Degradation and Mineralization of Carbamazepine Using an Electro-Fenton Reaction Catalyzed by Magnetite Nanoparticles Fixed on an Electrocatalytic Carbon Fiber Textile Cathode.

Pharmaceutical wastes are considered to be important pollutants even at low concentrations. In this regard, carbamazepine has received significant attention due to its negative effect on both ecosystem and human health. However, the need for acidic conditions severely hinders the use of conventional Fenton reagent reactions for the control and elimination of carbamazepine in wastewater effluents and drinking water influents. Herein, we report of the synthesis and use of flexible bifunctional nanoelectrocatalytic textile materials, Fe3O4-NP@CNF, for the effective degradation and complete mineralization of carbamazepine in water. The nonwoven porous structure of the composite binder-free Fe3O4-NP@CNF textile is used to generate H2O2 on the carbon nanofiber (CNF) substrate by O2 reduction. In addition, ·OH radical is generated on the surface of the bonded Fe3O4 nanoparticles (NPs) at low applied potentials (-0.345 V). The Fe3O4-NPs are covalently bonded to the CNF textile support with a high degree of dispersion throughout the fiber matrix. The dispersion of the nanosized catalysts results in a higher catalytic reactivity than existing electro-Fenton systems. For example, the newly synthesized Fe3O4-NPs system uses an Fe loading that is 2 orders of magnitude less than existing electro-Fenton systems, coupled with a current efficiency that is higher than electrolysis using a boron-doped diamond electrode. Our test results show that this process can remove carbamazepine with high pseudo-first-order rate constants (e.g., 6.85 h-1) and minimal energy consumption (0.239 kW·h/g carbamazepine). This combination leads to an efficient and sustainable electro-Fenton process.

Introduction

Personal care and pharmaceutical products (PPCPs) are becoming ubiquitous contaminants in aquatic and marine environments.[1] PPCPs are often regarded as pseudopersistent chemical contaminants due to their high consumption rates, partial degrees of biological transformation in conventional wastewater treatment plants, and their eventual discharge into natural waters.[2] For example, carbamazepine has been identified as one of the future emerging priority pollutant candidates by the European Union Water Framework Directive (WFD). Carbamazepine bioaccumulates in the living organism with subsequent endocrine disrupting effects, neurotoxicity, and developmental toxicity effects.[3−5] To eliminate carbamazepine from water, advanced oxidation processes (AOP) are needed.[6] Electrochemical oxidation is a unique class within the array of available AOPs, by which an oxidant can be generated onsite. To this end, we have previously investigated carbamazepine removal by electrochemically generated chloramine and achieved and high level of removal with a low energy consumption (13 kWh/log/m3); however halogenated byproducts were detected.[7,8] Therefore, a strong oxidant such as hydroxyl radical (·OH) is needed to ensure complete mineralization. However, electrochemical oxidation is energy intensive when employing dimensionally stable electrodes made from platinum group elements (PGE) as catalytic ohmic contacts. Platinum group metals, noble metal oxides, and boron-doped diamond electrodes are typically used for in situ reactive oxygen species generation.[9] Numerous attempts have been made to achieve comparable degradation efficiencies with inexpensive, earth abundant materials. We have recently reported on the use of SbSn/CoTi/Ir multilayer heterojunction anodes, which reduced the use of PGM.[10] Nevertheless, PGM-free electrodes represent a significant step toward green and sustainable use of AOPs. It has been demonstrated that carbon-based electrode material can be coupled in an electro-Fenton reaction to completely degrade PFOA.[11] However, a major drawback of this strategy is that large amount of FeSO4 needs to be dissolved in the electrolyte to facilitate the Fenton’s reagent reaction. Therefore, it is urgent to develop green and sustainable electrochemical AOPs that can work with minimal but stable non-precious metal catalysts.

