Local O₂ concentrating boosts the electro-Fenton process for energy-efficient water remediation

Significance

The electro-Fenton process which generates reactive oxygen species by O₂ reduction is one promising approach for water remediation, but to guarantee adequate O₂ supply, energy-intensive aeration is always needed, greatly increasing the costs for industrialization. Herein, inspired by the respiratory systems of fish, we developed a local O₂ concentrating (LOC) strategy to increase O₂ utilization efficiency (OUE) by in situ constructing O₂ concentrating reticular frameworks (RF) on the cathode. Amino-functionalization of RF further enhanced the interactions between organic linkers and O₂ molecules, rendering the cathode much higher OUE. The LOC approach represents an innovative, simple, yet effective strategy for reducing or even eliminating the need for an external O₂ supply, providing a conceptual advance to design an energy-efficient electro-Fenton process.

Abstract

The electro-Fenton process is a state-of-the-art water treatment technology used to remove organic contaminants. However, the low O₂ utilization efficiency (OUE, <1%) and high energy consumption remain the biggest obstacles to practical application. Here, we propose a local O₂ concentrating (LOC) approach to increase the OUE by over 11-fold compared to the conventional simple O₂ diffusion route. Due to the well-designed molecular structure, the LOC approach enables direct extraction of O₂ from the bulk solution to the reaction interface; this eliminates the need to pump O₂/air to overcome the sluggish O₂ mass transfer and results in high Faradaic efficiencies (~50%) even under natural air diffusion conditions. Long-term operation of a flow-through pilot device indicated that the LOC approach saved more than 65% of the electric energy normally consumed in treating actual industrial wastewater, demonstrating the great potential of this system-level design to boost the electro-Fenton process for energy-efficient water remediation.

The electro-Fenton process is one of the most promising technologies for the removal of organic contaminants from wastewater, as it produces highly reactive radicals from green molecular O₂ and electrons (1–3). However, high energy consumption is the major constraint to practical application (4, 5). Due to the low O₂ solubility (~8 mg L^–1) and slow diffusion (~1.96 × 10^–9 m² s^–1) in water at room temperature and pressure, O₂ mass transport from the bulk solution to the reaction interface is extremely inhibited (6, 7). For this reason, an energy-intensive aeration process is conventionally applied to guarantee a sufficient O₂ supply (8–10). In this way, however, over 75% of the electric energy is consumed by the aeration process (11), most of the supplied O₂ is wasted in the form of gas bubbles, and only a small amount of O₂ is dissolved into the water for electrochemical reactions, leading to an extremely low O₂ utilization efficiency (OUE) of less than 1% (12–14). Moreover, the energy consumed will grow exponentially when scaling up the process by increasing the aeration rate because the OUE is negatively correlated with the aeration rate (13, 15, 16). An enormous amount of energy is lost in a pilot-scale process because aeration consumes more than 95% of the electric energy (17). Therefore, improving the OUE is required to increase the overall energy efficiency of the electro-Fenton process.

Over the years, many researchers have sought to enhance the OUE by modulating O₂ diffusion near the electrode. Since hydrophobic microenvironments are believed to attract gas species, several highly hydrophobic electrodes have been developed to accumulate O₂ bubbles (18–21). However, the O₂ bubbles easily aggregate to form “dead areas” for the electrochemical reactions, which blocks the active sites and hinder mass transfer of the ionic reactants (22–24). Even worse, as mass transport of the gaseous/ionic reactants and electron transfer are disrupted by the formed microscopic solid–liquid–gas three-phase boundary, the extent to which hydrophobicity enhances the OUE is unclear (25–27). As a result, it is difficult to realize a satisfactory OUE simply by manipulating the hydrophobicity of the catalyst/electrode.

Inspired by fish that extract dissolved O₂ directly from the water to guarantee their O₂ supply, we propose a local O₂ concentrating (LOC) approach to increase the OUE. Unlike the conventional simple O₂ diffusion (SOD) route in which O₂ diffusion from the bulk electrolyte to the reaction interface is driven only by a concentration gradient, the LOC approach directly concentrates the dissolved O₂ from the bulk electrolyte via near-interface O₂ adsorption (Fig. 1). Just as fish breathe in water using their gills, the LOC electrochemical interface continuously extracts dissolved O₂ from the water by the near-interface frameworks. Specifically, when water passes through the narrow lamellar channels within gills, the dissolved O₂ is immediately taken by the capillary networks. In a similar way, the frameworks extract O₂ molecules from the water by the highly exposed O₂-adsorptive sites within their ordered channels. As a result, the local O₂ concentration near the reaction interface will increase and the local O₂ diffusion to the reactive center will be significantly accelerated. By using the LOC approach, OUE of the electro-Fenton process will be increased, reducing the need for an external O₂ supply and thereby increasing the energy efficiency.

Fig. 1.

Schematic illustration of the mechanism for an energy-efficient electro-Fenton process using the advanced LOC approach and comparison with the conventional SOD route. (C₀ and C_i are the O₂ concentration in the bulk solution and near the electrochemical interface, respectively.)

