Accepted Manuscript

Electrochemical sulfide removal and caustic recovery from spent caustic streams

Eleni Vaiopoulou, Thomas Provijn, Antonin Prévoteau, Ilje Pikaar, Korneel Rabaey

PII: S0043-1354(16)30038-0

DOI: 10.1016/j.watres.2016.01.039

Reference: WR 11793

To appear in: Water Research

Received Date: 6 October 2015

Revised Date: 30 December 2015

Accepted Date: 18 January 2016

Please cite this article as: Vaiopoulou, E., Provijn, T., Prévoteau, A., Pikaar, I., Rabaey, K.,Electrochemical sulfide removal and caustic recovery from spent caustic streams, Water Research(2016), doi: 10.1016/j.watres.2016.01.039.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

1

1

2

3

4

5

Electrochemical sulfide removal and caustic recovery from spent caustic 6

streams 7

Eleni Vaiopoulou1, Thomas Provijn1, Antonin Prévoteau1, Ilje Pikaar2, Korneel Rabaey1* 8

1Laboratory of Microbial Ecology & Technology, Faculty of Bioscience Engineering, University 9

of Ghent; Coupure Links 653, 9000 Ghent, Belgium 10

2School of Civil Engineering, The University of Queensland, Brisbane QLD 4072, Australia 11

*Corresponding author. Tel.: +32 9 264 5985; fax: +32 9 264 6248; 12

E-mail address: Korneel.rabaey@UGent.be 13

14

15

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

2

ABSTRACT 16

Spent caustic streams (SCS) are produced during alkaline scrubbing of sulfide containing sour 17

gases. Conventional methods mainly involve considerable chemical dosing or energy 18

expenditures entailing high cost but limited benefits. Here we propose an electrochemical 19

treatment approach involving anodic sulfide oxidation preferentially to sulfur coupled to 20

cathodic caustic recovery using a two-compartment electrochemical system. Batch experiments 21

showed sulfide removal efficiencies of 84 ± 4% with concomitant 57 ± 4% efficient caustic 22

production in the catholyte at a final concentration of 6.4 ± 0.1 wt% NaOH (1.6 M) at an applied 23

current density of 100 A m-2. Subsequent long-term continuous experiments showed that stable 24

cell voltages (i.e. 2.7 ± 0.1 V) as well as constant sulfide removal efficiencies of 67 ± 5 % at a 25

loading rate of 47 g(S) L-1 h-1 were achieved over a period of 77 days. Caustic was produced at 26

industrially relevant strengths for scrubbing (i.e. 5.1 ± 0.9 wt% NaOH) at current efficiencies of 27

96 ± 2 %. Current density between 0-200 A m-2 and sulfide loading rates of 50-200 g(S) L-1 d-1 28

were tested. The higher the current density the more oxidized the sulfur species produced and the 29

higher the sulfide oxidation. On the contrary, high loading rate resulted in a reduction of sulfide 30

oxidation efficiency. The results obtained in this study together with engineering calculations 31

show that that the proposed process could represent a cost-effective approach for sodium and 32

sulfur recovery from SCS. 33

34

Keywords: electrochemical treatment; spent caustic; sulfide; sodium hydroxide; recovery 35

36

37

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

3

1. Introduction 38

Hydrogen sulfide is a toxic, malodorous and corrosive compound. The removal of sulfide 39

dissolved in wastewater and off-gases from chemical and petrochemical industrial activities 40

represents a considerable cost (Maugans et al., 2010; Paulino and Alfonso, 2012; Veerabhadraiah 41

et al., 2011). The resulting wastewater is known as spent caustic stream (SCS), named after the 42

wasted or used caustic soda. A typical SCS contains 5-12 wt% NaOH and 0.1-4 wt% S2- and can 43

be characterized as sulfidic, cresylic or naphthenic depending on their origin and composition 44

(Alnaizy, R., 2008; Veerabhadraiah et al., 2011). The high pH and sulfide toxicity of SCS limit 45

direct biological treatment, whereas neutralization and dilution may release H2S(g). SCS is a 46

strong reducing agent and has a high oxygen demand (2 mol O2 per mol HS-) (Henshaw and Zhu, 47

2001), resulting in dissolved oxygen depletion. 48

The most commonly used methods to treat SCS involve physico-chemical processes including 49

wet air oxidation and incineration (Alnaizy, R., 2008; Veerabhadraiah et al., 2011), oxidation 50

with oxidant agents addition, precipitation and neutralization/acidification (Tanaka and 51