The application of nanostructured catalysts is widespread[12−14] However, many of these catalytic materials are used in colloidal suspension, which significantly limits their use in practical applications. Therefore, supported metal nanoparticle (NP) have recently gained attention due to low energy consumption and ease of preparation.[15] Typically, supported metal NPs can be prepared by thermal or chemical decomposition of metal precursors on heterogeneous supports.[16,17] In addition, the reactivity of such heterogeneous catalytic systems can be tuned by controlling the topology, size, distribution of metal NPs, and the chemical composition of the structural support.[18] To date, however, metal NPs with heterogeneous support are rarely explored for electrochemical AOPs due to the difficulty in (1) achieving a uniform dispersion of metal NPs on a conductive substrate, (2) formation of an appropriate bond between a metal NP and the substrate to facilitate interfacial electron transfer, and (3) providing structural stability. Indeed, well-dispersed nanostructures on supporting substrates have been demonstrated to have superior mechanical and electrical properties compared to random dispersed counterparts.[19,20]

Herein, we describe the application of metal (M) and nitrogen (N) codoped graphite (C) MNC flexible electrodes which can act as a multiphasic, bifunctional electro-Fenton system for the degradation of pharmaceutical byproducts such as carbamazepine. The functionalized flexible MNC electrodes consist of Fe3O4 nanoparticles (NP) bonded to a carbon nanofiber (CNF) matrix, subsequently termed Fe3O4-NP@CN. This composite electrode is synthesized via pyrolysis of electrospun nanofiber (NF) which allows for the precision control of the Fe3O4-NP dispersions even at low concentration (<1 wt % loading). The degradation of carbamazepine on Fe3O4-NP@CNF electrodes is investigated to minimize the Fe3O4-NP loading, while maximizing ·OH production via the electro-Fenton reaction.

Materials and Methods

Materials

All reagent-grade chemicals were obtained from Sigma-Aldrich and were used without any further purification. Solvents used were HPLC grade unless otherwise stated.

Preparation of CNF and Fe3O4 NP@CNF

CNF was prepared by carbonization of polyacrylonitrile (PAN) NF, which was prepared using an electrospinning method.[21] In a typical electrospinning synthesis, 1.5 g of PAN (molecular weight = 150000 g/mol) was dissolved in 15.6 mL of N,N-dimethylformamide (DMF) and mixed thoroughly for 12 h. The solution was fed through a stainless needle using a syringe pump (New Era pump system) at a rate of 0.6 mL/h. An aluminum foil collector was placed at a distance of 10 cm from the needle. A voltage of 10 kV was applied (EDSEMC high-voltage power supply) to produce a nonwoven mat of PAN nanofibers. The prepared PAN-NF mats were separated from the collectors and heated at 240 °C for 2 h to promote cross-linking to enhance mechanical strength. In a final step, the cross-linked PAN-NF mats were carbonized at 900 °C for 0.5 h under a N2 atmosphere to produce flexible CNF mats.

Fe3O4-NP@CNF composites were prepared using a similar process. DMF solutions with an appropriate amounts of iron acetylacetonate (Fe(acac)3) and PAN were used to produce nonwoven, yellow Fe-doped PAN mats. Further treatment was employed to promote cross-linking via pyrolysis at 900 °C for 0.5 h under a N2 atmosphere to produce the flexible Fe3O4-NP@CNF mats. Fe3O4-NP@CNF samples were prepared from solutions with the ratio of 0.05, 0.1, 0.3, 0.5, 1.0, and 5.0 wt % of Fe(acac)3 to PAN; they were denoted as FeCNF0.05, FeCNF0.1, FeCNF0.3, FeCNF0.5, FeCNF1, and FeCNF5, respectively.

Preparation of Electrodes

Even though CNF and Fe3O4-NP@CNF mats are conductive, maintaining a constant anode–cathode distance within a batch reactor is a difficult challenge due to their flexibility and deformability. To overcome this problem, rigid carbon paper was coated with the active catalyst to function as the working electrode. The electroactive coating was prepared by suspending ground CNFs or Fe3O4-NP@CNF powders (10 mg) in to 2.2 mL of Milli-Q water, 0.75 mL of isopropanol, and 50 μL of Nafion solution. The mixture was then spray-coated onto carbon paper (AvCarb MGL 190) at a catalyst loading of 0.5 mg/cm2. Both sides of carbon paper were uniformly coated (total geometric surface area of 8 cm2) to avoid electrolyte exposure of the supporting carbon paper substrate.

