 10 times (125 Bq mL⁻¹) using the same procedure as for the adsorption of Re. After adsorption, 3 mL of the supernatant was mixed with 10 mL of Ultima Gold AB cocktail to measure the active concentration, and the Tc removal ratio was calculated using Eq. (2), $\text{Removal ratio}=\left(1-\frac{A_{\text{t}}}{A_0}\right)\times 100\text{\%,}$ (2) where A₀ (Bq) and A_t (Bq) are the radioactivity of ⁹⁹Tc in solution before and after adsorption, respectively.

Synthesis of the modified Reillex 425

Reillex 425 resins were modified via a simple quaternization reaction, as illustrated in Scheme 1. The operation strongly affected the morphologies of the obtained products. The resin granules became powders when the reaction was performed under stirring (Fig. S1), which is undesirable for column treatment of radioactive sewage. To avoid pulverization and maintain the granular appearance of the products, modification must be carried out using a shaker. As shown in Figs. 1a–1e, although modification in a shaker roughened the resin surface, the modified Reillex 425 resins retained their spherical shape with an average bead size around 0.7 mm (Table 1). The resultant Reillex 425-Cn (n = 2, 3, and 5, representing the number of carbon atoms between the pyridinium and ammonium moieties) showed the same color as the pristine Reillex 425 resin, whereas Reillex 425-Me turned yellow because of adsorption of iodine during modification [33]. All modified resins displayed the same type II N₂ adsorption–desorption isotherms as pristine Reillex 425 (Fig. 1f). In the high-relative-pressure region (P/P₀ > 0.8), the N₂ adsorption quantity increased sharply, confirming the macroporous structure [34]. The pore size distributions demonstrated that all the resins possessed hierarchically porous structures, integrating pores at different size scales, from macropores to micropores (Fig. 1g). The modification did not alter the specific surface areas and total pore volumes of the Reillex resins (Table 1), suggesting that the porous structure, which is beneficial for adsorption, was retained after modification. Moreover, the average pore width slightly increased after modification, which might have enhanced the mass transfer of the adsorbates.

Scheme 1Full-screen View

(Color online) Schematic illustration for the synthesis of Reillex 425-Cn and Reillex 425-Me.

Fig. 1Full-screen View

(Color online) Digital photographs and SEM images of Reillex 425 (a), Reillex 425-C2 (b), Reillex 425-C3 (c), Reillex 425-C5 (d), and Reillex 425-Me (e). All the scale bars in the insets of SEM images are 300 μm. N₂ adsorption–desorption isotherms (f) and the corresponding pore size distributions (g) of different resins.