Takenaka, 1995; Sheu and Weng, 2001), electrochemical (Hariz et al., 2013; Nuñez et al., 2009; 52

Paulino and Alfonso, 2012), biological (De Graaf et al., 2012) or bio-electrochemical processes 53

(Zhang et al., 2013). 54

Despite the variety of available methods to treat SCS, the key limitations that restrict their 55

application are cost, complexity, high consumption of chemicals, safety / handling issues 56

(Alnaizy, R., 2008; Veerabhadraiah et al., 2011) and most importantly the lack of recovered 57

product. Biological processes can alleviate some of these issues, as well as delivering 58

hydrophilic sulfur as recovery product, but require SCS pre-treatment, biomass acclimation and 59

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

4

sludge handling to overcome limitations imposed by high toxicity, pH and COD load. 60

Therefore, there is a general interest in more cost-effective, energy efficient and chemical free 61

methods such as (bio)electrochemical treatment. Several studies showed its feasibility via in situ 62

production of e.g iron (Hariz et al., 2013), hypochlorous acid (Martinie et al., 2006), oxygen or 63

other alike oxidizing agents, and possible coupling of sulfide removal to energy recovery (Kim 64

and Han, 2014; Wei et al., 2012, 2013; Zhang et al., 2013). While these above mentioned studies 65

revealed the potential and can be considered a step forward, they come with some disadvantages 66

including sacrificial anodes, high-energy input to generate oxidizing agents and short life 67

expectancy of materials. 68

Here we propose a novel method that could avoid these concerns. The method relies on the 69

simultaneous anodic oxidization of sulfide coupled to cathodic caustic generation in a two-70

compartment electrochemical cell. In the anode, sulfide is oxidized to elemental sulfur and other 71

sulfur oxyanions, while in the cathode water is reduced to hydroxide anions. In order to maintain 72

electroneutrality, sodium from the anode migrates through a cation exchange membrane (CEM) 73

that separates the two chambers and allows the selective migration of sodium from the anode to 74

the cathode chamber. Key advantages of this approach would be 1) elimination of chemical 75

dosing for sulfide oxidation and thus, less operational, transport, handling and storage cost of 76

potentially hazardous chemicals, which reduces occupational health and safety concerns, 2) 77

recovery of sodium and oxidized sulfur species that can be re-used in situ or be sold, 3) 78

straightforward process design, 4) potentially low energy demand that can be sourced from 79

renewable supply and 5) a neutralized stream with lower salinity and sulfide is generated towards 80

discharge. Therefore, the overall objective of this study is to investigate the feasibility of this 81

approach and identify the key operational aspects. 82

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

5

2. Materials and Methods 83

2.1 Reactor setup and operation 84

2.1.1 Batch-fed reactor 85

The electrochemical cell consisted of two parallel Perspex frames with internal dimensions of 20 86

× 5 × 2 cm separated by a CEM (Fumasep FKB-PK-130, Fumatech GmbH, Germany) according 87

to (Pikaar et al., 2011). A reference electrode (Ag/AgCl (3M KCl), ALS, Japan, + 0.210 V vs. 88

SHE at 25 °C) was placed in the anode compartment. A flattened mesh shaped tantalum-iridium 89

mixed metal oxide (TaO2/IrO2 : 0.65/0.35) coated titanium electrode (Magneto Anodes BV, The 90

Netherlands) with a projected surface area of 100 cm2 was used as anode material. Stainless steel 91

fine mesh (projected surface area of 100 cm2) was used as cathode (mesh width 44 mµ, wire 92

thickness: 33 mµ, Solana, Belgium) and a stainless steel frame was serving as current collector. 93

A spacer (ElectroCell Europe A/S, Tarm, Denmark) was placed between the electrodes and the 94

CEM to prevent membrane contact with the electrodes. The batch electrochemical cell was 95

galvanostatically controlled using a power source (type PL-3003D, Protek) at a current density of 96

100 A m-2. 97

The anolyte consisted of 4 wt% NaOH and 1 wt% Na2S-S simulating a typical SCS. The 98

catholyte was 4 wt% NaOH at the onset of the experiment which ensured sufficient initial 99

conductivity and avoided putative non-electrically driven diffusion of sodium across the CEM. A 100

recirculation flow of 6 L h-1 was applied to obtain sufficient mixing by a peristaltic pump 101