Characterization of the Electrocatalysts

The surface morphology and elemental composition of the catalytic composites were examined using a ZEISS 1550 VP field-emission scanning electron microscope (FE-SEM) equipped with an Oxford X-Max SDD energy-dispersive X-ray spectrometer (EDS). TEM and TEM-EDS measurements were obtained on a FEI Tecnai F30ST (300 kV) transmission electron microscope (TEM) equipped with Oxford ultrathin window EDS detector. X-ray powder diffraction patterns were collected from a Panalytical X’pert Pro diffractometer with Cu Kα radiation (λ = 1.5418 Å). The elemental composition of the electrocatalysts was characterized by X-ray photoelectron spectroscopy (XPS) using a Surface Science Instruments M-Probe ESCA surface spectrometer. Monochromatic Al Kα radiation (1486.6 eV) was used.

Electrochemical Experiments

Electrochemical experiments were performed using a BioLogic VSP potentiostat with a three-electrode reactor cell at room temperature. A platinum (Pt) wire and saturated calomel electrode (SCE) were used as a counter electrode and a reference electrode, respectively. The recorded potentials were converted to the reversible hydrogen electrode (RHE) scale (RHE = SCE + 0.059 × pH + 0.241 V). Cyclic voltammetry (CV) was performed in an Ar- or O2-saturated electrolyte (0.1 M Na2SO4) at a scan rate of 10 mV/s. Constant potentials were applied to investigate H2O2 production on CNF and the electro-Fenton degradation of carbamazepine using Fe3O4-NP@CNF. An O2-saturated electrolyte (0.1 M Na2SO4) was used, while the solution pH was adjusted over the range of 4–10 using a combination of sulfuric acid and sodium hydroxide. H2O2 production yields were determined in triplicate. Electro-Fenton experiments were performed in duplicates with average values reported.

Analytical Methods

The concentration of H2O2 in samples collected at given time intervals were analyzed by titanium oxalate spectrophotometric method.[22] In this procedure, the collected reaction sample was acidified with 3 M sulfuric acid solution and then mixed with a potassium titanium oxalate solution (0.5 wt % in water) to produce the yellow pertitanic acid complex. The absorbance of the solution was measuring using a NanoDrop spectrophotometer at 400 nm. H2O2 solutions with known concentrations were used to construct a calibration curve.

During the electro-Fenton treatment process, the concentrations of carbamazepine and terephthalic acid (TPA) were analyzed by an ultrahigh performance liquid chromatography (Waters Acquity UPLC) system coupled to a UV detector (Acquity PDA) and a time-of-flight mass spectrometer (Waters XEVO GS-2 TOF). Acquity BEH C18 column (2.1 × 50 mm2; 1.7 μm particles) was used. (A) 0.1% formic acid and (B) acetonitrile were used as mobile phase with flow rate of 0.5 mL min–1. The mass spectrometer was operated with a source temperature of 120 °C, a solvent removal temperature of 400 °C, and a capillary voltage of 0.2 kV.