The ¹³C SSNMR spectra provided direct evidence for the successful modification of Reillex 425 resins. The characteristic peaks of poly(4-vinylpyridine) and polystyrene are shown in Fig. 2a. After tertiary pyridyl was quaternized with alkyltrimethylammonium, Reillex 425-C2 displayed three new peaks, ascribed to the carbon atoms in the spacer (C_h1 and C_h2) and trimethylammonium moiety (C_i), at 45.2, 63.6, and 55.4 ppm, respectively. Similarly, the characteristic peaks of all other modified Reillex 425 could be identified in the corresponding ¹³C SSNMR spectra. From the FT-IR spectra in Fig. 2b, the quaternization of pyridyl in Reillex 425 shifted the absorption peak for the in-plane C=C stretching vibration of the pyridyl ring from 1603 to 1637 cm⁻¹ [35, 36]. Other peaks at 1419 (aromatic semicircle C=N stretching of the pyridyl ring), 2927, and 2972 cm⁻¹ (stretching vibration of saturated C–H in the polymer main chain) remained the same [37]. Reillex 425-C2 simultaneously displayed two peaks at 1603 and 1637 cm⁻¹, and the other modified Reillex 425 only exhibited the latter peak, which implied that Reillex 425-C2 had a smaller quaternization efficiency than the other modified resins. XPS analysis of the valence states of the N atoms was performed to quantitatively estimate the modification efficiency. From the N 1s spectrum (Fig. 2c), Reillex 425 only contains two types of nitrogen — N atoms at tertiary (399.0 eV) and protonated (400.4 eV) pyridyl. After quaternization, all the Reillex 425-Cn displayed a new peak at higher binding energies around 401.5 eV, which was attributed to the quaternary N atoms in both pyridinium (N_c) and ammonium (N_d) moieties (Figs. 2d–2f) [3]. The modification efficiency, defined as the ratio of N_c to (N_a + N_b + N_c) (i.e., all N atoms in tertiary pyridyl, protonated pyridyl, and quaternary pyridinium), was estimated at 13.0%, 21.2%, and 26.6% for Reillex 425-C2, Reillex 425-C3, and Reillex 425-C5, respectively. Evidently, longer spacers reduce the electrostatic repulsion between the two positive N atoms in one DCP group, thus increasing the modification efficiency [38]. For comparison, Reillex 425-Me, in which the pyridyl groups were functionalized with methyl groups, was prepared. The smaller steric hindrance and weaker electrostatic repulsion of methyl-endowed Reillex 425-Me with a significantly higher modification efficiency of 48.6% (Fig. 2g). The modification densities (ρ_m, mmol g⁻¹, representing the moles of N_c per gram of resin) were estimated according to the modification efficiencies and the density of pyridyl in Reillex 425 (Table 1).

Fig. 2Full-screen View

(Color online) ¹³C CP-MAS SSNMR (a), FT-IR (b), and high-resolution XPS N 1s spectra of Reillex 425 (c), Reillex 425-C2 (d), Reillex 425-C3 (e), Reillex 425-C5 (f), and Reillex 425-Me (g).

Adsorption of Re(VII) on the modified Reillex 425 resins

Re(VII) primarily exists as a perrhenate anion (ReO₄⁻) over a pH range of 0 to 12 [39]; thus, its adsorption behavior is determined by the surface charge of the adsorbents. As shown in Fig. 3a, the zeta potential of the pristine Reillex 425 resin was positive at pH < 3 and negative at pH > 5 because of deprotonation of the pyridyl groups in the polymer (pK_a = 5.6) [40]. Consequently, the electrostatic repulsion between the resin and ReO₄⁻ gradually decreased when the pH decreased from 9 to 2. The q_e of Re on Reillex 425 reached a maximum of 194.8 mg g⁻¹ at pH = 2 (Fig. 3b). However, if the pH is further decreased, the competitive adsorption of concentrated Cl⁻ anions would suppress ReO₄⁻ adsorption [41, 42]. After the tertiary pyridyl groups on Reillex 425 were quaternized, all the modified resins exhibited positive zeta potentials at pH 1–9 owing to the intrinsic positively charged N atoms. Although all Reillex 425-Cn resins showed the same pH-dependent adsorption pattern as that of pristine Reillex 425 because of some remaining tertiary pyridyl, their q_e values were significantly enlarged under neutral conditions, remaining above 135.5 mg g⁻¹. Particularly, Reillex 425-C3 and Reillex 425-C5 exhibited larger q_e values than Reillex 425-C2 because of their higher modification efficiencies (Table 1). Reillex 425-Me showed almost the same adsorption amount as Reillex 425-C5 owing to its higher modification efficiency, even though the methyl pyridinium group had only one binding site. Notably, the morphology of the modified Reillex 425 greatly impacted the adsorption performance. The modified Reillex 425 powders obtained under stirring exhibited a dramatically decreased q_e under neutral conditions compared to the Reillex 425-Cn beads (Fig. S2). Maintaining the granular shape during the modification not only increases the adsorption amount but also benefits column treatment for sewage.

Fig. 3Full-screen View

Zeta potentials (a) and Re adsorption amounts (b) of different Reillex 425 resins at different initial pH (adsorbent dosage = 0.4 g L⁻¹, [Re]₀ = 100 mg L⁻¹, T = 298 ± 1 K, t = 12 h).

