Impact of thermal modification of carbon felt on the performance of oxygen reduction reaction and mineralisation of dye in on-line electro fenton system

Carbon felt (CF) is frequently used as an electrode in various processes of oxygen reduction reactions (ORR) and water treatment. Therefore, it is critical to increasing the performance of this material in ORR. In this study, the effect of the thermal modification, which is a simple and low-cost process, was examined. For testing the hydrogen peroxide generation efficiencies and the mineralisation of Acid Orange 7 (AO7) dye, the modified CF electrodes (under 0–900°C and N 2 atmosphere for 1 hour) were used as cathodes with Ti/RuO 2 -IrO 2 -TiO 2 as anode and saturated calomel electrode (SCE) as a reference electrode. In addition, to find the optimum Electro-Fenton (EF) process parameters, the potential of the cathode (0.8–3.0 V), pH (3.0–9.0), Fe 2+ concentration (0.2–1.0 mM) and electrolyte (sodium sulphate) concentration (10– 500 mM) were investigated. The highest efficiencies in dye and TOC removal were achieved at 0.8 V, 50 mM of electrolyte concentration with 0.3 mM of Fe 2+ by using CF modified at 900°C as 96.26% at 10 min and 91.83% at 120 min, respectively. Besides, the modification process was increased the electroactive surface area (41.70%) and crystallite size (22.69%) of the electrode. Thus, more effective treatment was achieved. Thanks to the modification, the electrical energy and electrochemical energy saving are approximately 23.87% and 70.22%, respectively. On the other hand, a kinetic study was conducted with a large number of data coming from the on-line analysis system, which is different from most systems in the literature. At the end of the kinetic study, it was revealed that EF systems could not be explained by a single kinetic equation. It was found that the EF system has two different rate equations (first order and pseudo first order) with high correlation coefficients (R 2 = 0.9965 and 9920)


Introduction
Due to its impressive specifications such as relatively low cost and high mechanical stability, porosity, electrical conductivity, and surface area, Carbon felt (CF) is a substance primarily used in processes such as energy [1], oxygen reduction reactions (ORR), and wastewater treatment as an electrode or supporting materials in electrodes [2].For wastewater treatment, CF electrodes were utilised in various electrochemical treatment processes such as the electro-reduction of heavy metals (Cr +6 ) [3], electron-ion exchanger (fluoride) [4], electrosorption desalination [4], and electro-Fenton [5].Inside these processes, one of the most exciting areas of CF is EF process.EF process is inspired by the classical Fenton process (CFP), which depends on the formation of hydroxyl radicals ( • OH) by the presence of H 2 O 2 and catalyser Fe 2+ at acidic conditions.EF has been developed as a solution to the transport and storage problems of peroxide in the CFP.
EF is based on the formation of hydrogen peroxide (H 2 O 2 ) at the cathode by an oxygen reduction reaction (ORR).(Equation 1.) and the formation of • OH, one of the most powerful oxidising agents in the presence of iron as catalyst (Equation 2) [6][7][8].The regeneration of Fe 2+ added externally on the cathode (Equation 3) is another advantage of EF over the CF process [9], and this advantage causes less usage of iron during the treatment process.
Various modification methods (chemical and/or physical (thermal)) have been adopted to increase the electrodes' efficiency in the CF process.However, chemical modification or a mixture of chemical and physical processes poses problems such as high material criteria, a high degree of manual procedure, and the availability of a robust chemical activator.Thermal improvement is an easy way to modify the CF samples where the samples are annealed in the furnace by gas flow, including gases such as O 2 and/or N 2 [10][11][12][13][14].However, there is not yet a detailed study in the literature on the efficiency of thermal modification CF in ORR performance and EF process.The effect of the temperature (0-900°C) under N 2 atmosphere on the performance of ORR and electrochemical properties of CF electrodes in an EF system was investigated in this study.The AO7 was selected as a model pollutant, and the effects of the cathodic potential (0.8-3 V), pH (3)(4)(5)(6)(7)(8)(9), Fe 2+ concentration (0.2-1 mM), and electrolyte concentration (10-500 mM) on the EF process were investigated.Furthermore, an on-line EF system was used for the investigation of the removal kinetics of EF.