Results and Discussion

Design Approach

Carbon-based electrodes (graphite, graphene, carbon nanotubes (CNTs), etc.) are utilized due to their ability to produce H2O2 via the selective oxygen reduction reaction (ORR). They serve as effective and economic alternatives for the state-of-art platinum group metals for the ORR, although the inherent chemical reactivity of H2O2 is insufficient to achieve complete mineralization of aqueous chemical contaminants such as PPCPs and perfluorinated compounds (PFCs). However, the O–O single bond strength (e.g., 142 kJ mol–1) is relatively weak, and thus in H2O2 the O–O bond is easily broken to produce hydroxyl radical. Incorporation of electroactive metals into carbon substrates provides a convenient method to generate ·OH of H2O2. The electro-Fenton process using Fe(III)-doped CNT has been shown to be able assist in the degradation of recalcitrant organic compounds such as trifluoroacetic acid or trichloroacetic acid.[23] However, weak electrostatic interactions between metal ion dopant and carbon substrate result in a significant metal loss by leaching; in addition, regeneration of the active electrocatalyst is required.[23] To overcome this problem, we have designed the Fe3O4-NP@CNF electrocatalyst based on the MNC approach. We believe that the ideal MNC electrocatalyst must meet certain criteria: (1) the metal-based catalyst must be well dispersed on the support substrate to allow sufficient contact of the electrolyte with both the catalyst and the substrate, (2) the metal-based catalyst (e.g., Fe3O4) must form a bond with the substrate, and (3) the electrocatalyst should exhibit flexibility. The first criterion allows an optimal utilization of the metal-based nanocatalytic material, which tends to aggregate into clusters in the absence of the substrate since the surface free energy is increased due to the decreasing particle size. The third criterion allows the construction of roll-type or foldable electrodes with associated form-factor advantages, which can maximize the electrode surface to electrolyte volume ratio.

To develop MNC based on these criteria, the formation of a fibrous scaffold was chosen. Electrospun PAN nanofibers were chosen for this purpose due to the abundance of nitrile (−C≡N) functional groups. Magnetite (Fe3O4) incorporation was achieved using a gelation-free coelectrospraying method. This method involves the mixing the nanofibers and iron acetylacetonate (Fe(acac)3), which serves as the Fe3O4 precursor. The electrospun Fe(acac)3@PAN nanofiber mat was heated to 240 °C for 30 min to enhance the physical strength of the mat. Two conversion process occurred simultaneously: (1) the thermal decomposition of Fe(acac)3 into Fe3O4 nanoparticles (NP) and (2) the formation of a thermoset ladder structure of the thermoplastic PAN nanofibers via cyclization.[24] The initial product produced, Fe3O4-NP@PAN, had a low conductivity; however, after calcination at 800 °C under a nitrogen atmosphere a conductive Fe3O4 NP@CNF was produced. Figure shows that a macroscopic nonwoven “nano-electrocatalytic textile” is produced (vide supra).

Figure 1

Digital photograph of folded nonwoven (A) PAN, (B) Fe(acac)3@PAN, (C) Fe3O4 NP@PAN, and (D) Fe3O4 NP@CNF (FeCNF0.1) mats.

Morphology and Structure

SEM micrographs show that both CNF and Fe3O4-NP@CNF are composed of interwoven and porous networks (Figure ) that are fully accessible to electrolytes and dissolved gas molecules in an aqueous electrolyte solution. These open networks should allow for effective mass transfer and gas dispersion during electrolysis. Because of the low concentrations of Fe3O4-NPs the underlying CNF substrate is unaffected by the presence of the magnetite nanoparticles. Both CNF and Fe3O4-NP@CNF have an average diameter of 180 nm. Because of the simplicity of self-assembling Fe3O4-NPs within the electrospun PAN-NF, control over the Fe3O4-NP loading level can be achieved. As shown in Figure S1, a series of Fe3O4-NP@CNF assemblages were obtained using different Fe(acac)3 precursor concentrations (0.1–5.0 wt %). The actual Fe loading in the Fe3O4-NP@CNF arrays was determined using SEM-EDS elemental analysis; these results are summarized in Table S2. EDS mapping provided direct evidence that the Fe3O4-NPs were evenly dispersed in CNF support substrate (Figure S1). Smooth nanofibers were observed except for the FeCNF5 sample; the SEM images indicate that the Fe3O4-NPs aggregate at high loadings (i.e., ≥5 wt %). To identify the crystal shapes of Fe3O4 NP doped onto CNF, TEM measurements of FeCNF1 were conducted (Figure S2). However, because of low loading levels (1 wt %), Fe3O4-NPs were indistinguishable from the CNT substrates. However, TEM-EDS characterization (Figure S3) confirms a uniform dispersion of Fe3O4-NPs.