To elucidate the adsorption mechanism for ReO₄⁻ on the modified Reillex 425 resins, the adsorption isotherms were investigated and analyzed using the Langmuir and Freundlich models (Fig. 4). The linear forms are given by Eqs. S(1) and (2), respectively in the Supporting Information. From Table 2, the Langmuir model describes the adsorption of ReO₄⁻ on all the modified Reillex 425 resins, which suggests monolayer adsorption. The maximum adsorption capacities (q_m) were calculated to be 344.8, 416.7, 588.2, and 625.0 mg g⁻¹ (1.852, 2.240, 3.159, and 3.360 mmol g⁻¹) for Reillex 425-C2, Reillex 425-C3, Reillex 425-C5, and Reillex 425-Me, respectively. Compared with other adsorbents listed in Table S1, the higher q_m of modified Reillex 425 favors an extended lifetime of the packed column. It is evident that q_m on Reillex 425-Me equaled its ρ_m, suggesting that all positively charged N atoms were occupied by ReO₄⁻. Meanwhile, all q_e values for Reillex 425-Cn were higher than their ρ_m values, ranging from 1.5 to 2 times, indicating that ammonium and pyridinium sites should be capable of adsorbing ReO₄⁻ simultaneously. The q_m of Reillex 425-C2 was twice its ρ_m, demonstrating that all adsorption sites on the DCP groups could be utilized. However, it seemed that not all the positively charged sites on Reillex 425-C3 and Reillex 425-C5 were accessible at q_m/ρ_m ratios below 2, indicating that the adsorption of ReO₄⁻ anions inside the micropores might be hindered.

Fig. 4Full-screen View

The adsorption isotherms of ReO₄⁻ on Reillex 425-Cn and Reillex 425-Me (a) fitted by Langmuir (b) and Freundlich models (c). The solid curves in (a) are the fitting curves by the Langmuir model (adsorbent dosage = 0.4 g L⁻¹, [Re]₀ = 10–1400 mg L⁻¹, initial pH = 6.6–6.8, T = 298 ± 1 K, t = 12 h).

The adsorption thermodynamics were further investigated to gain a better understanding of the different adsorption patterns between the different DCP groups. Reillex 425-C2 and 425-C5, which contained the shortest and longest spacers, respectively, were selected as representatives. The changes in the standard Gibbs free energy (ΔG⁰), enthalpy (ΔH⁰), and entropy (ΔS⁰) of adsorption were calculated using Eq. S(3) and S(4) (Supporting Information), respectively. From Fig. 5 and the calculated results in Table 3, the negative ΔH⁰ and ΔG⁰ values demonstrated that the spontaneous adsorption of ReO₄⁻ on both Reillex 425-C2 and 425-C5 was exothermic. A lower temperature benefits the adsorption process, as q_e and ΔG display negative and positive correlations with temperature, respectively. Adsorption of ReO₄⁻ on Reillex 425-C5 was more thermodynamically favorable as its ΔG⁰ was more negative than that of Reillex 425-C2. However, the adsorption isotherms showed that Reillex 425-C5 had a lower utilization rate for binding sites. Thus, ReO₄⁻ adsorption on Reillex 425-C5 is controlled by kinetics rather than thermodynamics. Meanwhile, kinetics had less effect on ReO₄⁻ adsorption for Reillex 425-C2.

Fig. 5Full-screen View

Adsorption amounts of Re on Reillex 425-C2 and Reillex 425-C5 at different temperatures (a). The fitting plots of ln K_d versus T⁻¹ (b) (adsorbent dosage = 0.4 g L⁻¹ [Re]₀ =100 mg L⁻¹, initial pH = 6.6–6.8, t = 12 h).

