Removal of organic micropollutants from drinking water by a novel electro-Fenton filter: Pilot-scale studies

Removal of organic micropollutants from drinking water by a novel electro-Fenton filter: Pilot-scale studies

Accepted Manuscript Removal of organic micropollutants from drinking water by a novel electro-Fenton filter: Pilot-scale studies Konstantinos V. Plaka...

2MB Sizes 3 Downloads 50 Views

Accepted Manuscript Removal of organic micropollutants from drinking water by a novel electro-Fenton filter: Pilot-scale studies Konstantinos V. Plakas, Stella D. Sklari, Dimitrios A. Yiankakis, Georgios Th. Sideropoulos, Vassilis T. Zaspalis, Anastasios J. Karabelas PII:

S0043-1354(16)30013-6

DOI:

10.1016/j.watres.2016.01.013

Reference:

WR 11767

To appear in:

Water Research

Received Date: 13 November 2015 Revised Date:

4 January 2016

Accepted Date: 8 January 2016

Please cite this article as: Plakas, K.V., Sklari, S.D., Yiankakis, D.A., Sideropoulos, G.T., Zaspalis, V.T., Karabelas, A.J., Removal of organic micropollutants from drinking water by a novel electro-Fenton filter: Pilot-scale studies, Water Research (2016), doi: 10.1016/j.watres.2016.01.013. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Graphical Abstract

1st cycle

1.0

M AN U

0.4

3rd cycle 0V 2V

SC

CDCF/CDCF,0

0.8

0.6

2nd cycle

RI PT

1.2

0.2

0.0

AC C

EP

TE D

0

1000

2000

3000

Time (min)

4000

5000

ACCEPTED MANUSCRIPT

Removal of organic micropollutants from drinking water by a novel electro-

2

Fenton filter: Pilot-scale studies

3

Konstantinos V. Plakas a,*, Stella D. Sklari a, Dimitrios A. Yiankakis b, Georgios Th. Sideropoulos b,

4

Vassilis T. Zaspalis a,c, Anastasios J. Karabelas a

5

a

6

(CERTH), 6th km Charilaou-Thermi Road, Thermi, Thessaloniki, GR 57001, Greece

7

b

TEMAK AETE, DA13 Street, Industrial Area Thessaloniki, Sindos, Thessaloniki, GR 57022, Greece

8

c

Department of Chemical Engineering, Aristotle University, Thessaloniki, GR 57001, Greece

9

Abstract

RI PT

1

M AN U

SC

Chemical Process and Energy Resources Institute, Centre for Research and Technology – Hellas

To assess the performance of a novel ‘filter’-type electro-Fenton (EF) device, results are reported from

11

pilot-scale studies of continuous water treatment, to degrade diclofenac (DCF), a typical organic micro-

12

pollutant, with no addition of oxidants. The novel ‘filter’ consisted of three pairs of anode/cathode

13

electrodes made of carbon felt, with cathodes impregnated with iron nanoparticles (γ-Fe2O3/F3O4

14

oxides). The best ‘filter’ performance was obtained at applied potential of 2V and low water superficial

15

velocities (~0.09 cm/s), i.e., the mineralization current efficiency (MCE) was >20%, during continuous

16

steady state treatment of tap water with low DCF concentrations (16 µg/L). The EF ‘filter’ exhibited

17

satisfactory stability regarding both electrode integrity (no iron leaching) and removal efficiency, even

18

after multiple filtration/oxidation treatment cycles, achieving (under steady conditions) DCF and TOC

19

removal 85% and 36%, respectively. This performance is considered satisfactory because the EF

20

process took place under rather unfavorable conditions, such as neutral pH, low dissolved O2

21

concentration, low electrical conductivity, and presence of natural organic matter and inorganic ions in

22

tap water. Ongoing R&D is aimed at ‘filter’ development and optimization for practical applications.

EP

AC C

23

TE D

10

24

Keywords: water decontamination; neutral electro-Fenton; pilot-unit; iron-impregnated carbon felt;

25

diclofenac

26 27 28

*Corresponding Author e-mail: [email protected] (K.V. Plakas) tel.: +30 2310498476 1

ACCEPTED MANUSCRIPT 1. Introduction

30

In recent years, electrochemical advanced oxidation processes (EAOPs) are considered environment-

31

friendly emerging alternatives for purification of water polluted by a multitude of organic compounds,

32

commonly encountered at very small concentration (Oturan et al., 2013; Sirés et al., 2014; Bocos et al.,

33

2015; Thiam et al., 2016). Removal of such micropollutants (threatening human health), in a reliable

34

and cost-effective manner, is still a challenging task. Among EAOPs, the electro-Fenton (EF) method is

35

perhaps the most popular one, forming the basis for a variety of related processes (Brillas et al., 2009).

36

The EF method combines the classical Fenton treatment with the electrochemical oxidation. Thus, the

37

electrical current induces the in situ generation of Fenton reagent, thus avoiding the use of significant

38

quantities of H2O2 and iron in the form of ferrous (or ferric) salts. The catalytic reaction is sustained by

39

Fe2+ regeneration, which takes place by reduction of Fe3+ with H2O2, hydroperoxyl radical, organic

40

radical intermediates, or directly at the cathode (Brillas et al., 2009). Iron is either added to the polluted

41

water (homogeneous EF) or embedded onto suitable supporting materials (heterogeneous EF), in order

42

to catalyze H2O2 - splitting to produce the oxidizing agent •OH via Fenton reactions. The iron

43

immobilization onto a support provides several advantages over the iron addition; i.e., a) the polluted

44

water can be treated at its physical pH (typical neutral to basic), and not at acidic pH as in the case of

45

homogeneous EF (the process can effectively occur within the range 2.5
46

accumulation of iron sludge that must be removed at the end of treatment, and c) the heterogeneous

47

catalyst is retained in the system, avoiding iron release in the treated water (Brillas et al., 2009; Iglesias

48

et al., 2015).

49

Although significant research on EF has been performed at bench-scale (Brillas et al., 2009; Feng et al.,

50

2013; Sirés et al., 2014; Babuponnusami & Muthukumar, 2014), for the effective elimination of toxic

51

and/or persistent organic pollutants from a large variety of wastewaters, pilot or full-scale applications

52

of this technology in the water industry are limited. Inadequately explored issues, responsible for this

53

gap between research results and practical EF applications, are environmental, technological and

AC C

EP

TE D

M AN U

SC

RI PT

29

2

ACCEPTED MANUSCRIPT economic, that need to be addressed with additional work on a) the optimum design of the

55

electrochemical cells/reactors, b) the development of novel electrodes (high-performance anodes and

56

cathodes with enhanced electrocatalytic properties), and c) the determination of optimal operating

57

conditions (electric potential, water flux, pH), taking into consideration the effect of the water matrix

58

on the treatment efficiency. To address these challenges, Brillas and collaborators (El-Ghenymy et al.,

59

2013; Garcia-Segura et al., 2014; Garcia-Segura and Brillas, 2016) have recently developed hybrid

60

systems integrating EF with UV-A light or solar irradiation, to accelerate the mineralization rate of

61

organics via photoelectro-Fenton (PEF) or solar photoelectron-Fenton (SPEF) processes, respectively.

62

Lately, the use of boron doped diamond (BDD) anode significantly enhanced the oxidation power and

63

mineralization efficiency of the EF process, thus paving the way for the development of pilot-scale

64

reactors and their application to treatment of large volumes of wastewaters. In this field,

65

commercialized products with BDD anodes have been developed, such as Oxineo® and Sysneo® that

66

have been applied for the automated disinfection of swimming pool water, as well as CONDIACELL®

67

and Diamonox® cells with typical applications in water disinfection and wastewater treatment (Sirés et

68

al., 2014).

