With the advancement of living standards, a variety of artificial chemicals have been consumed and consistently entered municipal sewage systems [1]. Many artificial chemicals, especially pharmaceuticals and personal care products (PPCPs), are environmentally persistent and resistant to degrade in traditional wastewater treatment processes [2, 3], resulting in long-lasting negative impacts on human health and ecological safety [4, 5]. Furthermore, the environmental risks of bio-refractory PPCPs have been recently intensified due to their heavy use in overwhelming pandemics [5-7]. To this end, it is urgent to develop viable advanced treatment technologies to effectively remove PPCPs from water environment.

Anodic oxidation technology receives increasing attention due to its advantages of high efficiency, simple operation process, and no chemical dose [8-10]. The anodic oxidation mechanisms mainly include direct electron transfer (DET) and indirect oxidation mediated by stronger oxidants (e.g., HO^•) [8, 11-13]. The processes of DET and indirect oxidation on the surface sites (termed []) at the anode (termed M) are shown in Eqs. 1-5 [9], where R stands for organic pollutants (e.g., PPCPs):

$ \mathrm{M}^n[]+\mathrm{H}_2 \mathrm{O} \rightarrow \mathrm{M}^{n+1} \mathrm{O}+2 \mathrm{H}^{+}+2 \mathrm{e}^{-} $

(1)

$ \mathrm{M}^{n+1} \mathrm{O}+\mathrm{R} \rightarrow \mathrm{M}^n[]+\mathrm{RO} $

(2)

$ \mathrm{M}^n[]+\mathrm{H}_2 \mathrm{O} \rightarrow \mathrm{M}^n\left[\mathrm{HO}^{\bullet}\right]+\mathrm{H}^{+}+\mathrm{e}^{-} $

(3)

$ \mathrm{M}^n[\mathrm{HO} \cdot] \rightarrow \mathrm{M}^n[]+\mathrm{HO}^{\bullet} $

(4)

$ \mathrm{HO}^{\bullet}+\mathrm{R} \rightarrow \mathrm{ROH} $

(5)

In essence, contaminants need to be in contact with the anode to leverage the DET process [8]. However, due to the presence of boundary layers, the mass transfer efficiency of pollutants near the anode surface (in the vicinity of 100 µm) is greatly limited [14, 15]. For indirect oxidation, the energy barrier of adsorbed HO^• (termed [HO^•]) leaving the anode surface is typically high (Eq. 4), resulting in a high operating potential and detrimental side reactions [16]. Therefore, developing novel strategies to optimize the anodic oxidation efficiency and energy consumption is of great significance for water purification.

The yield of oxidants and their interaction with aqueous pollutants in anodic oxidation processes are greatly influenced by the electrode materials and microstructures [17-20]. Generally, in order to inhibit the oxygen evolution side reaction at the anode, electrode materials with high oxygen evolution potentials are often selected [21, 22]. Anodes with this feature, such as boron-doped diamond (BDD), PbO₂, and Sb-SnO₂, have exhibited satisfactory performance treating various refractory pollutants [9, 23-26]. However, due to their high cost or/and the leaching of toxic metal ions [9, 27], these anodes are subject to challenges from larger-scale environmental applications. Ti₄O₇, a type of suboxide of TiO₂, possesses low biotoxicity of its constituent elements and exhibits high electrical conductivity owing to its sub-stoichiometric oxygen composition [28, 29]. Additionally, it possesses a relatively high oxygen evolution potential (~2.5 V vs. Ag/AgCl), making it a promising alternative for anodic oxidation [9]. Furthermore, the anode microstructure plays a crucial role in the mass transfer between the electrochemically generated oxidants and the aqueous pollutants [30, 31]. For instance, reactive electrochemical membranes can reduce the boundary layer to several micrometers, significantly addressing the limitations of slow interfacial charge transfer at Ti₄O₇ [15, 32, 33], nevertheless, this membrane structure potentially faces scaling issues. Alternatively, Xie et al. have recently proposed a flow anodic oxidation system that use flowable Ti₄O₇ electrode to degrade refractory pollutants, which enables a more energy-efficient oxidation pathway mediated by adsorbed hydroxyl radicals [27, 34]. However, the flow anodic oxidation inevitably resulted in an intermittent operation as well as an accessional unit for electrode separation and recovery [35]. Because of the replenishment of flowable electrode in batch-mode operation, the passivation of Ti₄O₇ flow anode under continuous operation has yet to be evaluated [28, 36]. Therefore, it is of great significance to establish a continuous flow anodic oxidation process and comprehensively evaluate the stability of anode material.