Thermal modification of CF cathodes
The CF (8.6 cm x 21.5 cm; 12.7 g (±0.05; the total area of 185 cm 2 ) was washed in acetone (1 h), and deionised water several times, then kept in an oven (100°C for 3 hours) and labelled as raw carbon felt (RCF).RCF was modified in a horizontal cylindrical furnace (Protherm, STF650-3Z, Turkey).For this, first, a purge process was performed with N 2 for 1 h.After the purge process, while the N 2 flow continued (400 cm 3 /min), the temperature of the furnace was increased to the desired value (400-900°C) with an increased rate of 5°C/min, and the electrode was kept in the oven for 1 hour at the final temperature.The cooling process of the furnace was carried out under N 2 flow, and it was named modified carbon felt (MCF).

Characterisation of CF cathodes
The structural characterisations and surface morphologies of CF cathodes were monitored by a Scanning Electron Microscopy (SEM, Hitachi, Germany), an optical tensiometer (KSV-CAM 200, Finland).The contact angle and wettability measurements under air were done with water drops with a volume of 5 μl using a computer-controlled motorised syringe.An X-ray diffractometer (XRD, Rigaku model, with Cu Kα (λ = 1.54059) with radiation flux at a scanning rate of 5°/min in the 2θ range of 10-90°, accelerating voltage of 40 kV and applied current of 40 mA).The average crystallite size was calculated from the XRD data using Scherre equation (Equation 4) [14,15].
Where L is crystallite size (Å), α is a constant equal to 0.94, β is the full width at half maximum in radians, and λ (Å) is the wavelength of the X-rays.
Furthermore, the spectra of the CF electrodes were recorded by a Fourier Transform Infrared Spectroscopy (FT-IR, PerkinElmer, Spectrum 100, USA).For C, H, N, S (%) determination, an elemental analyser (LECO-CHNS 628, USA) was used with the method of dynamic flash combustion (1000-1100°C) of the samples.
Electrochemical techniques were used at room temperature in a conventional threeelectrode cell with a Potentiostat (Gamry Interface 1000, England).In these techniques, CF electrodes (1cm 2 ), Ti/RuO 2 -IrO 2 -TiO 2 plate (3 × 4 cm), and a saturated calomel electrode (SCE) were used as working, counter, and reference electrodes, respectively.A 50-mM Na 2 SO 4 solution (pH 3) was used in LSV and EIS between 0.1 and −2.2 V at a scan rate of 50 mV/ s in an N 2 atmosphere with a flow rate of 16 L/h.A 10-mM K 3 [Fe(CN) 6 ], and 1.0-M KCl solution (pH 3) were used in CV between −0.4 and 0.8 V at a scan rate of 50 mV/s [16].
Besides, the Randles-Sevcik equation (Equation 5) was used for the calculation of the electroactive surface area of the CF samples (10 mM of K 3 Fe(CN) 6 as a test solution and 0.1 M of KCl as the supporting electrolyte at different scan rates) [2,14,17,18].
In the Randles-Sevcik equation, n, A, D, C, and γ are the number of electrons transferred, the electrode area (cm 2 ), the diffusion coefficient of the molecule (K 3 Fe(CN) 6 ) in solution (7.6x10 −6 cm 2 /s), the concentration of K 3 Fe(CN) 6 (mol/cm 3 ), and a scan rate of the potential perturbation (V/s), respectively.

EF system and procedure
The on-line EF experimental set-up is shown in Figure S1.The EF system was designed similarly to the previous study [19].CF cathode (185 cm 2 ), a Ti/RuO 2 -IrO 2 -TiO 2 plate anode (12 cm 2 ), and an SCE were placed into a 600-mL undivided cylindrical glass cell.The cell was filled with the 400 mL aqueous solution with 35 mg/L (0.1 mM) of AO7 at 25°C and mixed with a magnetic stirring (700 rpm) (Fig. S1).Na 2 SO 4 and FeSO 4 .7H 2 O were added to the cell, and solution pH was adjusted.A potentiostat (Gamry Interface 1000, England) was connected with electrodes, and oxygen (O 2 ) was supplied by bubbling dry air (16 L/h) into the cell.The cell was connected to a UV-vis spectrophotometer to obtain a large number of reliable kinetic data through a pump with a flow of 30 (±0.1) cm 3 /min (60 rpm), and the absorbance values of the model pollutant were monitored at its specific wavelength of 483 nm.

