Article pubs.acs.org/est
Carbon Nanotube Membrane Stack for Flow-through Sequential Regenerative Electro-Fenton Guandao Gao,*,†,‡ Qiaoying Zhang,‡ Zhenwei Hao,† and Chad D. Vecitis‡ †
Key Laboratory of Pollution Processes and Environmental Criteria (Ministry of Education), Tianjin Key Laboratory of Environmental Remediation and Pollution Control, College of Environmental Science and Engineering, Nankai University, Tianjin 300071, China ‡ School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138, United States S Supporting Information *
ABSTRACT: Electro-Fenton is a promising advanced oxidation process for water treatment consisting a series redox reactions. Here, we design and examine an electrochemical filter for sequential electro-Fenton reactions to optimize the treatment process. The carbon nanotube (CNT) membrane stack (thickness ∼200 μm) used here consisted of 1) a CNT network cathode for O2 reduction to H2O2, 2) a CNT-COOFe2+ cathode to chemical reduction H2O2 to •OH and HO− and to regenerate Fe2+ in situ, 3) a porous PVDF or PTFE insulating separator, and 4) a CNT filter anode for remaining intermediate oxidation intermediates. The sequential electro-Fenton was compared to individual electrochemical and Fenton process using oxalate, a persistent organic, as a target molecule. Synergism is observed during the sequential electro-Fenton process. For example, when [DO]in = 38 ± 1 mg L−1, J = 1.6 mL min−1, neutral pH, and Ecell = 2.89 V, the sequential electro-Fenton oxidation rate was 206.8 ± 6.3 mgC m−2 h−1, which is 4-fold greater than the sum of the individual electrochemistry (16.4 ± 3.2 mgC m−2 h−1) and Fenton (33.3 ± 1.3 mgC m−2 h−1) reaction fluxes, and the energy consumption was 45.8 kWh kgTOC−1. The sequential electro-Fenton was also challenged with the refractory trifluoroacetic acid (TFA) and trichloroacetic acid (TCA), and they can be transferred at a removal rate of 11.3 ± 1.2 and 21.8 ± 1.9 mmol m−2 h−1, respectively, with different transformation mechanisms.
■
INTRODUCTION Fenton chemistry has been researched for over a century,1 and the overall reaction mechanism is well understood,2,3 with the primary reactions in eqs 1 and 2. Fenton’s is commonly used as an advanced oxidation process for water treatment since it produces the hydroxyl radical (•OH; E0 = 1.9−2.7 V), which is one of the strongest aqueous oxidants that reacts with most species at diffusion controlled rates.4
Electro-Fenton (E-Fenton) is a second generation Fenton process that continuously generates H2O2 in situ via cathodic reduction of O22,5−12 following eq 3. O2 + 2H+ + 2e− → H 2O2 ,
(3)
The in situ H2O2 production of during E-Fenton avoids the risks associated with handling, transport, and storage of H2O2 and generally results in increased contaminant degradation rates as compared to classical Fenton chemistry. The E-Fenton process has been reported to be effective for the degradation of pesticides,13,14 dyestuffs,15,16 pharmaceuticals,17−19 and personal care products (PPCPs)20,21 as well as wastewaters from the food, 22,23 tanning, 24,25 and petrochemical26 industries. Although E-Fenton is effective for the degradation of many contaminants, the energy consumption ranges from 87.7 to 275 kWh kg TOC−1.2,27−30 In regards to further increasing the EFenton efficiency, recent research has focused on hybrid
H 2O2 + Fe 2 + + H+ → Fe3 + + H 2O + •OH, k1 = 63 M−1 s−1
(1)
Fe3 + + H 2O2 → Fe2 + + HO2• + H+, k 2 = 0.01 M−1 s−1
(2)
However, there are practical drawbacks to Fenton chemistry:2 1) the cost and risk associated with H2O2 storage and transportation, 2) the acid/base required to keep Fe dissolved during reaction and/or to neutralize reaction solutions before disposal, 3) the disposal of the produced iron sludge, and 4) the regeneration of Fe2+ i.e., k2 ≪ k1. © XXXX American Chemical Society
E 0 = 0.475 V vs Ag/AgCl
Received: November 20, 2014 Revised: January 13, 2015 Accepted: January 20, 2015
A
DOI: 10.1021/es505679e Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. Sandwiched electro-Fenton system based on carbon nanotube membrane stacks. A) Images of the unfolded sandwich membrane stacks including four layers and B) schematic of main roles of every layer in membrane stacks, and [P] and [P]m are pollutants and their oxidation intermediates, respectively.