A kinetic study demonstrated that Reillex 425-C2 showed a faster adsorption rate than Reillex 425-Me and all other Reillex 425-Cn, reducing the equilibrium time for adsorption from 5 h to 1 h (Fig. 6a). This agrees with the deduction that the adsorption of ReO₄⁻ on Reillex 425-C2 was less controlled by kinetics than on other Reillex 425-Cn. Pseudo-first-order and pseudo-second-order kinetic models were used to quantitatively investigate the adsorption kinetics, the linear forms of which are given by Eqs. S(5) and (6) in Supporting Information, respectively. Figs. 6b and 6c show the linear fitting curves of these two models, whose fitting results are listed in Table 4. The adsorption behavior of ReO₄⁻ can be described better by the pseudo-second-order kinetic model with a higher R², revealing that the adsorption is chemisorption [43]. Reillex 425-C2 exhibited a high adsorption rate constant k₂ of 0.00260 g mg⁻¹ min⁻¹, which was more than six times larger than that of the other modified Reillex 425.

Fig. 6Full-screen View

(Color online) The dynamic adsorption amounts of ReO₄⁻ by different modified Reillex 425 resins (a). Linear fitting by pseudo-first-order (b) and pseudo-second-order kinetic models (c). Analysis of the kinetics for ReO₄⁻ adsorption on Reillex 425-C2 and Reillex 425-C5 by intraparticle diffusion model (d) (adsorbent dosage = 0.4 g L⁻¹, [Re]₀ = 100 mg L⁻¹, initial pH = 6.6–6.8, T = 298 ± 1 K). The proposed transport pathways of ReO₄⁻ inside the micropores of Reillex 425-C2, Reillex 425-C5, and Reillex 425-Me (e).

The adsorption process consists of four sequential stages: bulk diffusion, boundary layer diffusion, intraparticle diffusion, and surface reactions [44]. In batch adsorption experiments, the resistance to bulk diffusion is negligible because of the small concentration gradient of the adsorbate caused by vigorous shaking. The final stage of the adsorption reaction is nearly immediate for ion exchange, and can be neglected. Therefore, the rate of the entire adsorption process is typically limited by the boundary layer and/or intraparticle diffusion stages. The following Eq. (3) depicts the intraparticle diffusion model proposed by Weber and Morris[45]. $q_{\text{t}}=k_{\text{p}}{t}^{1/2}+c_{\text{p}},$ (3) where the slope k_p (mg g⁻¹ min^−1/2) and intercept c_p (mg g⁻¹) are the rate constants of intraparticle diffusion and a constant related to the thickness of the boundary layer, respectively.

Piecewise linear regression suggested that the q_t vs. t^1/2 plots of both Reillex 425-C2 and Reillex 425-C5 showed a two-stage adsorption pattern. The first stage agrees with the intraparticle diffusion model, with an intercept of almost 0, indicating that the adsorption of ReO₄⁻ on modified Reillex 425 is only controlled by intraparticle diffusion (Fig. 6d) [46]. The second stage displayed almost constant adsorption amounts, demonstrating that adsorption approached equilibrium. Thus, the higher adsorption rate of ReO₄⁻ on Reillex 425-C2 is attributed to the faster diffusion of ReO₄⁻ anions inside the micropores.