Furthermore, reticular chemistry provides a fascinating guide for the assembly of ordered frameworks with pre-designed structures, high porosity, and organic functionality (28). Benefiting from the highly porous frameworks and tunable metrics, reticular frameworks (RF) are showing great application prospects in gas adsorption and storage (29). More intriguingly, high gas affinity of the RF endows them with adsorbing gas molecules in the near-electrode region, which kinetically boosts various gas-involving electrochemical reactions including carbon dioxide reduction reaction (CO₂RR) and nitrogen reduction reaction (30–33). These groundbreaking studies also intrigued us to design our LOC electrochemical interface using reticular chemistry.

In this proof-of-concept study, we designed an electrochemical interface with the LOC effect to increase the OUE, accelerate the reaction kinetics and increase the energy efficiency of the electro-Fenton process. By using reticular chemistry, a well-defined amino-functionalized RF (NH₂-RF) with precisely designed molecular structure was built on a model carbon felt electrode. In situ Raman spectra investigations unraveled that the O₂ molecules were concentrated from the bulk solution to the reaction interface. By quantifying the diffusion thickness from the electrochemical impedance spectroscopy (EIS), we found that the increased O₂ concentration near the reactive center further accelerated the mass transfer. To elucidate the O₂ concentrating mechanism, we employed density functional theory (DFT) calculations and grand canonical Monte Carlo (GCMC) simulations to investigate the activities of functional groups and possible O₂ interaction sites. Using the fabricated C-NH₂-RF electrode, up to 36.35% of OUE was achieved, surpassing the performance reported in the literature. Moreover, a series of experiments were performed to degrade refractory pollutant sulfamethoxazole (SMX). The outstanding removal kinetics and low energy consumption further supported the critical role of NH₂-RF in elevating the efficiency of the electro-Fenton process. Finally, a flow-through LOC pilot reactor was constructed to treat industrial wastewater and a 65.56% energy conservation was achieved, revealing the feasibility of the proposed strategy in practical applications.

Results and Discussion

Design of a Cathode with LOC Effect Using Reticular Chemistry.

Reticular chemistry enables the design of porous channels and O₂ interaction sites, and it was used to construct an LOC electrochemical interface. Fundamentally, both facilitated O₂ mass transfer and enhanced O₂ binding are highly desired. As such, a starting RF with two typical types of micropores was screened. As displayed in Fig. 2A, the one-dimensional hexagonal tunnels with open apertures provided entry pathways for the O₂-containing electrolyte, and the triangular bipyramidal cages provided abundant accessible active sites for O₂ adsorption. More importantly, as the core of the building unit, the organic linker 1,4-benzenedicarboxylic acid (BDC) played a crucial role in O₂ adsorption (Fig. 2B). Therefore, an amino (−NH₂) group, which provided stronger interactions with oxygen species, was introduced to functionalize the BDC linker (Fig. 2C) (30).

Fig. 2.

Design of an LOC C-NH₂-RF electrode with reticular chemistry. (A−C) Design strategies for amino-functionalized RF; SEM (D) and TEM (E) images of the C-NH₂-RF; (F−K) Elemental mapping images of the NH₂-RF; XRD patterns (L), FTIR spectra (M), contact angles, and BET surface areas (N) of the CF, C-RF, and C-NH₂-RF, respectively.

As a result, the NH₂-RF was uniformly decorated on the carbon fibers via a “build-on” strategy (Fig. 2D and SI Appendix, Figs. S1−S4). As expected, the highly porous structure of the coated NH₂-RF facilitated mass transport of both O₂ and the ionic reactants to the carbon fiber (Fig. 2E, with more analysis in SI Appendix, Text). Moreover, the atomically dispersed −NH₂ groups trapped the transported O₂ molecules (Fig. 2F−K). The X-ray diffraction (XRD) pattern indicated the high crystallinity of the integrated NH₂-RF (Fig. 2L). The Fourier transform infrared (FTIR) spectra and X-ray photoelectron spectroscopy (XPS) confirmed the successful functionalization of the BDC linkers with −NH₂ groups (Fig. 2M and SI Appendix, Figs. S5−S8). Notably, the Brunauer–Emmett–Teller (BET) surface area was significantly increased after −NH₂ functionalization, while the in-air water contact angle changed negligibly (Fig. 2N); this endowed the integrated C-NH₂-RF electrode with LOC capability based on chemical interactions rather than simple hydrophobic effects.

Investigation of LOC Effect Using In Situ Electrochemical Characterizations.

To confirm the LOC effect of the C-NH₂-RF in electrochemical reactions, in situ Raman spectroscopy was used to monitor the binding of molecular O₂ at the electrochemical interface (Fig. 3A). Notably, with a continued feed of the O₂-saturated electrolyte, an intensive peak representing the stretching vibrations of adsorbed O₂ molecules was observed in the spectrum of the C-NH₂-RF (34, 35), demonstrating the superior LOC performance (Fig. 3 B and C). Moreover, compared with C-RF, the characteristic peak for SO₄²⁻ on C-NH₂-RF declined significantly, which was attributed to repulsion by the concentrated O₂ at the interface. Unlike the SOD route with a pristine CF electrode, for which an intense SO₄²⁻ peak was generated by the concentrated ionic electrolyte at the electrode surface (SI Appendix, Fig. S9), a high local O₂ concentration eventually formed near the C-NH₂-RF interface based on the LOC effect. As revealed by the previous studies (31–33), suitable electrolyte repelling and superior N₂ concentrating effects of the RF can greatly boost N₂ transport to the catalyst surface. Inevitably, O₂ diffusion should be dramatically altered near this LOC interface.