processes such as photoelectro-Fenton (PEF)4,31−35 and sonoelectro-Fenton (SEF)36,37 processes due to hybrid synergism. Another method to improve E-Fenton would be to sequence the individual electrochemical processes i.e., O2 reduction to H2O2 followed by H2O2 reduction to •OH.38−41 For sequential electrochemistry, three-dimensional porous electrodes are utilized in the flow-through configuration. The flow through electrode configuration will also result in increased electrochemical kinetics and efficiencies due to convective mass transport to the electrode surface that promotes direct electron transfer.41−43 A recently developed electrochemical carbon nanotube (CNT) filter may have the potential to support a sequential flow-through E-Fenton process. For example, CNT cathodes have been observed to be effective and efficient for the in situ generation of H2O2 via the 2-electron reduction of O2,44,45 and the oxidized CNT that contains surface carboxylate groups has the potential to strongly chelate Fe2+ that can reduce H2O2 to HO•. The bound CNT-Fe2+ may also undergo E-Fenton chemistry i.e., as in situ regeneration of Fe2+, thus avoiding the loss of Fe2+, reducing production of Fe2+ sludge,2 and allowing operation at neutral pH in turn improving EFenton performance. In this study, we designed, constructed, and evaluated a 4layer CNT-based membrane stack for sequential flow-through E-Fenton chemistry. The fluid to be treated flowed through 1) a CNT network cathode for O2 reduction (eq 3), 2) a CNTCOOFe2+ network cathode for H2O2 reduction to •OH and Fe2+ electroregeneration in situ, 3) a PVDF or PTFE insulating separator between the cathode and anode, and 4) a CNT network anode for oxidation of remaining intermediates. First, the electrogeneration of H2O2 was examined as a function of influent pH, CNT doping, and cathode potential. Then, various methods for making CNT-COOFe2+ network were evaluated in regards to H2O2 reduction, lifetime, and electrochemical regeneration in situ. The synergism of the sequential flowthrough E-Fenton process was compared to flow-through electrochemistry and flow-through Fenton’s using oxalate as a
target molecule. Finally, the sequential E-Fenton was challenged with trifluoroacetate (TFA) and trichloroacetate (TCA).
■
EXPERIMENTAL SECTION Chemicals, Materials, CNT Prereatment, and Filter Preparation. All chemicals43 were purchased from SigmaAldrich, and their detailed information was listed in the SI. The undoped carbon nanotubes (C-CNT), nitrogen-doped carbon nanotubes (N-CNT), and boron-doped carbon nanotubes (BCNT) were purchased from NanoTechLabs, Inc. (Yadkinville, NC). The average diameters were 15 ± 6.6 nm (C-CNT), 18.6 ± 5.9 nm (B-CNT), and 25.1 ± 13.6 nm (N-CNT), and the average length was ∼100 μm. Pretreatment of the CNT was completed to remove any amorphous or other non-CNT carbon impurities, and the CNT filters were prepared following previously described methods as detailed in the SI.43,46 Preparation of Fe Coated CNT Network. The CNT were first oxidized to generate surface carboxylates (CNT-COO−) for Fen+ chelation.47,48 The CNT oxidation was completed by addition of 0.5 g CNT to 0.5 L of stirred 70 °C concentrated HNO3 in a round-bottom flask with a condenser for at least 12 h. The CNT-COO− solution was cooled to room temperature and 15 mL was vacuum filtered through a 5-μm PTFE membrane (Omnipore; Millipore) to form a CNT-COO− network.43,49,50 Then 1 M NaOH (100 mL) was flowed through the CNT-COO− filter to generate CNT-COONa, which was subsequently rinsed with DI water until the effluent pH was neutral. Finally, the CNT-COOFen+ filter (n = 2 or 3) was prepared by flowing either a 1 M FeCl2 or FeCl3 solution (100 mL) through the CNT-COONa filter. The CNTCOOFe3+ was transformed into CNT-COOFe2+ by electrochemical reduction at cathode potentials of −0.5 V, −0.8 V, or −1.0 V vs Ag/AgCl for 10 min using a CHI 604D (CHI Co., USA) electrochemical workstation. The CNT-based networks were characterized by XPS as detailed in the SI. B