Analytical procedures
Total Organic Carbon (TOC) analyses were realised by combustion method with Shimadzu TOC-analyser (Japan).All TOC measurements were repeated twice, and average data were reported if the test error was below 0.3%.The iodide method [20] was used for H 2 O 2 production by UV-vis spectrophotometer (DR 6000 model, Hance-Lange GmbH, Germany) at 351 nm.H 2 O 2 solutions were prepared in a concentration range of 0-10 mg/L, and a calibration curve at 351 nm was developed for H 2 O 2 with high regression constants (≥ 0.9979).
The current efficiency (CE) for H 2 O 2 production, the consumption of electrical energy (CEE, Equation 7, kWh/m 3 ), electrochemical energy consumption (EEC, Equation 8., kWh/ m 3 g TOC), and the mineralisation current efficiency (MCE, Equation 9) were defined as follows [21]: In these equations, z is the number of electrons transferred for ORR (2e − ), F is the Faraday constant (96.485C/mol), C H2O2 , is the H 2 O 2 concentration (mg/L), V is the solution volume (0.4 L), M H 2 O 2 is the molar mass of H 2 O 2 (34.01 g/mol), I is the applied current (A), t is the operating time (2 h), U is the cell voltage (Volt), Δ TOC ð Þ t represents the decrease of TOC at t time [21], 4.32 × 10 7 represents the conversion factor used to homogenise the units (3600 s/h × 12,000 mg C/mol), m is the number of carbon atoms ( 16) [22] and n, which was 84, represents the number of electrons consumed during mineralisation reaction of model pollutant given in Equation 10.

Structural characterisation of CF samples
The phase structure of the CF samples was characterised by using XRD.Carbon (C) peaks of the CF samples were similar and appeared at 2θ around 25°, which corresponded to the (002) peak and (100) peaks of the hexagonal carbon structure.From the XRD spectra given in Figure 1, it was found that CF samples contained approximately ≥ 99% carbon according to the carbon peak locations (2θ: 26.603).The peaks demonstrated the graphitisation occurred during the modification process and that the graphitisation of MCF samples was higher than that of RCF due to its wide peak.The wide (002) peak and the tiny peak of (100) could be attributed to the disorder graphite layers formed during the modification process.Besides, the crystalline size of the CF samples was calculated using the Scherrer equation (Equation 4).For MCF samples (MCF400, MCF600, MCF900) and RCF, the crystalline size values were 15.4 Å, 15.7 Å, 17.3 Å and 14.1 Å, respectively.This increase in crystalline size is temperature-dependent.Therefore, the selection of an appropriate temperature in a modification process is important to obtain suitable properties.
The surface morphology of CF samples before and after the modification process observed by SEM are shown in Figure 2. As shown in Figure 2(d), the microstructure of the MCF900 sample was characterised by the roughest surface among the samples, whereas the MCF600 sample exhibited moderate surface roughness.Besides, the contact angles of CF samples were found between 150° and 110°.Among the CF samples, the lowest contact angle value was obtained by the MCF900 sample (110°) as opposed to the 150° contact angle on the RCF (Figure 2).Compared to other samples, the hydrophilic surface of MCF900 is more effective in producing peroxide [19].However, it was observed that full hydrophilicity could not be achieved according to the contact angle measurements in the study.
FT-IR analysis was performed to determine the functional groups of CF samples.Figure 3 shows the FT-IR spectra obtained from the CF before and after modification.In the spectra, 3600 and 3200 cm −1 vibration frequency show the hydroxyl (OH) group, and the peak at 2329 cm −1 is related to the carbon triple bond.The C = O or C = N bonds give peaks at 1750-1640 cm −1 .Besides, the peaks at around 1403 cm −1 and 1565 cm −1 are due to the CH 2 and N-H bending vibration frequency.At the end of the modification, the spectra at different temperatures portrayed similar patterns, but the intensity of the absorption bands changed depending on the modification temperature.The elemental analysis table (Table 1) showed elements of CF samples.Elemental analysis results show that RCF does not contain oxygen-containing functional groups (C = O etc.), but carbon atoms are oxygenised as the temperature increases with thermal modification.On the other hand, if Table 1 and FT-IR spectra (Figure 3) are evaluated together, the stretching vibration seen at 1750-1640 cm −1 in FT-IR spectra may be C = O bond, not C = N.The modification temperatures were insufficient to break the existing bonds of the carbon, and therefore C = N bonds were not formed as a result of the modification.However, it has been reported in some studies that the excess amount of oxygen has a positive effect on ORR reactions and creates an active zone on the surface for H 2 O 2 production [23,24].