(Watson-Marlow Inc., Massachusetts, US). Masterflex Norprene tubing with an internal diameter 102

of 6 mm was used for both anolyte and catholyte and recirculation lines. H2 produced in the 103

cathode was collected in the cathode effluent bottle. 104

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

6

Figure 1 105

2.1.2 Continuous reactor operation 106

The continuous-mode electrochemical cell (Fig. 1 red lines) was set up as a low-volume cell 107

(internal dimensions of the cell compartments were 7 × 2 × 1 cm with 5 × 2 cm effective 108

membrane area). SCS of the same composition as for batch mode was used as anolyte. The 109

catholyte was initially 4 wt% sodium hydroxide and then distilled water was fed continuously at 110

a flow rate of 82 ± 17 mL d-1 (HRT 4 h). Recirculation flow was set at 2 L h-1 to provide 111

sufficient mixing in both compartments. Peristaltic pumps (Watson-Marlow Inc., Massachusetts, 112

US) and flows were verified daily to assure accuracy and calculate standard deviations. 113

Three different sets of experiments were performed. In the first one, the reactor was run in a 114

continuous mode to determine a long-term operation performance at a fixed current density of 115

100 A m-2. The operation time was run in four periods: 1) day 0-25 and 35-49; the anolyte flow 116

rate was 128 ± 5 mL d-1 and sulfide loading rate (SLR) of 40 ± 3 g(S) L-1 d-1, 2) day 25-35; anode 117

flow rate was decreased to 73 ± 6 mL d-1 and SLR was 26 ± 2 g(S) L-1d-1, 3) day 49-104; batch 118

mode operation at 0.5 A m-2 (reactor remains assembled) and 4) day 104-132; anolyte flow rate 119

was increased to 134 ± 6 mL d-1 and SLR was 47 ± 2 g(S) L-1d-1. The catholyte flow rate was kept 120

constant at 82 ± 17 mL d-1. Along with flow rates, cell voltage, sulfur species, sodium and 121

hydroxide concentrations were monitored on a daily basis. The second experiments assessed the 122

impact of current density on reactor performance and sulfur speciation. Experiments were run at 123

50, 100, 150 and 200 A m-2 and the values presented herein are the ones recorded in triplicates 124

once a new steady state was reached (typically following the 5 times the hydraulic residence time 125

thumb rule and as long as concentrations remained constant). Anolyte and catholyte flow rates 126

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

7

were 121 ± 10 mL d-1 at SLR of 42 ± 4 g(S) L-1d-1 and 73 ± 3 mL d-1 respectively. The third 127

experiments aimed to investigate the impact of SLR - by applying different flow rates ranging 128

from 135 to 530 mL d-1. Experiments were run at 50, 100, 150 and 200 g(S) L-1d-1 at 100 A m-2 129

and the values presented herein are the ones recorded in triplicates once a new steady state was 130

reached. The ratio between the smallest SLR and smallest flow rate is slightly different than the 131

ratio of the highest SLR and highest flow rate due to slight differences in HS- concentration 132

when preparing feeding. An open circuit replicate was run to confirm that sulfide removal and 133

NaOH recovery are only driven by the applied current. To assess whether sulfur species were 134

crossing the membrane, catholyte samples were taken periodically. 135

2.3 Chemical analysis 136

Samples from the reactor were immediately preserved in previously prepared Sulfide 137

Antioxidant Buffer solution prior to analysis as suggested by Keller-Lehmann et al. (2006). 138

Sulfide, sulfite (SO32-) and thiosulfate (S2O3

2-) concentrations were measured by ion 139

chromatography (IC), using an IC930 compact Metrohm IC system (Metrohm, Switzerland), 140

according to Keller-Lehmann et al. (2006). The eluent consists of 3.5 mM Na2CO3 and 3mM 141

NaHCO3 at a flow rate of 0.8 mL min-1. A 0.1 M NaOH solution is used to produce a pH 142

gradient needed for thiosulfate detection in the IC system. Sulfate (SO42-) was determined on an 143

IC761 compact Metrohm IC system (Metrohm, Switzerland) equipped with a Metrosep A Supp 144

5-150 anion exchange column and a conductivity detector, according to Standard Methods 145