Fig. 3.

In situ electrochemical characterizations used to investigate the LOC effect of the C-NH₂-RF. (A) Schematic illustration of the tailor-made device for in situ Raman spectroscopy; Raman spectra of the C-RF (B) and C-NH₂-RF (C) in an O₂-saturated electrolyte; Nyquist plots (D) and Bode plots (E) for the CF, C-RF, and C-NH₂-RF in the O₂-saturated electrolyte (the solid lines and symbols show the experimental data and fitting results, respectively); (F) diffusion layer thicknesses for the CF, C-RF, and C-NH₂-RF in the O₂-saturated electrolyte.

EIS is an effective and noninvasive method for investigating the electrical, electrochemical, and physical processes occurring in electrochemical systems (36). In particular, the Nernstian diffusion at the diffusion layer is characterized by an equivalent impedance in the low-frequency region of EIS (37–39). An equivalent circuit model is proposed to quantify the O₂ diffusion layer that extends from the electrochemical interface to the bulk solution (SI Appendix, Fig. S10) (19, 40–42). Based on the resulting Nyquist and Bode plots (Fig. 3 D and E), the diffusion layer thickness was estimated to be 3.15 μm for C-NH₂-RF, which was 6.47 times lower than that for CF (Fig. 3F). As indicated by the previous study (42), the reduced diffusion layer thickness can decrease the distance that CO₂ must diffuse through to reach the catalyst, thereby improving CO₂ mass transport and CO₂RR performance. Predictably, the thinner diffusion layer for C-NH₂-RF would also accelerate O₂ mass transfer, which is consistent with its much stronger LOC effect relative to CF.

As expected, a high local O₂ concentration was achieved near the C-NH₂-RF electrode surface via the strong LOC effect. As a result, the surface coverage of the *O₂ (θ_O2) adsorbed on the catalyst surface for electrochemical reactions was also increased significantly, since *O₂ is proportional to the local concentration of O₂ ([O₂]) as follows: θ_O2 = θ*·[O₂]·exp(−E_O2/RT) (43). A recent study also highlighted the benefit of O₂-carrying microporous nanocrystals in enhancing oxygen reduction reaction (ORR) kinetics owing to the increased O₂ density in the vicinity of the catalyst (44). Similarly, the high local O₂ concentration near the C-NH₂-RF would promote the ORR rate as well as increase the H₂O₂ selectivity by stabilizing the intermediates and inhibiting protonation within the reaction microenvironment (19). To reveal the ORR pathway, rotating ring-disk electrode tests were conducted (SI Appendix, Text S2). As can be seen from SI Appendix, Fig. S11, the disk currents of both C-RF and C-NH₂-RF are larger than that of CF, suggesting their higher ORR activities. In addition, the higher H₂O₂ selectivity of C-NH₂-RF compared to that of C-RF confirms that the -NH₂ groups not only facilitate LOC, but also favor a two-electron ORR pathway.

Mechanism Insight into the LOC Effect Enhancement by Organic Linker Tailoring.

To gain microscopic insights into the interactions between the concentrated O₂ and the C-NH₂-RF electrode, DFT calculations and GCMC simulations were conducted. First, the charge density difference between the absorbed O₂ molecules and the NH₂-RF structure was analyzed (Fig. 4 A and B). Obviously, the C atoms of the NH₂-BDC linker within the NH₂-RF structure had a much larger charge transfer with the absorbed O₂ molecule than the BDC linker within the RF structure, suggesting that the functional −NH₂ groups significantly enhanced the binding affinity of the BDC linker for O₂ (SI Appendix, Fig. S12). The interactions occurring between the C atom of the linker and molecular O₂ were investigated with an electronic density of states (DOS) analysis. A large overlap between the C-p states and O-p states was observed for the NH₂-RF structure, while no such phenomenon was observed for the RF structure (SI Appendix, Fig. S13). The enhanced hybridization between C and O after introduction of the −NH₂ groups promoted the interaction between the linker and O₂, which was consistent with the charge density difference analysis (Fig. 4C). In continuing this investigation, we used GCMC simulations to obtain the center of mass (COM) probability distributions for the O₂ molecules within the RF structures. Obviously, the NH₂-RF structure showed a higher probability density distribution than the RF structure, suggesting that the −NH₂ groups facilitated adsorption of the O₂ (Fig. 4 D and E). The significantly higher adsorption energy and O₂ density within the NH₂-RF structure both verified the crucial role of the −NH₂ groups in elevating the LOC effect at the electrochemical interface (Fig. 4F and with more details in SI Appendix, Table S3). It should be noted that NH₂-functionalization of BDC is also effective in increasing local CO₂ concentration near the catalyst and promoting CO₂RR performance, as revealed by a previous study (30), indicating the versatility of organic linker functionalization in tuning local gas environment during electrocatalysis.

Fig. 4.

