The phosphorus and fluoride wastewater used in this experiment was collected from the phosphate fertilizer factory (Zhuzhou, China), and the detailed information are: pH 3.53, 82.7 mg/L TF, 121.2 mg/L TP, 112.5 mg/L COD, 1800 mg/L SO₄ ^2-, 45 mg/L NH₃-N, and 3,320 μs/cm conductivity. All experiments were conducted using ultrapure water (UPH, 18.25 MΩ·cm). The pH of the wastewater was adjusted by 0.1 mM NaOH solution.

The electrode preparation process was shown in Figure 1. The commercial ACF was cut into 5 × 10 cm squares, placed in a beaker, boiled with boiling water for 2 h, and soaked in absolute ethanol for 2 h to remove ash impurities in the ACF. Then, the ACF was soaked in 30% HNO₃ solution for 6 h, washed with ultrapure water until the pH reached neutrality, and dried. Next, the acid-treated ACF was immersed in 200 ml of FeCl₃ solution (0%–3%) for 12 h. After taking it out to dry, the ACF was put in a tube furnace for calcination at different temperatures under an N₂ atmosphere for 4 h to obtain the A@Fe-ACF electrode.

The properties of the prepared ACF electrodes were characterized using different analytical techniques, including X-ray diffraction (XRD, Ultima IV, Japan), Scanning electron microscopy (SEM, Hitachi S4800, Japan), X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250XI, United States), Fourier transform infrared absorption spectroscopy (FTIR, Niolet iN10, United States), surface area and pore size analyzer (BET, ASAP2460, United States), and electrochemical impedance spectroscopy measurements (EIS, CHI 660E, China).

The CDI process was performed in a special acrylic cuboid reactor (L = 10 cm, W = 5 cm, H = 12 cm), with A@Fe-ACF as the anode attached to a stainless-steel plate and graphite plate as the cathode (Figure 2; Supplementary Figure S1). The electrodes are connected to a direct current (DC) regulated power supply (MESTEK DP3030) and maintained at a constant distance. After applying a voltage to the electrodes to initiate the reaction, key parameters such as voltage intensity, initial pH, plate spacing, and phosphorus and fluorine (P/F) ratio on phosphorus and fluorine removal efficiency were investigated by a conventional parametric approach. At preselected time intervals, the extracted solutions were transferred to 5 ml sampling tubes and analyzed uniformly.

The concentration changes of total phosphorus (TP) and total fluorine (TF) were used to represent the removal efficiency of phosphorus and fluorine. Molybdenum antimony anti-spectrophotometry was used to determine the concentration of total phosphorus. The ion-selective electrode method was used to determine the TF concentration (the indicator electrode was a fluoride ion-selective electrode, and the reference electrode was a saturated calomel electrode).

Data used in this study were entered in Excel, and all data were tested for normality (Shapiro-Wilk test) and homogeneity (Levene test) before analysis, and Duncan multiple comparisons were performed on the data means. TP and TF removal efficiency under different conditions were analyzed by one-way ANOVA. All data were statistically analyzed using IBM SPSS Statistics 23.0 software.

The morphologies of ACF before and after modification are shown in Figure 3. The initial ACF surface was smooth and flat (Figure 3A), and it became rough after HNO₃ modification (Figure 3B) due to the strong oxidizing property of HNO₃ corroding the ACF surface. The surface morphology of ACF was changed after acid treatment, and the pore structure was increased, which significantly increased the specific surface area of ACF (Figure 4A). Furthermore, the acid-treated ACF could provide more loading sites for Fe, conducive to the tighter bonding between the loading metal and the carrier. It can be seen that the iron oxide was covered on the ACF in a sheet shape when the iron loading modification was carried out alone (Figure 3C). The smooth carbon fiber surface does not provide an excellent binding point for the Fe element and will lead to the iron oxide difficult to bonded on the ACF and is easy to fall off. When ACF was treated with acid and modified with Fe loading, it could be observed that Fe oxides were tightly bound on ACF in irregular granular form compared with the modification of Fe alone (Figures 3D–H). Although Fe loading leads to a slight decrease in the specific surface area (Figure 4A), the synergistic effect of Fe oxides has further improved the TP and TF removal. The calcination temperature is another key factor affecting the CDI performance of A@Fe-ACF, as the adsorption efficiency gradually increased with the temperature. However, the removal efficiency of TP and TF decreased when the temperature was higher than 400°C (Supplementary Figures S2A, 2B). It can be seen from SEM that the Fe oxide on the surface of ACF is evenly distributed at 400°C, while the Fe oxide begins to sinter, and the load area becomes smaller when the temperature increases to 500°C–600°C (Figures 3E, F), thus affecting the CDI efficiency of A@Fe-ACF. In addition, the load of Fe is also an essential factor. The introduction of iron improved the adsorption efficiency of phosphorus and fluorine in wastewater, while the adsorption capacity also increased significantly with the loading concentration (Supplementary Figure S2C). However, the higher loading concentration would lead to Fe oxide blocking pores and inhibit the effect of CDI process. Therefore, 2% Fe loading concentration and the 400°C calcination temperature were the optimal conditions for A@Fe-ACF. The electrochemical properties of the electrodes were compared using EIS (Supplementary Figure S3), and it can be seen that the transfer resistance of the ACF decreases significantly with Fe loading. The presence of Fe accelerated the electron transfer efficiency and improved the conductivity of the electrode, which facilitated the capacitive deionization process.