(APHA, 1992). The eluent, consisting of 3.2 mM Na2CO3 and 1 mM NaHCO3, had a flow rate of 146

0.7 mL min-1. To measure the polysulfide and elemental sulfur concentrations, all sulfur species 147

were oxidized to sulfate with excess H2O2 as described elsewhere (Dutta et al., 2010). The 148

difference in sulfur equivalent between the sulfate after H2O2 oxidation and other species 149

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

8

measured before H2O2 oxidation (i.e. sulfide, sulfate, thiosulfate and sulfite) was regarded as the 150

sum of polysulfides and elemental sulfur. 151

Alkalinity, as indicator of NaOH concentration, is measured by titrating the cathode effluent with 152

a 1 M HCl solution. Sodium was determined following procedures outlined in Standard Methods 153

(APHA, 1992), using an IC761 compact Metrohm IC system (Metrohm, Switzerland) equipped 154

with Metrosep C6-250/4.0 cation-exchange column and a conductivity meter. The eluent, 155

consisting of 1.7 mM HNO3 and 1.7 mM dipicolinic acid, ran at a flow rate of 0.9 mL min-1. 156

All samples were run in triplicates. Experimental values are provided as the mean +/- standard 157

deviation. 158

2.4 Electrochemical measurements and calculations 159

During the continuous mode operation, cell voltage was monitored every 3 min with a VSP 160

multichannel potentiostat (Princeton Applied Research, France). The electrical resistance of the 161

cells was monitored by the current interrupt technique as described in the Supplementary 162

Material. Current densities are reported with respect to the projected surface area of the anode. 163

Depending on sulfide oxidation product, the oxidation process can involve different amount of 164

electrons per sulfide consumed. Anodic reactions of sulfide oxidation are described elsewhere 165

(Dutta et al., 2010). The coulombic efficiency (CE) for sulfide conversion was calculated 166

assuming only the 2-electron conversion of sulfide to elemental sulfur as described elsewhere 167

(Dutta et al., 2010). CE for sodium recovery is only restricted by membrane transport and was 168

calculated as the ratio of sodium theoretically transferred due to current applied to hydroxyl 169

anions produced (based on the assumption sodium cations and hydroxyl anions are equal, since 170

no other cations are in solution). 171

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

9

3. Results and Discussion 172

3.1 Batch mode reactor 173

At a fixed current density of 100 A m-2, 84 ± 4% of the sulfide was converted (Fig.2) at a 174

coulombic efficiency (CE) of 75 ± 4%. A NaOH solution was recovered at an efficiency over the 175

batch of 57 ± 4% at a CE of 91 ± 5%. The final concentration was 6.4 ± 0.1 wt% NaOH (1.6 M) 176

after 8 h experiments. The pH of the anolyte was initially 13.7 but decreased to 13.2 ± 0.1 at the 177

end of the experiment due to proton production from water splitting in the anode compartment 178

(Fig.2). The remaining high pH cannot be directly discharged, but ensures that residual sulfide 179

remains into solution rather than stripping off (pKa (H2S/HS-) = 6.9). In terms of energy input, 180

cell voltage evolution shows a moderate increase over time from 1.79 to 2.47 V (Fig. S1). This 181

increase is mainly because of sulfide depletion in the anolyte (low conductivity measured); 182

eventually leading to more energy demanding O2 evolution once HS- mass transfer cannot 183

sustain the current applied. This assumption is further based on the low internal resistance of the 184

cell, which was 0.10 ± 0.03 Ω (SI). At this current (1 A), this implies an ohmic drop of 0.10 ± 185

0.03 V accounting for about 4% of the operating voltage. The relatively low ohmic drop is 186

attributed to the high conductivity of both electrolytes. 187

Figure 2 188

The effluent NaOH concentration of 1.6 M (at a fixed current density of 100 A m-2) is considered 189

high, when compared with an electrodialysis SCS treatment that recovered up to 0.15 M NaOH 190

at higher current densities (800 A m-2) (Wei et al., 2012). In the same study, experiments at fixed 191

current density of 300 A m2 resulted in 0.05 M NaOH in 2 h batch experiments at CE of 100%. 192