69

In view of the above-mentioned challenges and considering the high cost of the BDD electrodes, a

70

novel electrochemical device has been developed in the form of a ‘filter’ comprised of a stack of carbon

71

anodic and cathodic electrode pairs for operation in continuous mode (Plakas et al., 2013). The design

72

and construction of this device is characterized by the following main innovative features: a) the

73

application of three-dimensional carbon electrodes (such as carbon felt) of high specific surface area

74

and low cost (in comparison to BDD) both as anode and cathode, b) the impregnation of catalytic nano-

75

particles in the porous cathodic electrodes, which is a key element for the realization of the

76

heterogeneous electro-Fenton reaction, and c) the development of a multi-layer electrode ‘filter’ (with

77

pairs of anodes and cathodes). This ‘filter’-type design facilitates the scale-up of the device and

78

promotes the uniform distribution of the water throughout the electrodes, thus ensuring the effective

AC C

EP

TE D

M AN U

SC

RI PT

54

3

ACCEPTED MANUSCRIPT fluid contact with the in-situ produced oxidants. Bench-scale tests with one pair of anode/cathode

80

electrodes, using diclofenac (DCF) as a model micropollutant (Sklari et al., 2015), resulted in optimized

81

cathodic electrodes (employing carbon felts and optimum procedures of iron nanoparticle

82

impregnation), ‘filter’ design and operating conditions (cathodic potential, feed flow). These results

83

paved the way for the design and construction of a fully automated laboratory-scale pilot system,

84

presented herein, which was tested under real conditions, as a necessary step toward applications.

85

The results of pilot testing are reported in this paper, for the removal of DCF from water by an electro-

86

Fenton ‘filter’ consisting of three pairs of anode/cathode electrodes, operating under near optimum

87

conditions (applied potential, water velocity). Specifically, the effect of the initial DCF feed

88

concentration and of the water matrix (deionized water with added sodium sulfate and tap water) on

89

process efficiency are examined in this work on pollutant removal (adsorption and molecular

90

degradation/mineralization) and ‘filter’ stability (integrity of electrodes, iron leaching) after multiple

91

adsorption/electrolysis cycles.

TE D

92

M AN U

SC

RI PT

79

2. Experimental work

94

2.1. Materials

95

A carbon felt with high specific surface area, coded here as CFm, was obtained from MAST Carbon

96

International Ltd (Great Britain). The physical parameters of this carbon felt are summarized in Table

97

S1 (Supplementary Material). CFm has a thickness of ~2 mm and was cut in discs of 62 mm diameter to

98

be used both as anode and cathode electrode. Iron (II) chloride tetrahydrate (FeCl2.4H2O, Panreac

99

Química S.A.U), iron (III) chloride hexahydrate (FeCl3.6H2O, Panreac Química S.A.U) and ethanol

100

absolute (CH3CH2OH, Scharlab SL), reagent grade, were used during the preparation of the cathodes.

101

Diclofenac sodium salt (DCF) and anhydrous sodium sulfate (Na2SO4) were of analytical grade (Sigma-

102

Aldrich) and used as received. All chemicals (Sigma-Aldrich) used for the analysis of DCF and H2O2

103

were reagent grade. Information on the chemical structure and the physicochemical properties of DCF

AC C

EP

93

4

ACCEPTED MANUSCRIPT are included in Table S2 (Supplementary Material). Two different water types, Deionized Water (DW)

105

and Tap Water (TW) were employed in this study. Water quality parameters of these feed water types

106

are included in Table S3 (Supplementary Material). DW was produced by passing tap water through

107

successive ion exchange columns, in which 2.82 mM Na2SO4 was added in order to increase DW

108

conductivity at approx. 700 µS/cm, similar to that of the TW. The latter was natural water from a well

109

(serving author’s laboratory facilities) which undergoes filtration and chlorination.

RI PT

104

SC

110 2.2. Preparation of the cathodic electrodes

112

The disk shaped electrodes were first cleaned with ethanol absolute to remove the impurities from their

113

surface, and then dried at room temperature for 24 h. This pretreatment procedure was found to affect

114

positively the adsorption of DCF molecules on the CFm electrodes, being seven times greater than that

115

recorded in the case of the non-pretreated CFm electrodes (Sklari et al., 2015). The three anodic

116

electrodes (A1, A2 and A3) were kept in a desiccator for further use, while three discs of CFm

117

electrodes (C1, C2 and C3) were treated according to the procedure described in our previous work

118

(Sklari et al., 2015), so as to prepare cathodic electrodes (designated here as CFm-nFe) impregnated

119

with mixed valence iron oxides (γ-Fe2O3/F3O4 oxides). The final loading was approximately 0.4 mg Fe

120

per m2 of CFm.

TE D

EP

AC C

121

M AN U

111

122

2.3. Characterization methods

123

The structural properties of the electrodes were determined by X-ray diffraction using a D-500

124

diffractometer from Bruker equipped with a Cu Kα radiation (λ = 1.5418 Å) source. The signal was

125

recorded for 2θ between 5 o and 80 o with a recording step of 1 o per min. Phase identification was made

126

through comparison with the JCPDS database. The N2 adsorption-desorption measurements were

127

performed in a Micromeritics TriStar porosimeter. The specific surface area (SBET) of the electrodes

128

was measured at liquid N2 temperature (−196 oC), using the Brunauer–Emmett–Teller (BET) method in 5

ACCEPTED MANUSCRIPT 129

the relative pressure P/Po range of 0.01 to 0.30. Prior to analysis, the sample was heated under high

130

vacuum at 200 oC for 18 h. SEM images were recorded using a JEOL JSM6300 microscope operating

131

at 20 kV. The samples were gold sputtered to avoid charging effects on the images.

RI PT

132 2.4. Electro-Fenton ‘filter’ – Pilot plant

134

The electro-Fenton ‘filter’ consists of three pairs of carbon felt electrodes of 4.5 gr total mass

135

(Graphical Abstract). Plastic spacers of ~1 mm thickness are placed in-between anode/cathode

136

electrodes to prevent short circuiting. All electrodes are in contact with thin metal perforated disks

137

made of stainless steel (SS316) which act as current carriers to the electrodes. The voltage applied

138

between each couple of anode/cathode electrodes is regulated by the operator and can vary between 0

139

and 6 Vdc. In order to prevent water leaks and create a narrow inter-electrode gap capable of

140

minimizing cell voltage and IR loss, 1 mm thick gaskets (made of nitrile butadiene rubber) of 86 mm

141

and 58 mm outer and inner diameter, respectively, are placed between the SS316 discs. The feed

142

solution entering from the top of the electrode stack flows axially so that the contaminant (DCF) and all

143

electroactive species cascade through the porous carbon electrodes, thus maximizing their contact time

144

and consequently the adsorptive interactions and EF reactions within the ‘filter’. Considering the CFm

145

porosity, the inter-electrode gap (defined by the gaskets) and the geometric characteristics of the

146

stainless steel current carriers, the effective volume of the ‘filter’ is estimated to be ~49 cm3 (Table S4

147

in Supplementary Material). According to this volume, the residence time of DCF within the ‘filter’

148

varies in the range 4.3 to 17.4 sec for flow rates of 10 to 40 L/h, respectively (Table S4 in

149

Supplementary Material).

150

It should be noted that no O2 is externally added into the feed solution (through either compressed air or

151

pure oxygen gas); i.e., the electrogeneration of H2O2 solely depends on the dissolved O2 and the oxygen

152

generated by water oxidation over the anodes (Eq. 1). The current efficiency (CE) of the new ‘filter’,

AC C

EP

TE D

M AN U

SC

133

6

ACCEPTED MANUSCRIPT defined as the ratio of the electricity consumed by the electrode reaction of interest (generation of H2O2

154

by reaction 2, Eq. 2) over the total electricity passing through the ‘filter’, can be calculated by Eq. 3.

155

H2O  ½ O2 + 2H+ + 2e-

(1)

156

O2(g) + 2H+ + 2e-  H2O2

(2)

157

CE =

158

where n represents the stoichiometric number of electrons transferred in reaction (1), F is the Faraday

159

constant (96485 C/mol), CH 2O2 is the concentration of H2O2 at the outlet of the ‘filter’ (µg/L), V is the

160

effective volume of the ‘filter’ (0.049 L), 106 is a factor for conversion of units (106 µg/g) M H 2O2 the

161

molecular weight of H2O2 (34 g/mol), and Q the charge consumed during the electrolysis (C). CE

162

values were calculated in preliminary tests with three pairs of CFm anode/cathode electrodes, as

163

described below.