In this study, we developed a continuous water purification system by integrating flow Ti₄O₇ anode with a porous Ti cathode. The schematic diagram of the continuous water purification system based on the flow anode is shown in Fig. 1a, Fig. S1 and Section S1 (Supporting information), containing an anode current collector, flow anode and a porous cathode. Herein, Ti₄O₇, possessing excellent conductivity, high chemical stability and catalytic activity for H₂O to HO^• conversion [31], was selected as the flow anode (Section S2 in Supporting information). The porous design of a cathode rather than an anode current collector was intended to avoid the fouling (e.g., via surface polymerization) of the porous structure during contaminant degradation. Since the anolyte was constantly acidified during the reaction, hardness ion deposition was not noted in treating real water samples. Moreover, according to our previous study, the non-uniform polarization characteristics of a porous electrode results in the rapid potential drop beneath the surface [37, 38]. Since the cathode terminal post (titanium wire) was connected to the external circuits at the outer layer (Fig. S1), the direct contact of flow anode materials with the inner layer of the cathode did not result in short circuit, and as a result, the separator was waived in this system setup. According to the requirements of continuous flow anodic oxidation, the screening principle of a porous cathode is that it has a smaller pore size than the flow anode particle and excellent electrical conductivity. Thus, porous Ti tube was used as the cathode due to its low cost, good machinability, great conductivity, and excellent corrosion resistance [39, 40].

To achieve the effective interception of the flow anode for continuous operation, the particle size of the flow anode and the pore size of the Ti cathode need to match. Scanning electron microscope (SEM) images showed that the pores on the surface of Ti tube were mostly in irregular shape, and the surface aperture size was less than 20 µm (Figs. 1b–d). Mercury intrusion method was further employed to evaluate the pore size distribution in the porous Ti tube. The results in Fig. 1e indicated that the pore size was significantly less than 10 µm. Moreover, the flow anode materials with a particle size greater than 12.9 µm exceeded 90% of the total, and the mass loss of the anode materials following 7-h operation was ~2% (Table S1 in Supporting information). Moreover, the cyclic stability of the flow anode was scrutinized during continuous operation of 72 h, and there was no significant deterioration in the pollutant removal performance and electrochemical characteristics (Fig. S2 in Supporting information). These findings have confirmed the capability of the proposed system for continuous operation of flow anodic oxidation.

It is important to note that the porous structure on the surface of the Ti tubes and Ti₄O₇ particles (Figs. 1b–d) may result in strong adsorption properties for carbamazepine (CBZ), a common and model PPCP widely detected in natural water environment [41-43]. As shown in Fig. 1f, when no potential was applied to the water purification system (termed Ad-flow anode), the effluent CBZ concentration decreased by 38% in the first hour, gradually increased with the extension of the operation time, and was the same as the inlet CBZ concentration after 6-h operation. Direct electrooxidation (at 2.0 V vs. Ag/AgCl) on the anode current collector (termed Flat plate anode) can reduce the CBZ effluent concentration by 20%, which however still possesses potential environmental risks [41]. Upon the addition of 10 g/L Ti₄O₇ particles (termed Flow anode), the effluent CBZ concentration was reduced by 99% at a hydraulic residence time (HRT) of 0.5 h. In addition, the mass transfer and current efficiencies of the plate anode and the flow electrode system were calculated according to Section S3 (Supporting information). The current efficiency of the flowing anode (0.4660) was about three times that of the flat plate anode (0.1554) system (Table S2 in Supporting information), and the active species generated by the flow anode system were more likely to contact and degrade contaminants, while the active species generated by the flat electrode may be quenched by the background substance. The mass transfer efficiency of the two systems was comparable, but the surface area of the flowing anode was much larger than the flat plate anode. As such, the degradation of contaminants by the active species was more efficient in the flow anode system. These results demonstrated that the addition of flow anode into the electrochemical oxidation system can significantly enhance the oxidation efficiency of contaminants.