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

10

The higher CE in this case can be explained by the smaller electrode (7 cm2) and bigger 193

catholyte volume leading to lower cathode concentrations ( > 500 cm2). 194

3.2 Continuous mode reactor 195

The electrochemical cell was operated in a continuous mode for 77 d exhibiting stable 196

performance and reproducibility as confirmed by stable cell voltage, sulfide removal efficiency 197

and restored values when disturbed (Fig.3). Cell voltage was constant at 2.74 ± 0.10 V for a 198

sulfide loading rate (SLR) of 47 ± 2.2 g(S) L-1 h-1 and flow rate of 0.13 L d-1 in the anodic 199

chamber. A SLR decrease from 40 ± 3.2 (day 0-25) to 26 ± 2.2 g(S) L-1 h-1 (day 25-34) resulted in 200

higher cell voltage of 3.08 ± 0.26 V. The cell voltage was restored after the SLR increased again 201

to 47 g(S) L-1 h-1 (day 35 - 45) (Fig.3). The same reproducible behavior was recorded after the 202

batch mode operation period (day 49 - 104) and unexpected single day lab implications (day 47 203

and day 124). Sulfide conversion efficiency for the overall 77 d of continuous operation was 67 204

± 5 % at a CE of 54 ± 7 % for both SLR of 47 and 26 g(S) L-1 h-1. For the different SLR applied, 205

sulfide conversion efficiency was 68 ± 5 % at a CE of 69 ± 3 % for a SLR of 47 g(S) L-1 h-1. 206

When the SLR was decreased to 26 g(S) L-1 h-1, these efficiencies were 71 ± 8% and 45 ± 6% 207

respectively, 208

Figure 3 209

Influent sulfide of 10.8 ± 0.3 g(S) L-1 was oxidized for 30% to thiosulfate (3.2 ± 0.3 g(S) L

-1), then 210

to sulfate (1.4 ± 0.1 g(S) L-1) for 13 %, to polysulfide and elemental sulfur (≈ 0.6 ± 0.7 g(S) L

-1) for 211

5.5 %, while 3.7 ± 0.1 g(S) L-1 of sulfide remained unconverted. These values come from analysis 212

data and do not include the elemental sulfur that has been deposited on the electrode or other 213

parts of the reactor. 214

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

11

Clean NaOH solution was recovered in the catholyte via hydroxide electrogeneration and Na+ 215

migration across the CEM. As Na+ was the only cation present, alkalinity analysis results can be 216

linked to NaOH concentrations. NaOH was produced at high CE (96 ± 2%) with a catholyte 217

effluent concentration of 1.3 ± 0.1 M (5.1 ± 0.4 wt%). 218

The iridium-tantalum oxide coated titanium anode showed stable performance through the whole 219

experimental period of about 5 months, despite the harsh conditions of high sulfide 220

concentrations and high pH. A study applying higher current densities than our range of 0-200 A 221

m-2 reported corrosion and limited lifetime of a mixed iridium-tantalum oxide coated titanium 222

electrode used as anode (Behm & Simonsson, 1997b). Corrosion in this case was attributed to 223

the hydrodynamics, as it occurred mainly on points of turbulence, and at high applied potentials. 224

Electrochemical sulfide oxidation at carbon electrodes can result in an elemental sulfur layer on 225

the anode, causing electrode passivation (Dutta et al., 2008). Here the anode remained fully 226

functional through the whole experimental period due to the different electrode material used 227

(higher oxidizing power) and higher pH (neutral vs highly alkaline). In our study, possible 228

mechanisms of continuous reactivation of the anode could be a de-flaking process by 229

concomitant oxygen generation on the anode surface, sulfur dissolution by polysulfide anions in 230

alkaline condition or further oxidation of sulfur in contact with the electrode to dissolved 231

thiosulfate or sulfate. A joint mechanism could be possible if the inner layer of the elemental 232

sulfur is oxidized to sulfur oxyanions while the remaining sulfur in the outer layer reacts with 233

sulfides to form polysulphide (Behm & Simonsson, 1997a). This appears the most plausible 234

mechanism as there was no observation of flakes or sulfur particles in the reactor or the effluent. 235

3.3.Impact of current density and sulfide loading rate on sulfide oxidation products 236