164

A schematic diagram and photos of the EF laboratory-scale pilot plant are included in Fig. 1. The plant

165

consists of a 200 L feed tank (TN1) where DCF is dissolved in DW or TW with the aid of a mixer

166

(MX1). The feed water is fed continuously to the EF ‘filter’ by a magnetic pump (PU1); a 60 µm

167

cartridge filter (FI1) is placed upstream of the EF ‘filter’, to protect the latter from suspended solids

168

(e.g. in experiments with tap water). The performance of the EF ‘filter’ can be compared with that of a

169

GAC filter (FI2) placed in parallel with the EF ‘filter’. The flow and pressure of feed water is measured

170

and controlled by a flowmeter (FM1) and a control valve (NV1), respectively. The pilot plant is

171

equipped with sensors at the inlet and outlet of the two filters recording the following parameters:

172

pressure (PT1, PT2), conductivity (CM1, CM2), pH (pH1, pH2) and redox potential (Rx1, Rx2). All

173

signals are transferred to a PLC (Unitronics OPLC V1210, comprising a 12-inch color TFT touch

174

screen), which enables data acquisition and monitoring of all parameters in real time, as well as the

175

control of system operation. PLC signals/data such as inputs and outputs are displayed online (to

nFC H 2 O 2 V

x100%

(3)

AC C

EP

TE D

M AN U

SC

106 M H 2 O 2 Q

RI PT

153

7

ACCEPTED MANUSCRIPT 176

monitor trends and make possible observations) and recorded onto a microSD memory card with a

177

selectable time stamp for future processing.

178

Fig. 1. a) Flow sheet. b) Front and side views of the electro-Fenton pilot plant; 1: feed pump, 2:

180

cartridge filter 60 µm, 3: EF ‘filter’, 4: control valve, 5: flowmeter, 6: PLC, 7: voltage regulator, 8:

181

powered agitator, 9: feed tank.

RI PT

179

SC

182 2.5. Experimental procedures

184

Considering the high capacity of the carbon felt electrodes for DCF adsorption (Sklari et al., 2015), the

185

experimental protocol consisted of two distinct operations: a) continuous flow of ~1 mg/L DCF feed

186

solution at a constant flow rate of 40 L/h without applying electric voltage (0 V), until saturation of the

187

carbon electrodes with DCF (adsorption step); b) continuous treatment of a DCF feed solution of

188

different initial DCF concentration (0.02 - 0.60 mg/L) at a constant flow rate of 10 L/h and applied

189

potential of 2.0 V per pair of anode/cathode electrodes (electrolysis step). The latter operating

190

conditions (2.0 V applied potential and 10 L/h flow rate) were found to be near optimum for H2O2

191

electrogeneration with the pilot EF ‘filter’, as described below. The performance of the EF ‘filter’

192

during multiple cycles of adsorption/oxidation of DCF solutions was also assessed, following the same

193

experimental protocol. All pilot tests were performed at room temperature (23±2 oC).

194

Samples were collected from the feed (sampling after the cartridge filter-FI1, Fig. 1a) and the outlet of

195

the EF ‘filter’ at certain time intervals for determining DCF, TOC and H2O2 concentrations. The

196

possible leaching of iron from the cathodic electrodes was also determined by measuring the iron

197

content in several samples collected during both adsorption and oxidation steps. After termination of

198

the experiment, the electrode samples were left to dry at room temperature and kept in a desiccator for

199

subsequent analysis (SEM). New carbon electrodes were used for each test.

AC C

EP

TE D

M AN U

183

200 8

ACCEPTED MANUSCRIPT 2.6. Analytical methods and measurements

202

Before analysis, samples withdrawn from electrolyzed solutions were filtered through 0.45 µm PTFE

203

Millipore membranes. Total organic carbon concentration of all samples was determined with a TOC

204

analyzer (TOC-5000A, Shimadzu Co.). The repeatability of these measurements was checked by

205

standard deviation (SD) and coefficient of variation (CV) and the measurement error was estimated to

206

be within 2-8% for all DW and TW samples. From these data, the mineralization current efficiency

207

(MCE %) for each treated solution at a given time t (h) was calculated by the equation:

208

MCE % =

209

where n is the number of transferred electrons, F is the Faraday constant (96485 C/mol), Vs is the

210

solution volume (L), ∆TOC is the solution TOC decay (mg/L), 12x103 is a factor for conversion of

211

units (12000 mg/mol), m is the number of carbon atoms of DCF (14 C atoms), I is the applied current

212

(A) and t the electrolysis time (s). Regardless of the feed water type (DW or TW) the number of

213

electrons (n) consumed was taken as 58 assuming that the mineralization reaction involves mainly DCF

214

conversion into CO2, with Cl- and NH +4 as the main inorganic ions (Brillas et al., 2010). This overall

215

reaction can be written as follows:

216

C14 H10 Cl 2 NO 2− + 26H 2 O → 14CO 2 + 2Cl − + NH +4 + 58H + + 58e −

217

The temporal variation of DCF concentration was followed by reversed-phase HPLC using a Shimadzu

218

(LC-10AD VP) liquid chromatograph coupled with a diode array detector (DAD-M20A) set at 270 nm

219

according to the analytical method described elsewhere (Sarasidis et al., 2014). The degradation of DCF

220

within the EF ‘filter’ was verified by the appearance of new peaks in UV/Vis spectra of the samples

221

(due to aromatic organic intermediates formed), from the consumption of the electrogenerated H2O2 as

222

a result of the Fenton reactions occurring on the catalytic CFm-nFe electrodes, as well as by GC-MS

223

measurements based on the method described elsewhere (Sarasidis et al., 2014).

RI PT

201

(4)

(5)

AC C

EP

TE D

M AN U

SC

nFVs ∆TOC x100 12 x10 3 mIt

9

ACCEPTED MANUSCRIPT The concentration of hydrogen peroxide was determined spectrophotometrically (UV-1700

225

Pharmaspec, Shimadzu) by the iodide method (detection limit of H2O2 concentration 20 µg/L) (Klassen

226

et al., 1994). The elemental analysis of iron was performed using a Perkin Elmer Optima 4300DV

227

Inductively Coupled Plasma Atomic Emission Spectrometer (Sklari et al., 2015).

RI PT

224

228 3. Results & Discussion

230

Table 1 includes the main experimental conditions in eleven continuous experiments concerning H2O2

231

electrogeneration in the EF ‘filter’ (Exp. No #1-4), removal of background TOC (included in TW) in

232

the absence of DCF (Exp. No #5) and DCF/TOC removal (adsorption/mineralization) by the EF ‘filter’

233

(Exp. No #6-11). In this series of experiments, the effects of feed DCF concentration and water matrix

234

are mainly investigated.

M AN U

SC

229

235

237

Table 1. Summary of main experimental conditions.

TE D

236

3.1. H2O2 electrogeneration in the EF ‘filter’

239

The efficiency of the EF ‘filter’ to electrogenerate H2O2 with three pairs of CFm anode/cathode

240

electrodes was optimized first, by experimenting with the two major operating parameters, namely

241

applied potential and feed water flow rate. Fig. 2 shows the time-dependent changes of H2O2

242

concentration measured in the outlet of the ‘filter’ in the case of continuous flow of DW (with the

243

addition of 2.82 mM Na2SO4) or TW at various applied potentials. As expected, the higher the applied

244

potential the higher the H2O2 generation for both types of water, with H2O2 concentration exhibiting a

245

tendency for stabilization within the first 30 min. This trend is probably associated with the increase in

246

the rate of all electrode reactions related to the synthesis of H2O2 (Eq. 1 and Eq. 2) and the necessary

247

time needed for the conditioning of the carbon electrode surfaces. However, the application of voltages

248

higher than 2 V (or 2.2V in the case of DW) resulted in decreased H2O2 concentrations, probably due to

AC C

EP

238

10

ACCEPTED MANUSCRIPT the promotion of parasitic reactions. Specifically, the configuration of the electrolytic cell requires the

250

feed water to flow through successive pairs of anode/cathode electrodes, thus bringing in contact the

251

H2O2 produced in the leading pair(s) of anode/cathode electrodes with the surface of other downstream

252

anodes and cathodes of the ‘filter’. This may result in the oxidation of H2O2 to O2 at the anode by

253

reaction 6 (Eq. 6) and/or its possible reduction to H2O at the cathode surface by reaction 7 (Eq. 7),

254

and/or the reduction of O2 in the cathodes directly to H2O in a four-electron exchange reaction (Eq. 8).

