Fate and transport of sulfidated nano zerovalent iron (S-nZVI): A field study

Journal Pre-proof Fate and transport of sulfidated nano zerovalent iron (S-nZVI): A field study Ariel Nunez Garcia, Hardiljeet K. Boparai, Cjestmir V. de Boer, Ahmed I.A. Chowdhury, Chris M.D. Kocur, Leanne M. Austrins, Jose Herrera, Denis M. O'Carroll PII:

S0043-1354(19)31093-0

DOI:

https://doi.org/10.1016/j.watres.2019.115319

Reference:

WR 115319

To appear in:

Water Research

Received Date: 16 July 2019 Revised Date:

14 November 2019

Accepted Date: 16 November 2019

Please cite this article as: Garcia, A.N., Boparai, H.K., de Boer, C.V., Chowdhury, A.I.A., Kocur, C.M.D., Austrins, L.M., Herrera, J., O'Carroll, D.M., Fate and transport of sulfidated nano zerovalent iron (SnZVI): A field study, Water Research (2019), doi: https://doi.org/10.1016/j.watres.2019.115319. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. 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. © 2019 Published by Elsevier Ltd.

Fate and Transport of Sulfidated Nano Zerovalent Iron (S-nZVI): A Field Study

Abstract Art CMC-S-nZVI

1

Fate and Transport of Sulfidated Nano Zerovalent Iron (S-nZVI): A Field Study

2 3

Ariel Nunez Garcia,1 Hardiljeet K. Boparai,1,2 Cjestmir V. de Boer,1,3 Ahmed I. A. Chowdhury,1,4

4

Chris M.D. Kocur,1,5 Leanne M. Austrins,6 Jose Herrera,7 Denis M. O’Carroll*1,8

5 6 7 8 9 10 11

1

Department of Civil and Environmental Engineering, Western University, 1151 Richmond Rd., London, Ontario, N6A 5B8, Canada 2

Department of Civil and Mineral Engineering, University of Toronto, 27 King’s College Circle, Toronto, ON M5S 1A1, Canada 3

12 13 14

Netherlands Organization for Applied Research, TNO, Princetonlaan 6, 3584 CB, Utrecht, The Netherlands 4

15 16 17

Institute of Water and Flood Management, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh

18 19 20

5

OHSU-PSU School of Public Health, Oregon Health & Science University, 3181 SW Sam Jackson Park Road, Portland, OR 97239, United States 6

21 22

Arcadis, 28550 Cabot Dr #500, Novi, 48377, MI, US

7

23 24 25

Department of Chemical and Biochemical Engineering, Western University, 1151 Richmond Rd., London, Ontario, N6A 5B8, Canada

26 27 28 29

8

30 31 32 33 34 35 36 37

School of Civil and Environmental Engineering, Connected Water Initiative, University of New South Wales, Sydney NSW 2052, Australia

*Corresponding author School of Civil & Environmental Engineering Civil Engineering Building (H20), Room CE303 Kensington Campus, University of New South Wales Sydney, NSW 2052 Australia Phone: (+61 2) 8071 9822 Email: [email protected]

1

38

Abstract

39

Treatment of nano zerovalent iron (nZVI) with lower valent forms of sulfur compounds

40

(sulfidation) has the potential to increase the selectivity and reactivity of nZVI with target

41

contaminants and to decrease inter-particle aggregation for improving its mobility. These

42

developments help in addressing some of the long-standing challenges associated with nZVI-

43

based remediation treatments and are of great interest for in situ applications. Herein we report

44

results from a field-scale project conducted at a contaminated site. Sulfidated nZVI (S-nZVI)

45

was prepared on site by first synthesizing carboxymethyl cellulose (CMC) stabilized nZVI with

46

sodium borohydride as a reductant and then sulfidating the nZVI suspension by adding sodium

47

dithionite. Transmission electron microscopy (TEM) coupled with energy dispersive X-ray

48

spectroscopy (EDS) of CMC-S-nZVI, from synthesis barrels, confirms the presence of both

49

discrete spherical nZVI-like particles (~ 90 nm) as well as larger irregular structures (~500 nm)

50

comprising of iron sulfides. This CMC-S-nZVI suspension was gravity fed into a sandy material

51

and monitored through multiple multi-level monitoring wells. Samples collected from upstream

52

and downstream wells suggest very good radial and vertical iron distribution. TEM-EDS analysis

53

from the recovered well samples also indicates the presence of both nZVI-like particles as well

54

as the larger flake-like structures, similar to those found in the injected CMC-S-nZVI

55

suspension. This study shows that S-nZVI stabilized with CMC can be safely synthesized on site

56

and is highly mobile and stable in the subsurface, demonstrating for the first time the field

57

applicability of S-nZVI.

58 59

Keywords: Nano Zerovalent Iron, Sulfidation, Dithionite, Subsurface Transport, Field

60

Application, Characterization

2

61

1.

Introduction

62

Nano zerovalent iron (nZVI) is the most commonly applied nanomaterial for water and soil

63

remediation and has been tested for several field-scale applications across the world (O'Carroll et

64

al. 2013, Phenrat et al. 2019, Stefaniuk et al. 2016). Successful remediation projects using nZVI-

65

based technologies rely on the delivery of the nanoparticles to the targeted area and the

66

establishment of a treatment zone. Despite advances in the design of nZVI, key technical

67

challenges remain, limiting its more widespread acceptance as a viable and competitive

68

remediation technology (Fan et al. 2016a, Fan et al. 2016b). These challenges mainly include

69

poor selectivity and low subsurface mobility. Though the development of stabilizers (He and

70

Zhao 2005, He et al. 2007, Phenrat et al. 2010, Schrick et al. 2004, Sun et al. 2007, Tiraferri et al.

71

2008) has led to its increased mobility (Kocur et al. 2013), limited research has been conducted

72

to improve the selectivity of nZVI and to decrease its reaction with natural reductant demand

73

processes in the subsurface. Most notably, these include hydrogen evolution reactions with water

74

(Eq. 1):

75



+2

→

+2

+

(1)

76

Decrease in hydrogen evolution has been observed after treating nZVI with lower valent

77

forms of sulfur compounds (i.e., sulfidation), indicating that its reaction with water is inhibited to

78

some extent (Fan et al. 2016a, Fan et al. 2016b, Nunez Garcia et al. 2016, Rajajayavel and

79

Ghoshal 2015). This has led to improvements in the longevity (Nunez Garcia et al. 2016) and

80

selectivity (Fan et al. 2016a) of nZVI particles. For example, during a sulfidated nZVI (S-nZVI)

81

treatability study, 63% of the iron was still in the zerovalent state after 400 days (Nunez Garcia

82

et al. 2016). Sulfidation of nZVI has increased the removal efficiency of target pollutants such as

83

trichloroethylene (Han and Yan 2016, Kim et al. 2013, Rajajayavel and Ghoshal 2015), 1,23

84

dichloroethane (Nunez Garcia et al. 2016), tetrabromobisphenol (Li et al. 2016), 4-nitrophenol

85

(Tang et al. 2016), diclofenac (Song et al. 2017), and metal ions (Cumbal et al. 2015, Fan et al.

86

2013, Su et al. 2015). Sulfidation methods can be classified as aqueous-aqueous and aqueous-

87

solid, depending on when the sulfur compound is introduced to the synthesis solution (Fan et al.

88

2017, Han and Yan 2016). These methods have been reported to yield particles with varying

89

physico-chemical and structural properties but similar reactivity in dechlorinating TCE (Han and

90

Yan 2016). It has also been reported that sulfidation decreases magnetic attractions between the

91

particles which results in decreased aggregation and sedimentation (Song et al. 2017, Su et al.

92

2015). Among the various sulfidation precursors tested so far, sodium dithionite (Na2S2O4) has

93

been widely used in other field-scale remediation applications such as In Situ Redox

94

Manipulation (ISRM) (Amonette et al. 1994).

95

Despite the increasing number of laboratory-based S-nZVI studies, no study has yet

96

investigated field-scale S-nZVI transport. A number of studies have reported success related to

97

the injectability and mobility of traditional polymer-coated nZVI at the field scale (Bennett et al.

98

2010, He et al. 2010, Johnson et al. 2013, Kocur et al. 2014), however, there is a need to assess

99

the field applicability of emerging reactive formulations, such as S-nZVI. This study presents a

100

field-scale demonstration showing the subsurface mobility of carboxymethyl cellulose (CMC)

101

stabilized and Na2S2O4 doped S-nZVI suspension. The specific objectives of this study were to:

102

1) scale up and develop an on-site field-scale CMC-S-nZVI synthesis method, 2) investigate

103

changes to in situ geochemistry following CMC-S-nZVI injection, 3) quantify in situ CMC-S-

104

nZVI transport, and 4) characterize field-synthesized CMC-S-nZVI suspension before injection

105

and in the multi-level monitoring wells after injection. This study builds upon our previous

106

understanding of nZVI field studies (Chowdhury et al. 2015, He et al. 2010, Kocur et al. 2014)

4

107

by investigating the horizontal as well as vertical distribution of the injected CMC-S-nZVI

108

suspension.