Figure 2

SEM images of (A) CNF and (B) Fe3O4 NP@CNF (FeCNF0.1).

The structure and phase composition of the CNFs and Fe3O4-NP@CNFs were characterized using powder X-day diffraction (PXRD). Because of the low detection limit of PXRD (3%), Fe3O4-NP@CNF at the highest magnetite loading (FeCNF10) was used for PXRD analyses. As shown in Figure S4, the metal nanocatalyst in Fe3O4-NP@CNF is indeed magnetite, Fe3O4. The XRD pattern is also consistent with cubic Fe3O4.

XPS analysis was used to determine other features of the surface chemistry of CNF and Fe3O4-NP@CNF (Figure S5). Because of the sensitivity of the XPS instrument used, the loadings of magnetite on FeCNF0.05 and FeCNF0.1 were too low to obtain suitable Fe 2p spectra; however, the FeCNF1 sample gave a resolved XPS spectrum. The high-resolution XPS spectra confirmed the presence of N, O, C, and Fe in FeCNF1 sample (Figure S6), while Fe is clearly absent in the CNF spectrum (Figure S6). The N 1s peak of CNF can be fitted to pyridinic N, pyrrolic N, graphitic N, and oxidized nitrogen at 398.0, 399.2, 400.6, and 403.9 eV, respectively (Figure S6). When compared with the N 1s of CNF spectrum, the FeCNF1 spectrum showed similar peaks to CNF (Figure S7), but an additional peak at 398.8 eV was also detected. We thus assigned the speak at 398.8 eV to Fe–N. A recent report suggested that Fe–N sites can serve as active sites for inducing the Fenton reaction.[25] The relative abundance of Fe–N sites in FeCNF1 suggests a moderate to high reactivity in terms of the Fenton reaction. To confirm our hypothesis, we utilized a method for identifying Fe species in a high-resolution XPS spectrum of Fe 2p.[26] It is evident that CNF does not contain Fe (Figure S6). In contrast, characteristic Fe 2p3/2 and Fe 2p1/2 peaks at 711.7 and 724.9 eV confirmed that both Fe2+ and Fe3+ are present in FeCNF1 (Figure S7). The O 1s spectrum of CNF consisted of three multiplets including O=C, O–C, and oxygen atoms adsorbed onto the CNF surface at 531.0, 532.3, and 535.1 eV, respectively. In contrast, the additional peak at 530.2 eV from FeCNF1 can be attributed to presence of lattice oxygen in Fe3O4. Table S1 shows that the oxygen levels are significantly higher in FeCNF1 than in CNF, thus providing additional conformation for the presence of magnetite nanoparticles, Fe3O4-NP.

Electrolytic Production of H2O2 on CNF

The electrochemical activity of the CNF substrate for the oxygen reduction reaction (ORR) was determined in part by using cyclic voltammetry (CV) to follow the oxygen reduction versus potential. As shown in Figure , the CNF electrode shows a pronounced reduction peak in an O2-saturated electrolyte solution (0.1 M Na2SO4), while the CV curve for CNF in an Ar-saturated 0.1 M Na2SO4 is featureless. The onset potential for CNF at pH 4, 7, and 10 was 0.314, 0.570, and 0.766 V vs RHE, respectively. The solution pH did not detectably change after the ORR reaction. In addition, the ORR peak of CNF electrode at pH 4 had the highest current density at 1.055 mA/cm2, while the current densities at pH 10 (0.5807 mA/cm2) and pH 7 (0.5775 mA/cm2) are similar. These results suggest that the CNF electrode can electrochemically catalyze ORR over a wide range of pH.

Figure 3

CV curves of CNF substrate in O2- and Ar-saturated electrolyte solution (0.1 M Na2SO4) at pH 4 (A), pH 7 (B), and pH 10 (C) with a scan rate of 10 mV/s.