Quenching experiments and electrochemical characterization were conducted to evaluate the contribution of hydroxyl radical oxidation and direct oxidation in the flow anodic system. Tertiary butanol (TBA) is a common HO^• quencher in the electrochemical advanced oxidation [19, 44], and unlikely compete with the CBZ for adsorption sites (Table S3 in Supporting information). As shown in Fig. 2a, when the adsorption and degradation of CBZ in the flow anodic oxidation system were carried out at the same time in the initial 1-h phase, TBA had little effect on the effluent CBZ concentration. As the adsorption capacity of the system was exhausted, the CBZ concentration in the effluent was mainly controlled by the degradation process. Thereafter, the effluent CBZ concentration gradually increased, suggesting that TBA quenched HO^• for indirect oxidation. According to Eq. 3, the direct oxidation of contaminants at the anode would produce corresponding oxidation peaks on its LSV curves. However, no obvious oxidation peak was generated in the LSV curve (Fig. 2b, Figs. S3, S4a and b in Supporting information). Therefore, we expect that HO^• should be the dominant reactive species in the flow anodic oxidation system.

It was noticed that the current intensity of the system was enhanced after the addition of the flow anode. This phenomenon may be ascribed to the formation of charge percolating network consisting of concentrated conductive particles [27]. The Nyquist plots from the impedance spectroscopy showed that the addition of flow anode materials resulted in a decrease of approximately 40 ohms in the internal resistance, with similar improvement noted for the interfacial charge transfer (Fig. S4c in Supporting information) [45]. Overall, it could be concluded that the charged flow anode particles extend the removal of contaminants from the vicinity of the anode surface to the entire charge percolating network, thus facilitating the contaminant removal (i.e., higher percentage removal and rate as shown in Fig. 1f).

The potential has an important effect on the formation of active species during anodic oxidation process [9, 19, 46]. Therefore, the removal of contaminants in the flow anodic oxidation system at different potentials was investigated (Fig. 2c). High water purification performance (99% CBZ removal) was obtained by applying a 2.0 V vs. Ag/AgCl potential to the anode current collector. When the anode potential was decreased to 1.5 V vs. Ag/AgCl, the effluent CBZ concentration increased dramatically (to 80% of the influent CBZ concentration) at the same hydraulic retention time. This phenomenon should be attributed to the reduced indirect oxidation via oxygen species or/and direct oxidation at lower potentials. When the anode was polarized at a high potential (2.5 V vs. Ag/AgCl), the oxidation rate and percentage contaminant removal were not further improved, probably due to the side reaction of oxygen evolution. As summarized in Table S4 (Supporting information), the energy consumption per order of magnitude removal (EEO) of this technique (0.12 kWh/m³) (Section S3) was lower than counterparts (1−70 kWh/m³), exhibiting a considerable application potential. According to the influent quality, the flow anodic oxidation process should be regulated to balance the treatment performance and energy consumption. Unless otherwise stated, 2.0 V vs. Ag/AgCl was chosen as the optimal operating condition in the following experimentation.

The number of reactive sites for oxidative species production is positively related to the load/mass concentration of anode materials [27, 34]. As shown in Fig. 2d, the load of 5 and 10 g/L flow anode could both reduce the effluent CBZ concentration by more than 99% in the steady state (e.g., after 7-h operation). In spite of the very similar effluent quality, the stable effluent CBZ concentration was reached faster with 10 g/L Ti₄O₇ dosage compared to 5 g/L Ti₄O₇ dosage, indicating that increasing the flow anode load in a certain range can enhance the reaction kinetics in the system. However, when the flow anode load was increased to 15 g/L, the effluent CBZ concentration decreased. On one hand, due to the agglomeration of the flow anode during the reaction (Fig. 1e), the higher flow anode load may result in the loss of reactive site and sedimentation of the electrode materials. On the other hand, it has been reported that over dosage can induce self-quenching effects in the advanced oxidation processes with the produced reactive species (e.g., HO^•) being transformed into non-active species (e.g., surface or/and lattice oxygen) [47, 48]. Thus, with regard to the capital cost and operating performance, the flow anode load should be optimized (i.e., 10 g/L Ti₄O₇ dosage under the conditions tested in this study) to enhance the contaminant removal. Additionally, the inlet flow rate would also affect the capacity of the flow anode. As the effluent flow rate increased from 0.70 mL/min to 0.98 mL/min, the pollutant removal efficiency could still be greater than 90% (Fig. S5 in Supporting information). When the effluent flow rate further increased to 1.58 and 1.92 mL/min, the effluent CBZ concentration increased significantly. Therefore, we can conclude that the effluent pollutant removal efficiency could be maintained within a certain HRT range (> 20.7 min), and the effluent pollutant concentration would increase when the treatment capacity of the device was exceeded.