DOI: 10.1021/es505679e Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology Table 1. Representative Carbon Nanotube Network Properties
a
CNT type
weight (mg)
deptha (μm)
pore ⟨d⟩ (nm)
O/C (%)
ΔH2O2- @ 10 min (%)
surficial Fe cont.b
CNT-HCl (as anode or cathode) CNT-COO− CNT-COOFe3+ CNT- COOFe3+, flow H2O2 CNT-COOFe3+-1.0 V CNT-COOFe3+-1.0 V, flow H2O2 CNT-COOFe3+-1.0 V, flow oxalate + H2O2
15.2 15.1 15.5 15.3 15.6 15.5 15.6
49.7 50.2 52.4 53.3 51.5 53.2 52.6
109.6 101.3 102.5 n/a 103.2 n/a n/a
2.24 4.08 4.08 n/a n/a n/a n/a
n/a n/a n/a 9 n/a 53 22
undetected 0.032% 6.89% 2.71% 7.29% 1.69% 7.89%
Measured by microcaliper. bMeasured by XPS.
Figure 2. H2O2 production using a CNT filter as cathode. Experiments were completed using C-CNT (black), B-CNT (red), and N-CNT (blue) and at pH 5.99 (square), 3.25 (circle) and 8.73 (triangle)) as a function of cathode potential. Electrochemical conditions were J = 1.6 mL min−1, [DO] = 38 ± 1 mg L−1 and [Na2SO4] = 10 mM. A) CV of CNT at pH 5.99 at a scan rate of 10 mV s−1, B) effluent H2O2 (mg L−1) after 10 min as a function of cathode potential, C) the corresponding DO conversion efficiency (%), and D) current efficiency (%).
Preparation of the Sequential Flow-through E-Fenton Membrane Stack. The E-Fenton membrane stack was prepared by mechanically pressing at 10 MPa for 5 min (Carver) the four different porous layers in the sequence displayed in Figure 1: 1) CNT cathode, 2) CNT-COOFen+ cathode, 3) PVDF or PTFE, and 4) CNT anode. The membrane stack thickness was ∼200 μm with individual layer thicknesses of 45 ± 10 μm (Table 1) and a diameter of 35 mm. The membrane stack was placed into an electrochemistry modified filtration cell (Whatman) with a perforated Ti current collector inserted from the top and bottom of the casing for electrical connection to potentiostat (CHI Inc., CHI604D) or a DC power supply (Agilent E3634) (Figures 1, S1, and S2). The appropriate influent solution was then pumped (Masterflex) through the CNT filter stack at the flow rate of 1.6 mL min−1.43 Flow-through H2O2 Electrogeneration and Analysis. A CNT network was used as both cathode and anode, and a PVDF or PTFE membrane was used to separate the electrodes. The influent solution contained 100 mM Na2SO4 as electrolyte and was sparged with pure oxygen (dissolved oxygen (DO) ∼38 mg L−1; Thermo-Orion DO meter), and H2O2 electrogeneration was examined as a function of influent pH and
cathode potential. The H2O2 content was measured by colorimetry (iodide),51 and oxygen transformation efficiency (ηO2) and current efficiency (ηi) were calculated using the following equations: ηO2 = (c[H 2O2 ]/34)/(Δ[O2 ]/32)*100
(4)
where ηi = (nFc[H 2O2 ]V/(1000*34*c[H 2O2 ]Q))*100
(5)
c[H2O2] is H2O2 produced, Δ[O2] is O2 consumed, and 32 and 34 are the molar mass of O2 and H2O2, respectively, n = 2 represents the stoichiometric number of electrons required for electrochemical H2O2 production, F is Faraday’s constant (96,485 C mol−1), V = t × J × 10−3 is the total volume treated where t is the reaction time (min) and J = 1.6 mL min−1 is the flow rate, and Q = i × t is the total electrons consumed during H2O2 electrogeneration.2 Flow-through Fenton Performance and Regeneration. To assess the CNT-COOFen+ activity as a function of Fen+ coating type (Fe2+, Fe3+, or Fe2+ via Fe3+ electroreduction), 14 mg L−1 of H2O2 was flowed through the various filters, and the time-dependent [H2O2]ef/[H2O2]in was monitored (see C