Theoretical insight into the LOC effect of the C-NH₂-RF. Charge density differences for O₂ adsorption on RF (A) and NH₂-RF (B) (isosurface value: 0.003 e/Bohr³); (C) DOS for O atoms adsorbed on the RF and NH₂-RF structures; contour plots of the COM probability density distributions for O₂ adsorbed on the RF (D) and NH₂-RF (E), respectively (298 K and 1.0 bar); (F) O₂ adsorption energy and average numbers of O₂ molecules adsorbed by the RF and NH₂-RF structures.

H₂O₂ Production Performance and Pollutant Degradation Capability of C-NH₂-RF.

To quantitatively evaluate the oxygen reduction kinetics and enhancement by the LOC effect, we first analyzed the influence of the O₂ diffusion process on the H₂O₂ yield performance from the SOD route. Theoretically, a higher stirring speed would facilitate diffusion of the reactants from the bulk solution to the catalyst surface. The H₂O₂ yield was sharply increased with stirring speeds raising, indicating that O₂ diffusion was the limiting factor influencing the H₂O₂ production efficiency in the SOD process with a CF electrode. In addition, the H₂O₂ yield on CF was also increased with a higher O₂ flow rate because the higher O₂ concentration in the bulk electrolyte facilitated O₂ diffusion to the catalyst surface with a higher concentration gradient (Fig. 5A). Compared to that for the CF, a much higher H₂O₂ yield was seen for the C-RF even without O₂ pumping, which was attributed to the LOC effect of the C-RF. The H₂O₂ yield was further increased at the C-NH₂-RF electrode owing to the thinner diffusion layer and enhanced LOC effect at the interface of the C-NH₂-RF (Fig. 5B). After normalization to the electrochemical active surface area (ECSA), the C-NH₂-RF exhibited a H₂O₂ yield of 0.42 mg h^–1 cm_ECSA⁻² (SI Appendix, Fig. S15). Notably, due to the significant LOC effect, the C-NH₂-RF exhibited a high OUE of 36.35%, which was 11.18 times that of the CF. Accordingly, the electric energy consumption (EEC) for H₂O₂ production was significantly reduced from 70.03 to 12.37 kWh kg⁻¹ (SI Appendix, Fig. S16). Moreover, the H₂O₂ yield positively correlated with the O₂ flow rate and showed nearly no correlation with the stirring speed in the C-NH₂-RF system, which indicated that the main factor influencing H₂O₂ yield is the local O₂ concentration, other than reactant diffusion (Fig. 5C and SI Appendix, Table S2). The H₂O₂ production on the C-NH₂-RF electrode increased with the applied potential increasing (SI Appendix, Fig. S17). Besides, the H₂O₂ production and FE (Faradaic efficiency) under natural diffusion conditions were almost stable and showed only small fluctuations during 10 continuous cycles of 600 min, indicating the high stability and sustainability of the C-NH₂-RF electrode (Fig. 5D). Similar conclusions can also be obtained from the insignificant change on the physicochemical properties of the C-NH₂-RF electrode after electrocatalysis (SI Appendix, Figs. S18−S22). It should be noted that the dissolved O₂ concentration in the electrolyte slightly increased during the H₂O₂ production experiments without aeration, which can be attributed to the quick supplement of O₂ evolving on the anode (SI Appendix, Fig. S23). Moreover, the dissolved O₂ concentration for CF is higher than that for C-NH₂-RF, which may be attributable to the higher OUE of C-NH₂-RF. These results indicate that the rate-limiting factor is local O₂ utilization rather than the O₂ concentration in bulk solution in an undivided reactor, for the O₂ evolved on the anode can supplement O₂. As the OUE and EEC of the proposed electrochemical system surpassed those of most reported systems (SI Appendix, Table S4), this system-level design shows great promise for increasing the energy efficiency of the electro-Fenton process.

Fig. 5.

Electrocatalytic H₂O₂ production and SMX degradation with the C-NH₂-RF. (A) The effects of stirring speed and O₂ flow rate on H₂O₂ production by the CF; (B) H₂O₂ production without aeration (Left part) and OUEs (Right part) of the CF, C-RF, and C-NH₂-RF systems; (C) colored contour map showing H₂O₂ production by the C-NH₂-RF as a function of stirring speed and O₂ flow rate; (D) stability tests of the C-NH₂-RF during H₂O₂ production without aeration; (E) SMX removal for the CF/Fe²⁺ and C-NH₂-RF/Fe²⁺ systems; (F) ESR spectra for detection of •OH with DMPO used as a trapping agent in the CF/Fe²⁺, C-NH₂-RF/Fe²⁺ and C-NH₂-RF/Fe²⁺ (N₂) systems.