255

Consequently, the accumulation of H2O2 in the outlet of the ‘filter’ could be smaller than expected

256

(based on the results with a single pair of anode/cathode electrodes).

257

H2O2  O2(g) + 2H+ + 2e-

258

H2O2 + 2H+ + 2e-  2H2O

259

O2(g) + 4H+ + 4e-  2H2O

260

Taking into consideration that the presence of catalytic iron on cathode’s surface enables the

261

implementation of Fenton reactions, it is foreseen that the aforementioned parasitic reactions will be of

262

lesser importance, since most of the H2O2 produced in each pair of anode/cathode will be instantly

263

consumed to generate oxidative species (●OH). This means that each pair of anode/cathode electrodes

264

is likely acting as an autonomous EF cell, being connected in series with other successive EF cells.

(7) (8)

SC

TE D

M AN U

(6)

EP

265

RI PT

249

Fig. 2. Generation of H2O2 as a function of electrolysis time at various applied potentials for (a) DW

267

(Exp. No #1-2), (b) TW (Exp. No #3).

268

AC C

266

269

From Fig. 2 it becomes obvious that the feed water composition significantly affects the

270

electrogeneration of H2O2. For example, at a flow rate of 40 L/h (corresponding to a residence time of

271

4.3 sec) and optimum applied potentials, the average current efficiency of the ‘filter’ was 18.6% and

272

8.3% for the DW and the TW, respectively. This difference can be justified by the higher pH (~7.8) of

11

ACCEPTED MANUSCRIPT the TW (less available protons in Eq. 2), and to the presence of organic (TOC, UV254) and inorganic

274

(HCO3-, Cl-, SO42-) species that may consume or negatively affect the electrogeneration of H2O2 in the

275

‘filter’.

276

The influence of feed water flow rate on H2O2 electrogeneration was studied in the case of TW at the

277

near optimum applied potential of 2 V (Exp. No #4). According to Fig. 3a, reduced flow rates result in

278

larger accumulation of H2O2 in the outlet of the ‘filter’. For example, at a flow rate of 10 L/h the H2O2

279

concentration measured after 180 min of continuous flow (~980 µg/L) was 60% larger than the one

280

measured at 40 L/h (~610 µg/L). At the same time a remarkable increase in current efficiency was

281

observed, with values >13% measured in the case of 10 L/h flow rate, at steady-state conditions (Fig.

282

3b). Obviously, low flow rates (in the range of values studied here) promote the interaction of the

283

electroactive species with the CFm anode/cathode electrodes, thus, enhancing the kinetics of the

284

reactions related to the generation of H2O2 (Eq.1 and Eq. 2). It is noted that negligible differences in the

285

current density where observed when varying the TW flow rate (data not shown here).

M AN U

SC

RI PT

273

TE D

286

Fig. 3. a) H2O2 concentration and b) current efficiency as a function of electrolysis time, for four

288

different TW flow rates (Exp. No #4).

289

EP

287

3.2. DCF adsorption in the EF ‘filter’

291

In this study, DCF adsorption was assessed by continuous flow experiments, during the electrode

292

saturation step applied in all experiments, according to the aforementioned protocol. All filtration runs

293

where performed with the same values of operating parameters, which can have a significant influence

294

on the performance of the adsorption. Specifically, DCF adsorption was recorded for similar volumetric

295

flow rate (40 L/h), effective volume (~49 cm3) and feed solute concentration (~1 mgDCF/L), but for

296

different water matrices (Table 1). Under these conditions, the breakthrough curves for DCF removal

297

are depicted in Fig. 4a.

AC C

290

12

ACCEPTED MANUSCRIPT Breakthrough times, corresponding to CDCF/CDCF,o = 0.10 were found to be 133 and 38.4 min for the

299

‘filter’ operating with three couples of CFm/CFm-nFe electrodes during the treatment of TW (Exp. No

300

#6) and DW (Exp. No #8), respectively. Saturation times (corresponding to CDCF/CDCF,o = 0.90) were

301

found to be 754.6 and 656.6 min, respectively. The differences observed in breakthrough times between

302

the two water matrices are obviously related to their organic and/or inorganic composition (Table S3 in

303

Supplementary Material) which seems to affect the mass transfer and consequently the adsorption of

304

the DCF.

SC

RI PT

298

305

Fig. 4. a) Breakthrough curves, and b) amount of DCF adsorbed per unit mass of electrodes as a

307

function of filtration time ([DCF]feed ~1 mg/L, flow rate 40 L/h).

M AN U

306

308

As can be seen in Fig. 4a, DCF is removed to a larger extent when TW is used, especially during the

310

first two hours of solute filtration. One would expect that the organic and/or inorganic species present

311

in TW would compete with DCF molecules for adsorption sites on the carbonaceous electrodes, thus

312

increasing the diffusion rates of the latter, and consequently decreasing their removal. Obviously, the

313

possible occupation of adsorption sites by the organics of the TW, as evidenced by the decreased TOC

314

in the outlet of the ‘filter’ (data not shown here), enhanced their interactions with the free DCF

315

molecules, thus increasing the contact time of DCF in the ‘filter’, leading to higher percentage removal

316

values. Similar interactions with adsorbed inorganic ions is not possible, since slight differences were

317

observed in the conductivity values recorded in the inlet and the outlet of the ‘filter, which is evidence

318

of a negligible adsorption of ions in the ‘filter’ (data not shown here).

319

When CFm electrodes where used both as anodes and cathodes (reference experiment No #11), the

320

breakthrough times for 10% and 90% removal of DCF were 93 and 476.4 min, respectively. Obviously,

321

in the absence of iron nanoparticles on the cathode electrodes, the available electrode surface area for

322

DCF adsorption is larger, thus increasing the breakthrough time. However, the profile of the

AC C

EP

TE D

309

13

ACCEPTED MANUSCRIPT breakthrough curve generated in this case is steeper, which indicates that a smaller resistance to mass

324

transfer prevails in the adsorption process, resulting in a faster saturation rate.

325

The differences observed in breakthrough times between TW and DW are not reflected in the amount

326

of DCF adsorbed by the electrodes. According to Fig. 4b, the adsorption capacities followed a similar

327

pattern, with the equilibrium values amounting to 35.6±1.1 mgDCF/gCFm. The decreased adsorption

328

capacity reported in this study in comparison to that of the batch tests described previously (70.7

329

mgDCF/gCFm for recirculation of electrolytic solutions of 30 mg/L DCF with a flow rate 3 L/h) (Sklari et

330

al., 2015), was expected, considering the lower feed DCF concentration (1 mg/L) and the higher water

331

flow rate (40 L/h) applied in this study. For the same reasons, the amount of DCF adsorbed in the case

332

of the reference Exp. No #11 (CFm anode/cathode electrodes), was 45.4 mgDCF/gCFm, which is 35.8%

333

smaller than the one obtained in previous batch tests (Sklari et al., 2015).

334

M AN U

SC

RI PT

323

3.3. DCF removal during the electrolysis

336

After the preliminary saturation of the EF ‘filter’ with DCF, the electrolysis step took place at near

337

optimum operating conditions (applied potential of 2V and water superficial velocity of ~0.092 cm/s),

338

by varying the initial DCF concentration (0.02-0.60 mg/L). The EF ‘filter’ performance was assessed in

339

several pilot tests by employing one or multiple cycles of adsorption/oxidation of DCF dissolved in TW

340

or DW.