As demonstrated above, all Reillex 425-Cn samples retained an almost unchanged pore structure. The difference between the –Py⁺C₂H₄N⁺Me₃ and –Py⁺C₅H₁₀N⁺Me₃ groups, or more accurately, the different spacers are presumed to drastically affect anion transport inside the microporous channels of Reillex 425-Cn. The longer alkyl spacers were more flexible, which enabled greater changes in the conformation of the DCP groups. It is possible to rearrange the pyridinium and ammonium sites vertically owing to the electrostatic repulsion between them and the hydrophobic aggregation of the long carbon spacers, as illustrated in Fig. 6e. Under such circumstances, the adsorption of ReO₄⁻ at the first cationic site impedes the diffusion of ReO₄⁻ towards the second cationic moiety [47]. Furthermore, the aggregation of long and flexible carbon chains forms hydrophobic microphases inside microporous channels, which increases the resistance of intraparticle diffusion of hydrophilic species such as hydrated anions [48]. Therefore, steric hindrance and hydrophobic repulsion generated by the longer spacer of the –Py⁺C₅H₁₀N⁺Me₃ groups retard ReO₄⁻ transport inside the micropores of Reillex 425-C5. For Reillex 425-C2, the two positive charges connected by shorter alkyl spacers tended to be laterally arranged (Fig. 6e), leading to smaller steric hindrance. Moreover, hydrophobic microphases are not formed due to repulsion among the positive charges. As a result, the transport of ReO₄⁻ anions inside the microporous channels modified with Py⁺C₂H₄N⁺Me₃ groups was enhanced. Reillex 425-Me functionalized with monocationic methyl pyridinium also exhibited a slower adsorption rate than Reillex 425-C2, although no spacers were present. This is because the quaternization of pyridyl with methyl increases the hydrophobicity of the surface of the microporous channels, hindering the intraparticle diffusion of ReO₄⁻ (Fig. 6e) [49]. According to the above discussion, dicationic –Py⁺C₂H₄N⁺Me₃ groups with short spacers can promote the transportation of ReO₄⁻ inside the microporous channels.

To evaluate the contribution of DCP groups to practical Tc removal, the adsorption kinetics of TcO₄⁻ on different modified Reillex 425 samples were further investigated. As displayed in Fig. 7a, all the modified Reillex 425 resins remove approximately 99% of the Tc due to its trace concentration of 125 Bq mL⁻¹ (0.2 mg L⁻¹ of ⁹⁹Tc). Reillex 425-C2 still exhibited the fastest adsorption rate, with the highest 1-hour removal ratio of 93.5%. Tc removal accelerated by the –Py⁺C₂H₄N⁺Me₃ groups was verified by kinetic analysis. The adsorption kinetics followed a pseudo-second-order kinetic model (Fig. 7b and 7c), similar to that of ReO₄⁻. Reillex 425-C2 showed the largest k₂ of 0.211 g mg⁻¹ min⁻¹ among all three Reillex 425-Cn samples (Table 5).

Fig. 7Full-screen View

The dynamic adsorption amounts and removal ratios of TcO₄⁻ on different modified Reillex 425 resins (a). Linear fitting by pseudo-first-order (b) and pseudo-second-order kinetic models (c). Analysis of the kinetics for TcO₄⁻ adsorption on Reillex 425-C2 and Reillex 425-C5 by intraparticle diffusion model (d). (adsorbent dosage = 0.4 g L⁻¹, [Tc]₀ = 0.2 mg L⁻¹, initial pH = 6.6–6.8, T = 298 ± 1 K).

TcO₄⁻ adsorption on Reillex 425-Me was faster than that on Reillex 425-C2, which was different from the adsorption of ReO₄⁻. Piecewise linear regression of the kinetic data using the intraparticle diffusion model validated that the adsorption of TcO₄⁻ on Reillex 425-Cn showed the same pattern as that of ReO₄⁻, only controlled by intraparticle diffusion (Table 6). However, TcO₄⁻ adsorption on Reillex 425-Me is controlled by both stages of the boundary layer and intraparticle diffusion. After obtaining the intercept c_p in the intraparticle diffusion model by linear fitting of the first stage, an initial adsorption factor (R_i), defined by the following Eq. (4) can be calculated to distinguish the contributions from different diffusion processes. R_i was 0.772 for Reillex 425-Me, implying that 77.2% of the TcO₄⁻ adsorption was governed by intraparticle diffusion. The remaining 22.8% occurs almost instantaneously, as determined by boundary layer diffusion [50]. Reillex 425-C2 exhibited a 52.5% higher k_p than Reillex 425-Me, demonstrating that the –Py⁺C₂H₄N⁺Me₃ groups also accelerated the adsorption of TcO₄⁻ by promoting intraparticle diffusion. Generally, higher mass transfer rates shorten the mass transfer zone when adsorbents are packed in a column. In other words, the difference between breakthrough and equilibrium times is reduced, resulting in a steeper breakthrough curve. Consequently, fixed-bed columns can be utilized more efficiently. $R_i=1-\frac{c_p}{q_e}$ (4)