109 110

2.

Materials and Methods

111

2.1

Site Description

112

Field work was conducted in Sarnia, Ontario at a site adjacent to a demolished chlorinated

113

solvents production facility. During operation, the study area was home to various utilities and

114

sewers, some of which might have been the cause of underground spillage. Groundwater at the

115

site is contaminated with a mixture of chlorinated solvents, including tetrachloroethene (PCE),

116

trichloroethene (TCE), carbon tetrachloride (CT), and chloroform (CF). The presence of dense

117

non-aqueous phase liquid (DNAPL) was also visually confirmed between 4 and 5 m below

118

ground surface (bgs). After operation, the infrastructure was covered with a porous, sandy

119

material, stretching east to west (Fig. 1), and capped with an engineered clay layer. Hence

120

borehole logs, advanced with an AMS PowerProbe drill rig, revealed three major sections

121

composed of a clayey fill at the top, non-native sandy material in the middle, and the native clay

122

at the bottom of the geologic system. These sections are not continuous, containing in some

123

instances traces of gravel, brick and wood fragments. The underlaying natural geologic unit is

124

composed of brown weathered clay (until 6.7 m bgs), followed by a nearly impervious

125

unweathered clay with hydraulic conductivities of 2.1×10-9 and 1.6×10-10 m s-1 (Kocur et al

126

2014). The water table was measured to be ~1.5 m bgs throughout the study area.

127 128

2.2

Background Conditions

129

A reducing environment was already prevalent within the targeted depth of 4.12 - 4.42 m

130

bgs. Oxidation-reduction potential (ORP) values ranged between -5 mV and -72.3 mV and pH 5

131

was near-neutral (7.0 ± 0.56). Daughter products from PCE and CT dechlorination (i.e.,

132

dichloroethenes (DCEs) and dichloromethane (DCM)) were present in background samples,

133

suggesting the occurrence of natural attenuation or the influence of previous CMC-nZVI

134

injection performed at nearby wells in November 2010 (Kocur et al. 2014). Background

135

conditions might have been impacted by this injection. However, due to the locations and layout

136

of the current monitoring wells, the past field trial is not considered to have affected the current

137

CMC-S-nZVI injection (conducted in November 2014). Summarized data for the background

138

geo-chemical conditions are shown in Table S1. Total iron and boron concentrations ranged from

139

0.06 to 0.17 and 0.25 to 0.36 mmol L-1, respectively, in the background samples. Injected iron

140

and boron concentrations were greater than 100 times of the background concentrations;

141

therefore, a background correction was not applied to the total iron and boron concentrations.

142 143

2.3

Installation of Monitoring Wells

144

Eight multi-level monitoring wells (MWs) and one injection well (NIW) were installed.

145

Each multi-level MW consisted of seven color-coded intervals 0.305 m apart, with each having a

146

screen length of 0.127 m (Fig. 1). The screens were installed within a range of 2.9 to 4.9 m bgs.

147

During drilling, well logs were obtained for transects corresponding to the wells NA3, NIW,

148

NA1, and NB2. NA4 could not be logged but this well is known to be sandy backfill due to its

149

proximity to a sump that was emplaced in 2010. Fig. 1 shows a plan and a cross-sectional view

150

of well locations including a legend denoting the color codes for multiple well levels. Six

151

intervals, from Black to White, were monitored for NB1 and NB2. However, groundwater

152

monitoring targeted only depths between 4.12 and 4.42 m bgs which correspond to the blue and

153

white intervals for most of the other wells (Fig. 1). Locations of greater depths were not sampled

6

154

(e.g., the red interval, not shown in Fig. 1, located at 4.73 m bgs) due to their close proximity to

155

the source zone. This configuration was selected to maximize capturing of horizontal as well as

156

vertical extent of CMC-S-nZVI breakthrough, downgradient of the injection well.

157 158

2.4

CMC-S-nZVI Synthesis and Injection

159

620 L of S-nZVI, stabilized with ~0.77% weight/volume CMC, were synthesized on site in

160

four distinct batches (155 L each). The mass and volume of each reagent used for the synthesis is

161

reported in Table S2. All reagents were dissolved in deoxygenated, deionized water. nZVI

162

suspension was prepared by reducing ferrous sulfate (A&K Petrochem Ind. Ltd., Vaughn, ON)

163

with sodium borohydride (GFS Chemicals Inc., Columbus, OH) in a process modified and

164

optimized for field applications (Bennett et al. 2010, He et al. 2010, Johnson et al. 2013, Kocur

165

et al. 2014). Anoxic conditions were maintained by continuously purging the solutions and the

166

headspace with high purity nitrogen gas (Praxair Canada Inc., Sarnia, ON). The freshly

167

synthesized nZVI suspension was then treated with Na2S2O4 (Alfa Aesar, Ward Hill, MA) to

168

yield final dithionite and nZVI concentrations of 22 mM and 18 mM (1 g L-1), respectively.

169

The Fe0 content of the field synthesized batches was tested using acid digestion with

170

hydrochloric acid (1 ml 32% HCl), as reported previously (Nunez Garcia et al. 2016). Before

171

sulfidation, the Fe0 content was ~85.0% of the theoretical total iron concentration. After

172

sulfidation, acid digestion of CMC-S-nZVI formed a cloudy suspension and no H2 formation was

173

observed (Fig. S1). Excess dithionite and its decomposition products (e.g., thiosulfate) remain in

174

the CMC-S-nZVI suspension and cannot be decanted out during acid digestion. Thiosulfate

175

consumes acid to form a cloudy/white suspension of colloidal sulfur (Eq. 2) (Johnston and

176

McAmish 1973).

7

177

+

→

( )

+

(2)

178

Therefore, quantification of the Fe0 content was not possible for the CMC-S-nZVI solution

179

(aqueous-solid sulfidation with dithionite), likely due to the presence of thiosulfate or other

180

dithionite decomposition products.

181

Large scale on-site synthesis of S-nZVI with dithionite can present unique health and

182

safety (H&S) challenges and considerations. Though dithionite decomposition products can

183

generate hydrogen sulfide (H2S) (de Carvalho and Schwedt 2001), it has never been reported

184

during the ISRM applications in the field. However, during a preliminary CMC-S-nZVI

185

synthesis trial for the current study, H2S (6-7 ppm) was detected near the vicinity of the synthesis

186

barrel after adding dithionite, although not near the breathing zone. Generation and off-gassing

187

of H2S might have come from the reaction of dithionite decomposition products with the excess

188

hydrogen gas (H2) formed from the borohydride hydrolysis during the nZVI synthesis.

189

Borohydride is typically added in stoichiometric excess (four times the amount needed) for rapid

190

and uniform growth of nZVI particles (Zhang and Elliott 2006). To ensure sufficient time for the

191

nucleation of nZVI particles and to allow the dissipation of H2 gas to minimize the H2S

192

formation during the on-site synthesis, the nZVI suspension was continuously mixed for about an

193

hour before adding dithionite. The final CMC-S-nZVI suspension was mixed for at least one

194

hour before injecting into the subsurface. Because of the potential of H2S generation in large

195

quantities during the aqueous-aqueous sulfidation process (as a result of direct reaction with

196

borohydride), the aqueous-solid method is recommended as the preferred sulfidation approach

197

for the on-site synthesis of CMC-S-nZVI when using dithionite. CMC-S-nZVI suspension was

198

then injected by gravity over 16 hours, maintaining a constant head in the injection well.

199

Injection rates ranged between 0.85-1 L min-1. 8

200

During injection, the hydraulic gradient was controlled for 32 hours by using two

201

recirculation wells (3.91 m downstream and 15.8 m upstream of injection well, Fig. 1) to

202

increase the advective flux throughout the study area. The induced gradient between the

203

recirculation wells was 0.088, compared to a natural gradient of 0.009.

204 205

2.5

Sample Collection and Analytical Methodology

206

To allow for the simultaneous sampling of all selected wells and intervals, a multiple-port

207

set up was constructed. Quality assurance / quality control field protocols were followed

208

including notes recording observations in the field, using dedicated and decontaminated

209

sampling equipment, and minimizing the aeration during sample collection. All samples were

210

collected in pre-cleaned laboratory supplied bottles with suitable preservatives. All bottles were

211

clearly labeled with a designated sample identifier number, analytical parameters, and date and

212

time of sampling. The samples were immediately stored in insulated coolers with ice packs to

213

maintain low temperature and shipped to the laboratory as soon as possible, not exceeding the

214

recommended maximum holding times.