341

3.3.1. Effect of DCF concentration

342

The temporal variation of DCF degradation/mineralization was studied for three different feed

343

concentrations: 0.034, 0.260 and 0.590 mg/L using DW as feed water (Fig. 5). According to Table 1,

344

the respective TOC concentrations varied from 0.13 to 0.34 mg/L, corresponding to DCF and to other

345

dissolved organics already present in DW. Fig. 5a shows that a sharp decrease in DCF’s concentration

346

is recorded within the first 30 min of electrolysis, with DCF removal efficiency stabilized after 180 min

347

of continuous treatment. As expected, the higher the initial DCF concentration the lower the stabilized

AC C

EP

TE D

335

14

ACCEPTED MANUSCRIPT 348

DCF removal. This is due to the reduced interaction of the dissolved DCF molecules with both the

349

electrodes and the oxidant species produced in the EF ‘filter’ at a given residence time (17.4 sec in all

350

experiments).

RI PT

351 Fig. 5. Temporal variation of a) DCF concentration, b) MCE% and applied DC current (I), c) H2O2

353

concentration, and d) pH and Redox, at different feed DCF and TOC concentrations (in brackets),

354

during the electrolysis step. Experimental conditions: DW, flow rate ~10 L/h, applied voltage 2 V (Exp.

355

No #8 - #10).

SC

352

M AN U

356

According to a previous study (Sklari et al., 2015), the efficiency of the EF ‘filter’ to remove DCF/TOC

358

is linked with two different mechanisms, a) the electrosorption of the feed water organics (DCF and

359

other dissolved organics present in DW or TW) on the surface of the charged electrodes, and b) their

360

heterogeneous electro-Fenton degradation/mineralization by the strongly oxidizing •OH radicals

361

produced during the H2O2 decomposition by the mixed valence iron oxides on the CFm-nFe electrodes

362

(γ-Fe2O3/F3O4 oxides). Considering the relatively small residence time of all dissolved species in the

363

EF ‘filter’, the reduced DCF values measured during the first minutes of operation may be attributed to

364

the enhanced electrosorption (Sklari et al., 2015). On the other hand, EF oxidation may have affected

365

the adsorbed (on the carbon surfaces) organics, as also evidenced by the low H2O2 concentrations

366

measured in the outlet of the ‘filter’ (Fig. 5c), the UV spectra and the respective GC/MS measurements

367

of samples collected at longer electrolysis times (data described below). The suggestion of DCF

368

removal due to electrosorption is also supported by the results of Exp. No #11 (Fig. S1, Supplementary

369

Material), in which CFm electrodes were used both as anode and cathode electrodes. The operation of

370

the ‘filter’ with a low applied direct current voltage (2V in each pair of CFm electrodes) has increased

371

the adsorption rate and capacity of the ‘filter’ probably through the manipulation (i.e. according to

372

process requirements) of the interfacial potential of the conductive CFm electrodes. Considering the

AC C

EP

TE D

357

15

ACCEPTED MANUSCRIPT rather hydrophobic nature of DCF and its rather negative charge (pKa
374

electrosorption is presumed to be induced via π–π dispersive interactions with the two aromatic rings

375

approaching the carbon surface and electrostatic attraction or repulsion forces with the anode or cathode

376

electrodes, respectively. Moreover, the possible oxidation of DCF at the specific applied potential is not

377

evidenced in this case by the UV spectra of the samples collected in the outlet of the ‘filter’ (Fig. S2 in

378

Supplementary Material) (no new peaks observed).

379

The temporal variations of the mineralization efficiencies calculated from Eq. 4 are presented in Fig.

380

5b. Specifically, the MCE% values of the two larger TOC concentrations (Table 1) are presented (Exp.

381

No #9 and #10), due to the limited accuracy of the TOC analyzer in measuring concentrations below

382

0.1 mg/L. The mineralization efficiency measured after the first 30 min of electrolysis is consistent with

383

the concentration decay of DCF (Fig. 5a), since MCE% stabilizes fast at values varying from 8 to 10%,

384

for similar initial TOC concentrations (0.28 - 0.34 mg/L). It is interesting to note that the stability

385

observed in the mineralization efficiency of the ‘filter’ is not in line with the measured current density

386

of the applied DC voltage, and therefore with the electric charge consumed; interestingly, the latter

387

appears to decrease with time, falling after 10 hours of continuous electrolysis by 35-48% (from around

388

900 mA to 470-580 mA). This decrease in electric density was also observed in previous bench scale

389

experiments (Plakas et al., 2013). A likely explanation is the gradual development of gases within the

390

‘filter’ (formation of oxygen on the anodes or CO2 due to organic’s mineralization) that tend to

391

gradually increase the resistance in the system.

SC

M AN U

TE D

EP

AC C

392

RI PT

373

393

3.3.2. Effect of the water matrix

394

According to the preceding discussion, the feed water composition (pH, organic/inorganic content) may

395

affect significantly the performance of the EF ‘filter’ in terms of H2O2 electrogeneration and DCF

396

adsorption on the carbon electrodes. The influence of these effects on the oxidation/mineralization

397

performance of the EF ‘filter’ during the electrolysis of DCF is depicted in Fig. 6, in which the results 16

ACCEPTED MANUSCRIPT of the pilot tests performed with TW (Exp. No #6 and #7) are depicted. Results in Fig. 6a show that an

399

increased DCF initial concentration may not necessarily affect the removal/degradation rate of the

400

pollutant, as in the case of DW (Fig. 5a); however, it appears to have a pronounced effect on TOC

401

mineralization (Fig. 6b). Specifically, MCE% increases from 10 to 20% when treating TW with low

402

concentration of DCF, which is justified by the lower electric charge consumed (decreased DC current,

403

I – Fig. 6b) and the larger TOC reduction observed in the latter case. On the other hand, the negligible

404

effect of DCF feed concentration on solute removal could be explained by the increased interactions

405

between the DCF molecules and the organics that are already adsorbed on the carbon electrodes (pre-

406

saturation step), under positive polarization conditions. In the case of the low feed DCF concentration,

407

the MCE% values are essentially due to the removal of the naturally occurring organics present in the

408

TW (Table S3 in Supplementary Material). This is confirmed by the results of a reference experiment

409

conducted with single TW, in the absence of DCF (Exp. No #5), where TOC reduction varied from 72

410

to 95%, during 10 h of continuous electrolysis (Fig. S3 in Supplementary Material), with the respective

411

MCE% values varying from 16.9% to 19.3%.

412

It is recalled that the calculation of the MCE% values is based on the hypothesis that the mineralization

413

is solely linked to the molecules of DCF. Considering that the natural organics of TW consists of a

414

heterogeneous mixture of organic molecules of relatively large molecular weight (mainly humic

415

substances), the ratio n/m in Eq. (4) may strongly differ from that assumed for DCF (n/m=4.14).

416

Depending on the redox properties of the humic substances, the ratio n/m could be smaller (if humic

417

substances act as electron acceptors) or even larger (if humic substances act as electron donors) than

418

that of the DCF. Moreover, humic substances may participate in redox reactions with the iron oxides on

419

the cathodes, thus altering the number of electrons consumed in the ‘filter’ for TOC mineralization.

420

Therefore, the results presented in Fig. 6b are only indicative of the overall performance of the EF

421

‘filter’ to remove DCF from a real water matrix.

AC C

EP

TE D

M AN U

SC

RI PT

398

17

ACCEPTED MANUSCRIPT Results in Fig. 6c show a reduced accumulation of H2O2 in the outlet of the ‘filter’, as a result of the

423

Fenton reactions occurring therein, or even the decomposition of H2O2 by the adsorbed and/or

424

dissolved organics. The pH of the treated water remains unchanged throughout the electrolysis, whereas

425

its redox potential drops sharply within the first minutes of electrolysis, increasing thereafter to its feed

426

value within 10 h of filter operation (Fig. 6d). Such a behaviour is similar between the TW and DW

427

(Fig. 5d) and is indicative of less oxidative conditions. This means that the dissolved oxygen originally

428

present in the feed water (~7.8 mg/L) is drastically consumed in the ‘filter’ by electrode reactions,

429

including reaction 2, Eq. 2 (oxygen reduction to H2O2).