Electrochemical Characterisation of CF Samples
Electrochemical techniques (CV, LSV, EIS) were conducted for RCF and MCF samples.The test graphics are shown in Figure 4.
The CV study was conducted to identify the electroactive surface characterisation of RCF and MCF samples.The CV graphics are displayed in Figure 4(a).The reduction degrees of RCF and MCF samples were observed as a function of temperature.The peak current density observed in the CV graphics significantly increased with an increase in temperature from 0 to 900°C.Besides, the electroactive surface areas of the CF samples were calculated according to the Randles-Sevcik formula (Equation 5) from the slope of the Ip vs. γ ½ graphic.The area values of RCF, MCF400, MCF600, MCF900 were 1047.8cm 2 /g, 1219.0 cm 2 /g, 1385.1 cm 2 /g, and 1484.7 cm 2 /g respectively.The MCF900 with the highest electroactive surface area value indicated 41.69% more the area than that of the RCF.
LSV provides to examine the ORR activity of CF samples for H 2 O 2 production.The LSV graphic of the CF samples is shown in Figure 4(b).According to Figure 4(b), MCF samples more sufficiently represent the electrode ORR activity for the H 2 O 2 production than the RCF does.A higher net current density means a higher ORR activity.In this LSV study, the MCF900 electrode received the highest current density between −0.1 V and −2.2 V. EIS is an efficient method to investigate the interface properties and determine the electrical conductivity of CF samples [25].EIS tests were measured as a function of frequency.Figure 4(c) shows the impedance spectra for the CF samples.The spectra were recorded at frequencies from 50 kHz to 10 MHz.Accordingly, the highfrequency region reflects the electrolyte properties of the CF samples, while the lowfrequency region reflects the surface properties.The impedances values of RCF, MCF400, MCF600, MCF900 in low frequency region (9.9 Hz) were 8.62, 8.0, 6.91 and 4.61 ohm respectively.The surface impedance of MCF900 was found low than that of the other CF samples in the applied frequency range.This lower surface impedance value of MCF900 demonstrated that the electrosorption efficiency is higher than that of the other CF samples.