With this in mind, we selected SMX, a widely used antibiotic in the livestock industry, as a model contaminant with which to evaluate the pollutant removal efficiency (RE) of the C-NH₂-RF electrode in the electro-Fenton process. Notably, almost 100% of the SMX was removed by the C-NH₂-RF/Fe²⁺ system within 30 min and without aeration, while only 68.89% SMX removal was achieved with the CF/Fe²⁺ system under the same conditions (Fig. 5E). In addition, the SMX removal proceeded via pseudo-first-order kinetics, and the apparent rate constant (k_obs) for the C-NH₂-RF/Fe²⁺ system was 0.18 min⁻¹, which was 4.61 times higher than that for the CF/Fe²⁺ system (SI Appendix, Fig. S24A). Electron spin resonance (ESR) spectroscopy was used to identify the reactive oxygen species and 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) was used as a trapping agent. An intense quartet signal with an intensity ratio of 1:2:2:1, which corresponded to the DMPO-•OH adduct, was observed in both the CF/Fe²⁺ and C-NH₂-RF/Fe²⁺ spectra, indicating that •OH was the main oxidant in both systems (Fig. 5F). This was confirmed by the fact that only 12.56% of the SMX was removed by the C-NH₂-RF/Fe²⁺ system when tert-butanol was added as a •OH scavenger (SI Appendix, Fig. S24B). Notably, minimal SMX removal was achieved via physical adsorption or electrochemical reduction under a N₂ atmosphere, suggesting that SMX removal by the C-NH₂-RF/Fe²⁺ system proceeded via H₂O₂ generated by oxygen reduction. In addition, the specific energy consumption (SEC) of the C-NH₂-RF/Fe²⁺ system was 0.49 kWh g^–1 TOC (total organic carbon), which corresponded to only 60.92% of the energy consumed by the CF/Fe²⁺ system (SI Appendix, Fig. S25). Based on the degradation intermediate products detected, the possible SMX degradation pathways were proposed (SI Appendix, Text S3 and Fig. S26). The SMX molecule was converted into low-molecular-weight intermediates after hydroxylation, deamination, demethylation, dehydroxylation, and ring-opening reactions and finally mineralized into CO₂ and H₂O. Of note, the k_obs for SMX removal by the C-NH₂-RF/Fe²⁺ system without aeration were comparable to that for the traditional electro-Fenton processes with O₂/air pumping, suggesting ample potential for increasing the energy efficiency by reducing or even eliminating the energy consumed by the aeration process (SI Appendix, Table S5). The H₂O₂ concentration during SMX degradation was also measured (SI Appendix, Fig. S27). About 3 mg L⁻¹ H₂O₂ remained in the solution, as most H₂O₂ was decomposed by Fe²⁺.

Energy Efficiency Analysis of the Electro-Fenton Process with LOC Effect.

In view of the excellent organic pollutant degradation performance shown by the C-NH₂-RF/Fe²⁺ system under natural air diffusion conditions, we amplified the LOC effect to enhance the electro-Fenton process for treating actual industrial wastewater. As displayed in Fig. 6A, our scalable flow-through wastewater treatment system included two steps: 1) The H₂O₂ used to degrade the organic pollutants was generated in the electrolyzer with air and electricity and 2) the wastewater mixed with the H₂O₂ produced in situ flowed through a Fenton filter where decomposition of the residual H₂O₂ was catalyzed (SI Appendix, Text S4). Unlike the conventional electro-Fenton process in which aeration is indispensable, no pump equipment was needed for our LOC system. We envisioned that increasing the OUE instead of the aeration rate was the best way to decrease the overall energy consumption of the electro-Fenton process, as indicated by a technoeconomic analysis (TEA) (Fig. 6B and SI Appendix, Text S4). More promisingly, by reducing the need for aeration, the LOC approach will also bring appreciable environmental benefits, including reducing operation costs, chemical consumption, and possible release of harmful substances to the surroundings.

Fig. 6.

Energy-efficient electro-Fenton process enhanced by the LOC approach in treating actual industrial wastewater. (A) Schematic drawing of the energy-efficient LOC electro-Fenton process for wastewater treatment; (B) TEA of the energy consumed by wastewater treatment with the electro-Fenton process; (C) photograph of the flow-through device used to treat industrial wastewater, (D) COD of the treated industrial effluent and (E) SEC of the flow-through device.

Using the versatile build-on strategy, five C-NH₂-RF electrodes were successfully fabricated with the desired dimensions (7 cm × 14 cm), and they were used as cathodes in the pilot-scale flow-through device. Five commercially available RuO₂–IrO₂/Ti mesh electrodes were placed 5 mm from the cathodes and served as anodes. In particular, a Fenton filter with uniformly immobilized catalytic magnetite nanoparticles was assembled into a tailored device located behind the electrochemical reactor to catalyze the decomposition of the residual H₂O₂ (SI Appendix, Figs. S28–S30). Subsequently, we used the flow-through system and continuously treated actual industrial effluent from a chemical industrial park (Fig. 6C and SI Appendix, Tables S6 and S7). After the oxidation process, the chemical oxygen demand (COD) decreased from 215.32 mg L⁻¹ to 42.53 mg L⁻¹ (Fig. 6D). In the meantime, 80.47% of the TOC was removed with an SEC of 0.39 kWh g^–1, which was lower than those of other advanced electrochemical oxidation processes (SI Appendix, Table S8). Moreover, the EEC was 65.56% less than that of the pristine CF electrode without the LOC effect (Fig. 6E and SI Appendix, Figs. S31 and S32). This verified the feasibility of the proposed LOC approach in optimizing the electro-Fenton process for energy-efficient water purification. Although additional aspects, such as clogging of the LOC layer over time and uneconomical electrode regeneration after long-term operation, must be accounted for when assessing the potential of the present electrode in scale-up applications, we envision that, based on the LOC interface design, advanced electrode materials exhibiting better treatment performance, higher stability, and reusability will be discovered soon.