215

Geochemical parameters, including ORP and pH, were measured on-site using a water

216

quality analyzer (YSI 556 MPS, Yellow Spring, OH). Samples for total iron, sulfur, and boron

217

were digested and diluted with hydrochloric acid or nitric acid and then analyzed using

218

inductively coupled plasma-optical emission spectroscopy (Varian Vista-Pro Axial, Santa Clara,

219

CA). Analysis for sulfate was performed using a high-performance liquid chromatograph

220

equipped with a conductivity detector (Model 432, Waters, Milford, MA), a 4.6 × 50 cm IC-Pak

221

Anion column (#Wat007355) and 12% acetonitrile in water eluent. Colloidal stability of CMC-

222

nZVI suspension was determined for the samples collected before and after addition of

9

223

dithionite, using a UV-Vis spectrophotometer (Helios Alpha, Thermo-Fischer, Waltham, MA).

224

Measurements were taken at 508 nm for 88 hours at 10 min intervals. It should be noted that all

225

the samples were vacuum sealed immediately after collection, to minimize oxidation during

226

transportation to the laboratory and storage before analysis (6-24 hours). The UV-Vis absorption

227

spectra were also obtained for CMC-nZVI and CMC-S-nZVI suspensions from synthesis batches

228

and for selected MW samples in the wavelength range of 200-900 nm. Zeta (ζ) potential and

229

effective hydrodynamic diameter (quantified by Dynamic Light Scattering (DLS)) for the

230

samples from synthesis batches and selected MWs were determined using a Zeta Plus particle

231

analyzer (BIC, Brookhaven, Holtsville, NY) and Zeta Plus software.

232

To determine the particle size and morphology of the nanoparticles, transmission electron

233

microscopy (TEM) analysis was conducted in both bright (Philips CM10 TEM, Philips

234

Electronics, Eindhoven, Netherlands) as well as dark (FEI Titan 80-300 TEM, FEI Technologies

235

Inc., Oregon, USA) field modes along with selected area electron diffraction (SAED). Samples

236

were prepared by diluting the nanoparticle suspensions with deoxygenated water and then a drop

237

of the diluted sample was dried on a 400 mesh Formvar/Carbon copper grid (Tedpella Inc.,

238

Redding, CA) in the anaerobic glove box. Elemental composition of these samples was

239

determined by energy dispersive x-ray spectroscopy (EDS) using the INCA detector (Oxford

240

Instruments, Abingdon, UK) attached to the FEI Titan TEM.

241 242

3.

Results and Discussion

243

3.1

Horizontal Mobility

244

The CMC-S-nZVI suspension was quite mobile with significant transport to NA4-Blue and

245

NA3-White, 2.7 m downgradient and 1.71 m upgradient from the injection well, respectively, at 10

246

the first sampling time of 4.75 hours (Fig. S2). With time, increased iron concentrations were

247

also found in two other downgradient monitoring wells at a distance of 0.86 m (NB1-White) and

248

0.91 m (NC1-White) from the injection well. The highest iron concentration was detected in

249

NB1-White at 18 hours after the start of injection, reaching 1.31 mmol L-1 (8.8% of the injected

250

14.9 mmol L-1) (Fig. 2a & Fig. S2).

251

In these four wells, sulfate, sulfur and boron concentrations often followed a similar trend

252

as iron (Fig. 2c & Fig. S3a-b). Notable ‘upgradient’ migration of the CMC-S-nZVI suspension to

253

NA3-White might have occurred due to a localized gradient caused by the injection well. Higher

254

iron concentrations at NB1-White, NA4-Blue, NA3-White, and NC1-White are also clearly

255

illustrated in Fig. 3 where all the sampled wells were compared for their Fe to B molar ratios

256

(Fe/B) against the Fe/B ratio (~0.39) of injected CMC-S-nZVI suspension. Both boron and iron

257

would undergo dispersion, diffusion, and dilution but only iron particles are presumed to be

258

removed from the aqueous phase due to filtration. Boron is stable in aqueous environments and

259

can be found as boric acid (H3BO3) and borate anion ([BO3]3-) (Leenhouts et al. 1998).

260

Uncharged H3BO3 is the dominant species found in most of the natural water systems and its

261

surface adsorption is deemed unlikely due to the direct competition with water for available

262

surface sites (Quast et al. 2006). It has been shown that boron is conserved during groundwater

263

transport and can be considered as a conservative tracer under many conditions (Leenhouts et al.

264

1998, Quast et al. 2006). However, borate ions are found to precipitate on the outer layer of the

265

nZVI during its synthesis with borohydride (Nurmi et al. 2005). It is possible that some boron

266

might have retained in this manner and the boron concentrations reported herein are an

267

underestimation. For this reason, boron is rather operationally defined as a conservative tracer in

268

this study. Therefore, deviation from the calculated Fe/B ratio, represented by a straight line in

11

269

Fig. 3, corresponds to the extent of retention of iron particles during transport to each well. Fig. 3

270

shows that approximately 50% of the iron particles were retained during subsurface transport.

271

Delivery of iron nanoparticles to greater distances have been previously reported for various

272

nZVI formulations. For example, Busch et al. (2015) injected 1.2 kg of Carbon-Iron-Colloids

273

(CIC), an activated carbon supported treatment containing 15% Fe0 by weight and suspended in

274

CMC. Through visual observations and gravimetric analysis, it was estimated that 12.5% of the

275

mass injected was delivered to the extraction well located at a distance of 5.3 m, with a peak total

276

iron concentration of 0.013 mmol L-1 (0.74 mg L-1). However, mobility results from the current

277

study can be considered an improvement when compared to previous traditional CMC-nZVI

278

gravity-fed injections in sandy media. For example, CMC-nZVI was detected only 1 m

279

downstream of another injection well in a previous trial at the current site (Kocur et al. 2014).

280

Another study conducted in an engineered aquifer reported only 4% of the total injected iron at a

281

well 1 m downstream (Johnson et al. 2013). He et al. (2010) observed a normalized peak total

282

iron concentration of 15% and 3% in monitoring wells located 1.5 and 3 m downstream of the

283

injection well, respectively. In other studies, indirect measurements have been employed to

284

confirm the breakthrough of iron. Using total solids (TS) and suspended solids (SS) as an

285

indicator for nZVI, Wei et al. (2010) reported travel distances to 3 m, with minor changes at the

286

furthest monitoring distance of 5 m. One reason for improved mobility of CMC-S-nZVI could be

287

the reduced inter-particle magnetic attractions that would otherwise lead to sedimentation

288

(discussed in Section 3.4.5).Normalizing iron breakthrough to boron (obtained by calculating the

289

area under the C C0-1 vs time curve) enables quantification of the extent of CMC-S-nZVI particle

290

retention. 53.4% and 55.9% particle breakthrough was observed at NA4-Blue and NA3-White

291

respectively, the wells further upstream and downstream (Fig. S4). Good particle mobility was

12

292

also observed at the other monitoring locations. For example, normalized breakthrough of 46.9%

293

and 30.1% was quantified at NB1-White and NC1-White, respectively. Using a similar

294

normalization approach, He et al. (2010) reported 37.4% iron breakthrough in a well 1.5 m away

295

from the injection point. Kocur et al. (2014) reported a peak breakthrough of 75% when

296

normalized to a tracer, before decreasing to 50% for most of the injection period in a well 1 m

297

downstream of the injection well. This analysis should be used cautiously when both iron and

298

boron are present in low concentrations. For example, NB2-White would seem to yield the

299

highest transport, with 73.6% breakthrough based on the normalized Fe/B areas (Fig. S4), even

300

though concentrations of iron did not exceed 1.4% (0.2 mmol L-1) of the injected solution (Fig.

301

S2). Data for NA2-Blue, NA1-White, and NB3-White are not discussed here due to their low

302

iron and boron concentrations.