M AN U

430

SC

RI PT

422

Fig. 6. Temporal variation of a) DCF concentration, b) MCE% and applied DC current (I), c) H2O2

432

concentration, and d) pH and Redox, at different initial DCF and TOC concentrations (in brackets)

433

during the electrolysis. Experimental conditions: TW, flow rate ~10 L/h, applied voltage 2 V (Exp. No

434

#6 and #7).

435

TE D

431

3.3.3. DCF degradation and identification of oxidation products

437

In order to verify that oxidative reactions take place within the ‘filter’, samples collected in the outlet of

438

the ‘filter’ were analyzed by means of UV/Vis spectroscopy and GC/MS. The UV/Vis spectra were

439

recorded in the region from 190 to 600nm, while filtrates, after their preliminary SPE concentration,

440

were analyzed at the GC/MS system operating in full-scan mode. All mass spectra collected (Total Ion

441

Chromatograms – TIC) were analyzed based on the NIST/EPA/NIH ’05 library which suggested

442

possible chemical structures. The purpose of these analyses was not the elucidation of possible DCF

443

degradation pathways, but rather the identification of organic intermediates formed during the

444

electrolysis. Fig. S4 (Supplementary Material) illustrates indicative UV/Vis and GC/MS measurements

445

of samples collected during the electrolysis of TW spiked with DCF at a rather high feed concentration

446

(Exp. No #6). A decreased UV absorption at 276 nm (characteristic of DCF) is reported in parallel with

AC C

EP

436

18

ACCEPTED MANUSCRIPT the formation of new peaks, the intensities of which, decrease also with the electrolysis time. These

448

results, in conjunction to the observed fluctuations in UV absorbance throughout the wavelength range,

449

are indicative of products formed during the Fenton oxidation of DCF and of the natural organics

450

present in the TW. According to the GC/MS measurements (typical full scan chromatograms depicted

451

in Fig. S4b; mass spectra of parent and degradation products depicted in Fig. S5-S9, Supplementary

452

Material), these products are mostly cyclic and aliphatic organics (elution at lower retention times than

453

DCF) with characteristic oxygen groups (carboxylic acids, acrylics), formed possibly by the ●OH attack

454

(or other activate forms of oxygen) on the chlorine-benzene ring of the DCF molecules or the carbon

455

rings of the humic substances present in the TW, the nature of which is unfortunately unknown. It is

456

possible, also, that other organic intermediates (e.g. short-chain acids, etc.) are formed in the process, as

457

evidenced by the residual TOC values measured at longer electrolysis times, although such recalcitrant

458

species cannot be identified by the analytical method employed here (GC/MS).

M AN U

SC

RI PT

447

459

3.3.4. Examination of performance in long-term continuous operating mode

461

In order to test the performance and the stability of the EF filter, TW solutions of relatively high DCF

462

feed concentration (~0.625 mg/L) were treated in multiple filtration/oxidation cycles lasting 80 h (Exp.

463

No #6; Fig. 7). The performance of the filter is shown to be stabilized after the second electrolytic

464

treatment, exhibiting at steady state conditions DCF and TOC removal rates of 85% (Fig. 7a) and 36%

465

(Fig. 7b), respectively. This performance is very promishing, considering that all experiments were

466

performed under conditions considered rather unfavourable for the EF oxidation; i.e. neutral pH, low

467

concentration of dissolved O2, small residence time of the electroactive species in the filter (~14.6 sec),

468

presence of other organics in TW.

AC C

EP

TE D

460

469

19

ACCEPTED MANUSCRIPT 470

Fig. 7. Temporal variation of a) DCF concentration and b) TOC concentration, in three successive

471

adsorption/electrolysis cycles. Experimental conditions: TW, flow rate ~10 L/h, applied voltage 2 V,

472

[DCF]feed ~0.62 mg/L (Exp. No #6).

RI PT

473 It is interesting to note that after the first treatment cycle, the adsorption capacity of the ‘filter’ is

475

reduced, showing also a tendency to desorb DCF. This behaviour is indicative of a decreased

476

regeneration of the carbon electrodes at the applied electrolytic periods. Obviously, a longer

477

electrolysis/oxidation time is needed prior to the operation of the ‘filter’ as adsorbent (under zero

478

potential conditions). This observation dictates the establishment of an operation protocol, in which the

479

EF ‘filter’ works as a simple adsorbent until a specified breakthrough time (e.g. when CDCF/CDCF,0

480

=0.1), after which, the potential can be applied so as to fully regenerate the ‘filter’ through Fenton

481

reactions. The exact time periods required for each step depends on the feed water quality and the

482

concentration of the organic micropollutants. Ongoing R&D on this concept is carried out at CERTH,

483

toward the EF ‘filter’ development and optimization for relatively small-scale applications; e.g.

484

domestic filters for potable water treatment, or similar tasks.

TE D

M AN U

SC

474

485

3.4 Electrode characterization after pilot testing

487

The XRD patterns of the used electrodes show the presence of carbon, titanium oxide and iron oxide in

488

the structure of the electrodes (Fig. S10, Supplementary Material). Titanium oxide is impurity of the

489

initial CFm material, whereas the presence of iron oxide confirms the stability of iron species on the

490

cathode electrode surface after the pilot scale studies regardless of the type of water that is used and the

491

sequence of the electrode in the ‘filter’. This result confirms the sustainability of the purposed electro-

492

Fenton process.

493

The specific surface area of the electrodes was determined using adsorption data points in the relative

494

pressure P/Po range 0.01 to 0.30. The anode electrodes were identified as micro-porous, while the

AC C

EP

486

20

ACCEPTED MANUSCRIPT cathode electrodes displayed a type- IV isotherm, containing also a fraction of meso-pores, while

496

hysteresis was observed indicating the occurrence of capillary condensation in the pores (Fig. S11,

497

Supplementary Material). Using these data, the specific surface area (SBET) of materials used in this

498

study was calculated (Table 2). As expected, the existence of iron oxide at the electrode matrix (CFm-

499

nFe) results in a reduced SBET since iron oxide exhibits lower SBET than CFm, with the former covering

500

part of the CFm surface.

RI PT

495

502

SC

501

Table 2. Specific surface area of electrode materials used in Exp. No #7 and Exp. No #8.

M AN U

503

For Exp. No #7 with TW, the SBET values of anode electrodes are reduced after the test by

505

approximately 10% (Fig. S11, Supplementary Material and Table 2). This reduction is not considered

506

particularly important because the BET theory is valid for mesoporous materials which have SBET

507

values around 250 m2/g (Gregg & Sing, 1982). However, using synthetic water (Exp. No #8), the

508

decrease of anode electrode specific area values is higher than that observed in Exp. No #7. This can be

509

due to the existence of impurities which exist at the surface of electrode and especially at the surface of

510

the 1st pair anode (Fig. 8). The impurities were identified with EDX analysis and are attributed to sulfur

511

species which originated from Na2SO4. The SBET values of cathode electrodes are invariable after the

512

pilot scale tests regardless of the type of water that is used. This remark and the results of ICP

513

measurements (Table 3) prove that the iron species are stable on the cathode electrode surface.

515

EP

AC C

514

TE D

504

Table 3. Iron concentration at the ‘filter’ effluent during Exp. No #7 and Exp. No #8.

516 517

Scanning electron micrographs of the electrode materials used in Exp. No #7 and Exp. No #8 are

518

presented in Fig. 8. The presence of iron oxide is evident in these images, while the initial ratio Fe/C =

519

30% w/w was confirmed using EDX analysis of the electrodes. 21

ACCEPTED MANUSCRIPT 520 521

Fig. 8. Scanning electron micrographs of Exp. No #7 ((a) anode: A and (b) cathode: C) and Exp. No #8

522

((c) anode: A and (d) cathode electrode: C) (magnification 2x103).