Conclusions

In summary, we have proposed an innovative LOC approach for the electro-Fenton process, which alleviated the O₂ mass transfer problems caused by the low O₂ solubility and slow diffusion in the conventional SOD route. The LOC effect allowed direct extraction of O₂ from the bulk solution, which reduced the diffusion layer thickness and the O₂ concentration gradient near the electrochemical interface. As a result, the high local O₂ concentration formed at the interface significantly increased the OUE, which was 11.18 times that of the SOD route and higher than those of the state-of-the-art systems. Due to the remarkably high OUE (36.35%) and rapid H₂O₂ production, the LOC effect provided outstanding degradation kinetics with a rate constant of 0.18 min^–1 for decomposition of the refractory organic pollutant SMX via the electro-Fenton process run without pumping O₂/air. Compared with the conventional SOD route, this system-level design conserved over 65% of the electric energy normally used in treating industrial wastewater without aeration at the pilot scale. This work provides a strategy for O₂ concentrating at the reaction interface and constitutes a fundamental breakthrough in boosting the energy-efficient electro-Fenton process for advanced water remediation.

Materials and Methods

Electrode Preparation.

A piece of CF (2.0 cm × 3.0 cm × 0.5 cm, with more details in SI Appendix, Table S1) was immersed in acetone and ultrasonicated for 30 min to remove impurities and oil stains and then thoroughly rinsed with deionized water. Then, the RF layer was grown in situ on the cleaned CF via the previously reported method with some modification (45). Pluronic F127 (180 mg), 200 mg of FeCl₃·6H₂O, and 0.3 mL of CH₃COOH were dissolved in 15 mL of deionized water, followed by stirring for another hour. Subsequently, the cleaned CF and 68 mg of NH₂-BDC were added to the above solution and stirred for another 2 h. The resulting mixture was then sealed in a Teflon-lined stainless-steel autoclave and heated at 110 °C for 24 h. After cooling to room temperature, the modified CF piece was washed with ethanol 5 times and dried at 80 °C in a vacuum oven overnight to obtain the C-NH₂-RF. Similarly, C-RF was synthesized in the same way except that BDC was used as the organic linker. Detailed information on the materials and chemicals used is presented in SI Appendix, Text S1.

Electrode Characterization.

The electrode morphology and elemental distribution were observed with a field emission scanning electron microscope (SEM, ZEISS Merlin Compact; Jena; Germany) and transmission electron microscope (TEM, JEOL JEM-2100; Tokyo; Japan). The crystal structures of the integrated RF were investigated with a D8 Advance X-ray diffractometer (Bruker; Karlsruhe; Germany) using Cu Kα radiation. The chemical compositions and added functional groups were identified by FTIR spectroscopy recorded on an FTIR spectrometer (Nicolet 5700, Thermo Fisher Scientific; Waltham; USA) and XPS spectra obtained from a Thermo Scientific K-Alpha spectrometer with Al Kα radiation. The porosities and specific surface areas were evaluated with nitrogen adsorption-desorption isotherms obtained with a surface area and porosimetry system (ASAP 2460; Micromeritics; Georgia; USA) and the BET method. The hydrophobicity was evaluated by measuring the in-air water contact angles on a goniometer (JY-82C; Chengde Dingsheng; Chengde; China). The LOC effect was determined by in situ Raman tests with a Horiba HR Evolution spectrometer (LabRamHR; HOBIN Co.; Paris; France) with an excitation wavelength of 532 nm.

Electrochemical Measurements.

Cyclic voltammetry (CV) and EIS were recorded on an electrochemical workstation (CHI660E, Chenhua Instrument Co., Ltd.; Shanghai; China) with a three-electrode system. Carbon felt-based electrode, SCE and Pt wire were used as working electrode, reference electrode and counter electrode, respectively. To be specific, CV and EIS were measured in 0.05 mol L⁻¹ Na₂SO₄ solution and O₂-saturated 0.05 mol L⁻¹ Na₂SO₄ solution respectively. The CV was recorded at the scan rates of 40, 80, 120, 160, and 200 mV s⁻¹. The EIS analyses were conducted at 100 mHz to 100 kHz with a 5-mV amplitude. An equivalent circuit model was proposed to describe the impedances of the porous carbon electrode (40), where R_s is the internal resistance, R_trns is the resistance to electron transport in the porous carbon layer, R_ct is the charge transfer resistance at the electrode/electrolyte interface, C_trap is the trap capacitance in the carbon layer, C_dl is the double-layer capacitance at the electrode/electrolyte interface, and Z_d is the impedance of the diffusion layer (SI Appendix, Fig. S10). At low frequencies, the impedance of the diffusion boundary layer can be expressed as (41):

$$Z_d\cong R_d-j\times w\times R_d^2{\times C}_d,$$ [1]

where j is an imaginary unit, ω is the angular frequency, and R_d and C_d are the equivalent resistance and capacitance, respectively.

The diffusion layer thickness δ was determined with the following equation (41):

$$\delta =\sqrt{3\times D_0\times R_d\times C_d},$$ [2]

where D₀ is the O₂ diffusion coefficient (1.96 × 10^–9 m² s^–1).