303

CMC-S-nZVI transport was also quantified at 3 and 17 days after injection. Temporal

304

changes in iron concentration were sometimes non-monotonic. Following the cessation of

305

injection, iron concentrations decreased for NA3-White, NA4-Blue, and NC1-White on day 3

306

(Fig. 2b). Similarly, the iron concentrations also decreased for NB2-Green, NB2-Blue, and all

307

the intervals of NB1 (Table S3). However, on day 17, concentrations remained relatively

308

constant for NA3-White, NC1-White, and NB2-Blue; increased for NA4-Blue, NB2-Green, and

309

NB2-Clear; and decreased for all the intervals of NB1 from Black to White (Fig. 2b and Table

310

S3). Increase in concentration from day 3 to day 17 at NA4-Blue may be attributed to the

311

preferential transport of iron from the injection well, which still had retained high concentration

312

of CMC-S-nZVI suspension, under a natural gradient. It is also noted that during injection,

313

CMC-S-nZVI reached NA4-Blue before the other wells (Fig. 2 & Fig. S2). Given that NA4-Blue

314

is the farthest monitoring well (Fig. 1a), it is suspected that site heterogeneities, including

13

315

preferential flow paths, allowed the CMC-S-nZVI suspension to initially bypass NC1-White and

316

NB2-White. 2D laboratory studies (Phenrat et al. 2010) as well as field injections (Bennett et al.

317

2010, Kocur et al. 2014) have reported that polymer-modified nZVI suspensions preferentially

318

travel through the more conductive hydraulic pathways. The sharp decrease on day 17 at NB1

319

(all intervals), which is closest to the injection well, is likely affected by similar processes,

320

cessation of the CMC-S-nZVI source as well as by CMC-S-nZVI particle deposition on the

321

porous media (Kocur et al. 2014). It is important to note that the relatively high iron

322

concentrations on day 3 represent a significant improvement in the stability of the particles when

323

compared to previous field trials. For example, Kocur et al. (2014) reported a decrease in the

324

normalized iron concentration from 50% during injection to 7.8% at 16 hours after the injection.

325 326

3.2

Vertical Migration and Visual Observations

327

Concentrations of both iron and boron increased with depth in the NB1 well, with the

328

highest recorded in the direct pathway of the injected suspension (i.e., 4.4 - 4.5 m bgs) (Fig. 4a-

329

b). The color of the NB1-White groundwater sample was found to be dark black (image not

330

shown), further supporting the highest concentrations of iron at this location. The presence of

331

CMC-S-nZVI suspension was also clearly visible in the upper levels of NB1 (Fig. S5b),

332

indicating that particles travelled vertically (~1.7 m), up to the Black interval which is 2.9 m bgs.

333

For NB2, located 1.78 m from the injection well, the highest concentrations were measured at the

334

Blue interval (i.e., 4.1 - 4.2 m bgs) at 21.25 hours, also visually indicated by the dark black color

335

of the NB2-Blue groundwater sample (Fig. 4c-d & Fig. S5b). The metal concentrations were

336

lower in the Clear interval (just above the Blue interval) as compared to the NB2-Green, with no

337

significant change in the iron concentrations at the Yellow and Black intervals (Fig. S5c). A

14

338

similar trend was observed visually where NB2-Green has a darker black color than NB2-Clear

339

(Fig. S5b). This could be due to the subsurface heterogeneity and preferential flow paths. In

340

contrast, significant increase in boron concentrations was quantified at the Yellow and Black

341

intervals of the NB2 well. The presence of boron at the upper most intervals suggests that iron

342

particles were retained during vertical transport of the injected solution to these sampling points.

343

Though iron concentrations were not measured for all the levels for NA4, the dark black color of

344

groundwater samples indicates that the CMC-S-nZVI suspension had travelled vertically

345

upwards even to the uppermost level of this well (Fig. S5b). Similarly, the upward migration of

346

CMC-S-nZVI has been visually noticed for NC1 and NA3 up to Green and Clear intervals,

347

respectively. Other studies have also reported vertical migration of ZVI/nZVI (Quinn et al. 2005,

348

Velimirovic et al. 2014).

349

Particles also remained in suspension at the Blue and Clear levels of NB1 at day 3 and day

350

17 (Fig. S5c). The presence of injected suspension is also clearly visible in NA4-Blue, NC1-

351

White, NB1-White, NB3-White, and NB3-Blue on day 17 (Fig. S5d). In fact, suspended particles

352

remained in the injection well for months following the injection period, as observed from a

353

sample collected at 196 days (Fig. S5e) having 13.8% of the injected iron.

354 355

3.3

Changes in Groundwater pH and ORP

356

pH did not change significantly at any of the monitoring wells and remained near neutral

357

(7.0 ± 1.0) throughout the injection period (Fig. S6). Though pH is expected to increase due to

358

the corrosion of iron by water, other nZVI field studies have also reported less than anticipated

359

increases (Kocur et al. 2014, Wei et al. 2010). This suggests that dilution of the injected

360

suspension and the buffering capacity of the subsurface media would moderate the effect of

15

361

CMC-S-nZVI on the groundwater pH. Another reason could also be the limited corrosion of S-

362

nZVI by water (He et al. 2018, Rajajayavel and Ghoshal 2015) as the negatively charged S atoms

363

on the nZVI surface would weaken its interaction with the H2O molecules by repelling the O

364

atoms of H2O molecules (Gu et al. 2019).

365

ORP was measured for discrete samples using a conventional platinum (Pt) electrode.

366

During CMC-S-nZVI injection, a sharp drop in ORP was observed for NA4-Blue, NA3-White,

367

and NB2-White at t = 4h (Fig. S7) which aligned with the arrival of CMC-S-nZVI suspension at

368

these locations (Fig. S2). There was also a gradual decrease in the ORP for NB1-White and

369

NC1-White. The lowest ORP (-175 mV) was recorded for NA4-Blue, six hours after injection.

370

This was followed by NA3-White at -155 mV. Due to time constraints in the field, it was not

371

possible to allow for enough equilibration time for all the ORP measurements. Therefore, for

372

some cases, including NB1-White, measurements were recorded within 5-10 minutes and actual

373

ORP values might be much lower than reported here. Decrease in ORP was concurrent with the

374

iron and sulfate breakthrough, as shown in Fig. S8 for NB1-White. These results indicate a

375

significant influence of the CMC-S-nZVI injection on the aquifer geochemistry. However, the

376

decrease in the ORP was not as noticeable as reported in the previous nZVI field studies

377

(Johnson et al. 2013, Kocur et al. 2014, Wei et al. 2010). Changes in ORP are a function of the

378

redox couples contributing to the measured mixed potentials (Emix) in the system. For nZVI

379

suspensions, the major redox couples are H2/H+ and dissolved Fe2+/Fe3+ species (Shi et al. 2015).

380

Past research has reported lower H2 evolution for S-nZVI during its corrosion by water (Fan et

381

al. 2016a, Rajajayavel and Ghoshal 2015). Thus, the relatively mild decrease in ORP in the

382

current study might be due to the lower generation of H2 in the CMC-S-nZVI treatment as

383

compared to the pristine nZVI. The Fe2+ speciation and concentrations would also be different

16

384

for the (CMC-)S-nZVI treatment compared to pristine nZVI (Nunez Garcia et al. 2016,

385

Rajajayavel and Ghoshal 2015). Moreover, nZVI adsorption on the electrode and the type of

386

electrode used also play a significant role in the ORP measurements (Johnson et al. 2013).

387 388

3.4

Characterization

389

3.4.1 TEM-EDS

390

Detailed particle characterization can be used to help elucidate operative transport

391

processes as well as to evaluate changes in nZVI particles due to sulfidation and subsurface

392

transport.

393

TEM-EDS analysis of CMC-nZVI (before sulfidation) suspension from synthesis batch

394

shows that particles were primarily small discrete spheres and mainly composed of iron (Fig.

395

S9). However, larger irregular structures representing iron (oxy)(hydr)oxides were also present.

396

Though the synthesis vessels were constantly purged with nitrogen, some level of oxidation is to

397

be expected during the field synthesis. The SAED pattern for CMC-nZVI consists of diffused

398

rings indicating its amorphous structure.

399

TEM micrographs of CMC-S-nZVI from the synthesis batches suggest that particles with

400

two different morphologies were present following the sulfidation process (Figs. S10-S11). Most

401

of the particles were small discrete spheres that resembled the nZVI particles from the

402

unsulfidated nZVI suspension (Fig. S9a). These particles, with an average size of 90 ± 13 nm (n

403

= 82), were also found to be similar in size to those previously reported for the field-synthesized

404

nZVI (Kocur et al. 2014). Han and Yan (2016) also reported that the appearance of spherical

405

particles in S-nZVI, formed during treatment of nZVI with thiosulfate, was akin to that of the

406

unmodified nZVI. The presence of oxygen, observed in the EDS scans of the smaller particles,

17

407

indicates that peripheral oxidation by water might have resulted in the formation of an iron oxide

408

coating (Figs. S10-S11). There was low, or in some cases, no sulfur (S) present in these nZVI-

409

like particles indicating either the presence of a small amount of iron sulfides (FeSX) or their

410

complete absence. This suggests that a major portion of CMC-S-nZVI suspension was still

411

comprised of nZVI-like particles, which would mainly be Fe0. Nunez Garcia et al. (2016) found

412

most of the iron to be still preserved in the zerovalent state after nZVI sulfidation. Han and Yan

413

(2016) also detected the presence of Fe0 in their S-nZVI particles after thiosulfate treatment. The

414

SAED pattern (Fig. S11) for nZVI-like particle in the CMC-S-nZVI shows that the rings are

415

sharper and more distinct than that for the original CMC-nZVI (Fig. S9). Some spots were also

416

present in the CMC-S-nZVI sample. This suggests that the amorphous nZVI particles have

417

started turning crystalline. The second type of particles were larger flake-like structures (Figs.