In the anodic oxidation process, the solution pH has multiple effects on the contaminant removal [49]. The solution pH determines the charge of contaminants in the reaction system, and a low pH value resulting in protonation significantly inhibits the contact between contaminants and anode surface due to electrostatic repulsion [23]. Moreover, the solution pH affects the oxygen evolution potential (OEP) of the electrode materials [50]. As shown in Fig. S4b, the onset potential of oxygen evolution reaction was reduced as the solution pH increased from 5 to 6 and 8. Therefore, the deterioration of pollutant removal performance under alkaline conditions can be ascribed to the enhancement of oxygen evolution side reaction and the inhibition of reactive species generation at higher pH (Fig. 2e) [51].

Water background substances have a significant impact on water treatment efficiency [10, 46]. The effects of Cl⁻, HCO₃⁻ and humic acid (HA) on CBZ removal were further investigated in the flow anodic oxidation system, and natural surface water (Table S5 in Supporting information) was used as a complex matrix to assess its practical application potential. Cl⁻ has negligible impacts on the effluent CBZ concentration (Fig. 2f). Although background Cl⁻ may compete with pollutants for active species, the secondary oxidant originated from Cl⁻, chlorine free radicals (Cl^•), has considerable oxidizing activity [9, 52], thus contributing to the comparable treatment performance. Similar to Cl⁻, HCO₃⁻ also competes with CBZ for active species, however, the reactivity of resultant HCO₃⁻ radical (HCO₃) with CBZ was much lower than that of Cl^•− (k_Cl, CBZ = (3.7 ± 0.3) × 10¹⁰ L mol⁻¹ s⁻¹ and k_HCO, CBZ = (3.4 ± 0.4) × 10⁶ L mol⁻¹ s⁻¹) [53], deteriorating the process performance. The unsaturated bond contained in HA has affinity for many active species [54], which obviously led to the decrease of the removal efficiency of CBZ. Due to the quenching effect of natural organic matters and background ions, while the CBZ removal efficiency was reduced in the actual water, the effluent concentration was still removed by 65% over 7 h operation.

Intermediate products formed during the CBZ degradation in the flow anodic system were identified using LC-MS/MS analysis (Section S4, Table S6 and Fig. S6 in Supporting information), and the time-course results of the intermediate concentrations are shown in Fig. 3. Combing the information on the identified oxidation products, their relative abundances and possible oxidation pathways of CBZ in this process are proposed in Fig. 3. The hydroxyl radicals first attack C═C on the N-heterocyclic ring in CBZ, because of the high frontier electron density (FED) of the olefinic double bond of CBZ [55-57]. After various reactions of oxygen addition, hydroxy substitution, dealkylation, and ring-opening, CBZ eventually transform into small molecule substances followed by mineralization.

Moreover, the aquatic ecotoxicity effects of CBZ and its degradation intermediates were assessed using quantitative structure-activity relationships (QSAR) method and the toxicity estimation software tool (T.E.S.T.). The LC₅₀ of CBZ for many aquatic organisms is about 5 mg/L (Figs. 4b–d). While our water purification system reduced the effluent CBZ concentration to 0.2 µmol/L (about 0.047 mg/L), the significant bioaccumulation of CBZ might result in CBZ concentrations in organisms being higher than those in the environment (Fig. 4a). In addition, it was noted that its degradation intermediates, P253-2, P253-3 and P251, accumulated during the process. These intermediates exhibit similar biological toxicity to CBZ. Therefore, in the process of evaluating the effluent ecotoxicity, the ecotoxicity of the main intermediate products should also be included in addition to the parent compound. Although no equilibrium concentrations of the three main intermediates were obtained in this study, the detection of secondary products such as P269 indicated that our flow anode system could further degrade toxic intermediates to reduce the biological toxicity of the effluent. Overall, the flow anode system has been demonstrated to alleviate the ecotoxicity, which is crucial to control the environmental risks thereby.