RI PT

523 4. Conclusions

525

Following the satisfactory results obtained from bench-scale testing, the performance of a scaled-up

526

version of a novel EF ‘filter’-type device, was assessed at pilot scale for the efficient removal of

527

micropollutants present in drinking water sources; the novel ‘filter’, operating under no addition of

528

oxidants, combines the attributes of pollutants’ adsorption and heterogeneous Fenton oxidation. Pilot

529

tests were carried out with two water types (tap water, deionized water with added Na2SO4), under near

530

optimum operating conditions (applied potential of 2V in each pair of anode/cathode electrodes, and

531

water superficial velocity of ~0.09 cm/s), by varying the feed concentration of the pharmaceutical

532

diclofenac (DCF).

533

The continuous once-through operation of the pilot electro-Fenton ‘filter’ presented herein, exhibited

534

remarkable performance both in terms of process efficiency and stability of the anode/cathode

535

electrodes. Specifically, the data show a rather significant adsorption of the hydrophobic DCF

536

molecules on the carbon felt electrodes (35.6±1.1 mgDCF/gCFm), which are subsequently oxidized under

537

positive polarization conditions due to Fenton reactions taking place in the presence of catalytic iron

538

nanoparticles (γ-Fe2O3/F3O4 oxides) on the cathode electrodes. The effectiveness of the Fenton

539

oxidation and consequently the regeneration of the carbon electrodes depend on the operation time, the

540

feed water quality and the concentration of DCF. For relatively high DCF feed concentration (0.625

541

mg/L) in tap water, the saturated EF ‘filter’ reaches steady state operation after approx. 1 hour,

542

removing more than 80% of the DCF. Even after multiple filtration/oxidation treatment cycles, the

543

novel ‘filter’ achieves (under steady conditions) DCF and TOC removal 85% and 36%, respectively.

544

This performance is considered satisfactory because the EF process takes place under rather

AC C

EP

TE D

M AN U

SC

524

22

ACCEPTED MANUSCRIPT unfavorable conditions, such as neutral pH, low dissolved O2 concentration, low electrical conductivity,

546

and presence of natural organic matter and inorganic ions in tap water. The developed electrodes (anode

547

and cathode) are stable after the pilot scale studies regardless of the type of water used and the sequence

548

of the electrode in the ‘filter’. This result confirms the robustness of the electro-Fenton process

549

regarding the development of novel electrodes: high-performance, durable anodes and cathodes with

550

enhanced electrocatalytic properties.

551

Ongoing R&D on this novel EF ‘filter’ proceeds along two main lines; i.e. pilot studies toward

552

equipment performance optimization and materials’ processing to improve the catalytic cathodes

553

characteristics. Future research will target also to pilot tests under galvanostatic conditions (constant

554

current in each pair of electrodes), as a means of maintaining/improving the current efficiency of the

555

filter during the electrodes regeneration. Finally, an important R&D task that merits further

556

investigation is the implementation of toxicological assessment studies, focusing on the formation of

557

mostly unknown oxidation intermediates (especially when treating real water with dissolved natural

558

organics).

559

TE D

M AN U

SC

RI PT

545

Acknowledgments

561

The authors wish to thank the personnel of Analytical Services Unit of the Chemical Process and

562

Energy Resources Institute (ASU-CPERI) for analytical support. Financial support from the General

563

Secretariat for Research and Technology, Greek Ministry of Education, through the programme EPAN-

564

II/ESPA: “SYNERGASIA” (project 09-SYN-42-630) is gratefully acknowledged.

AC C

565

EP

560

566

References

567

Babuponnusami, A., Muthukumar, K., 2014. A review on Fenton and improvements to the Fenton

568

process for wastewater treatment. J. Env. Chem. Eng. 2 (1), 557–572.

23

ACCEPTED MANUSCRIPT 569

Bocos, E., Alfaya, E., Iglesias, O., Pazos, M., Sanromán, M.Á., 2015. Application of a new sandwich of

570

granular activated and fiber carbon as cathode in the electrochemical advanced oxidation treatment

571

of pharmaceutical effluents. Sep. Purif. Technol. 151, 243-250.

573

Brillas, E., Sirés, I., Oturan, M.A., 2009. Electro-Fenton process and related electrochemical

RI PT

572

technologies based on Fenton’s reaction chemistry. Chem. Rev. 109 (12), 6570−6631.

Brillas, E., Garcia-Segura, S., Skoumal, M., Arias, C., 2010. Electrochemical incineration of diclofenac

575

in neutral aqueous medium by anodic oxidation using Pt and boron-doped diamond anodes.

576

Chemosphere 79, 605−612.

SC

574

El-Ghenymy, A., Cabot, P.L., Centellas, F., Garrido, J.A., Rodríguez, R.M., Arias, C., Brillas, E., 2013.

578

Mineralization of sulfonamide by electro-Fenton and solar photoelectro-Fenton in a pre-pilot plant

579

with Pt/air-diffusion cell. Chemosphere 91, 1324-1331.

M AN U

577

Feng, L., van Hullebusch, E.D., Rodrigo, M.A., Esposito, G., Oturan, M.A., 2013. Removal of residual

581

anti-inflammatory and analgesic pharmaceuticals from aqueous systems by electrochemical

582

advanced oxidation processes. A review. Chem. Eng. J. 228, 944–964.

TE D

580

Garcia-Segura, S., Cavalcanti, E.B., Brillas, E., 2014. Mineralization of the antibiotic chloramphenicol

584

by solar photoelectron-Fenton: From stirred tank reactor to solar pre-pilot plant. Appl. Catal. B:

585

Environ. 144, 588-598.

EP

583

Garcia-Segura, S., Brillas, E., 2016. Combustion of textile monoazo, diazo and triazo dyes by solar

587

photoelectron-Fenton: Decolorization, kinetics and degradation, routes. Appl. Catal. B: Environ.

588

181, 681-691.

589 590

AC C

586

Gregg, S.J., Sing, K.S.W., 1982. Adsorption, Surface Area and Porosity, second ed., Academic Press, London, p. 42-61 & 285.

591

Iglesias, O., Fernández de Dios, M.A., Tavares, T., Sanromán, M.A., Pazos, M., 2015. Heterogeneous

592

electro-Fenton treatment: preparation, characterization and performance in groundwater pesticide

593

removal. J. Ind. Eng. Chem. 27, 276–282. 24

ACCEPTED MANUSCRIPT 594 595

Klassen, N.V., Marchington, D., McGowan, H.C.E., 1994. H2O2 determination by the I3− method and by KMnO4 titration. Anal. Chem. 66 (18), 2921−2925. Oturan, N., Wub, J., Zhang, H., Sharmac, V.K., Oturan, M.A., 2013. Electrocatalytic destruction of the

597

antibiotic tetracycline in aqueous medium by electrochemical advanced oxidation processes: Effect

598

of electrode materials. Appl. Catal. B: Environ. 140–141, 92–97.

RI PT

596

Plakas, K.V., Karabelas, A.J., Sklari, S.D., Zaspalis, V.T., 2013. Toward the development of a novel

600

electro-Fenton system for eliminating toxic organic substances from water. Part 1. In situ generation

601

of hydrogen peroxide. Ind. Eng. Chem. Res. 52, 13948−13956.

SC

599

Sarasidis, V.C., Plakas, K.V., Patsios, S.I., Karabelas, A.J., 2014. Investigation of diclofenac

603

degradation in a continuous photo-catalytic membrane reactor. Influence of operating parameters.

604

Chem. Eng. J. 239, 299−311.

M AN U

602

Sirés, I., Brillas, E., Oturan, M.A., Rodrigo, M.A., Panizza, M., 2014. Electrochemical advanced

606

oxidation processes: today and tomorrow. A review. Environ. Sci. Pollut. Res. 21 (14), 8336–8367.

607

Sklari, S.D., Plakas, K.V., Petsi, P.N., Zaspalis, V.T., Karabelas, A.J., 2015. Toward the development

608

of a novel electro-Fenton system for eliminating toxic organic substances from water. Part 2.

609

Preparation, characterization, and evaluation of iron-impregnated carbon felts as cathodic electrodes.

610

Ind. Eng. Chem. Res. 54, 2059−2073.