418

S10-S11), composed mainly of iron and sulfur, with an average particle size of 505.2 ± 81.4 nm

419

(n = 11). These were comparatively fewer in number (Fig. S10b) and were not observed in the

420

unsulfidated CMC-nZVI suspension (Fig. S9). The morphology of these particles suggests that

421

FeSX distribution on some of the nZVI particles did not occur uniformly to form the typical core-

422

shell structure (Fan et al. 2013). Rather these particles were either overgrown into FeSX flakes or

423

the Fe0 core was abundantly covered by the flaky FeSX shell (Fan et al. 2013, Su et al. 2015, Su

424

et al. 2018). There is also a possibility that the few iron (oxy)(hydr)oxide particles present in the

425

unsulfidated CMC-nZVI suspension would have transformed to the larger FeSX structures after

426

dithionite addition.

427

Particles recovered from NB1-White and NB1-Clear (~0.86 m from NIW) were similar in

428

morphology to those found in the injected CMC-S-nZVI suspension (Figs. 5a-c), possessing both

429

spherical and flake-like structures. In a previous nZVI trial in the adjoining area, Kocur et al.

18

430

(2014) also did not notice any significant morphological changes between the injected particles

431

and those recovered from the monitoring well. Fig. 5 shows that the nZVI-like particles, present

432

as small discrete spheres (≥100 nm size), had either low or no sulfur content as depicted in the

433

EDS spectra (Fig. 5b-c: E1, E2, and E4). The EDS spectra of larger flake-like particles, showing

434

significant S peaks along with Fe, indicate the presence of FeSX phase (Fig. 5b-c: E3 and E5 &

435

Fig. S12: E1-E3). These distinct FeSX particles were also easily found in the NB1-Blue, NB2-

436

Green and NB2-Clear (the latter two ~ 1.78 m from NIW) (Figs. S12-S13). However, the small

437

nZVI-like particles were not easily detectable in these monitoring well samples. The lower iron

438

concentrations in these samples might have made it difficult to locate these particles during TEM

439

analysis. These particles could also have oxidized/sulfidized or retained as nZVI but aggregated

440

into larger clusters. For example, a TEM image of NB1-Blue shows the presence of small

441

spherical nZVI-like particles that are clustered together and covered by a thick sheet of possibly

442

FeSX (Fig. S12a1). TEM-EDS data indicates that particles in the monitoring well samples,

443

collected at 18 and 72 hours after injection, are similar to those collected from the original CMC-

444

S-nZVI synthesis batches. SAED pattern for the NB1-Clear (Fig. S12b Inset) shows well-defined

445

spots and visible rings, indicating the polycrystalline nature of the particles.

446

It was interesting to see that the samples collected from NIW, at 196 days after the

447

injection, were still black in color indicating the stability of the injected suspension (Fig. S5e).

448

Over time, exposure to oxygenated water from upstream would result in the oxidation of iron

449

sulfides. However, the ORP data shows that reduced conditions were still prevalent at this site

450

(Fig. S7). The presence of Fe3+, Cl-, and SO42- would also favor sulfide oxidation (Bibi et al.

451

2011). In Fig. S14, TEM images show the presence of nano-spheres, plate-like particles, and

452

larger flake-like structures presumably representing iron oxides, (oxy)hydroxides, and sulfides.

19

453

An et al. (2017) reported the formation of lepidocrocite (γ-FeOOH) and magnetite (Fe3O4) from

454

oxidative dissolution of amorphous FeS. Formation of akaganéite (FeO(OH,Cl)) was observed

455

during the oxidation of iron sulfides from chloride-rich sulfidic sediments (Bibi et al. 2011). The

456

presence of Cl peaks in the EDS spectra of NIW suggests the formation of FeO(OH,Cl) and

457

sulfur peaks indicate the presence of sulfides (Fig. 5: E6-E7). During oxidation, FeO(OH,Cl)

458

might have coated the surface of FeSX to form FeSX-FeO(OH,Cl), thus suppressing further

459

oxidation of FeSX (Jeong et al. 2010). Moreover, the black color of the suspension also suggests

460

the presence of iron sulfides or FeII oxides (FeO, Fe3O4). FeIII (oxy)(hydr) oxides, if present,

461

might be a minor species.

462 463

3.4.2 UV - Vis Spectra

464

Fig. S15a shows the absorbance spectra for CMC-nZVI, CMC-S-nZVI and NB1-White.

465

Spectra for samples from other monitoring wells, collected during active injection, did not

466

deviate from NB1-White (Fig. S15b). Therefore, NB1-White was chosen as a representative

467

sample for comparison with CMC-nZVI and CMC-S-nZVI. The absorbance of CMC-nZVI gave

468

small peaks between 300 and 372 nm, followed by a uniform decline up to 900 nm. In contrast,

469

the CMC-S-nZVI spectrum sharply declined between 290 to 310 nm and then gradually decayed

470

from 310 to 900 nm. The spectrum obtained from NB1-White resembles that of CMC-S-nZVI,

471

though at a lower absorbance which would be due to its lower concentration. It is also interesting

472

to note that the NB2-Blue spectrum showed significantly higher absorbance and resembles

473

CMC-S-nZVI quite well. NB2-Blue was one of the samples with the darkest black color (Fig.

474

S5b) and its iron concentration increased from 210 µM to 637 µM at the end of the injection

475

(Table S3).

20

476

To further compare these particles to the injected CMC-S-nZVI, another groundwater

477

sample from NB1-White was intentionally oxidized in the laboratory (labeled ‘NB1-Whiteox’) by

478

exposing it to air. The NB1-Whiteox spectrum differs from the CMC-S-nZVI and unoxidized

479

NB1-White, showing a more gradual decrease up to 550 nm and very little absorbance

480

afterwards. This data supports results from the TEM micrographs suggesting that the recovered

481

samples from monitoring wells were similar to the injected CMC-S-nZVI suspension,

482

experiencing minimal oxidation during the subsurface transport.

483 484

3.4.3 Particle Size Distribution (PSD)

485

TEM provides information about the inner electron-dense metal core and excludes the

486

outer CMC layer. Thus, DLS was used to quantify the hydrodynamic diameter which includes

487

both the metal particle as well as its outer CMC layer yielding the overall size of the particle.

488

The hydrodynamic diameter of unsulfidated CMC-nZVI was 355.8 ± 1.8 nm with a monomodal

489

particle size distribution (PSD) (Fig. S16a). However, the PSD changed to bimodal after addition

490

of dithionite indicating the presence of two types of particles (Fig. S16b). These results are in

491

alignment with the findings from the TEM analysis. For the multimodal PSD, the calculated

492

hydrodynamic diameter does not give accurate particle size information. Thus, the DLS data for

493

the samples with bimodal distributions is discussed in terms of size range rather than the median

494

or mean diameter. The size of smaller particles for CMC-S-nZVI ranged from 357.4 to 438.7 nm

495

which is close to the hydrodynamic diameter of unsulfidated CMC-nZVI particles. The size of

496

larger particles ranged from 881 to 1038 nm. Similarly, the MW samples also showed a bimodal

497

distribution supporting the presence of smaller nZVI-like particles along with the larger FeSX

498

structures, as shown in TEM analysis (Fig. S16c-d). Some differences in the DLS sizes of

21

499

smaller particles, for MW samples from the CMC-S-nZVI, could be due to the variability

500

associated with the field sampling. However, the size of larger particles for MW samples was

501

significantly greater than that of CMC-S-nZVI particles. This could be due to the formation of

502

some larger clusters as seen in the TEM micrographs for MW samples.

503 504

3.4.4 Zeta (ζ) Potential

505

Zeta (ζ) potential (-51.8 ± 1.0 mV, Fig. S17) for the unsulfidated CMC-nZVI in the

506

synthesis batches was consistent with those in the literature (Chowdhury et al. 2012, Kocur et al.

507

2013). A less negative ζ-potential (-44.9 ± 2.4 mV) was observed after adding dithionite.