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

12

Current density affects anode potential and thus, further affects anode sulfide oxidation and the 237

formation of more oxidized sulfur species. The higher the current density, the higher the 238

proportion of sulfur oxyanions such as thiosulfate and sulfate became (Fig.4A). Controls in 239

absence of current (OCV) showed no sulfide oxidation. At 50 A m-2 no sulfate but elemental 240

sulfur and polysulfide were produced, at 100 and 150 A m-2 sulfide oxidation was gradually 241

enhanced and concentrations of thiosulfate and sulfate increased, whereas at 200 A m-2 no steady 242

state was achieved due to constantly increasing cell voltage (Fig.S2). Cell voltage attained the 243

same values on the current density steps backwards (Fig.S2). The same trends were also 244

confirmed for sulfur species production when repeating the experiments applying the reverse 245

scheme of current densities, i.e. from 200 to 0 by steps of 50 A m-2 (Fig.4A, S2). The production 246

of more oxidized sulfur species in higher current density is in agreement with the previously 247

described sulfide oxidation mechanism. The same behaviour in high alkaline conditions has been 248

observed elsewhere (Behm & Simonsson, 1997a; Kim and Han, 2014). 249

Figure 4 250

Sulfide removal efficiency reached a maximum of 86 ± 3 % at 150 A m-2. The lowest recorded 251

sulfide removal was 73 ± 1 % at 50 A m-2. The CE of sulfide oxidation decreased with increasing 252

current density, as expected. A CE higher than 100 % for sulfide oxidation at 50 A m-2 is likely 253

related to polysulfide formation occurring at high pH at the electrode interface (Fig.4A) via the 254

chemical reaction between previously formed S0 and dissolved sulfide (Behm & Simonsson, 255

1997a). This reaction is thermodynamically and kinetically favored at high pH (Steudel and 256

Eckert, 2003), which in our study is above 13. High sulfide loading rates (SLR) due to high 257

influent flow rates resulted in lower overall sulfide removal and less oxidized sulfur species, 258

which further induced elemental sulfur and polysulfide formation at high pH (Fig.4B). Higher 259

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

13

SLR implied lower steady state voltage values and higher CE (Fig.S3). This is an apparent effect 260

as it is mostly linked to the lower anode potential and low mass transfer limiting current for 261

sulfide oxidation. Although it was not possible to differentiate analytically between elemental 262

sulfur and polysulfide, at lower current densities the anode effluent had a characteristic intense 263

yellow color that is indicative of polysulfide (Steudel and Eckert, 2003). Results regarding sulfur 264

speciation, as far as polysulfide and elemental sulfur are concerned, were mostly qualitative 265

based on the color and texture of the anode effluent. 266

The NaOH production rate increased linearly with the current density increase at high CE (98 ± 3 267

%). Effluent concentration of NaOH increased from 2.7 ± 0.1 to 9.8 ± 1.4 wt% at current 268

densities from 50 to 200 A m-2. Theoretically expected NaOH concentration values were 269

calculated according to the current applied to drive the crossing of Na+ from anode to cathode 270

compartment and catholyte flow rate, which was recorded daily (73 ± 3 mL d-1). Production 271

efficiency of NaOH was 76 ± 2 % for current densities of 50 to 150 A m-2, based on the ratio of 272

expected concentrations to actual sodium measurements. These findings imply, as expected, that 273

the current density has within the range tested had no significant effect on CE, whereas caustic 274

strength can have a greater impact as at higher NaOH concentrations, sodium and hydroxyl ion 275

back diffusion might happen. Since the SLR experiments are based on flow rate changes in the 276

anolyte, they have also no impact on the NaOH recovery. Effluent concentration of NaOH ranges 277

from 5.2 ± 0.2 to 6.4 ± 0.8 wt% when SLR ranged from 50 to 150 g(S) L-1 h-1 and current density 278

remained fixed at 100 A m-2. The fluctuations are explained by influent catholyte flow rate 279

fluctuations ranging between 54 and 81 mL d-1. The measured values correspond well with the 280

theoretically predicted values and they are well reproducible on the backwards step from 200 to 281

50 g(S) L-1 h-1. CE is recorded close to 100 ±7 % for all experiments. 282

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

14

3.5 Implications for practice 283

Electrochemical SCS treatment was technologically feasible and resulted in recovery of sodium 284

as a clean sodium hydroxide solution and different sulfide oxidation products, such as elemental 285

sulfur, polysulfide, thiosulfate and sulfate. To implement such process into practice, there are a 286

number of considerations to be addressed as a first step. These include design issues, economic 287

viability, and optimization in terms of anode potential, higher caustic strengths, different 288

membranes to increase stability of the membrane and electrode over longer period of time etc. 289