Although the contaminated seawater in Fukushima Daiichi nuclear power station has been desalinated during the treatment, the treated water still contains Cl⁻ (1900 mg L⁻¹, or 5.4 × 10⁻² mol L⁻¹) and SO₄²⁻ (610 mg L⁻¹, or 6.4 × 10⁻³ mol L⁻¹) [51], the two main anions in seawater [52], which may hamper the adsorption of TcO₄⁻. In 0.05 mol L⁻¹ of Cl⁻ or SO₄²⁻, the q_e of Re on Reillex 425-C2 decreased to 64.9% and 49.7% of the original value, respectively, owing to competitive adsorption by the DCP groups (Fig. 8a). With an increase in the anion concentration, Cl⁻ and SO₄²⁻ had different effects on Re adsorption. q_e declined to 25.8% in 1 mol L⁻¹ of Cl⁻, while it remained almost the same (44.8%) in 1 mol L⁻¹ of SO₄²⁻. Reillex 425-C5 showed a similar trend; q_e decreased from 72.1% to 28.3% in Cl⁻ but remained constant at approximately 65.0% in SO₄²⁻ as the anion concentration rose from 0.05 to 1 mol L⁻¹(Fig. 8b). Because of the hydrophobic nature of the Reillex 425 resin matrix, the inner walls of the porous channels should retain a certain hydrophobicity even after the dication modification. The hydration energies for the different anions follow the order SO₄²⁻ (−1080 kJ mol⁻¹) < Cl⁻ (−340 kJ mol⁻¹) < ReO₄⁻ (−330 kJ mol⁻¹) [53]. Therefore, SO₄²⁻, which is more hydrophilic, has less influence on ReO₄⁻ adsorption. Reillex 425-C5, containing hydrophobic microphases formed by longer and more flexible carbon chains, is more resistant to these hydrophilic anions than Reillex 425-C2 for the same reason. In addition to Cl⁻ or SO₄²⁻, the treated water also contains trace amounts of other radioactive anions, such as ¹²⁹I⁻ and ¹²⁵Sb(OH)₆⁻ [54]. Reillex 425-C2 and Reillex 425-C5 showed separation factors (S) between Re and I (S_Re/I = K_d,Re/K_d,I) of over 4.6 and 5.0, and S_Re/Sb (S_Re/Sb = K_d,Re/K_d,Sb) of over 14.7 and 64.2, respectively, demonstrating their ability to remove TcO₄⁻ selectively from the treated water in Fukushima (Fig. 8c). After the adsorption of ReO₄⁻, the used Reillex 425-Cn could be easily regenerated by HNO₃, with q_e remaining over 95% after five adsorption–desorption cycles (Fig. 8d).

Fig. 8Full-screen View

Adsorption amounts of Re on Reillex 425-C2 (a) and Reillex 425-C5 (b) at different concentrations of Cl⁻ and SO₄²⁻ (adsorbent dosage = 0.4 g L⁻¹, [Re]₀ = 100 mg L⁻¹, T = 298 ± 1 K, initial pH = 6.6–6.8, t = 12 h). The adsorption selectivity of ReO₄⁻ towards I⁻ and Sb(OH)₆⁻ on Reillex 425-C2 and Reillex 425-C5 (c) (adsorbent dosage = 0.4 g L⁻¹, [Re]₀ = [I]₀ =[Sb]₀ = 50 mg L⁻¹, T = 298 ± 1 K, initial pH = 6.6–6.8, t = 12 h). Reusability of Reillex 425-C2 and Reillex 425-C5 (d) (adsorbent dosage = 0.4 g L⁻¹, [Re]₀ = 100 mg L⁻¹, T = 298 ± 1 K, initial pH = 6.6–6.8, t = 6 h).