Moreover, the stability of flow anode particles is important for its practical application. The crystal phases of flow anode particles before and after the oxidation operation were investigated by X-ray diffraction pattern (XRD). As shown in Fig. 5a, the flow anode particles were almost unchanged and consistent with the standard PDF card of Ti₄O₇ after the oxidation operation, indicating the great stability of Ti₄O₇ in this system. Moreover, SEM images of Ti₄O₇ before and after the oxidation operation also proved the dimensional stability of Ti₄O₇ in the system (Figs. 1c and d, Figs. S7 and S8 in Supporting information). While the crystalline structure of Ti₄O₇ remained stable, previous studies indicated that the Ti_xO_2x−1 surface gradually passivated under long-term operation, resulting in deterioration of the electrocatalytic performance [58]. Thus, high-resolution X-ray photoelectron spectroscopy (XPS) spectra was used to detect the valence evolution of the elements on Ti₄O₇ surface during the oxidation operation (Figs. 5b and c). Two pair double peaks were observed at 457.1 and 462.9 eV as well as 458.3 and 464.3 eV, which were attributed to Ti⁴⁺ 2p and Ti³⁺ 2p in Ti_xO_2x−1 XPS spectra before operation. After the oxidation operation, the binding energy of Ti 2p shifted to higher region, indicating the trend of electron loss of Ti contained in Ti₄O₇ flow anode. O 1s high-resolution XPS spectra of Ti₄O₇ also supported the speculation. The binding energy and intensity of the peaks attributed to Ti⁴⁺-O and Ti³⁺-O were both increased, suggesting the oxidation of Ti and the increase of O element abundance [59]. Meanwhile, the ratio of Ti/O of the Ti₄O₇ flow anode decreased from 0.36 to 0.29 after the reaction, providing the straightforward evidence for the oxidation of Ti₄O₇ (Table S7 in Supporting information). While Ti₄O₇ flow anode was slightly oxidized, its water purification performance was stable during 7-h anodic oxidation operation. Moreover, it was reported that Ti₄O₇ could recover its electrochemical activity by cathodic polarization, which needs further investigation and optimization.

In summary, a porous cathode was introduced in this study to realize the continues water purification in flow anodic oxidation. The CBZ concentration in the effluent can be reduced by 99%. The quenching experiment and electrochemical characterization showed that the electrified flow anode could oxidize water to hydroxyl radicals, thus extending the contaminant removal site from the anode surface to the whole conductive network. While slight passivation of Ti₄O₇ particles was noted over long-time operation, their dimensional structures and water purification performance were stable. While bicarbonate and humic acid in actual water could affect the contaminant removal in the flow anodic oxidation process, its efficiency was largely sustained and contributed to reduced ecotoxicity following oxidation. These findings are expected to provide insight and guidance for the design of cost-effective flow anode system for water purification.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Runsheng Xu: Writing – original draft, Investigation, Formal analysis, Data curation. Haotian Wu: Investigation, Data curation. Daoyuan Zu: Supervision, Resources, Methodology. Kui Yang: Software, Resources, Methodology. Xiangtong Kong: Resources, Methodology. Jinxing Ma: Writing – review & editing, Supervision, Funding acquisition, Formal analysis, Conceptualization.

Acknowledgments

This study was financially supported by the Basic Science Center Project of the National Natural Science Foundation of China (No. 52388101), Program for Guangdong Introducing Innovative and Entrepreneurial Teams (No. 2019ZT08L213), and the National Natural Science Foundation of China (No. 52100030). The authors would like to thank Shiyanjia Lab (www.shiyanjia.com) for the XPS analysis.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110517.