DOI: 10.1021/es505679e Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology Figure S3C). Experiments were completed in the presence and absence of 0.12 mM oxalate to examine the extent of CNTCOOFen+ self-oxidation by Fenton chemistry. The ability to electroreduce and thus regenerate the CNT-COOFen+ activity as a function of cathode potential was investigated by first flowing 14 mg L−1 H2O2 through the filter until the effluent concentration was equivalent to the influent concentration and then applying a set cathode potential for 10 min. Comparison of Flow-through E-Fenton to Individual Electrochemistry and Fenton. For all experiments, the electrode order and configuration are listed in Figure S3, and the influent oxalate was 0.25 mM in 10 mM Na2SO4 with J = 1.6 mL min−1. For flow-through electrochemistry, the influent solution was sparged with N2 to remove O2 and eliminate H2O2 electrogeneration. For flow-through Fenton, [H2O2]in = 14 mg L−1 (equivalent to the H2O2 generated during E-Fenton) was flowed through a CNT-COOFen+ filter. For the E-Fenton, the influent solution was sparged with pure oxygen ([DO]in = 38 ± 1 mg L−1) and then flowed through the 4-layer stack at a cathode potential of −1.0 V. To further evaluate the E-Fenton process, the system was challenged with 0.25 mM of TFA or TCA. TFA and TCA of Analytical Procedures by UPLC/MS/ MS. The reaction intermediates were monitored and identified by the LC/MS/MS system (Waters Xevo TQ-S) with full scan from m/z 19−200 in negative electrospray ionization (ESI) mode with a dwell time of 400 ms. The analytical column used was a Waters BEH C18 (2.1 × 50 mm; 1.7 μm particle size) column, and the eluent flow rate was 0.450 mL min−1. A eluent gradient was used with an initial methanol/water ratio of 90/10 (v/v) that was decreased linearly to 50/50 (v/v) over the first 2 min, then decreased linearly to 0/100 (v/v) over the next 11 min, and finally increased linearly to 90/10 (v/v) over 1 min where it was held for the next 1 min. The LC/MS/MS parameters were set as follows: the desolvation temperature was 350 °C, the capillary voltage was 3 kV, and the cone voltage was set to 40 V.
Figure 3. Fenton reaction by flowing H2O2 through the CNTCOOFen+ filter. The influent H2O2 is 14 mg L−1 is flowed through CNT-COOFen+, exchanged with Fe2+ (red cycle), Fe3+ (black triangle), Fe2+ from Fe3+ reduction (−0.5 V (gray down triangle), −0.8 V (cyan square), and −1.0 V (blue diamond) cathode potential), respectively. H2O2 adsorption to the CNT (green line).
■
RESULTS AND DISCUSSION H2O2 Production by a Cathodic CNT Filter. Cyclic voltammetry (CV) using the C-CNT (black), B-CNT (red), and N-CNT (blue) was completed using a pure O2 sparged solution and compared to the C-CNT control (dashed black) sparged with pure N2 as displayed in Figure 2A. The O2 reduction peaks are noted with an arrow. The peak potentials (Epc) had the following order N-CNT (−0.47 V) < C-CNT (−0.31 V) < B-CNT (−0.30 V) vs 1 M Ag/AgCl which were all lower than the expected theoretical potential (0.108 V, calculated according to the Nerst equation of eq 3 with pH 5.99) due to the overpotential effect, and the peak current (ipc) ranged from high to low was N-CNT (33.3 mA) > C-CNT (9.8 mA) > B-CNT (2.9 mA). The order of the peak potentials and currents is consistent with the dopant e.g., the n-type N-CNT results in reducing electrons being the major carrier and pushes the Fermi level closer to the conduction band.46,52−54 In regards to H2O2 production displayed in Figure 2B, the NCNT had a maximum rate of 240 mg H2O2 m−2 h−1, the lowest among the three CNT samples likely due to continued reduction of H2O2 to •OH and H2O.55−58 The C-CNT had the highest max rate at −0.3 V of 1300 mg H2O2 m−2 h−1 (or steady-state effluent concentration of 14 mg L−1 H2O2). The BCNT had a moderate max rate of 810 mg H2O2 m−2 h−1.