The ECSA of the electrode was calculated based on the following equation:

$$\text{ECSA}=\frac{C_{\text{EDLC}}}{C_{\text{EDLC}, \text{carbon}}},$$ [3]

where C_EDLC is the electric double-layer capacitance of electrode and is determined by CVs at various scan rates, and C_{EDLC, carbon} is the ideal specific capacitance of carbon electrode (C_{EDLC, carbon} = 10 μF cm^–2).

Electrocatalytic H₂O₂ Production Experiments.

The electrocatalytic H₂O₂ production experiments were conducted in a 100-mL undivided cell filled with 0.05 mol L⁻¹ Na₂SO₄ as the supporting electrolyte. The C-NH₂-RF (electroactive area: 5 cm²), Pt plate (2 cm × 2.5 cm), and SCE were used as the working, counter, and reference electrodes, respectively. Unless specified, for all electrocatalytic tests, a constant cathodic potential of −0.9 V vs. SCE and a stirring speed of 300 rpm were applied with a CHI660E electrochemical workstation. The concentration of dissolved O₂ during electrolysis was also qualified by a dissolved oxygen analyzer (TN2509, Industrial Scientific Corporation).

Pollutant Degradation Experiments.

Specifically, the electrocatalytic SMX degradation experiments were carried out in a 50-mL undivided cell filled with 10 mg L⁻¹ SMX, 0.2 mmol L⁻¹ Fe²⁺, and 0.05 mol L⁻¹ Na₂SO₄ as the supporting electrolyte, and the pH of the electrolyte was adjusted to 3.0 with a 0.1 mol L⁻¹ H₂SO₄ solution. At the various time intervals, a 100 μL aqueous sample was withdrawn and quenched with tert-butyl alcohol (100 mmol L⁻¹) for further analysis. The H₂O₂ concentration during SMX degradation was also measured.

Pilot-Scale Experiments.

The flow-through pilot-scale reactor (21.5 cm × 15 cm × 9 cm) was home-made by acrylic material (SI Appendix, Fig. S30). The industrial effluent was flowed through this pilot device with a flow rate of 300 mL h⁻¹. Five large-scale C-NH₂-RF electrodes (7 cm × 14 cm) were placed 5 mm from corresponding RuO₂–IrO₂/Ti mesh electrodes with the same size. The working current density was 2.5 mA cm⁻².

Analytical Methods.

The H₂O₂ concentration was measured with the potassium titanium (IV) oxalate method and a UV–vis spectrophotometer (18). At various time intervals, approximately 1.5 mL of the aqueous sample was filtered through a polytetrafluoroethylene syringe filter (0.45 μm; Massachusetts; USA), then 1 mL of the filtered solution, 0.5 mL of a H₂SO₄ solution (3 mol L^–1), 0.5 mL of a potassium titanium (IV) oxalate solution (0.05 mol L^–1) and 1 mL of deionized water were thoroughly mixed. Subsequently, the absorbance of the mixed solution was measured at 400 nm with a UV–vis spectrophotometer (UV-2600, Shimadzu; Kyoto; Japan). The absorbance–concentration curve was calibrated with a series of standard H₂O₂ solutions (SI Appendix, Fig. S14).

The H₂O₂ generation rate (Y_H2O2, μmol h⁻¹ cm⁻²) was calculated according to:

$$Y_{{\mathrm{H}}_2{\mathrm{O}}_2}=\frac{3600\times V\times C\left({\mathrm{H}}_2{\mathrm{O}}_2\right)\times 1000}{t\times A\times M_{{\mathrm{H}}_2{\mathrm{O}}_2}},$$ [4]

where V (L) is the volume of electrolyte, C(H₂O₂) (mg L⁻¹) is the concentration of generated H₂O₂, t (s) is the time, A (cm²) is the electrode area, and M_H2O2 is the H₂O₂ molecular weight (34 g mol⁻¹).

The OUE (%) was calculated as follows:

$$\text{OUE}=\frac{\mathrm{n}({\mathrm{O}}_2, {2\mathrm{e}}^{-}\text{reduction})}{\mathrm{n}({\mathrm{O}}_2, \text{supplied})}\times 100\%,$$ [5]

where n(O₂, supplied) is the amount of O₂ supplied in moles by aeration, and n(O₂, 2e⁻ reduction) is the amount of O₂ utilized for H₂O₂ generation.

The n(O₂, supplied) was calculated by:

$$\mathrm{n}({\mathrm{O}}_2, \text{supplied})=\frac{R\left({\mathrm{O}}_2\right)\times t}{{60000\times V}_{\mathit{m}}},$$ [6]

where R(O₂) is the aeration rate of O₂ (mL min⁻¹), t is the aeration time (s), and V_m is the molar volume of gas (24.5 L mol⁻¹ at 25 °C).

The n(O₂, 2e⁻ reduction) was calculated according to:

$$\mathrm{n}\left({\mathrm{O}}_2, 2{\mathrm{e}}^{-} \text{reduction}\right)=\mathrm{n}\left({\mathrm{H}}_2{\mathrm{O}}_2\right)=\frac{C\left({\mathrm{H}}_2{\mathrm{O}}_2\right)\times V}{M_{{\mathrm{H}}_2{\mathrm{O}}_2}},$$ [7]

where M_H2O2 is the H₂O₂ molecular weight (34 g mol⁻¹).