EP

TE D

605

Thiam, A., Brillas, E., Garrido, J.A., Rodríguez, R.M., Sirés, I., 2016. Routes for the electrochemical

612

degradation of the artificial food azo-colour Ponceau 4R by advanced oxidation processes. Appl.

613

Catal. B: Environ. 180, 227–236.

614

AC C

611

25

ACCEPTED MANUSCRIPT Table 1

[DCF]feed (mg/L) b -

DW

-

[TOC]feed (mg/L) -

(+) CFm (-) CFm

DW

-

-

(+) CFm (-) CFm

TW

-

-

(+) CFm (-) CFm

TW

-

Flow (L/h) 40

Conductivity (µS/cm) 650

Average pHfeed 4.5

Eapplied (V) 1.4

40

790

4.9

1.4

40

RI PT

#2a #2b #2c #2d #2e #3a #3b #3c #3d #3e #4a #4b #4c

Water type a DW

-

700±50

SC

#1b

Anode/Cathode electrodes -c (+) CFm (-) CFm

4.6±0.1

40

660±10

7.8±0.1

10 20 30

670±5

7.8±0.1

M AN U

Exp. No. #1a

1.0 1.4 1.8 2.2 2.6 1.6 2.0 2.4 2.8 3.2 2.0

(+) CFm TW d 10 695 7.9 0/2 e (-) CFm-nFe (+) CFm #6 0.625 0.95 10 685 8.2 0/2 TW f (-) CFm-nFe (+) CFm #7 TW 0.016 0.56 10 675 7.9 0/2 (-) CFm-nFe (+) CFm #8 DW 0.034 0.13 10 760 5.7 0/2 (-) CFm-nFe (+) CFm #9 0.590 0.34 10 700 4.8 0/2 DW (-) CFm-nFe (+) CFm #10 DW 0.260 0.28 10 730 4.6 0/2 (-) CFm-nFe (+) CFm #11 DW 0.620 0.42 10 750 4.5 0/2 (-) CFm g a DW: Deionized water with added 2.82 mM Na2SO4, TW: Tap water, b Initial DCF concentration at the beginning of the electrolysis (after the saturation of the electrodes with DCF), c Reference experiment of H2O2 electrogeneration in the absence of CFm electrodes, d Removal of TOC included in the TW in the absence of DCF, e Two steps experimental protocol (adsorption at 0V, electrolysis at 2V); f Long-term experiment (80 h), g Reference experiment of DCF/TOC removal with CFm electrodes (no iron on cathodes).

AC C

EP

TE D

#5

ACCEPTED MANUSCRIPT Table 2

SBET (m2/g) Exp. No #7 1005

A1

907

A2

874

A3

928

846

CFm-nFe

800

800

C1

824

760

784

793

797

804

C2

EP

TE D

C3

M AN U

cathode

SC

CFm anode

AC C

Exp. No #8

RI PT

Electrode

1005 826 901

ACCEPTED MANUSCRIPT Table 3

time (min)

Fe concentration (mg/L) Exp.#8

0

0.000

0.099

120

0.000

-

600

0.000

total

0.000

RI PT

Exp.#7

SC

0.089

AC C

EP

TE D

M AN U

0.188

ACCEPTED MANUSCRIPT Figure 1

M AN U

SC

RI PT

(a)

(b)

TE D

7

8 6

2

EP

4 5

3

AC C

1

9

ACCEPTED MANUSCRIPT Figure 2

2400

1.0V 1.4V (without CFm electrodes)

2100

1.4V 1.8V 2.2V 2.6V

1800

1500

1.6V 2.0V 2.4V 2.8V 3.2V

SC

1200

600

300

0 50

100

150

0

TE D

Electrolysis time (min)

EP

0

M AN U

900

AC C

H2O2 (μg/L)

(b)

RI PT

(a)

50

100

150

200

ACCEPTED MANUSCRIPT Figure 3

1200

14

(a) 1000

SC

600

M AN U

400

0 50

100

150

0

TE D

Electrolysis time (min)

EP

0

50

8

4 10L 20L 30L 40L

100

10

6

10L 20L 30L 40L

200

AC C

H2O2 (μg/L)

800

12

150

2

0 200

CE (%)

RI PT

(b)

ACCEPTED MANUSCRIPT Figure 4

1.0

50

(a)

SC

0.6

M AN U

0.4

TW (Exp. No #6) DW (Exp. No #8) DW (Exp. No #11)

0.2

0.0 100

200

300

400

500

600

0

100

EP

TE D

Filtration time (min)

AC C

0

200

300

30

20

TW (Exp. No #6) TW (Exp. No #8) TW (Exp. No #11)

400

40

m

CDCF/CDCF,0

0.8

mgDCF/gCF

RI PT

(b)

500

600

10

0 700

ACCEPTED MANUSCRIPT Figure 5

1.0

24

0.8

1000

(b)

0.034 mgDCF/L 0.26 mgDCF/L 0.59 mgDCF/L

MCE (0.28 mgTOC/L) MCE (0.34 mgTOC/L) I (0.28 mgTOC/L) I (0.34 mgTOC/L)

RI PT

(a)

20

MCE (%)

0.4

12

0.2

4

0.0

pHout/pHin

200 150 100

100

200

300

TE D

0.034 mgDCF/L 0.26 mgDCF/L 0.59 mgDCF/L

50

400

500

AC C

EP

Electrolysis time (min)

600

1.2

1.2

1.0

1.0

0.8

0.8

0.6

0.6

0.4

0.4

pH (0.034 mgDCF/L) Redox (0.034 mgDCF/L)

0.2 0.0

0

100

200

300

400

Electrolysis time (min)

500

600

0.2 0.0

Redoxout/Redoxin

250

H2O2 (μg/L)

0

(d)

300

0

200

M AN U

0

(c)

0

400

SC

8

600

I (mA)

CDCF/CDCF,0

16 0.6

800

ACCEPTED MANUSCRIPT Figure 6

1.0

24

1000

(b)

0.016 mgDCF/L 0.625 mgDCF/L

RI PT

(a)

20

0.8

MCE (%)

CDCF/CDCF,0

0.4

12

0.2

4

0

0.016 mgDCF/L 0.625 mgDCF/L

(c)

pHout/pHin

60

20

100

200

300

400

500

Electrolysis time (min)

EP

0

AC C

0

TE D

40

600

0

(d)

1.0

1.0

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2 pH (0.016 mgDCF/L) Redox (0.016 mgDCF/L)

0.0

0

100

200

300

400

Electrolysis time (min)

500

600

0.0

Redoxout/Redoxin

H2O2 (μg/L)

80

200

M AN U

0.0

400

SC

8

600

I (mA)

MCE (0.56 mgTOC/L) MCE (0.95 mgTOC/L) I (0.56 mgTOC/L) I (0.95 mgTOC/L)

16 0.6

800

ACCEPTED MANUSCRIPT Figure 7

1st cycle

2nd cycle

1st cycle

3rd cycle

3rd cycle

2nd cycle

1.2

2.0 0V 2V

(b)

1.0

SC

0.4

0.2

0.0 1000

2000

3000

4000

0

TE D

Time (min)

EP

0

M AN U

0.6

AC C

CDCF/CDCF,0

0.8

1000

1.6

1.2

0.8

0.4

2000

3000

4000

0.0 5000

TOC/TOC0

RI PT

(a)

ACCEPTED MANUSCRIPT Figure 8

CFm

A1

A2

A3

M AN U

SC

(b)β)

RI PT

(a)

CFm-nFe (c)

(d)

AC C

EP

CFm-nFe

C2

C3

A1

A2

A3

TE D

CFm

C1

C1

C2

C3

ACCEPTED MANUSCRIPT

Research Highlights • Pilot testing of a novel ‘filter’-type electro-Fenton device for water treatment • Efficient removal of diclofenac by combined (electro)sorption-Fenton oxidation at

RI PT

neutral pH • Stable long term operation regarding efficiency and electrodes performance

AC C

EP

TE D

M AN U

SC

• Effect of feed water quality on diclofenac removal/mineralization