RCF studies
The effect of pH on H 2 O 2 production, TOC, and dye removal efficiency was investigated between 3 and 9 using RCF.It is clear from Figure 5 that RCF was quite effective for H 2 O 2 production, TOC, and dye removal efficiency over the pH range 3-9.The study was carried out for 120 min with 50 mM Na 2 SO 4 , at 2 V cathodic potential.However, dye removal measurement was performed for the first 10 minutes, unlike TOC measurements.When the pH increased, the TOC and dye removal efficiency decreased.Besides, the H 2 O 2 production capacity of the RCF was pH-dependent.RCF produced maximum peroxide (37.21 mg/L) at pH 3.0 in 120 minutes.However, RCF produced tiny H 2 O 2 amounts at pH 9.0 (10.86 mg/L).Therefore, a pH value of 3.0 was found as effective compatible with the literature [26][27][28][29].For this reason, the pH was kept constant at 3 in subsequent studies.
One of the most critical parameters in EF is the cathodic potential; according to Faraday laws, increased potential causes more electron transfer, increasing current and operational costs.So, the change of the cathodic potential was observed between 0.8 V and 3 V using RCF.The study was carried out for 120 min.in 50 mM Na 2 SO 4 solution (pH 3).The effect of the cathodic potential on H 2 O 2 production, TOC, and dye removal efficiency can be seen in Figure 6.This figure clearly shows that cathodic potential plays a vital role in EF systems.
The raising cathodic potential (U) and current (I) raise the CEE and the CE for H 2 O 2 production.The calculated CEE (kWh/m 3 ) and CE (%) according to Equations 7 and 6 are given in Table 2.According to Table 2, the minimum CEE value (0.4 kWh/m 3 ) and maximum CE value (78.6%) and were obtained at the cathodic potential of 0.8 V.The iron concentration is another important parameter in the EF process because it provides the formation of the • OH from H 2 O 2 catalytically (Equation 2) [30][31][32].Generally, up to a level, increasing catalyst concentration causes a positive effect on the classic Fenton process; however, the overdosing of catalyser causes consumptions of the hydroxyl radicals because of some parasitic reactions (Equation 11-13) [27,33,34].
Therefore, the impact of iron amount on removal efficiencies of TOC and dye was investigated by dosing Fe 2+ between 0.2 mM and 1.0 mM using RCF (Figure 7) 0.1 mM AO7 solution and 50 mM Na 2 SO 4 at pH 3 for 120 min.at 0.8 V.
This study obtained a fluctuation in dye and TOC removal efficiencies with varying initial Fe 2+ concentrations (Figure 7) regardless of a trend, demonstrating the importance of determining the optimum Fe 2+ concentration in the EF process.In the Electro-Fenton process, increasing Fe concentration increases the treatment efficiency to a specific point due to the catalytic effect in the first step.The treatment efficiencies may decrease in the second step due to parasitic reactions (Equations 11-13).In the third step, the increased Fe concentration may cause  coagulation and increase treatment efficiencies.These three steps may be the reason for obtained fluctuation in dye and TOC removal efficiencies.It is evident in Figures 6  and 7 that the optimum Fe 2+ concentration and cathodic potentials were 0.3 mM and 0.8 V, respectively.

MCF studies
Effect of thermal modification of CF samples on H 2 O 2 production, dye, and TOC efficiency was investigated in 50 mM Na 2 SO 4 , 0.1 mM dye, 0.3 mM Fe 2+ concentration (pH 3.0) at 0.8 V cathodic potential.However, the study was carried out for 120 min without dye and Fe +2 in H 2 O 2 production experiments.The effect of modification temperature in the EF system can be seen in Figure 8.This figure shows clearly that modification temperature played an essential role in H 2 O 2 production, dye, and TOC removal efficiency.Considering Figure 8, the optimum modification temperature was selected as 900°C due to high H 2 O 2 production and dye and TOC efficiency.The MCF900 electrode provides the highest treatment efficiency due to the superior features mentioned in the characterisation section.These superior features are the highest oxygen content that creates active zones, the lowest contact angle, the roughest surface, the low surface impedance that provides the highest electrosorption efficiency, and the highest electroactive surface.Effect of Fe 2+ (0.2-1 mM), electrolyte concentration (10-500 mM) and cathodic potential (0.8-3 V on dye and TOC efficiency was investigated in 0.1 mM dye solution (pH 3.0) by using MCF900 electrode.As seen in Figures 9, 0.3 mM, 0.8 V, and 50 mM were found to be the most effective Fe 2+ concentration, cathodic potential, and electrolyte concentration values, respectively.
The treatment efficiency obtained with the Voltage and Fe 2+ concentrations changing in the EF process shows a ripple.The reason for this ripple is discussed in detail in Section 3.3.The concentrations of electrolyte are a critical parameter because it affects both operational cost and efficiency in electrochemical treatment processes.However, the correlation between the amount of electrolyte and the treatment efficiency of the EF process has been reported differently in many studies.Low electrolyte concentrations do not provide adequate conductivity, and this increases the operational cost.The high electrolyte concentrations may cause electrode corrosion in the EF and reduce the reactivity of Fenton species by SO 2À 4 [35].At the end of RCF and MCF samples studies, the optimum experiment conditions for the kinetic investigations were found as 3 of pH, 0.8 V of potential, 0.3 mM of Fe 2+ concentration, and 50 mM of Na 2 SO 4 concentration with MCF900 electrodes.