The FE (%) for H₂O₂ production was calculated with the following equation:

$$\text{FE}\left(\%\right)=\frac{n\times F\times V{\times C}_{\mathrm{H}2\mathrm{O}2}}{M_{{\mathrm{H}}_2{\mathrm{O}}_2}\times 1000\times I\times t}\times 100\%,$$ [8]

where n is the electron transfer number (2 for oxygen reduction to H₂O₂), F is the Faraday constant (96,487 C mol⁻¹), and I (A) is the current.

The EEC (kWh kg⁻¹) for H₂O₂ production was calculated with the following equation:

$$\text{EEC}=\frac{U\times I\times t}{C_{{\mathrm{H}}_2{\mathrm{O}}_2}\times V},$$ [9]

where U and I are the applied voltage (V) and current (A), respectively.

The concentration of SMX was measured by high-performance liquid chromatography (e2695, Waters, USA) with a C18 reversed-phase column (4.6 mm × 150 mm) and a UV detector. The mobile phase was a mixture of ultrapure water and methanol with a volume ratio of 60:40 and a flow rate of 0.8 mL min⁻¹. The column temperature and the detection wavelength were set at 30 °C and 254 nm, respectively. The SMX RE was calculated with the following equation:

$$\text{RE}\left(\%\right)=\frac{C_0-C_{\mathit{t}}}{C_0}\times 100\%,$$ [10]

where C₀ and C_t are the concentrations of SMX at the beginning of the reaction and at a given time (t), respectively.

The SMX removal kinetics were analyzed with the pseudo-first-order kinetic model and the following equation:

$$ln\left(\frac{C_{\mathit{t}}}{C_0}\right)=k_{\mathit{o}\mathit{b}\mathit{s}}\times t,$$ [11]

where k_obs is the pseudo-first-order rate constant (min⁻¹) for SMX removal.

The TOC was determined with a TOC analyzer (TOC V-CPN; Shimadzu; Kyoto; Japan). For TOC analyses, 10 mL of the sample was treated immediately after degradation with the scavenging reagents (0.1 mol L⁻¹ Na₂SO₃, 0.1 mol L⁻¹ KH₂PO₄, 0.1 mol L⁻¹ KI, and 0.05 mol L⁻¹ NaOH) to completely remove the residual H₂O₂. The COD was measured with a visible spectrophotometer (DR 1900, HACH, Colorado, USA).

The concentrations of metal ions were measured by inductive coupled plasma optical emission spectrometer (ICP–OES, iCAP 7000 Series, Thermo Fisher Scientific; Waltham; USA).

The SEC (kWh g⁻¹) for TOC removal was calculated with the following equation:

$$\text{SEC}=\frac{U\times I\times t}{V\times ({\text{TOC}}_0-{\text{TOC}}_t)}\times {10}^3,$$ [12]

where U is the voltage (V), I is the current (A), and TOC₀ and TOC_t are the COD values (mg/L) at the initial time and time t, respectively.

Computational Details.

All computations were implemented with the spin-polarized DFT method in the Vienna ab initio simulation package (46). The projector augmented wavefunction pseudo-potentials were used to describe the ionic potentials (47). The exchange correlation energy was calculated with the Perdew–Burke–Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) (48). During geometry relaxation, the energy cutoff was chosen as 500 eV. In the structural optimizations, the Brillouin zone was sampled by 3 × 3 × 2 mesh points in k-space based on the Monkhorst-Pack scheme for both the RF and NH₂-RF structures (49). The force convergence criterion for atomic relaxation was 0.02 eV Å^–1. The energy (ΔE_ad) for adsorption of O₂ on the RF and NH₂-RF structures was defined as:

$$\mathrm{\Delta }{E}_{\mathit{ads}}=E_{str+mol}-E_{\mathit{str}}-E_{\mathit{mol}},$$ [13]

where E_{str + mol} represents the energy of the O₂ molecule adsorbed on the RF and NH₂-RF structures. E_str represents the total energy of the RF and NH₂-RF structures, and E_mol is the energy of the O₂ molecule calculated in bulk.

GCMC simulations were performed with the Material Studio 8.0 package. The ground states of O₂ were obtained by DFT as implemented in the Dmol3 module (50). The double numerical plus d-functions were used as the basis set, and the PBE (51) GGA functions were used to capture the exchange correlation effects between electrons. The O₂ sorption locations were searched with GCMC simulations and the Metropolis method (52, 53). Condensed-phase Optimized Molecular Potentials for Atomistic Simulation Studies forcefields (54) were used to describe the chemical interactions among the frameworks and gas molecules. The Lennard-Jones interactions were cut off at 18.5 Å. Electrostatic interactions and van der Waals interactions were calculated with Ewald and atom-based summation rules. The GCMC sampling was modeled at 298 K, and the pressure of O₂ was fixed at 1.0 bar. A total of 1 × 10⁷ steps of GCMC were used to reach thermal equilibrium, and 5 × 10⁷ production steps were used to determine the conduct statistics.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.