508

Interestingly, this was in contrast with the findings of previous studies where ζ-potential was

509

reported to be more negative after sulfidation and it was attributed to the presence of FeSX on the

510

surface of S-nZVI particles (Kim et al. 2013, Rajajayavel and Ghoshal 2015, Tang et al. 2016).

511

However, these studies used bare S-nZVI particles that were washed multiple times with

512

deoxygenated water and thus the supernatant from the synthesis was not retained. In contrast, S-

513

nZVI particles used in the current study were stabilized with CMC before sulfidation and the

514

supernatant from the synthesis containing excessive dithionite and its decomposition products

515

(e.g., thiosulfate, sulfate) was retained in the suspension. Moreover, the Na+ ions from the

516

NaBH4, Na2S2O4, and Na-CMC were also present in the suspension. Even though the CMC-S-

517

nZVI particles in the current study had FeSX on the surface, the presence of these anions and

518

cations would significantly increase the ionic strength of the CMC-S-nZVI suspension resulting

519

in less negative ζ-potential (Saleh et al. 2008, Suponik et al. 2016). The average ζ-potential for

520

multiple monitoring well samples collected during injection was -20.4 ± 1.3 mV (Fig. S17),

521

which is less than half of the measured ζ-potential for the injected CMC-S-nZVI suspension.

22

522

This differs from the study of Kocur et al. (2014) where the particles recovered from monitoring

523

wells had similar ζ-potential (-48.3 ± 2.3 mV) to that of the synthesis batches (-49.2 ± 1.5 mV).

524

The further less negative ζ-potential for monitoring well samples in the current study could be

525

due to dilution (Tantra et al. 2010).

526 527

3.4.5 Colloidal Stability

528

Particle stability is a prerequisite for optimal nZVI delivery to the contaminated source

529

zones as the settling of nZVI particles in the synthesis vessel, before injection, would limit the

530

mass delivered to the subsurface (Kocur et al. 2013). In the current study, 50.5% of the

531

unsulfidated CMC-nZVI, synthesized on site, aggregated and settled after 72 hours (Fig. S18).

532

This is consistent with the sedimentation curves from a similar study where only 50% of the

533

CMC-nZVI remained in the suspension at 24-32 hours after on-site synthesis (Kocur et al. 2014).

534

In contrast, only 7.7% sedimentation was observed for the on-site synthesized, dithionite-treated

535

CMC-nZVI (i.e., CMC-S-nZVI) after 72 hours (Fig. S18). This data suggests that the injected

536

CMC-S-nZVI particles would stay suspended for much longer periods in the groundwater,

537

increasing their mobility and delivery to the contaminated areas. Recent studies have suggested

538

that sulfidation of nZVI can effectively inhibit its aggregation and sedimentation (Gong et al.

539

2017, Song et al. 2017, Su et al. 2016, Su et al. 2015). The rate of aggregation depends on a

540

range of factors including particle concentration, Fe0 content, particle size distribution, and the

541

thickness of adsorbed polymer in the case of stabilized nZVI (Phenrat et al. 2010). Within the

542

core-shell structure of nZVI, the content of Fe0 dictates the magnitude of inter-particle magnetic

543

attractions. Decreasing the Fe0 content (oxidation to less magnetic particles) decreases nZVI

544

aggregation and favors its transport, resulting in enhanced mobility (Phenrat et al. 2010). The

23

545

greater stability of S-nZVI has been attributed to the lower magnetic attractions between iron

546

sulfides (FeSX) (Su et al. 2016, Su et al. 2015). For example, Gong et al. (2017) reported a

547

decrease in the saturation magnetization from 165.6 emu/g for nZVI to 78.0 emu/g for S-nZVI.

548

Though Fe0 content of CMC-S-nZVI in the current study could not be measured (Details in

549

Section 2.4) our past research (Nunez Garcia et al. 2016) has shown a decrease in Fe0 content

550

and formation of iron sulfides after sulfidating the bare nZVI with dithionite. Thus, the decreased

551

aggregation and sedimentation of CMC-S-nZVI in this study can be attributed to the decrease in

552

Fe0 content and the formation of iron sulfides, resulting in lower inter-particle magnetic

553

attractions.

554 555

4.

Conclusions

556

CMC-stabilized S-nZVI was successfully synthesized on-site by the borohydride reduction

557

method and aqueous-solid sulfidation with sodium dithionite. For field-scale synthesis of

558

relatively large quantities of S-nZVI, it is suggested to perform an aqueous-solid sulfidation over

559

an aqueous-aqueous approach in order to reduce the possible generation of H2S. To minimize

560

health and safety concerns associated with side reactions between chemical precursors, it is

561

recommended to work near the stoichiometry amount necessary for borohydride and to optimize

562

the concentration of dithionite (by S/Fe ratio) for the intended application. In the present study,

563

the suspension was mixed for 1 to 2 hours to allow time for the dissipation of any H2S generated.

564

TEM images of CMC-S-nZVI synthesis batches revealed the presence of both discrete spherical

565

nZVI particles as well as larger flake-like structures, associated with iron sulfides. CMC-S-nZVI

566

was found to possess better colloidal stability than CMC-nZVI which could possibly contribute

567

to its better transport in the subsurface. Approximately 620 L of CMC-S-nZVI was fed under 24

568

gravity into a sandy aquifer by an injection well. CMC-S-nZVI suspension was mobile in the

569

subsurface, achieving good horizontal and vertical distribution throughout the study area, with

570

detection in multiple monitoring wells both downstream and upstream of the injection well.

571

Travel distances ranged from 0.9 m to at least 2.7 m, which was the location of the farthest

572

monitoring well. TEM-EDS analysis confirmed the presence of both nZVI-like as well as FeSX

573

flaky structures in the MW samples, similar to those identified in the CMC-S-nZVI synthesis

574

batches. This is further supported by the DLS analysis which showed a bimodal particle size

575

distribution for the MW samples, similar to CMC-S-nZVI. Results reported herein demonstrate

576

CMC-S-nZVI is highly mobile at the field scale and very stable (both colloidal and chemical)

577

under subsurface conditions.

578 579

Acknowledgements

580

This research was supported by the Natural Sciences and Engineering Research Council

581

(NSERC) of Canada through the Remediation Education Network (RENEW) training program

582

and the Industrial Postgraduate Scholarship (IPS) program for Ariel Nunez Garcia.

583 584

Supplementary Data

585

The following information can be found in the Supporting Information: Mass and volume

586

of reagents used during synthesis, CMC-S-nZVI mobility measurements, sulfur and sulfate

587

concentrations in monitoring wells, ORP and pH results, particle characterization using TEM-

588

EDS, zeta potential, and sedimentation curves, and photographs of samples from the synthesis

589

batches and monitoring wells. This material is available free of charge via xxx.

590 25

591

Declarations of interest

592

None.

593

Author Contributions

594

All authors contributed to the preparation of this manuscript and have given approval to the final

595

version.

596 597

References

598 599 600

Amonette, J.E., Szecsody, J.E., Schaef, H.T., Gorby, Y.A., Fruchter, J.S. and Templeton, J.C. (1994) Abiotic Reduction of Aquifer Materials by Dithionite: A Promising In-Situ Remediation Technology. Gee, G.W. and Wing, N.R. (eds), pp. 851-881, Battelle Press, Columbus, OH.

601 602 603

An, X.-L., Huang, F.-G., Ren, H.-T., Wang, Y.-F., Chen, Y., Liu, Z.-M., Zhang, H.-W. and Han, X. (2017) Oxidative dissolution of amorphous FeS and speciation of secondary Fe minerals: Effects of pH and As(III) concentration. Chemical Geology 462, 44-54.

604 605 606

Bennett, P., He, F., Zhao, D.Y., Aiken, B. and Feldman, L. (2010) In situ testing of metallic iron nanoparticle mobility and reactivity in a shallow granular aquifer. J Contam Hydrol 116(1-4), 35-46.

607 608 609

Bibi, I., Singh, B. and Silvester, E. (2011) Akaganéite (β-FeOOH) precipitation in inland acid sulfate soils of south-western New South Wales (NSW), Australia. Geochimica et Cosmochimica Acta 75(21), 6429-6438.

610 611 612

Busch, J., Meißner, T., Potthoff, A., Bleyl, S., Georgi, A., Mackenzie, K., Trabitzsch, R., Werban, U. and Oswald, S. E. (2015) A field investigation on transport of carbon-supported nanoscale zero-valent iron (nZVI) in groundwater. J Contam Hydrol 181, 59-68.

613 614 615

Chowdhury, A.I.A., Krol, M.M., Kocur, C.M., Boparai, H.K., Weber, K.P., Sleep, B.E. and O'Carroll, D.M. (2015) nZVI injection into variably saturated soils: Field and modeling study. J Contam Hydrol 183, 16-28.