It has been demonstrated that the selectivity of the oxygen reduction reactions (ORR) can be altered by controlling the calcination temperature of the carbon precursor.[27] Literature reports suggest that the ORR current efficiency on carbon-based electrodes varies between 60 and 95%.[27−29] In addition, simulations and experimental results indicate that graphite and nitrogen-doped carbon materials are highly selective for the two-electron oxygen reduction reaction to form H2O2, with selectivity above 90%.[27,30] To determine the dominant reaction pathway for the ORR on the CNF electrode, we quantified the amount of H2O2 produced in O2-saturated electrolyte solutions. The H2O2 generation rate was found to be the highest at pH 7 (39.65 mmol/h/g), followed by pH 4 (32.35 mmol/h/g) and pH 10 (18.07 mmol/h/g) (Figure ). Even though CNF materials are considered to be nonporous with low reactive surface areas, the reported H2O2 production rates are greater than those obtained using porous carbon electrodes under similar conditions.[31] DFT calculations and experimental evidence suggest that carbon nitride and oxidized carbon favor the 2e– transfer pathway for O2 reduction leading to H2O2 formation.[32,33] Our high-resolution XPS data for CNF indicate the presence of both oxygen functional groups and carbon nitride on the CNF surface (Figure S6).

Figure 4

Concentration of H2O2 produced by CNF at −0.345 V at pH 4, 7, and 10.

Electro-Fenton Degradation of Carbamazepine

Carbamazepine was chosen as a probe compound to demonstrate the feasibility of utilizing the heterogeneous electro-Fenton reaction on Fe3O4-NP@CNF for the treatment of dilute pharmaceutical wastes in water. To distinguish between loss by sorption alone and hydroxyl radical oxidation of carbamazepine, the concentration versus time profiles for electrosorption on CNF and oxidation via H2O2 formation are compared in Figure . The initial carbamazepine concentration was 1 ppm in 50 mL electrolytes. All processes follow pseudo-first-order kinetics, but the removal rate of each differs significantly. Complete removal of carbamazepine was achieved within 30 min for the electro-Fenton process using the Fe3O4-NP@CNF electrode. On the contrary, carbamazepine removal efficiency was only 21.1 and 8.8% for CNF electrosorption and H2O2 oxidation, respectively. Degradation kinetics are shown in Figure S8, where the electro-Fenton process has a much larger first-order rate constant compared to loss by sorption or molecular oxidation by hydrogen peroxide.

Figure 5

Carbamazepine removal by electro-Fenton (FeCNF0.1, −0.345 V), electrosorption (FeCNF0.1, −0.345 V), electrosorption (CNF, −0.345 V), and H2O2 degradation (200 ppm). Electrosorption and H2O2 degradation are performed in Ar saturated electrolyte at pH 7.