Further, the presence of other compounds such as organic pollutants need to be considered. 290

Economics of the process would rely on the cost of investment, including materials and 291

engineering cost, the operational cost and possible savings from recovery of chemicals and 292

omission of other costly methods to remove sulfide. An estimation of this cost has been 293

calculated as an indication only (Table S1). Further process optimization and testing with real 294

wastewater is required before up-scaling this technology and calculating in more detail the 295

process cost. Another preliminary techno-economical approach estimated electrochemical NaOH 296

recovery cost from SCS at less than 1 USD per kg NaOH (Wei et al., 2012). 297

4. Conclusions 298

• Electrochemical spent caustic steams treatment has been shown herein to potentially be 299

an economically feasible and environmental friendly process that can recover valuable 300

products from waste. 301

• Sulfide conversion can be driven towards elemental sulfur, polysulfide, thiosulfate and 302

sulfate depending on the current density and sulfide loading rate. 303

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

15

• Besides sulfur species, sodium can be recovered at high coulombic efficiencies and 304

alleviate investment and operational costs. 305

• Electrochemical sulfide oxidation and sodium recovery processes were robust and reactor 306

materials remained unaffected during operation. 307

• Before this concept can be applied for more industrial applications and thus, promote 308

implementation of sustainable technology and resource recovery, rigorous economic 309

assessment, process optimization and testing in field conditions are required. 310

311

312

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

16

References 313

Alnaizy, R., 2008. Economic Analysis for Wet Oxidation Processes for the Treatment of Mixed 314

Refinery Spent Caustic. Environmental Progress 27 (3), 295–301. 315

American Public Health Association (APHA), Greenberg, A.E., Clesceri, L.S. and Eaton, A.D 316

1992. Standard Methods for the Examination of Water and Wastewater, 18th ed. Washington, 317

DC, USA. 318

Behm, M. & Simonsson, D. 1997a. Electrochemical production of polysulfides and sodium 319

hydroxide from white liquor Part I: Experiments with rotating disc and ring-disc electrodes 320

Journal of applied electrochemistry 27, 507-518. 321

Behm, M. & Simonsson, D. 1997b. Electrochemical production of polysulfides and sodium 322

hydroxide from white liquor Part II: Electrolysis in a laboratory scale flow cell. Journal of 323

applied electrochemistry 27, 519-528. 324

Buisman, C.J.N, Vellinga, S.H.J., Janssen, G.H.R. and Dijkman, H. 1999. Biological sulfide 325

production for metal recovery. In TMS Congress 1999, Fundamentals of lead and zinc extraction 326

for metal recovery 1–7. 327

Hariz, I. Ben, Halleb, A., Adhoum, N., Monser, L. 2013. Treatment of petroleum refinery 328

sulfidic spent caustic wastes by electrocoagulation. Separation and Purification Technology 107, 329

150–157. 330

Dutta, P.K., Rabaey, K., Yuan, Z., Keller, J. 2008. Spontaneous electrochemical removal of 331

aqueous sulfide. Water Research 42, 4965 – 4975. 332

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

17

Dutta, P.K., Rabaey, K., Yuan, Z., Rozendal, R.A., Keller, J. 2010. Electrochemical sulfide 333

removal and recovery from paper mill anaerobic treatment effluent Water Research 44, 2563 – 334

2571. 335

Henshaw, P. F., Zhu, W. 2001. Biological conversion of hydrogen sulphide to elemental sulphur 336

in a fixed-film continuous flow photo-reactor Water Research 35 (15), 3605-3610. 337

Keller-Lehmann, B., Corrie, S., Ravn, R., Yuan, Z., Keller, J., 2006. Preservation and 338

simultaneous analysis of relevant soluble sulfur species in sewage samples. In: 2nd International 339

IWA Conference on Sewer Operation and Maintenance, Vienna, Austria. 340

Kim, K. and Han, J.-I., 2014. Performance of direct alkaline sulfide fuel cell without sulfur 341

deposition on anode International Journal of Hydrogen Energy 39, 7142 – 7146. 342

Paulino, J.F. and Alfonso, J. 2012. New Strategies For Treatment and Reuse of Spent Sulfidic 343