616 617

Chowdhury, A.I.A., O’Carroll, D.M., Xu, Y. and Sleep, B.E. (2012) Electrophoresis enhanced transport of nano-scale zero valent iron. Advances in Water Resources 40, 71-82.

618 619 620

Cumbal, L.H., Debut, A., Delgado, D.A., Jurado, C.B. and Stael, C. (2015) Synthesis of Multicomponent Nanoparticles for Immobilization of Heavy Metals in Aqueous Phase. NanoWorld J 1(4), 103-109.

26

621 622 623

de Carvalho, L.M. and Schwedt, G. (2001) Polarographic determination of dithionite and its decomposition products: kinetic aspects, stabilizers, and analytical application. Analytica Chimica Acta 436(2), 293-300.

624 625 626 627

Fan, D., Lan, Y., Tratnyek, P.G., Johnson, R.L., Filip, J., O’Carroll, D.M., Nunez Garcia, A. and Agrawal, A. (2017) Sulfidation of Iron-Based Materials: A Review of Processes and Implications for Water Treatment and Remediation. Environmental Science & Technology 51(22), 13070-13085.

628 629 630

Fan, D., O’Brien Johnson, G., Tratnyek, P.G. and Johnson, R.L. (2016a) Sulfidation of Nano Zerovalent Iron (nZVI) for Improved Selectivity During In-Situ Chemical Reduction (ISCR). Environmental Science & Technology 50(17), 9558-9565.

631 632 633

Fan, D., O’Carroll, D.M., Elliott, D.W., Xiong, Z., Tratnyek, P.G., Johnson, R.L. and Nunez Garcia, A. (2016b) Selectivity of Nano Zerovalent Iron in In Situ Chemical Reduction: Challenges and Improvements. Remediation Journal.

634 635 636 637

Fan, D.M., Anitori, R.P., Tebo, B.M., Tratnyek, P.G., Pacheco, J.S.L., Kukkadapu, R.K., Engelhard, M.H., Bowden, M.E., Kovarik, L. and Arey, B.W. (2013) Reductive Sequestration of Pertechnetate ((TcO4-)-Tc-99) by Nano Zerovalent Iron (nZVI) Transformed by Abiotic Sulfide. Environmental Science & Technology 47(10), 5302-5310.

638 639 640

Gong, Y., Gai, L., Tang, J., Fu, J., Wang, Q. and Zeng, E.Y. (2017) Reduction of Cr(VI) in simulated groundwater by FeS-coated iron magnetic nanoparticles. Science of The Total Environment 595, 743-751.

641 642 643

Gu, Y., Gong, L., Qi, J., Cai, S., Tu, W. and He, F. (2019). Sulfidation mitigates the passivation of zero valent iron at alkaline pHs: Experimental evidences and mechanism. Water Res 159, 233241.

644 645 646

Han, Y. and Yan, W. (2016) Reductive Dechlorination of Trichloroethene by Zero-valent Iron Nanoparticles: Reactivity Enhancement through Sulfidation Treatment. Environmental Science & Technology 50(23), 12992-13001.

647 648 649

He, F., Li, Z., Shi, S., Xu, W., Sheng, H., Gu, Y., Jiang, Y. and Xi, B. (2018). Dechlorination of Excess Trichloroethene by Bimetallic and Sulfidated Nanoscale Zero-Valent Iron. Environmental Science & Technology 52(15), 8627–8637.

650 651 652

He, F., Zhao, D. and Paul, C. (2010) Field assessment of carboxymethyl cellulose stabilized iron nanoparticles for in situ destruction of chlorinated solvents in source zones. Water Res 44(7), 2360-2370.

653 654 655

He, F. and Zhao, D.Y. (2005) Preparation and characterization of a new class of starch-stabilized bimetallic nanoparticles for degradation of chlorinated hydrocarbons in water. Environmental Science & Technology 39(9), 3314-3320.

27

656 657 658

He, F., Zhao, D.Y., Liu, J.C. and Roberts, C.B. (2007) Stabilization of Fe-Pd nanoparticles with sodium carboxymethyl cellulose for enhanced transport and dechlorination of trichloroethylene in soil and groundwater. Industrial & Engineering Chemistry Research 46(1), 29-34.

659 660

Johnston, F. and McAmish, L. (1973) A study of the rates of sulfur production in acid thiosulfate solutions using S-35. Journal of Colloid and Interface Science 42(1), 112-119.

661 662 663

Jeong, H.Y., Han, Y.-S. and Hayes, K.F. (2010) X-ray Absorption and X-ray Photoelectron Spectroscopic Study of Arsenic Mobilization during Mackinawite (FeS) Oxidation. Environmental Science & Technology 44(3), 955-961.

664 665 666 667

Johnson, R.L., Nurmi, J.T., Johnson, G.S.O., Fan, D.M., Johnson, R.L.O., Shi, Z.Q., SalterBlanc, A.J., Tratnyek, P.G. and Lowry, G.V. (2013) Field-Scale Transport and Transformation of Carboxymethylcellulose-Stabilized Nano Zero-Valent Iron. Environmental Science & Technology 47(3), 1573-1580.

668 669 670

Kim, E.-J., Murugesan, K., Kim, J.-H., Tratnyek, P.G. and Chang, Y.-S. (2013) Remediation of Trichloroethylene by FeS-Coated Iron Nanoparticles in Simulated and Real Groundwater: Effects of Water Chemistry. Industrial & Engineering Chemistry Research 52(27), 9343-9350.

671 672 673

Kocur, C.M., Chowdhury, A.I., Sakulchaicharoen, N., Boparai, H.K., Weber, K.P., Sharma, P., Krol, M.M., Austrins, L., Peace, C., Sleep, B.E. and O'Carroll, D.M. (2014) Characterization of nZVI Mobility in a Field Scale Test. Environmental Science & Technology 48(5), 2862-2869.

674 675

Kocur, C.M., O'Carroll, D.M. and Sleep, B.E. (2013) Impact of nZVI stability on mobility in porous media. J Contam Hydrol 145, 17-25.

676 677 678

Leenhouts, J.M., Bassett, R.L. and Maddock, T. (1998) Utilization of intrinsic boron isotopes as co-migrating tracers for identifying potential nitrate contamination sources. Ground Water 36(2), 240-250.

679 680

Li, D., Mao, Z., Zhong, Y., Huang, W., Wu, Y. and Peng, P.a. (2016) Reductive transformation of tetrabromobisphenol A by sulfidated nano zerovalent iron. Water Res 103, 1-9.

681 682 683

Nunez Garcia, A., Boparai, H.K. and O’Carroll, D.M. (2016) Enhanced Dechlorination of 1,2Dichloroethane by Coupled Nano Iron-Dithionite Treatment. Environmental Science & Technology 50(10), 5243-5251.

684 685 686 687

Nurmi, J.T., Tratnyek, P.G., Sarathy, V., Baer, D.R., Amonette, J.E., Pecher, K., Wang, C.M., Linehan, J.C., Matson, D.W., Penn, R.L. and Driessen, M.D. (2005) Characterization and properties of metallic iron nanoparticles: Spectroscopy, electrochemistry, and kinetics. Environmental Science & Technology 39(5), 1221-1230.

688 689 690

O'Carroll, D., Sleep, B., Krol, M., Boparai, H. and Kocur, C. (2013) Nanoscale zero valent iron and bimetallic particles for contaminated site remediation. Advances in Water Resources 51, 104-122.

28

691 692 693 694

Phenrat, T., Cihan, A., Kim, H.J., Mital, M., Illangasekare, T. and Lowry, G.V. (2010) Transport and Deposition of Polymer-Modified Fe-0 Nanoparticles in 2-D Heterogeneous Porous Media: Effects of Particle Concentration, Fe-0 Content, and Coatings. Environmental Science & Technology 44(23), 9086-9093.

695 696 697 698

Phenrat T., Lowry G.V., Babakhani P. (2019) Nanoscale Zerovalent Iron (NZVI) for Environmental Decontamination: A Brief History of 20 Years of Research and Field-Scale Application. In: Phenrat T., Lowry G. (eds) Nanoscale Zerovalent Iron Particles for Environmental Restoration. Springer, Cham.

699 700

Quast, K.W., Lansey, K., Arnold, R., Bassett, R.L. and Rincon, M. (2006) Boron isotopes as an artificial tracer. Ground Water 44(3), 453-466.

701 702 703

Quinn, J., Geiger, C., Clausen, C., Brooks, K., Coon, C., O'Hara, S., Krug, T., Major, D., Yoon, W.S., Gavaskar, A. and Holdsworth, T. (2005) Field demonstration of DNAPL dehalogenation using emulsified zero-valent iron. Environmental Science & Technology 39(5), 1309-1318.