Fe3O4-NP@CNF is a bifunctional electrochemical catalyst. The electro-Fenton reaction is catalyzed by the Fe3O4-NPs, while H2O2 is generated in situ by O2 reduction catalyzed by the CNF substrate. Therefore, the effect of Fe3O4-NP loading level on the CNF was investigated to optimize ·OH radical production. The synthesized series of Fe3O4-NP@CNFs at loading levels of 0.05–1.00 wt % were compared for carbamazepine removal at pH 7 as shown in Figure . The carbamazepine removal efficiency increased as Fe3O4-NP loading level was reduced from 1.0 to 0.1 wt %, while further reduction in the Fe3O4-NP loading resulted in a reduction in the carbamazepine removal efficiency. All of the Fe3O4-NP@CNF electro-Fenton followed pseudo-first-order kinetics (Figure S9). The kinetic constants for carbamazepine degradation were determined to be 1.79, 6.85, 2.43, 1.35, and 0.52 h–1 for FeCNF0.05, FeCNF0.1, FeCNF0.3, FeCNF0.5, and FeCNF1, respectively (Table S3). As shown in Figure S10, the carbamazepine removal efficiency is also a function of the solution pH. The kinetic constants for carbamazepine removal are 4.78, 6.85, and 3.30 h–1 for FeCNF0.1 electrode at pH 4, 7, and 10, respectively (Figure S11 and Table S3). Our data suggest that the electro-Fenton process catalyzed using the bifunctional Fe3O4-NP@CNF electrode can be performed over a wide range of solution pH. The moderately enhanced removal efficiency at pH 7 can be attributed to an increased H2O2 production at circumneutral pH (Figure ), which in turn increases ·OH production. The CVs of the CNF substrate show that the current density of the CNF electrode is directly proportional to the applied potential. This, in turn, should impact the H2O2 production rate (Figure ). Figure S12 shows that in fact the carbamazepine removal efficiency was significantly improved as the applied potential was increased from −0.145 to −0.545 V. The kinetic constants for carbamazepine removal are 4.81, 6.85, and 9.00 h–1 for FeCNF0.1 electrode at −0.145, −0.345, and −0.545 V, respectively (Figure S13 and Table S3). Figure shows that complete mineralization of carbamazepine is achieved within 3 h of the electro-Fenton reaction as catalyzed by FeCNF0.1 at pH 7. In contrast, H2O2 oxidation only removes 8.8% of carbamazepine under similar conditions.

Figure 6

Effect of Fe3O4 loading level on electro-Fenton removal efficiency of carbamazepine at pH 7.

Figure 7

TOC removal efficiency of electro-Fenton removal of carbamazepine (FeCNF0.1, −0.345 V, pH 7).

The stability of the electrocatalysts is important for practical applications. In this regard, we have examined the electro-Fenton degradation of carbamazepine using the recycled FeCNF0.1 electrode. Our results (Figure S15) show that the catalytic activity (>90%) is retained even after repeated usage.

Mechanism of Carbamazepine Mineralization

The high degree of carbamazepine mineralization efficiency obtained using a Fe3O4-NP@CNF cathode (Figure ) is clearly attributed to the production of ·OH via electro-Fenton reaction. CVs of Fe3O4-NP@CNF electrodes suggest the onset potential of oxygen reduction is not significantly affected by the Fe loading concentration (Figure S16).

To explain the observed differences in the carbamazepine degradation efficiency as a function of the loading level of magnetite on the Fe3O4-NP@CNF electrodes, the amount of ·OH generated for each loading level was probed using terephthalic acid (TPA) as a hydroxyl radical trap. TPA has been widely used as a chemical probe to detect ·OH, since the rate constant between TPA and ·OH (4.4 × 109 M–1 s–1) is orders of magnitude higher than those of other reactive oxygen species that might be involved in this study, such as H2O2 (3.0 × 107 M–1 s–1) or singlet oxygen (5.0 × 104 M–1 s–1).[34−36] In addition, TPA (pKa1 = 3.5 and pKa2 = 4.4), which predominately exists in doubly deprotonated carboxylate form at pH 7, is electrostatically repelled from the negatively charged Fe3O4@CNF cathode, and thus TPA is an excellent radical trap for probing solution phase production of ·OH. Because TPA reacts with ·OH to produce 2-hydroxyterephthalic acid (HTA), the steady-state concentration of ·OH can be determined by monitoring the TPA concentration in the electrolyte. Figure S14 shows the that amount of TPA adsorbed onto the platinum needle anode was negligible. This is due to the low surface area of the platinum needle anode in addition to the formation of oxygen gas bubbles generated via water oxidation on the Pt surface. As shown in Figure , the Fe3O4-NP loading level impact on ·OH production has a similar trend in activity as observed for carbamazepine degradation (Figure ). Again, the FeCNF0.1 composite electrode produced the most ·OH, followed by FeCNF0.05 > FeCNF0.5 > FeCNF1. In energy conversion, there is an interest in reducing the size of the electrocatalyst to maximize its utilization efficiency by providing the greatest number of accessible active sites.[37] Indeed, the Fenton reaction on crystalline particles of Fe3O4, which are nonporous, normally prevents nonsurface accessible Fe(II) to participate in the reaction. Thus, doping of the CNF substrate with a minimal amount of Fe3O4-NP decreases the formation of bulk-phase crystalline particles. This is evident from the PXRD spectrum of Fe3O4-NP@CNF with low Fe content (Figure S4). Therefore, Fe(II) from FeCNF0.1 embedded in the graphitic molecular support has approximately identical local coordination environments, which, in turn, maximize the active sites available for the electro-Fenton reaction. In the case of Fe3O4-NP@CNF with a lower Fe content (FeCNF0.05), in situ generated H2O2 is found in excess due to an insufficient amount of Fe(II) to generate the hydroxyl radical.