Figure 4. Fenton reaction and electrochemical regeneration of CNTCOOFe3+ to CNT-COOFe2+ in the absence or presence of oxalate. A CNT-COOFe3+ was first reduced at a cathode potential of −1.0 V and then is challenged with H2O2 (14 mg L−1) until the CNT-COOFen+ no longer degrades H2O2 (square); then the above inactivated CNTCOOFen+ is regenerated by reduction at a cathode potential −1.0 V and used again A) in the absence or B) presence of oxalate.
D
DOI: 10.1021/es505679e Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
time. For example, the CNT-COOFe2+ (red) reduced ∼70% of the [H2O2]in at 1 min (∼1,110 mg H2O2 m−2 h−1) and 20% at 5 min (∼319 mg H2O2 m−2 h−1) and lost activity after 10 min. As expected from slow kinetics (eq 2), the CNT-COOFe3+ (black up triangle) had negligible performance, similar to H2O2 sorption on C-CNT (green line). The reduction of the CNTCOOFe3+ at −0.5, −0.8, or −1.0 V for 10 min (>4.0 C electron flow; Figure S5) was able to reduce the adsorbed Fe3+ (∼0.5 mg; 8.9 μmol; 0.8 C equiv; eq 6; Figure S5) and increases the H2O2 reactivity to equal to or greater than the CNT-COOFe2+. Fe3 + + e− → Fe2 +,
E 0 = 0.55 V
(6)
3+
For example, the CNT-COOFe −1.0 V (CNT-COOFe 3+ reduced at −1.0 V for 10 min) can reduce ∼95% H2O2 (1,300 mg H2O2 m−2 h−1) for the first 5 min and ∼45% H2O2 at 10 min (∼585 mg H2O2 m−2 h−1), whereas CNTCOOFe2+ had lost all reactivity after 10 min. The improved performance of the CNT-COOFe3+−1.0 V is probably due to increased Fe3+ sorption to CNT-COONa from the stronger complexation interaction between Fe3+ and CNT-COO− as compared to Fe2+. The regeneration of the Fenton reactivity i.e., the conversion of CNT-COOFe3+ to CNT-COOFe2+ was also evaluated. A CNT-COOFe3+−1.0 V was first challenged with H2O2 (similar conditions Figure 3), then reduced at −1.0 V, and challenged again with H2O2 as displayed in Figure 4A. The regenerated CNT-COOFe3+−1.0 V lost a significant amount of reactivity as compared to the initial run, for instance, ∼95% of influent H2O2 was degraded in the first run at 5 min, whereas only ∼5% of influent H2O2 was degraded in the second, suggesting Fe loss from the filter. The loss of Fe was confirmed by XPS (Table 1 and Figure S6) of the CNT-COOFe3+−1.0 V sample before (7.3% Fe) and after (1.7% Fe) H2O2 filtration. Any produced hydroxyl radical will only have a few species to react with such as H2O2 or the CNTs themselves (eq 6).
Figure 5. Three kinds of process of oxidizing oxalate. Electrochemistry (N2) (blue square) means only electrochemically oxidizing oxalate under condition of removing DO by sparging N2; Fenton (H2O2) (red square) means that only chemical Fenton reaction plays a role by flowing 14 mg/L H2O2 solution and 0.25 mM oxalate through CNTCOOFe n+ filter; and E-Fenton (O 2 ) (dark square) means comprehensive process containing 1) electrochemical reaction of producing H2O2 on cathode by reducing O2, 2) electrochemical Fenton reaction between H2O2 produced in situ and CNT-COOFen+, then 3) oxidizing further pollutant via •OH from Fenton and other oxidant from anode, and, most importantly, 4) regenerating CNTCOOFen+ automatically and directly on cathode by electroreduction reaction. Fenton reaction rate decreases sharply (blank red square) during 30 min then it was electrochemically reduced ex situ then run the second cycle again (solid red square), as the comparison, E-Fenton and electrochemistry also were challenged by the second cycle without the electroreduction.