Figure 4B showed the FTIR images under different modification conditions. The characteristic peak at 3,460 cm⁻¹ is the stretching vibration peak of -OH, and the absorption peak here is significantly enhanced after modification, indicating that the hydrophilicity of the material has been improved. The new peak at 1711 cm⁻¹ may be the characteristic peak of C=O bond stretching vibration in the carboxylic acid group, representing the introduction of oxygen-containing functional groups after acid modification (Wu et al., 2015). The peak at 1,630 cm⁻¹ represents the backbone vibration of C=C, where the intensity gradually decreased with ACF modification. This phenomenon might be caused by the high temperature brought by the heat treatment (Wang et al., 1994; Georgakopoulos, 2003). The enhancement of the C-O bond at 1,200 cm⁻¹ and 1,062 cm⁻¹ could be attributed to the production of more abundant oxygen-containing groups (Meng et al., 2018). Moreover, the metal oxide peak at 560 cm⁻¹ is Fe-O (Verma et al., 2021), which further proved that Fe was efficiently loaded on the ACF surface.

From XPS analysis in Figure 4C, it could be observed that Fe in A@Fe-ACF-400 exists as Fe²⁺ and Fe³⁺, which could be attributed to the introduction of oxygen-containing functional groups promoted the redox on the electrode surface (Ren et al., 2020). The cycling of Fe²⁺/Fe³⁺ increased the current efficiency, thereby indirectly improving the effect of the CDI process (Wang et al., 2021). The iron element in A@Fe-ACF-400 mainly existed in Fe₂O₃ and Fe₃O₄ (shown in Figure 4D). The characteristics of Fe oxides could enhance the electron transfer in the electric field and improve the efficiency of electro adsorption to remove phosphorus and fluorine ions (Song et al., 2019).

Voltage intensity played a crucial role In the CDI process. It can be seen from Figures 5A, B that the removal of TP and TF showed a trend of first high and then low with the increase of voltage. When the voltage was 1.5 V, the removal of TP and TF reached 89.4% and 85%. The increase in voltage drives a higher current and establishes a steady-state faster, then the increase in electron mobility promotes the accumulation and adsorption of contaminant ions in the electric double-layer (Chai et al., 2020; Lin et al., 2022). However, the removal efficiency dropped to 83% and 78% when the voltage intensity was increased to 2 V. That is the excessively high voltage aggravating the hydrogen evolution side reaction Eq. 1, thus resulting in the desorption of PO₄ ^3- and F⁻ from the ACF and inhibiting the removal of phosphorus and fluorine (Ren et al., 2016). Therefore, 1.5 V was selected as the optimum voltage intensity. The solution pH is one of the essential factors to be considered in the removal of phosphorus and fluoride. Within a certain range, the removal efficiencies of TP and TF increased with the pH and reached the best at pH 7 (Figures 5C, D). The pH value significantly affects the existence of phosphate and fluoride ions in the solution Eqs. 2–4. When the pH is acidic, the uncharged H₃PO₄ occupies the dominant position, while when the pH is 7, the charged HPO₄ ^2- increases (Huang et al, 2017a), then the hydration radius decreases, which is easier to be absorbed by the electric field (Gabelich et al., 2002; Huang et al., 2013). Moreover, fluorine mainly exists in the F⁻ when the pH is more than 6, which benefits CDI treatment and stabilization, and a good electrostatic interaction under neutral conditions can positively influence the adsorption of F⁻ (Martinez-Vargas et al., 2021). However, when the pH was higher than 7, the removal rate of TP and TF decreased instead. This trend may be attributed to the enhanced competition between phosphate ions and hydroxyl ions for electrode adsorption sites at higher pH (Huang et al., 2013). A Higher pH environment could cause electrostatic repulsion to re-release the adsorbed fluoride, thus reducing the removal efficiency (Yu et al., 2018). In conclusion, pH seven would be more suitable for A@Fe-ACF electrode CDI treatment of phosphorus-fluoride wastewater. It can be seen from Figures 5E, F that the removal efficiency of TP and TF was the highest when the plate spacing is 1 cm. The closer plate spacing improved the rate of ion migration between the plates, enhanced the ions enriched on the electric double layer and the thickness of the electric double layer, further reduced the migration distance of HPO₄ ^2- and F⁻ in the electric field, and accelerated the adsorption of target ions. An excessively close plate spacing would interfere with the reactions between cathode and anode, and the rapidly increasing current intensity may lead to side reactions (Chai et al., 2020). In contrast, the long distance in the CDI reaction could limit ion mobility and exacerbates energy consumption (Yang et al., 2020). Therefore, 1 cm was determined as the optimal plate spacing.