Figure 8

Effect of Fe3O4 loading level on ·OH steady state concentration (−0.345 V, pH 7).

The electro-Fenton promoted degradation of carbamazepine is similar to the ·OH oxidation of carbamazepine.[38] On the basis of the intermediates detected in herein (Figure S17) and on other reports of carbamazepine degradation,[38,39] we propose a mechanistic pathway leading to carbamazepine mineralization by the electro-Fenton process as shown in Figure S18.

Energy Efficiency for Carbamazepine Degradation

During the Fe3O4-NP@CNF catalyzed electro-Fenton reactions, the applied potential was −0.345 V with a corresponding current density of 3.46 mA/cm2 at pH 7; this value can be compared to a previously reported current density of 190 mA/cm2 at a similar applied potential during electrochemical degradation of carbamazepine.[40] Given these conditions, the energy consumption required by the electro-Fenton process to degrade carbamazepine is 0.239 kW·h/g carbamazepine. When compared with other advanced oxidation processes such as electrolysis with a boron-doped diamond electrode (4.0 kW·h/g carbamazepine), electrolysis with PbO2 (4.4 kW·h/g carbamazepine), photodegradation using UV/peroxymonosulfate (933 kW·h/g carbamazepine), and photodegradation using UV/H2O2 (396 kW·h/g carbamazepine), electro-Fenton oxidation using Fe3O4-NP@CNF cathodes requires significantly less energy.[41,42] It should be noted that the previously reported results were obtained under different operational parameters, which could have an impact on apparent power consumption levels and treatment efficiencies. An in-depth study is necessary to provide a comprehensive understanding of the energy requirement of these alternative AOPs.

The unique characteristic of Fe3O4-NP@CNF arrays integrates multifunctional catalysts into a single stable material to maximize the catalytic efficiency using minimal amount of Fe3O4 as an iron(II)-containing catalyst. A loading level of 0.36 wt % Fe or 0.0012 mg Fe/cm2 of electrode surface area (equivalent to 0.48 mg Fe/L under the applied experimental conditions) is needed for Fe3O4-NP@CNF cathodes to maximize the ·OH production from H2O2. In contrast, 56 mg Fe/L of an Fe(II) electrolyte is required for the electro-Fenton reaction to fully convert H2O2 produced by MOFs derived carbon electrode into ·OH under similar conditions.[11] Our current results indicate that Fe3O4-NP@CNF can provide a cost-effective material to catalyze electro-Fenton reactions.

In summary, efficient degradation of carbamazepine has been achieved using the electro-Fenton process based on a bifunctional Fe3O4-NP@CNF electrocatalyst. In this novel electro-Fenton system, uniformly dispersed Fe3O4-NPs, which are bonded to the CNF cathode substrate, can obtain a high electro-Fenton efficiency with minimal Fe mass loading level (0.48 mg Fe/L). Given the relatively low-energy consumption of this system combined with its wide pH range, the Fe3O4-NP@CNF composite electrode potentially provides a cost-effective treatment method to achieve the total mineralization of carbamazepine in a relatively short period of time. Furthermore, this electro-Fenton system can be applied for the removal of other pharmaceutical and personal care product from wastewater and drinking water.