The C-CNT was utilized as cathode for all following experiments due to its best performance in O2 reduction to H2O2. The DO reduction and current efficiencies as a function of pH and cathode potential are displayed in Figure 2C and D, respectively. In regards to pH, the performance ranged from high to low was pH 6.0 > 8.7 > 3.2, which does not agree with the stoichiometry of eq 3,59,60 and the lowest performance at pH 3.2 is likely due to competition with water reduction to H2 (2H+ + e− → H2, E0 = −0.22 V).42 The excellent performance of the cathode at neutral pH solves a common Fenton chemistry issue of having to constantly acidify to keep the Fe dissolved. The max DO efficiency is near 50%, which is better than previous reports of 37%,2,59 and the max current efficiency is near 42%. Effluent DO and pH as well as i-E and V-E relations are displayed in Figure S4. The max rate and efficiency of H2O2 generated at a cathode potential of −0.31 V (total voltage of 2.1 V with C-CNT anode in Figure S2D) are lower than previous reports of −0.4 to −1.6 V of cathode potential.2,59 The energy consumption (9.4 kWh kgH2O2) for H2O2 production, the primary electrochemical H2O2 production cost, is lower than the price of commercial bulk H2O2 which is about 1.5 $ per kilogram H2O2 (Free on Board Price) without consideration of transportation and storage of fee and risk. Flow-through Fenton Performance and Electroregeneration of the CNT-COOFen+ Filter. The various Fen+ filters (n = 2, 3) were initially challenged with [H2O2]in = 14 mg L−1 (similar to max effluent H2O2 produced in previous section) at J = 1.6 mL min−1 in the absence of applied potential as displayed in Figure 3 with the system configuration in Figure S3B. In all cases, the extent of H2O2 reduction decreases with
H 2O2 + •OH → H 2O + HO2• , k 7 = 2.7 × 107 M−1 s−1
(7)
CNT‐COO−Fen + + •OH → CNT + CO2 + HO− + Fen + (8)
The reaction of •OH with a CNT carboxylate group would result in loss of the carboxylate group and in turn a Fe binding site within the filter, which can be validated by Fe content before (6.98% Fe) and after (2.71% Fe) H2O2 filtration. To test this hypothesis, a similar set of regeneration experiments was completed with the addition of 0.12 mM oxalate to compete with CNT-COOFe3+ (8.9 μmol Fen+) and H2O2 (0.4 mM) for hydroxyl radicals as displayed in Figure 4B. C2O4 2 − + •OH → CO2 + CO2•− , k 9 = 1.4 × 106 M−1 s−1
(9)
C2O4 2 − + H 2O2 → CO2 + H 2O, k10 < 3.6 V) greater than •OH (∼2.7 V).61 Both TFA and TCA can be transferred by our novel E-Fenton at removal rate of 11.3 ± 1.2 and 21.8 ± 1.9 mmol m−2 h−1, respectively; however, their transformation routes are much different as depicted in Figure 6. The flow-through E-Fenton system here was able to reduce TFA to DFA by defluorination; however, DFA was not further reduced to MFA. Also, the observation of fluoroform and formate as products indicates that the E-Fenton process can break the C−C bonds of TFA. However, no expected products of the reaction of TFA and •OH were observed, which is consistent with previous reports that •OH cannot oxidize TFA.61 Compared to TFA (C−F bond energy, 127 kcal mol−1), TCA is easily reduced due to weaker C−Cl (bond energy, 80.9 kcal mol−1). First, TCA can be electrooxidized resulting in C−C bond cleavage via •OH from EFenton and can also be electroreduced to DCA, MCA, and acetic acid via dechlorination. Then DCA, MCA, and acetic acid can be further oxidized to lower molecular weight compounds and eventually mineralized. In summary, regenerative flow through E-Fenton was investigated by synthesis of a novel carbon nanotube membrane stack. Compared with previous E-Fenton studies (Table S1), the flow-through CNT-based electrode stack here allowed for the following: 1) E-Fenton operation at neutral pH since the Fe was bound to the electrode, 2) production of a single multifunctional membrane-electrode stack of ∼200 μm thickness with convective mass transport, and 3) in situ Fe2+ electroregeneration and prevention of Fe2+ loss via addition of an electron donor and consequent Fe sludge production. As a result, the flow-through E-Fenton system has good efficiency and efficacy (∼45% current efficiency; Table S1) low cell voltage (