Caustic Streams from Petroleum Industry Quim. Nova 35 (7), 1447–1452 344

Pikaar, I., Rozendal, R.A., Yuan, Z., Keller, J., Rabaey, K. 2011. Electrochemical sulfide 345

removal from synthetic and real domestic wastewater at high current densities. Water Research 346

45, 2281 – 2289. 347

Martinie, G.D., Shabrani, F.M.A., Dabrousi, B.O. 2006. Process for treating a Sulfur-containing 348

spent caustic refinery stream using a membrane electrolyzer powered by a fuel cell. Pub. No.: US 349

2006/0254930 A. 350

Maugans, C., Howdeshell, M., De Haan, S. Apr 2010. Update: Spent caustic treatment. 351

Hydrocarbon Processing, 89 (4), 61. 352

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

18

Nuñez P., Hansen, H.K., Rodriguez, N., Guzman, J., Gutierrez C. 2009. Electrochemical 353

Generation of Fenton's Reagent to Treat Spent Caustic Wastewater. Separation Science and 354

Technology 44, 2223-2233. 355

Sheu, S.H. and Weng H.S. 2001. Treatment of olefin plant spent caustic by combination of 356

neutralization and Fenton reaction. Water Research 35 (8), 2017–2021. 357

Steudel, R. and Eckert, B. 2003. Solid Sulfur Allotropes. Topics in Current Chemistry, 230, 1–358

79. 359

Tanaka, N. and Takenaka, K., 1995. Control of hydrogen sulfide and degradation of organic 360

matter by air injection into a wastewater force main. Water Science and Technology 31 (7), 273-361

282. 362

Veerabhadraiah, G., Mallika, N., Jindal, S. 2011. Spent caustic management: Remediation 363

review. Hydrocarbon Processing 364

(http://www.hydrocarbonprocessing.com/Article/2925664/Spent-caustic-management-365

Remediation-review.html). 366

Wei, Y. Li, C., Wang, Y., Zhang, X., Li, Q., Xu, T. 2012. Regenerating sodium hydroxide from 367

the spent caustic by bipolar membrane electrodialysis (BMED) Separation and Purification 368

Technology 86, 49–54. 369

Wei, Y. Wang, Y., Zhang, X., Xu., T. 2013. Comparative study on regenerating sodium 370

hydroxide from the spent caustic by bipolar membrane electrodialysis (BMED) and electro-371

electrodialysis (EED) Separation and Purification Technology 118, 1–5. 372

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

19

Zhang, J. Zhang, B., Tian, C., Ye, Z., Liu, Y., Lei, Z., Huang, W., Feng, C. 2013. Simultaneous 373

sulfide removal and electricity generation with corn stover biomass as co-substrate in microbial 374

fuel cells. Bioresource technology 138, 198–203 375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

20

List of figure legends 391

Fig.1 Schematic diagram of the electrochemical cell. The cell was run in batch (proof-of-concept 392

experiments) and continuous (long term experiments) mode. Anode influent consisted of 4 wt% 393

NaOH and 1 wt% Na2S-S, and cathode influent was 4 wt% NaOH for the batch experiments and 394

distilled water for the continuous experiments. 395

Fig. 2 Sulfide removal in the anode at high pH in 8h batch experiments and CE of 75± 4%. 396

Fig. 3 Cell voltage evolution during the long term experiment. Without considering the 397

interruption period (day 49 - 104), the decrease in loading rate (day 25 - 34), and other single day 398

lab implications (day 47 and day 124), a stable cell voltage of 2.74 ± 0.10 V was maintained for 399

77 days of operation at 100 A m-2. 400

Fig. 4 (A) Impact of current density on sulfur speciation in the anolyte effluent. Influent SLR: 42 401

± 4 g(S) L-1d-1, flow rate: 121 ± 10 mL d-1. Higher current densities result in more oxidized 402

species and higher sulfide oxidation. (B) Effect of different sulfide loading rates 50 - 200 g(S) L-1 403

h-1 by a step of 50 g(S) L-1 h-1 on sulfur speciation at 100 A m-2. Higher loading rates result in less 404

sulfide removal, less oxidized sulfur species and induce polysulfide formation. 405

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

• Electrochemical spent caustic treatment allows economical product recovery

• Current density and sulfide loading rate determine sulfide oxidation products

• Sodium is recovered at high coulombic efficiencies and alleviates costs

• Reactor materials remained unaffected during long term electrochemical operation