Kinetics study in the EF system
The mineralisation curve obtained under optimum experimental conditions is shown in Figure 10.For the first 2.3 min.(stage I), the AO7 mineralisation curve is different from that of the remaining mineralisation time (stage II).Therefore, we interpreted the data using three kinetic models (Table S1) [33,[35][36][37].In Table S1, C 0 is the initial concentration of the AO7 solution (0.1 mM or 35 mg/L) after the adsorption equilibrium, C t is the AO7 concentration at a given reaction time t and k is the reaction rate constant.According to Table S1, the rate constant and correlation coefficient values (R 2 ) were obtained for each kinetical model (Table S2).According to the on-line analysis results, after the first 30 seconds, the increasing reaction rate fit the first-order kinetic equation up to 2.3 minutes (Stage I).After 2.3 min (Stage II), the reaction rate started to fit the pseudofirst-order kinetic.These results showed that the EF processes could not be explained by a single kinetic model.

Comparison of MCF and RCF cathodes
Two new experiments were carried out under optimum experiment conditions by using MCF900 and RCF cathode to see the effect of modification on the EF system (Table 3).The experiments performed under the same conditions were compared in terms of CEE (kWh/ m 3 ), EEC (kWh/m 3 g TOC), MCE (%), operating cost, and cathode lifetime.One of the most crucial cost items in the EF process is electrical energy.Another is electrochemical energy.Therefore, the energy efficiency of RCF and MCF900 at 0.8 V cathodic potential was investigated.The results of the investigation are shown in Table 3.The consumption of electrical energy (CEE) and electrochemical energy consumption (EEC) savings are approximately 23.87% and 70.22%, respectively, in MCF900 compared to RCF.Additionally, mineralisation current efficiency (MCE) is 15.41% higher in MCF900 than in RCF.On the other hand, when the cost ($/m 3 ) of treatment is compared between MCF900 and RCF, it is found that MCF900 treats at 30% more affordable cost.
The long-term stability of CF is critical for the practical use of the process.After 40 hours, the dye and TOC removal efficiency (%) of MCF900 were decreased 3.17% and 14.21%, while that of RCF were decreased 21.93% and 3.74%, respectively.The lifetime results of the MCF900 and RCF are shown in Fig S2 show no significant reduction in the first 40 hours, indicating the high potential for using the modified electrode [14,17].
In an EF experimental setup designed to investigate the mineralisation kinetics of AO7, unlike previous kinetic studies, the dye concentration was monitored on-line to more accurately calculate the rate constant.The kinetic analysis showed that the mineralisation kinetics consists of two stages, and this stage can describe by the first-order kinetic model and pseudo-first-order kinetic, following each other.
In order to compare MCF900 with the RCF cathode in terms of energy efficiency, operating cost, and long-term stability, the experiments were performed under the same experimental conditions.The results showed that studying with the MCF900 cathode increases the treatment efficiency while reducing the treatment cost.In addition, according to the long-stability experiments, it was found that the MCF900 cathode remained more stable than the RCF after 40 hours of operation.

Figure 1 .
Figure 1.XRD spectra of CF samples and carbon.

Figure 5 .
Figure 5.The effect of pH on H 2 O 2 production, dye, and TOC removal efficiency with RCF.

Figure 6 .
Figure 6.The effect of cathodic potential on dye and TOC removal efficiency with RCF.

Figure 7 .
Figure 7.The effect of Fe 2+ concentration on dye and TOC removal efficiency with RCF.

Figure 8 .
Figure 8.The effect of modification temperature on H 2 O 2 production, dye, and TOC removal efficiency (0°C indicates the RCF cathode).

Figure 9 .
Figure 9.The effect of Fe 2+ , electrolyte concentration, and cathodic potential on dye and TOC removal efficiency with MCF900.

Table 1 .
Elemental analysis results of CF samples.

Table 2 .
CE and CEE for H 2 O 2 production obtained in 120 min.

Table 3 .
The energy efficiency of RCF and MCF900 under the optimum process conditions.Electric energy cost is calculated according to the Turkish Market. *