Exploring the effect of the P/F ratio in wastewater on the CDI is important. The removal of TF increased with decrease in the relative concentration of phosphorus (Figures 5G, H). The adsorption rate constant of TP decreased from 0.1492 to 0.0782, while that of TF increased from 0.071 to 0.147 when the P/F ratio changed from 2:1 to 1:2. For HPO₄ ^2-, the value of Z/r (charge/radius) is higher than that of F⁻, which is beneficial to its adsorption and binding on ACF (Zuo et al., 2016). When the phosphorus concentration decreased, the binding competitiveness reduced, and the adsorption trend of F⁻ increased. Therefore, the control of the P/F ratio in wastewater could be a pivotal step to further improving the simultaneous treatment efficiency.

By studying the adsorption process of the coexistence of NO₃ ⁻, Cl⁻, SO₄ ^2-, HCO₃ ⁻, and SiO₃ ^2-, it can be found that different anions concentrations have different effects on the TP and TF removal efficiency (Supplementary Figures S4A, 4B). The presence of NO₃ ⁻ and Cl⁻ did not affect the removal of phosphorus and fluorine significantly, while the coexistence of SO₄ ^2-, HCO₃ ⁻, and SiO₃ ^2- had certain effect on the adsorption capacity. That could be attributed to the adsorption competition between coexisting ions with F⁻ and HPO₄ ^2- (Yu et al., 2018). These anions will lead to the precipitation of Ca²⁺ in the wastewater, blocking the pores of the ACF and passivating the electrodes, thereby reducing the adsorption capacity of the reaction system (Yoosefian et al., 2017). Results indicated that A@Fe-ACF can still maintain high adsorption efficiency in the presence of different ion concentrations, showing relatively excellent anti-interference performance. 2 H + + 2 e − → H 2 ↑ (1) H 3 P O 4 = H + + H 2 P O 4 − (2) H 2 P O 4 − = H + + H P O 4 2 − (3) H P O 4 2 − = H + + P O 4 3 − (4)

The removal efficiency of phosphorus and fluoride under different modified conditions is shown in Figures 6A, B, including the two cases of non-electric adsorption and CDI. It could be observed that under pure adsorption conditions, acid modification and iron impregnation slightly enhanced the adsorption of TP and TF. The removal efficiency of phosphorus and fluorine increased rapidly when the electric field was applied. The increase in the specific surface area caused by the etching of pores using HNO₃ and the presence of Fe oxides provide more adsorption sites for ACF, so the acid modification assisted Fe loading significantly promoted the CDI efficiency. The adsorption kinetics of the A@Fe-ACF electrode CDI process was further investigated (Supplementary Figures S5A, 5B). The precision of the pseudo-second-order kinetic equation is better than that of the pseudo-first-order kinetics, indicating that chemical adsorption played a dominant role in this process.

The possible reaction mechanism of simultaneous phosphorus and fluoride wastewater treatment based on A@Fe-ACF electrode CDI is shown in Figure 7. Anions such as HPO₄ ^2- and F⁻ would move towards the anode under the action of the electric field and subsequently adsorbed on A@Fe-ACF superior. Meanwhile, the loading of Fe can promote the complexation of HPO₄ ^2- and F⁻, and enhance the immobilization of pollutants on the electrodes.

The recycling ability of the electrode is important for evaluating its stability during the CDI process. It can be seen from Figure 8 that after five cycles of A@Fe-ACF by adsorption-desorption, the removal efficiencies of TP and TF dropped to 83% and 77.6%, which still keep a good adsorption performance. The above results show that A@Fe-ACF has good stability in phosphorus-fluoride wastewater treatment.

Gradual removal of phosphate and fluoride from wastewater using precipitation was the most widely reported strategy (Table 1). After pH adjustment, phosphate and fluoride might precipitate in the form of CaHPO₄, Ca₃(PO₄)₂, and CaF₂ (Grzmil and Wronkowski, 2006). However, the dosing of large amounts of neutralizing reagents required for higher pH and the adjustment of effluent pH can be costly and operationally burden (Xia et al., 2021). A large number of generated precipitates cannot be recovered entirely, and there is a risk of secondary pollution. Some new phosphorus and fluorine treatment technologies have been proposed and achieved certain results, but the longer treatment period and lower treatment efficiency are still enormous challenges (Wu et al., 2017; Yang et al., 2021). Efficient simultaneous removal of phosphate and fluoride from wastewater can be achieved by CDI using A@Fe-ACF. The presence of salt ions in wastewater could act as an electrolyte to improve the current efficiency, and the lower energy consumption and shorter reaction time will reduce the treatment cost. Based on the above advantages, the simultaneous removal of phosphorus and fluoride wastewater by CDI using an A@Fe-ACF electrode can be an innovative and feasible technology.