704 705

Rajajayavel, S.R.C. and Ghoshal, S. (2015) Enhanced reductive dechlorination of trichloroethylene by sulfidated nanoscale zerovalent iron. Water Res 78, 144-153.

706 707 708

Saleh, N., Kim, H.J., Phenrat, T., Matyjaszewski, K., Tilton, R.D. and Lowry, G.V. (2008) Ionic strength and composition affect the mobility of surface-modified Fe-0 nanoparticles in watersaturated sand columns. Environmental Science & Technology 42(9), 3349-3355.

709 710 711

Schrick, B., Hydutsky, B.W., Blough, J.L. and Mallouk, T.E. (2004) Delivery vehicles for zerovalent metal nanoparticles in soil and groundwater. Chemistry of Materials 16(11), 21872193.

712 713 714

Shi, Z.Q., Fan, D., Johnson, R.L., Tratnyek, P.G., Nurmi, J.T., Wu, Y.X. and Williams, K.H. (2015) Methods for characterizing the fate and effects of nano zerovalent iron during groundwater remediation. J Contam Hydrol 181, 17-35.

715 716 717

Song, S., Su, Y., Adeleye, A.S., Zhang, Y. and Zhou, X. (2017) Optimal design and characterization of sulfide-modified nanoscale zerovalent iron for diclofenac removal. Applied Catalysis B: Environmental 201, 211-220.

718 719

Stefaniuk, M., Oleszczuk, P. and Ok, Y.S. (2016) Review on nano zerovalent iron (nZVI): From synthesis to environmental applications. Chemical Engineering Journal 287, 618-632.

720 721 722

Su, Y., Adeleye, A.S., Huang, Y., Zhou, X., Keller, A.A. and Zhang, Y. (2016) Direct Synthesis of Novel and Reactive Sulfide-modified Nano Iron through Nanoparticle Seeding for Improved Cadmium-Contaminated Water Treatment. Scientific Reports 6, 24358.

723 724 725

Su, Y., Adeleye, A.S., Keller, A.A., Huang, Y., Dai, C., Zhou, X. and Zhang, Y. (2015) Magnetic sulfide-modified nanoscale zerovalent iron (S-nZVI) for dissolved metal ion removal. Water Res 74, 47-57.

29

726 727 728 729

Su, Y., Jassby, D., Song, S., Zhou, X., Zhao, H., Filip, J., Petala, E. and Zhang, Y. (2018) Enhanced Oxidative and Adsorptive Removal of Diclofenac in Heterogeneous Fenton-like Reaction with Sulfide Modified Nanoscale Zerovalent Iron. Environmental Science & Technology 52(11), 6466-6475.

730 731 732

Sun, Y.P., Li, X.Q., Zhang, W.X. and Wang, H.P. (2007) A method for the preparation of stable dispersion of zero-valent iron nanoparticles. Colloids and Surfaces a-Physicochemical and Engineering Aspects 308(1-3), 60-66.

733 734

Suponik, T., Lemanowicz, M. and Wrona, P. (2016) Stability of green tea nanoscale zero-valent iron. E3S Web Conf. 8, 01048.

735 736 737

Tang, J., Tang, L., Feng, H., Zeng, G., Dong, H., Zhang, C., Huang, B., Deng, Y., Wang, J. and Zhou, Y. (2016) pH-dependent degradation of p-nitrophenol by sulfidated nanoscale zerovalent iron under aerobic or anoxic conditions. J Hazard Mater 320, 581-590.

738 739

Tantra, R., Schulze, P. and Quincey, P. (2010) Effect of nanoparticle concentration on zetapotential measurement results and reproducibility. Particuology 8(3), 279-285.

740 741 742

Tiraferri, A., Chen, K.L., Sethi, R. and Elimelech, M. (2008) Reduced aggregation and sedimentation of zero-valent iron nanoparticles in the presence of guar gum. Journal of Colloid and Interface Science 324(1-2), 71-79.

743 744 745 746

Velimirovic, M., Tosco, T., Uyttebroek, M., Luna, M., Gastone, F., De Boer, C., Klaas, N., Sapion, H., Eisenmann, H., Larsson, P.O., Braun, J., Sethi, R. and Bastiaens, L. (2014) Field assessment of guar gum stabilized microscale zerovalent iron particles for in-situ remediation of 1,1,1-trichloroethane. J Contam Hydrol 164, 88-99.

747 748 749

Wei, Y.-T., Wu, S.-C., Chou, C.-M., Che, C.-H., Tsai, S.-M. and Lien, H.-L. (2010) Influence of nanoscale zero-valent iron on geochemical properties of groundwater and vinyl chloride degradation: A field case study. Water Res 44(1), 131-140.

750 751

Zhang, W.-x. and Elliott, D.W. (2006) Applications of iron nanoparticles for groundwater remediation. Remediation Journal 16(2), 7-21.

752

30

(a)

(b) 1 2 3 4 5

Fig. 1. (a) Plan and (b) cross-sectional views of the study area. The injection well is denoted as NIW.

1

6

1.4

1.4

(a)

Iron (mmol L-1)

NB1-White 1.2

NA4-Blue

1.0

NA3-White

(b) 1.2 1.0

NC1-White 0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0 7

0.0 0

5

10

15

20

7.0

0

5

10

15

(d)

Boron (mmol L-1)

(c) 6

6.0

5

5.0

4

4.0

3

3.0

2

2.0

1

1.0

0

0.0

0 7 8 9 10 11 12

5

10

15

20

Time (Hours after injection)

0

5

10

15

Time (Days after injection)

Fig. 2. Changes in iron and boron molar concentrations for NB1-White, NA4-Blue, NA3-White, and NC1-White (a) & (c) during and (b) & (d) after injection. Initial time refers to background samples collected 28.5 hours prior to injection.

2

13 2.5 NB1-White NA4-Blue

Iron (mmol L-1)

2.0

NA3-White



1.5



. .







NC1-White

:

NB2-White

0.39

NA2-Blue NA1-White

Fe/B = 0.18

1.0

NB3-White

0.5

0.0 0 14 15 16 17 18 19

1

2

3

4

5

6

Boron (mmol L-1) Fig. 3. Iron and boron concentrations up to 22 hours during the injection. Dashed and solid lines represent the molar ratio of Fe/B in the injected suspension and in the monitoring well samples, respectively.

3

2.8

2.8

(a) NB1

Depth (m bgs)

3.2

3.2

3.4

3.4

3.6 3.8

-28.5 hours (BG)3.6 18 hours 3.8

4.0

4.0

4.2

4.2

4.4

4.4

2.8 0 3.0

0

0

1

1

1

1

12.8 0

(c) NB2

Depth (m bgs) 21 22 23 24 25 26

1

2

3

4

2

3

4

3.0

3.2

20

(b) NB1

3.0

3.0

5

6

(d) NB2

3.2

3.4

3.4 -28.5 hours (BG)

3.6

21.25 hours

3.6

3.8

3.8

4.0

4.0

4.2

4.2

4.4

4.4

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Iron (mmol L-1)

0

1

5

6

Boron (mmol L-1)

Fig. 4. Depthwise profile of (a) & (c) Iron and (b) & (d) Boron concentrations for the NB1 and NB2 wells. Each data point denotes a depth interval from top to bottom: 1) Black (2.9 m), 2) Yellow (3.2 m), 3) Green (3.51 m), 4) Clear (3.81 m) 5) Blue (4.12 m), and 6) White (4.42 m) (the NIW screen lies between 4.27 and 4.88 m bgs).

4

(b) (a)

E1. nZVI-like, ~ FeSX coating

E1

E2

E2. nZVI-like, ~ FeSX coating

E3. FeSX

E3

(c)

(d)

E4. nZVI-like, no FeSX

E4

E7 E5. FeSX

E6

E6. ~FeSX-FeO(OH,Cl)

E7. ~FeSX-FeO(OH,Cl)

E5 27 28 29

Fig. 5. TEM and EDS of samples from (a) S-nZVI Synthesis batch, (b) NB1-White at t = 18 h, (c) NB1-Clear at t = 72 h, and (d) NIW at t = 196 d after injection. 5

Highlights •

620 L of CMC-S-nZVI was fed under gravity into a sandy aquifer by an injection well.

•

CMC-S-nZVI suspension was mobile in the subsurface, with travel distances of up to 2.7 m.

•

nZVI-like nanoparticles and flaky FeS structures were recovered from monitoring wells.

•

CMC-S-nZVI remained colloidally and chemically stable under subsurface conditions.

Declaration of interests ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: