← Back

Phosphorescent iridium(III)-bis-N-heterocyclic carbene complexes as mitochondria-targeted theranostic and photodynamic anticancer agents.

PMID: 25477176
UC Berkeley UC Berkeley Previously Published Works Title Bacteria attenuation by iron electrocoagulation governed by interactions between bacterial phosphate groups and Fe(III) precipitates Permalink https://escholarship.org/uc/item/27f797jr Journal Water Research, 103 ISSN 0043-1354 Authors Delaire, Caroline van Genuchten, Case M Amrose, Susan E et al. Publication Date 2016-10-01 DOI 10.1016/j.watres.2016.07.020 Peer reviewed eScholarship.org Powered by the California Digital Library University of California Elsevier Editorial System(tm) for Water Research Manuscript Draft Manuscript Number: Title: Bacteria attenuation by iron electrocoagulation governed by interactions between bacterial phosphate groups and Fe(III) precipitates Article Type: Research Paper Keywords: Iron electrocoagulation; bacteria attenuation; bacterial surface functional groups; specific interactions; bivalent cations; oxyanions Corresponding Author: Mrs. Caroline Delaire, Corresponding Author's Institution: University of California, Berkeley First Author: Caroline Delaire Order of Authors: Caroline Delaire; Case M van Genuchten, PhD; Susan E Amrose, PhD; Ashok J Gadgil, PhD Abstract: Iron electrocoagulation (Fe-EC) is a low-cost process in which Fe(II) generated from an Fe(0) anode reacts with dissolved O2 to form (1) Fe(III) precipitates with an affinity for bacterial cell walls and (2) bactericidal reactive oxidants. Previous work suggests that Fe-EC is a promising treatment option for groundwater containing arsenic and bacterial contamination. However, the mechanisms of bacteria attenuation and the impact of major groundwater ions are not well understood. In this work, using the model indicator Escherichia coli (E. coli), we show that physical removal via enmeshment in EC precipitate flocs is the primary process of bacteria attenuation in the presence of HCO3-, which significantly inhibits inactivation, possibly due to a reduction in the lifetime of reactive oxidants. We demonstrate that the adhesion of EC precipitates to cell walls, which results in bacteria encapsulation in flocs, is driven primarily by interactions between EC precipitates and phosphate functional groups on bacteria surfaces. In single solute electrolytes, both P (0.4 mM) and Ca/Mg (1-13 mM) interfered with the adhesion of EC precipitates to bacterial cell walls, whereas Si (0.4 mM) and ionic strength (2-200 mM) did not impact E. coli attenuation. Interestingly, P (0.4 mM) did not affect E. coli attenuation in electrolytes containing Ca/Mg, consistent with bivalent cation bridging between bacterial phosphate groups and inorganic P sorbed to EC precipitates. Finally, we found that EC precipitate adhesion is largely independent of cell wall composition, consistent with comparable densities of phosphate functional groups on Gram-positive and Gramnegative cells. Our results are critical to predict the performance of Fe-EC to eliminate bacterial contaminants from waters with diverse chemical compositions. Suggested Reviewers: Shankar Chellam Texas A&M University chellam@tamu.edu Dr. Chellam is an expert on water treatment technologies, and he has worked on iron and aluminum electrocoagulation for virus control. Chuanyong Jing Chinese Academy of Science cyjing@rcees.ac.cn Dr. Jing works on biogeochemistry and on environmental interfacial processes. He has recently co-authored a ATR-FTIR study about bacteriagoethtite adhesion Sharon Walker University of California, Riverside swalker@engr.ucr.edu Dr. Walker is an expert on bacteria-particle adhesion in sub-surface environments, and on the role of cell surface polymers in bacteria adhesion and transport Andreas Voegelin EAWAG andreas.voegelin@eawag.ch Dr. Voegelin is an expert on molecular environmental geochemistry and on the reactivity of iron(III) precipitates and interactions with cooccurring ions. Jon Chorover University of Arizona Chorover@email.arizona.edu Dr. Chorover is an expert on the investigation of the interactions between iron oxides and bacterial cell walls using spectroscopic methods Sanjay Mohanty University of Pennsylvania sanjay.mohanty@colorado.edu Dr. Mohanty has worked on bacteria-mineral interactions in the context of bacteria transport in biofilters Changa Lee Ulsan National Institute of Science and Technology clee@unist.ac.kr Dr. Lee has worked on pathogen inactivation and on the production of reactive oxidants from zero-valent and ferrous iron Cover Letter, For Editor only Mark van Loosdrecht Department of Biochemical Engineering Delft University of Technology KWR Watercycle Research Delft Netherlands April 21st, 2016 Dear Water Research Editor, On behalf of all coauthors, I am pleased to submit our manuscript “Bacteria Attenuation by Iron Electrocoagulation Governed by Interactions between Bacterial Phosphate Groups and Fe(III) Precipitates” enclosed for consideration as a research paper for Water Research. In this study, we investigate a specific application –bacteria attenuation- of iron electrocoagulation (Fe-EC), a promising technology for the treatment of arsenic-contaminated groundwater in low-resource settings. Simultaneous arsenic and bacteria attenuation in Fe-EC has been demonstrated in our previous study, and constitutes a significant advantage of this technology in areas where arsenic often concurs with fecal contamination. However, the processes leading to bacteria attenuation and the impact of groundwater composition are not well understood. This manuscript presents new results elucidating the molecular-scale mechanisms of bacteria attenuation in Fe-EC, and the role that major groundwater ions, such as HCO3-, P, Si, Ca and Mg, play in such mechanisms. Our work goes beyond presenting remediation results because we thoroughly investigate the processes leading to bacteria inactivation and removal in Fe-EC. In addition, our findings have significant implications for field treatment as they allow to predict the performance of Fe-EC to attenuate various types of bacterial contamination in different groundwater matrices. We showed that attenuation is independent of cell wall composition, which is critical to generalize our findings to all bacterial species relevant to water quality. Finally, the molecular mechanisms identified in this study can be used to discuss the potential of Fe-EC, and other Fe-based coagulation processes, to treat various water sources, such as surface water, agricultural runoff and wastewater. The results presented here are novel, as no other study to the authors’ knowledge has investigated the bacterial surface functional groups and the type of interaction involved in the adhesion of Fe(III) (oxyhydr)oxides to cell walls in water matrices representative of field conditions. Specifically, molecular-scale interactions between bacteria and Fe(III) oxides in systems containing bivalent cations and oxyanions, which alter the surface of bacterial cells and Fe(III) oxides, are not known. Finally, we believe that our approach to elucidate bacteria-precipitates interactions is innovative. In systems where bacteria are encapsulated inside flocs and in the complex groundwater-like electrolytes, spectroscopic techniques such as ATR-FTIR cannot adequately determine bacterial functional groups mediating bacteria-precipitate adhesion. Instead, we used a novel approach, where macroscopic data of bacteria attenuation in systematically varied electrolytes was combined with ζ-potential measurements to elucidate molecular-scale processes. Building on previous spectroscopic studies in more simple controlled systems, our approach allowed us to gain knowledge on bacteria-Fe(III) precipitate interactions in complex water matrices. We believe that this work will be relevant to a general audience interested in mechanistic aspects as well as in field applicability of drinking water treatment technologies. Sincerely yours, Caroline Delaire (corresponding author) Department of Civil and Environmental Engineering University of California, Berkeley Berkeley, California, USA 94720 Phone: +1 (510) 417-9491 caroline.delaire@orange.fr Case M. van Genuchten Institut de Dynamiques de la Surface Terrestre University of Lausanne Lausanne, 1015-CH, Switzerland cmvangenuchten@gmail.com Susan E. Amrose Department of Civil and Environmental Engineering University of California, Berkeley Berkeley, California, USA 94720 samrose@berkeley.edu Ashok J. Gadgil Energy Technologies Area Lawrence Berkeley National Laboratory Berkeley, California, USA 94720 ajgadgil@lbl.gov *Highlights (for review) Bacteria attenuation by iron electrocoagulation governed by interactions between bacterial phosphate groups and Fe(III) precipitates Caroline Delaire*,† , Case M. van Genuchten§ , Susan E. Amrose†, Ashok J. Gadgil†,‡ † Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720-1710, United States § Department of Earth Sciences – Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht 3508TA, The Netherlands ‡ Energy Technologies Area, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States Highlights  In natural waters, bacteria attenuation by Fe-EC is primarily due to physical removal with flocs  Bacterial phosphate groups govern the adhesion of EC precipitates to cell walls  Ca/Mg decrease removal, Si has no effect, and P decreases removal if Ca/Mg are absent  Fe-EC is equally effective for Gram positive and negative (rough and smooth) strains Graphical Abstract *Manuscript Click here to view linked References 1 Bacteria attenuation by iron electrocoagulation governed by interactions between bacterial 2 phosphate groups and Fe(III) precipitates 3 4 Caroline Delaire*,† , Case M. van Genuchten§ , Susan E. Amrose†, Ashok J. Gadgil†,‡ 5 6 7 8 9 10 11 † Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720-1710, United States 12 13 14 § Department of Earth Sciences – Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht 3508TA, The Netherlands 15 16 17 ‡ Energy Technologies Area, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States 18 19 20 21 22 23 24 * Corresponding author: Department of Civil and Environmental Engineering, University of California, Berkeley, CA 94720-1710, United States. Phone: (+1) 510-417-9491; email: caroline.delaire@orange.fr 25 26 27 28 1 29 30 Abstract Iron electrocoagulation (Fe-EC) is a low-cost process in which Fe(II) generated from an Fe(0) anode 31 reacts with dissolved O2 to form (1) Fe(III) precipitates with an affinity for bacterial cell walls and (2) 32 bactericidal reactive oxidants. Previous work suggests that Fe-EC is a promising treatment option for 33 groundwater containing arsenic and bacterial contamination. However, the mechanisms of bacteria 34 attenuation and the impact of major groundwater ions are not well understood. In this work, using the 35 model indicator Escherichia coli (E. coli), we show that physical removal via enmeshment in EC 36 precipitate flocs is the primary process of bacteria attenuation in the presence of HCO3-, which 37 significantly inhibits inactivation, possibly due to a reduction in the lifetime of reactive oxidants. We 38 demonstrate that the adhesion of EC precipitates to cell walls, which results in bacteria encapsulation in 39 flocs, is driven primarily by interactions between EC precipitates and phosphate functional groups on 40 bacteria surfaces. In single solute electrolytes, both P (0.4 mM) and Ca/Mg (1-13 mM) interfered with the 41 adhesion of EC precipitates to bacterial cell walls, whereas Si (0.4 mM) and ionic strength (2-200 mM) 42 did not impact E. coli attenuation. Interestingly, P (0.4 mM) did not affect E. coli attenuation in 43 electrolytes containing Ca/Mg, consistent with bivalent cation bridging between bacterial phosphate 44 groups and inorganic P sorbed to EC precipitates. Finally, we found that EC precipitate adhesion is 45 largely independent of cell wall composition, consistent with comparable densities of phosphate 46 functional groups on Gram-positive and Gram-negative cells. Our results are critical to predict the 47 performance of Fe-EC to eliminate bacterial contaminants from waters with diverse chemical 48 compositions. 49 50 Keywords 51 Iron electrocoagulation; bacteria attenuation; bacterial surface functional groups; specific interactions; 52 bivalent cations; oxyanions. 2 53 Highlights 54  In natural waters, bacteria attenuation by Fe-EC is primarily due to physical removal with flocs 55  Bacterial phosphate groups govern the adhesion of EC precipitates to cell walls 56  Ca/Mg decrease removal, Si has no effect, and P decreases removal if Ca/Mg are absent 57  Fe-EC is equally effective for Gram positive and negative (rough and smooth) strains 58 59 60 61 Graphical abstract 62 63 64 65 3 66 1. Introduction 67 Iron electrocoagulation (Fe-EC) is a process relying on the electrolytic dissolution of an Fe(0) anode 68 to generate Fe(II), which is oxidized by dissolved O2 to produce Fe(III) (oxyhydr)oxide precipitates with 69 an affinity for microbial and chemical contaminants (Delaire et al., 2015; van Genuchten et al., 2012). Fe- 70 EC can efficiently remove arsenic from contaminated groundwater (Amrose et al., 2014; Li et al., 2012), 71 and has also been shown to attenuate bacteria in a range of water matrices (Barrera-Dı́az et al., 2003; 72 Delaire et al., 2015; Ghernaout et al., 2008). In a recent study, we demonstrated that Fe-EC can attenuate 73 Escherichia coli (E. coli) from synthetic Bengal groundwater (SBGW) without detriment to arsenic 74 removal (Delaire et al., 2015), confirming that Fe-EC has promising applications for low-cost 75 groundwater remediation (Amrose et al., 2014). Two processes contributed to bacteria attenuation in Fe- 76 EC: (1) physical removal, caused by bacteria enmeshment in Fe(III) flocs and subsequent settling, and (2) 77 inactivation by reactive species produced upon Fe(II) oxidation by O2. Fundamental aspects of the 78 mechanisms underlying these two processes remain unknown. For example, the type of chemical 79 interactions governing bacteria enmeshment in flocs is not well understood. In addition, the effect of 80 major groundwater components, such as HCO3-, Ca, Mg, Si, and P, which can interfere with both 81 inactivation and removal, has not been investigated. Finally, the impact of bacteria surface structure 82 (Gram-positive versus Gram-negative, smooth versus rough Gram-negative) on attenuation has not been 83 elucidated. By addressing these knowledge gaps, this study can improve considerably our predictions of 84 Fe-EC performance in various water matrices containing different types of bacterial contamination. 85 Our previous work suggests that the adhesion of EC precipitates to cell walls is a key process in 86 bacteria enmeshment in flocs (Delaire et al., 2015). Specifically, the significantly higher bacteria removal 87 by Fe-EC in comparison to coagulation with pre-synthesized ferryhydrite (for the same Fe(III) 88 concentration) shows that removal cannot be solely attributed to the mechanical sweeping of bacterial 89 cells by Fe(III) flocs (sweep flocculation). In addition, increased removal at higher Fe dosages indicates a 90 stoichiometric relationship between Fe(III) precipitates and bacterial surfaces, consistent with the primary 4 91 role of precipitate adhesion to cell walls. However, important questions remain regarding the bacterial 92 functional groups involved in such adhesion, the type of interaction (electrostatic versus specific 93 bonding), and the effects of groundwater chemistry and cell wall structure. 94 Four types of surface functional groups are present on bacterial cell walls at comparable densities: 95 hydroxyl (pKa ~ 9.0), amine (pKa ~ 9.0), carboxyl (pKa ~ 4.7), and phosphate groups (pKa1 ~ 3.1, pKa2 ~ 96 6.6) (Borrok et al., 2005; Ngwenya et al., 2003). Hydroxyl and amine moieties do not have a strong 97 affinity for Fe(III) oxides (McBride and Kung, 1991; Norén et al., 2008) and therefore they are not 98 expected to strongly interact with EC precipitates. By contrast, carboxyl and phosphate moieties have 99 strong affinities for Fe(III) oxides (Arai and Sparks, 2001; Chassé et al., 2015; Filius et al., 2000; van 100 Genuchten et al., 2014a) and studies using Attenuated Total Reflectance Fourier-Transform Infrared 101 spectroscopy (ATR-FTIR) have shown direct bonding of bacterial phosphate and carboxyl groups to 102 hematite and goethite (Elzinga et al., 2012; Parikh and Chorover, 2006; Parikh et al., 2014). However, 103 these studies were performed in controlled laboratory systems and simple water matrices, and they cannot 104 be directly extrapolated to Fe-EC in groundwater, where precipitates and bacteria interact in an agitated 105 suspension and in the presence of bivalent cations (Ca and Mg) and oxyanions (P and Si), which can sorb 106 to bonding sites on bacteria (Beveridge and Koval, 1981; Johnson et al., 2007) and precipitates (van 107 Genuchten et al., 2014b), respectively, and may therefore interfere with adhesion. 108 In addition to electrolyte composition, a number of studies have shown that the biomolecular structure 109 of bacterial cell walls can affect their interactions with mineral surfaces through changes in surface 110 charge, hydrophobicity and steric hindrance (Chen and Walker, 2012; Jacobson et al., 2015; Walker et al., 111 2004). Because waterborne pathogenic bacteria and indicator organisms span the range of Gram-positive, 112 smooth and rough (with and without O-antigen) Gram-negative strains (WHO, 2011), understanding the 113 impact of cell wall structure on bacteria attenuation with Fe-EC is essential to generalize our findings to 114 all bacterial species relevant to water quality. 5 115 Spectroscopic techniques such as ATR-FTIR, X-ray fluorescence (XRF) and X-ray absorption 116 spectroscopy (XAS) have been used to study bacteria-Fe systems (Chan et al., 2009; Elzinga et al., 2012; 117 Miot et al., 2009; Yan et al., 2016). However, these techniques cannot adequately determine bacteria- 118 Fe(III) interactions in systems where Fe(III) is co-precipitated with bacteria in complex electrolytes 119 similar to groundwater. For example, P-Fe bonds from bacteria-precipitate interactions and from aqueous 120 P sorption to precipitates look very similar using ATR-FTIR (Elzinga et al., 2012) and would not be 121 distinguishable with P K-edge XAS (Kelly et al., 2008). Additionally, ATR-FTIR is not suited to 122 investigate interactions taking place inside large flocs due to the low penetration length of infrared beams 123 in aqueous medium (~1µm). To circumvent these limitations, the present study proposes an innovative 124 approach, where macroscopic data of bacteria attenuation in systematically varied electrolytes are 125 combined with ζ-potential measurements to elucidate the molecular interactions between bacteria and EC 126 precipitates. Although this approach can only provide indirect evidence for specific interactions between 127 bacteria and precipitates, it builds upon previous spectroscopic studies, which have identified bacteria-Fe 128 oxide bonding processes in simple controlled systems (Elzinga et al., 2012; Parikh and Chorover, 2006; 129 Parikh et al., 2014) and structures of Fe-EC precipitates in complex water matrices (van Genuchten et al., 130 2014a, 2014b), to gain information about bacteria removal mechanisms in groundwater-like electrolytes. 131 The goals of this study are to: (1) determine the impact of HCO3-, Ca, Mg, P and Si on bacteria 132 attenuation with Fe-EC, (2) identify the bacterial functional groups involved in the adhesion of EC 133 precipitates to cell walls and investigate the type of interaction (electrostatic versus specific), and (3) test 134 the generalizability of these conclusions to various bacteria types. To achieve these objectives, we first 135 compared Fe-EC with FeCl3 coagulation to distinguish the contributions of inactivation and removal via 136 enmeshment in flocs to overall bacteria attenuation in Fe-EC as a function of the HCO3- concentration. 137 Inactivation results were confirmed using live-dead staining. Second, we systematically investigated the 138 effect of ionic strength, Ca/Mg and P/Si on E. coli attenuation, both in single and multiple solute 139 electrolytes, to constrain the bacterial functional groups involved in precipitate adhesion to cell walls. ζ- 6 140 potential, a proxy for surface charge, was used to assess the interaction of major groundwater ions with 141 the surface of EC precipitates or E. coli cells. Third, we validated our proposed mechanism with 3 142 bacteria strains bearing different surface structures (smooth and rough Gram-negative, and Gram- 143 positive). Our results strongly suggest that Fe-EC can be used to remove various types of bacteria from a 144 wide range of water matrices representative of regions affected by arsenic and microbial contamination of 145 drinking water sources. More generally, this study can help predict the performance of Fe-EC, and other 146 Fe-based coagulation processes, to reduce bacterial contaminants from drinking water and wastewater. 147 148 149 2. Methods 2.1. Bacteria preparation and enumeration 150 One Gram-positive and two Gram-negative bacterial strains were used: Enterococcus faecalis (ATCC 151 19433, no antibiotic resistance), Escherichia coli K12 (NCM 4236, kanamycin-resistant), and Escherichia 152 coli ECOR 10 (from STEC center, ampicillin-resistant (Mazel et al., 2000)). K12 is a rough strain (no O- 153 antigen) (Stevenson et al., 1994) whereas ECOR 10 is a smooth strain (O-antigen present, serotype O6) 154 (STEC center, 2016). After three propagations in growth media amended with appropriate antibiotics, 155 stationary-phase bacteria were rinsed 3 times and resuspended in 100 mM NaCl as detailed in the 156 Supporting Information. Bacteria were spiked in Fe-EC electrolytes to achieve initial concentrations of 157 106.1-6.7 CFU/mL (105.0-5.8 CFU/mL for E. faecalis). Bacteria concentrations were enumerated in duplicate 158 in 0.1 mL aliquots as colony forming units (CFU) using the spread plate technique on agar amended with 159 appropriate antibiotics (detection limit of 10 CFU/mL), as described in the Supporting Information. 160 161 162 163 2.2. Electrolytes The list of electrolytes used in bacteria attenuation experiments is specified in Table S1. In summary, we first varied the concentration of HCO3- (0.1-8.0 mM) to examine its impact on bacteria inactivation. 7 164 Second, a range of ionic strengths was investigated by varying NaCl (in deionized water and in SBGW) 165 or NaClO4 (in 1 mM CaCl2). Then, concentrations of bivalent cations (Ca: 0-13.5 mM and Mg: 0-10.6 166 mM) and oxyanions (P: 0-0.4 mM and Si: 0-0.4 mM) were systematically varied, in single and composite 167 electrolytes, to elucidate their effect on bacteria removal. Finally, SBGW containing 8.2 mM HCO3-, 2.7 168 mM Ca, 2.0 mM Mg, 1.3 mM Si, 0.15 mM P, and 6.3 µM As(III), was prepared as described elsewhere 169 (Delaire et al., 2015) and used as the electrolyte in some experiments. All experiments were conducted at 170 pH 7.0 ± 0.3, except for the comparisons between the three bacterial strains, which were conducted at pH 171 7.5 ± 0.2. The pH was held constant throughout experiments by adding HCl, NaOH or NaHCO3 as 172 needed. Electrolytes were selected in part to overlap with previous work on the structure of EC 173 precipitates (van Genuchten et al., 2014a, 2014b, 2012), which we leverage in our interpretations of 174 bacteria attenuation and ζ-potential measurements. 175 176 2.3. Fe-EC and FeCl3 experiments 177 The procedure used for Fe-EC experiments has been described elsewhere (Delaire et al., 2015) and is 178 detailed in the Supporting Information. Briefly, two 1 cm × 8 cm Fe(0) electrodes were submerged in 200 179 mL of electrolyte spiked with bacteria (anodic submerged area of 3 cm2). In all experiments, a current 180 density of 10 mA/cm2 was applied for 11 min, resulting in a Faradaic Fe dosage of 0.5 mM. After the 181 electrolysis stage, suspensions were stirred open to the atmosphere for 90-180 min to allow for complete 182 Fe(II) oxidation and formation of Fe(III) precipitates. Suspensions were then left to settle overnight to 183 separate individual cells from cells associated with EC precipitates. When required for floc formation and 184 settling (Table S1), 5 mg/L-Al of Al2(SO4)3 (alum) was added at the end of the mixing period, along with 185 approximately 1.5 mM NaHCO3 to avoid a pH drop. Preliminary tests confirmed that the addition of alum 186 did not significantly modify bacteria attenuation (see Supporting Information). Solution pH was not 187 controlled during the settling stage. In a subset of experiments, coagulation by FeCl3 addition was used 188 instead of Fe-EC to isolate the contribution of removal from that of inactivation. In these experiments, 1 8 189 mL of a 100 mM FeCl3 solution was added to the electrolyte and the solution pH, which dropped to ~3 190 during FeCl3 addition, was re-adjusted to 7.0±0.1 in less than 5 min. 191 Unfiltered and filtered (0.45 μm nylon filters) samples were taken before Fe-EC, and before and after 192 overnight settling, for measurements of Fe, As, Ca, Mg, P, and Si by inductively coupled plasma optical 193 emission spectrometry (ICP−OES, PerkinElmer 5300 DV, measurement error typically < 5%). All 194 samples for ICP-OES analysis were digested in 0.2 M HCl. Filtered and unfiltered samples were used to 195 measure Fe(II) and total Fe (Fe(II) + Fe(III)), respectively (Delaire et al., 2015). Across the 113 bacteria 196 attenuation experiments reported here, the total Fe concentration after Fe-EC (Fe dosage) was 96% ± 7% 197 of the value predicted by Faraday’s law (0.5 mM). Unoxidized Fe(II) (before settling) and unsettled Fe 198 (after settling) were <1.2% and <4.7% of the total Fe dosed, respectively. Because the formation of 199 calcite, magnesite or hydroxyapatite in our experiments was limited if not negligible (see Supporting 200 Information), Ca/Mg/P removal measured by ICP-OES was used as a proxy for Ca/Mg/P uptake by EC 201 precipitates. Bacteria attenuation was calculated as the difference between log CFU concentrations before 202 Fe-EC and after settling (samples taken from the supernatant, ∼ 3 cm below the surface), and therefore 203 accounts for both inactivation and removal via enmeshment in flocs. Bacteria attenuation experiments 204 were generally replicated three or more times, except for 12 experiments conducted in duplicate or less 205 (see Table S2). We report average bacteria attenuations ± one standard deviation across replicates. 206 Finally, to assess the effect of P/Si on the uptake of carboxyl moieties by EC precipitates, we 207 performed citrate removal experiments using Fe-EC in the presence and absence of oxyanions under 208 conditions identical to E. coli removal experiments, using 10 mg/L-Al of alum before settling (Table S1). 209 Citrate concentrations were measured as total C with a TOC-VCSH analyzer (Shimadzu). 210 211 2.4. ζ-potential measurements and bacterial viability tests 212 In this study, ζ-potential measurements, which are a proxy for surface charge, were used to assess the 213 interaction of major groundwater ions with the surface of EC precipitates or E. coli K12 cells. ζ-potential 9 214 was measured by dynamic light scattering (Malvern Zetasizer Nano-ZS) at 633 nm. In addition, 215 qualitative assessments of membrane permeabilization, which were used as a proxy for bacteria 216 inactivation, were performed with the BacLight LIVE-DEAD kit (Invitrogen) used in conjunction with 217 fluorescent microscopy (Zeiss AxioImager, 63× Plan-Apochromat objective, EndoGFP and mCherry 218 filters, UC Berkeley CNR Biological Imaging Facility). Sample preparation and data collection 219 procedures are described in the Supporting Information. 220 221 2.5. Model of Ca/Mg complexation by bacterial cell walls 222 Drawing on previous work (Johnson et al., 2007; Ngwenya et al., 2003), we derived a simple 223 equilibrium surface complexation model, which included three bivalent cation adsorption sites on 224 bacterial cell walls: carboxyl groups, protonated and deprotonated phosphate groups. The model predicts 225 the percentage of bacterial phosphate and carboxyl groups complexed by Ca and Mg as: 𝐾 2+ 2+ 𝐾𝑃1,𝐶𝑎 [𝐶𝑎2+ ]+𝐾𝑃1,𝑀𝑔 [𝑀𝑔2+ ]+ 𝐴2 + (𝐾𝑃2,𝐶𝑎 [𝐶𝑎 ]+𝐾𝑃2,𝑀𝑔 [𝑀𝑔 ]) [𝐻 ] 226 %𝑃 𝑔𝑟𝑜𝑢𝑝𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑥𝑒𝑑 = [𝐻+] 227 %𝐶 𝑔𝑟𝑜𝑢𝑝𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑥𝑒𝑑 = [𝐻+] 𝐶,𝐶𝑎 𝐾 𝐾 [𝐶𝑎 2+ ]+𝐾𝑃2,𝑀𝑔 [𝑀𝑔2+ ]) + 1+ 𝐴2 +𝐾𝑃1,𝐶𝑎 [𝐶𝑎 2+ ]+𝐾𝑃1,𝑀𝑔 [𝑀𝑔2+ ]+ 𝐴2 (𝐾 𝐾𝐴1 [𝐻+ ] [𝐻+ ] 𝑃2,𝐶𝑎 𝐾 𝐾𝐴 [𝐶𝑎2+ ]+𝐾𝐶,𝑀𝑔 [𝑀𝑔2+ ] + 1+𝐾𝐶,𝐶𝑎 [𝐶𝑎2+ ]+𝐾𝐶,𝑀𝑔 [𝑀𝑔2+ ] ∗ 100 ∗ 100 (1) (2) 228 Deprotonation constants of bacterial surface functional groups and Ca adsorption constants were 229 obtained directly from the literature (Johnson et al., 2007). Mg adsorption constants were derived from a 230 relationship between metal-acetate and metal-bacteria complexation constants proposed by Johnson et al 231 (Johnson et al., 2007). Additional details regarding the derivation of this model, including equilibrium 232 constants, are given in the Supporting Information and Table S3. 233 234 235 3. Results and Discussion 3.1. Effect of HCO3- on the contributions of removal and inactivation 10 236 The effect of 8 mM HCO3- on E. coli attenuation by Fe-EC and FeCl3 coagulation is shown in 237 Figure 1a. Representative images of live-dead stained E. coli are presented in Figure 1b-e. Whereas 8 mM 238 HCO3- did not significantly affect E. coli attenuation by coagulation with FeCl3, the presence of HCO3- 239 decreased attenuation by Fe-EC by ~1.2 log. Because no reactive oxidants are produced from an Fe(III) 240 salt (Hug and Leupin, 2003), minimal inactivation occurs during FeCl3 coagulation (consistent with live- 241 dead staining, Figure 1b-c), which implies that attenuation via FeCl3 addition is exclusively due to 242 physical removal (enmeshment in flocs). Any difference in precipitate-bacteria adhesion between Fe-EC 243 and FeCl3 coagulation would lead to higher removal in the latter, because the precipitates generated by 244 FeCl3 coagulation in a HCO3- electrolyte are less crystalline and thus have a higher surface area than Fe- 245 EC precipitates (Schwertmann and Cornell, 2000; van Genuchten et al., 2014b; Voegelin et al., 2010). 246 Consequently, the difference in attenuations between Fe-EC and FeCl3 coagulation can conservatively be 247 attributed to inactivation. 248 As shown in Figure 1a, HCO3- did not affect physical removal, which is consistent with ζ-potential 249 measurements showing that HCO3- does not significantly interact with the surface of E. coli cells or 250 Fe(III) precipitates (Figure S1). By contrast, 8 mM HCO3- decreased inactivation substantially by ~1.2 251 log. We found a strong correlation between bacteria inactivation in Fe-EC (Figure 1a) and membrane 252 permeabilization (Figure 1d-e). Membrane damage may be caused by reactive intermediates such as O2●-, 253 H2O2, and Fe(IV), which are generated during Fenton-type reactions (Hug and Leupin, 2003; Keenan and 254 Sedlak, 2008) and have been associated with bactericidal effects (Alt et al., 1999; Ikawa et al., 2010; Kim 255 et al., 2010). The inhibition of inactivation by HCO3- might be explained by the formation of CO3●- 256 radicals, which are produced when HCO3- or Fe(II)-carbonate complexes react with H2O2 (Hug and 257 Leupin, 2003; Medinas et al., 2007). CO3●- is much more reactive than O2●-, H2O2, and Fe(IV) (Augusto 258 and Miyamoto, 2011; Jacobsen et al., 1998; Neta et al., 1988) (see Supporting Information), and is 259 therefore a much shorter-lived and less selective oxidant. Thus, we speculate that large HCO3- 260 concentrations reduce membrane damage and inactivation by shifting the nature of reactive species 11 261 produced during Fe-EC towards a shorter-lived oxidant (CO3●-) that is more likely to die off in the bulk 262 (e.g. reacting with Fe(II), Cl-, HCO3-) than to interact with cell membranes. 263 Overall, Figure 1 shows that both inactivation and removal (via enmeshment in flocs) contribute to 264 E. coli attenuation in Fe-EC, and that the concentration of HCO3- governs the amount of inactivation. In 265 the remaining sections of our study, we will focus on removal. Interactions between EC precipitates and 266 E. coli cells are investigated by varying levels of ionic strength, Ca, Mg, P, and Si. Because these ions are 267 not expected to react with oxidants such as O2●-, H2O2, or Fe(IV) (Hug and Leupin, 2003; Li et al., 2012; 268 Roberts et al., 2004), nor to interact with lipid aliphatic chains, which are the target of oxidants on cell 269 membranes (lipid peroxidation), they are assumed to have a negligible effect on inactivation. Therefore, 270 their potential impact on E. coli attenuation will be solely attributed to changes in removal. 271 272 273 3.2. Effect of ionic strength Increasing ionic strength over 2 orders of magnitude (2-200 mM), which results in increased charge 274 screening (Debye length decreased tenfold), did not significantly affect E. coli attenuation by Fe-EC, 275 regardless of the initial electrolyte composition (Figure S2). The negligible effect of ionic strength 276 suggests that electrostatic interactions play a secondary role compared to specific interactions in the 277 adhesion of EC precipitates to E. coli cells. In the following two sections, we investigate the bacterial 278 surface sites involved in these interactions by systematically varying the concentration of bivalent cations 279 and oxyanions in order to selectively complex adsorption sites on the surface of E. coli cells and EC 280 precipitates, respectively. 281 282 3.3. Effect of bivalent cations: Ca and Mg 283 3.3.1. 284 285 Single solute electrolytes (no oxyanions, no HCO3-) E. coli attenuation as a function of Ca and Mg concentrations is shown in Figure 2a. Ca and Mg both decreased E. coli attenuation, with a larger inhibitory effect observed for Mg (2.1 log decrease in 12 286 attenuation when Mg increased from 0 to 10.6 mM) than for Ca (1.3 log decrease in attenuation when Ca 287 increased from 0 to 12.9 mM). Because bivalent cations should not affect inactivation (see 3.1.), these 288 reductions in bacteria attenuation can be interpreted as reductions in E. coli removal. 289 Figure 2b shows the ζ-potential of EC precipitates and E. coli cells as a function of Ca/Mg 290 concentrations. In this single Ca/Mg solute electrolyte, EC precipitates were positively charged. 291 Increasing concentrations of Ca/Mg had a limited effect on the ζ-potential of precipitates, suggesting that 292 bivalent cations interacted minimally with their surface. This result was expected given the repulsive 293 electrostatic forces between bivalent cations and positively-charged EC precipitates, and is consistent 294 with previous work showing negligible uptake of Ca/Mg by Fe(III) (oxyhydr)oxides at circumneutral pH 295 in the absence of oxyanions (Kanematsu et al., 2013; Stachowicz et al., 2008). By contrast, Ca and Mg 296 caused a significant increase in the ζ-potential of E. coli cells, indicating a strong interaction between 297 bivalent cations and bacteria surfaces. Figure S3 shows the percentage of bacterial functional groups 298 complexed by bivalent cations, as predicted by our equilibrium surface model. According to this model, 299 raising Ca/Mg concentrations from 0 to 13 mM leads to a significant increase in the complexation of 300 carboxyl (from 0 to 70-80%) and phosphate (from 0 to 90-95%) groups, which is consistent with the 301 observed increase in E. coli ζ-potential (Figure 2b). 302 Figure 2c combines E. coli attenuation results (Figure 2a) and model outputs (Figure S3) to highlight 303 that E. coli removal decreases as the percentage of complexed bacterial carboxyl and phosphate groups 304 increases. Stronger inhibition of E. coli removal by Mg than by Ca (Figure 2a) is consistent with this 305 trend, because Mg has a higher affinity for bacterial surface functional groups (Beveridge and Koval, 306 1981) (Table S3 and Figure S3). 307 3.3.2. Groundwater-like electrolytes (with oxyanions and HCO3-) 308 Figure 2d shows the effect of Ca (0-13.5 mM) and Mg (2.4-10.5 mM) on E. coli attenuation in a 309 groundwater-like electrolyte containing 8 mM HCO3-, 1.2 mM Si, and 0.4 mM P. Similar to the single 13 310 Ca/Mg solute system, bivalent cations reduced E. coli attenuation, with Ca/Mg concentrations above 10 311 mM leading to a 1-2 log decrease in attenuation. 312 Figure 2e shows ζ-potentials of EC precipitates and E. coli cells as a function of Ca/Mg 313 concentrations in the groundwater-like electrolyte. Bivalent cations increased the ζ-potential of E. coli 314 cells, consistent with the complexation of phosphate and carboxyl groups on cell walls, as explained in 315 section 3.3.1. In this electrolyte, EC precipitates were negatively-charged due to the sorption of P and, to 316 a lesser extent, Si (P:Fe and Si:Fe molar solids ratios of 0.7 ± 0.1 and 0.06 ± 0.04, respectively) 317 (Appenzeller et al., 2002; Hamid et al., 2011). In contrast to previous experiments in the absence of 318 oxyanions, bivalent cations significantly interacted with the surface of EC precipitates in the 319 groundwater-like electrolyte, as indicated by a substantially higher ζ-potential at larger Ca/Mg 320 concentrations. This increase in precipitate surface charge coincided with increased Ca/Mg uptake, with 321 solids ratios going from 0.5 ± 0.1 to 1.2 ± 0.7 mol Ca:mol Fe, and from 0.3 ± 0.1 to 0.5 ± 0.4 mol Mg:mol 322 Fe, respectively. EC precipitates with similar chemical compositions (i.e. Ca/Mg:P:Fe molar ratios) have 323 been documented in previous studies performed in nearly identical electrolytes, but in the absence of 324 bacteria (van Genuchten et al., 2014a, 2014b). In these studies, Ca was shown to interact with P sorbed to 325 Fe(III) precipitates, via direct Ca-O-P bonds, and to a lesser extent, electrostatically. In the present study, 326 the observed increase in precipitate ζ-potential with Ca/Mg in the groundwater-like electrolyte is 327 consistent with such interactions of Ca/Mg with P sorbed to EC precipitates. 328 Figure 2f illustrates the inverse relationship between E. coli attenuation in the groundwater-like 329 electrolyte and the percentage of bacterial functional groups complexed by Ca/Mg (derived from our 330 model). Figure 2f also includes data from our previous study of E. coli attenuation in SBGW containing 331 2.6 mM Ca and 1.9 mM Mg (Delaire et al., 2015), which are consistent with this trend. Finally, we note 332 that E. coli attenuations in groundwater-like electrolytes (Figure 2f) were overall ~1 log lower than in 333 single solute systems (Figure 2c), which is consistent with the inhibition of inactivation by 8 mM HCO3- 334 shown in section 3.1. 14 335 Taken together, Figures 2a-f show that Ca/Mg decreases E. coli removal independent of the 336 electrolyte, and more specifically, independent of the surface charge of EC precipitates: whether Ca/Mg 337 increase (Figure 2b, no oxyanions) or decrease (Figure 2e, oxyanions present) the electrostatic barrier to 338 precipitate adhesion on cell walls, bivalent cations equally inhibit E. coli removal. Combined with the 339 limited impact of ionic strength (Section 3.2 and Figure S2), this result confirms the minimal role of 340 electrostatic interactions on E. coli removal and instead points to the importance of specific interactions 341 between EC precipitates and bacterial phosphate and/or carboxyl groups. These findings are in good 342 agreement with previous ATR-FTIR studies that provided evidence for direct bonding between Fe oxides 343 and bacterial phosphate/carboxyl groups in more simple and controlled systems (Elzinga et al., 2012; 344 Parikh and Chorover, 2006; Parikh et al., 2014). 345 346 3.4. Effect of oxyanions: P and Si 347 3.4.1. Single solute electrolytes (no bivalent cations, no HCO3-) 348 Figure 3a shows the effect of 0.4 mM Si/P on E. coli attenuation in electrolytes containing no Ca/Mg. 349 Whereas Si had no detectable effect, P reduced E. coli attenuation by 1.6 log. Because Si and P should not 350 affect inactivation, as explained in section 3.1, these effects correspond to changes in removal via 351 enmeshment in flocs. ζ-potential measurements of EC precipitates and E. coli cells as a function of P/Si 352 concentrations are presented in Figure 3b. Si and P had no detectable effect on the ζ-potential of E. coli 353 cells, reflecting the absence of interaction between these oxyanions and bacterial cell walls. By contrast, 354 Si and P significantly decreased the ζ-potential of EC precipitates, indicating oxyanion sorption 355 (Appenzeller et al., 2002; Hamid et al., 2011), which is supported by the uptake of Si and P measured by 356 ICP-OES (Si:Fe and P:Fe molar solids ratios of 0.3 and 0.6, respectively). Because electrostatic 357 interactions do not play a major role in E. coli removal, as demonstrated above, lower bacteria removal in 358 the presence of P cannot be explained by the decrease in precipitate surface charge. Rather, the results in 359 Figure 3a indicate that inorganic aqueous P competes with bacterial functional groups involved in 15 360 bonding to EC precipitates. By contrast, our results indicate that Si does not strongly compete with these 361 functional groups. 362 Because aqueous P and bacterial phosphate groups are structurally and chemically similar, they are 363 expected to compete for precipitate surfaces. However, the competition between P and carboxyl groups is 364 less straight-forward. To assess the effect of P on the adsorption of carboxyl moieties, we measured the 365 removal of citrate (a proxy for carboxyl groups) by Fe-EC in the presence and absence of P. As shown in 366 Figure 3c, P decreased citrate removal by nearly 54% (initial P:C molar ratio of 0.9). In E. coli attenuation 367 experiments, the molar ratio of aqueous P to bacterial surface carboxyl groups is ~ 2500 mol P: mol C 368 (see Supporting Information). Therefore, aqueous P is expected to strongly compete with bacterial 369 carboxyl groups in attenuation experiments. 370 Fe(III) (oxyhydr)oxides have a much higher affinity for P than for Si (Li et al., 2014; Roberts et al., 371 2004). Therefore, Si is not expected to effectively compete with bacterial phosphate groups for precipitate 372 surfaces. However, Figure 3c shows that Si decreased citrate removal in Fe-EC by nearly 20% (initial 373 Si:C molar ratio of 0.7). In E. coli attenuation experiments, where the molar ratio of Si to bacterial surface 374 carboxyl groups is orders of magnitude higher (~ 2500, see Supporting Information), it is thus likely that 375 Si would inhibit bacteria removal if carboxyl groups played an important role in the adhesion of EC 376 precipitates. Because Si had no detectable effect on E. coli attenuation (Figure 3a), we propose that 377 phosphate groups are the primary sites for the adhesion of EC precipitates to cell walls, with negligible 378 contributions from carboxyl groups. 379 3.4.2. Groundwater-like electrolytes (with bivalent cations, HCO3- and Si) 380 In Figure 3d, we show the effect of P (0-0.4 mM) on E. coli attenuation in the presence of Ca (2 and 9 381 mM) or Mg (8 mM) in a groundwater-like electrolyte containing 8 mM HCO3- and 1.2 mM Si. In contrast 382 to experiments in electrolytes free of bivalent cations, where P decreased E. coli removal by 1.6 log 383 (Figure 3a), 0.4 mM P had no effect on E. coli removal in the presence of Ca/Mg. We note that lower E. 16 384 coli attenuations in Figure 3d compared to Figure 3a (~ - 2 log) are due to the inhibition of inactivation by 385 8 mM HCO3- (shown in section 3.1) and to the reduction in removal caused by Ca/Mg (shown in section 386 3.3). 387 Figures 3e-f show ζ-potential measurements of EC precipitates and E. coli cells, respectively, as a 388 function of P concentration in the groundwater-like electrolyte containing bivalent cations. Figure 3e 389 shows that P did not interact significantly with bacterial cells, as expected. In contrast to single oxyanion 390 systems (Figure 3b), EC precipitates in the groundwater-like electrolyte were negatively-charged for all P 391 concentrations, due to sorbed Si/P. In addition, the ζ-potential of EC precipitates did not decrease when 392 the P concentration increased from 0 to 0.4 mM, despite substantial P uptake by precipitates (P:Fe molar 393 solids ratios of 0.6-0.8, see Table S4). This result stands in strong contrast with electrolytes containing no 394 Ca/Mg, where high concentrations of P (0.4 mM) and similar P:Fe solids ratios (0.6 mol:mol) 395 significantly decreased EC precipitate surface charge (Figure 3b). In the groundwater-like electrolytes, 396 ICP-OES measurements indicated that Ca/Mg uptake by EC precipitates increased by 20-200% – 397 depending on the initial Ca/Mg concentration– in the presence of 0.4 mM P (Table S4). This co-sorption 398 of Ca/Mg explains the negligible impact of P sorption on the surface charge of EC precipitates. 399 Based on the co-sorption of Ca/Mg and P, the behavior of precipitate and bacteria surfaces, and the 400 negligible effect of P on E. coli removal observed in our system, we propose that Ca/Mg can act as a 401 bivalent cation bridge between bacterial phosphate groups and P sorbed to EC precipitates. This Ca/Mg 402 configuration, which creates additional sites at the precipitate surface that can interact with bacterial cell 403 walls, is consistent with the Ca-P-Fe configurations documented previously in comparable systems (Senn 404 et al., 2015; van Genuchten et al., 2014a; Voegelin et al., 2010). 405 406 3.5. Attenuation of different types of bacteria 17 407 The attenuation of E. coli K12, E. coli ECOR 10 and E. faecalis in SBGW with an Fe dosage of 0.5 408 mM is shown in Figure 4. No significant difference between the log attenuations of the three different 409 bacterial strains was observed, despite their considerably different cell wall structures. For example, the 410 surface of Gram-positive E. faecalis is composed of a peptidoglycan layer topped with techoic acids, 411 whereas the surface of Gram-negative E. coli is made of phospholipids and lipopolysaccharides (LPS) 412 (Madigan et al., 2000). Furthermore, the two E. coli strains differ by the length of their LPS: ECOR 10 is 413 a smooth strain with a full-length LPS (with O-antigen), whereas K12 is a rough strain with a truncated 414 LPS (no O-antigen). Such differences in cell wall composition lead to differences in hydrophobicity, 415 surface charge, surface roughness and steric hindrance to approach mineral surfaces and nanoparticles 416 (Chen and Walker, 2012; Jacobson et al., 2015; Walker et al., 2004). 417 Previous studies have found that cell wall composition and LPS length affect the interactions of 418 bacteria with mineral surfaces (sand, iron-oxide coated sand, and gold nanoparticles) in systems governed 419 by non-specific interactions, such as electrostatic, steric, hydrophobic, and van der Waals forces (Chen 420 and Walker, 2012; Jacobson et al., 2015; Mohanty et al., 2013; Truesdail et al., 1998; Walker et al., 421 2004). In contrast to these studies, similar attenuation of E. coli K12, E. coli ECOR 10 and E. faecalis in 422 our system is likely due to the dominant role of specific interactions in bacteria-precipitate adhesion. 423 Phosphate functional groups, which we showed are the primary binding sites for EC precipitates, are 424 present in similar abundance on Gram-negative and Gram-positive bacteria (Borrok et al., 2005) (mainly 425 on phospholipids and on techoic acids, respectively), explaining similar removal of E. coli and E. faecalis. 426 In addition, negligible steric hindrance from longer LPS on E. coli is likely due to the small size of EC 427 precipitates compared to bacterial cells (Figure S4). 428 Based on these results, we expect that Fe-EC would be similarly effective for all waterborne 429 pathogenic bacteria, both Gram-negative (e.g. Vibrio cholera, Shigella, Salmonella, pathogenic E. coli) 430 and Gram-positive (e.g. E. faecalis, Bacillus cereus, Staphylococcus aureus). Finally, similar attenuation 431 of E. coli K12 and E. coli ECOR 10 suggests that fecal pathogens, which are typically smooth strains 18 432 (Felix and Pitt, 1935), would be as effectively removed as our model indicator E. coli K12. Overall, these 433 results are promising for the application of Fe-EC to drinking water or wastewater treatment. 434 435 4. Conclusions 436 In this study, we showed that bacteria inactivation, which can be significant in the absence of oxidant 437 scavengers, is largely suppressed by HCO3- concentrations characteristic of natural waters. Therefore, we 438 expect physical removal to be the primary process of bacteria attenuation in most water treatment 439 applications. Sludge sterilization before handling and disposal (e.g. via heat treatment) may therefore be 440 necessary as flocs may contain viable pathogens. 441 We have shown that removal is driven by the interactions of EC precipitates with bacterial phosphate 442 groups, which may bind to Fe(III) surfaces directly or via a Ca/Mg bridge to P sorbed on precipitates. In 443 light of these mechanisms, the contrasted effects of P and Si observed in this study can be generalized to 444 other strongly- (e.g., arsenate) and weakly- (e.g., borate, arsenite, nitrate) sorbing oxyanions, respectively. 445 Similarly, the observed impact of Ca/Mg (hardness) can be extrapolated to metallic bivalent cations that 446 may be present in wastewater, such as Cu2+, Cd2+, Pb2+, and Zn2+. 447 Consistent with the universal presence of phosphate groups on bacteria surfaces, Fe-EC is equally 448 effective towards Gram-positive and Gram-negative bacteria, rough and smooth alike. Our results 449 strongly suggest that Fe-EC, which is a technology applicable to decentralized arsenic remediation in 450 low-resource settings (Amrose et al., 2014; Holt et al., 2005), can also effectively remove all types of 451 bacterial contamination from a wide range of groundwater sources. Field validation of these promising 452 results as well as an investigation of virus attenuation are needed to confirm the potential of Fe-EC to 453 substitute for existing disinfection methods when applied to groundwater treatment. 454 455 Acknowledgements 19 456 This work was supported by the Development Impact Lab (USAID Cooperative Agreement AID-OAA- 457 A-13-00002) and the Andrew and Virginia Rudd Family Foundation Chair for Safe Water and Sanitation 458 administered by the Blum Center for Developing Economies. Additionally, C.M.v.G. acknowledges 459 funding support from the Netherlands Organization for Scientific Research (NWO) through a Veni grant. 460 We wish to express our gratitude to Kara Nelson, who provided useful discussions regarding the 461 complexation of Ca/Mg to bacterial functional groups, and to David Sedlak, Denise Schichnes, John 462 Wertz and Aidan Cecchetti for their kind assistance along various steps of this work. ζ-potential 463 measurements were conducted at the Molecular Foundry of Lawrence Berkeley National Laboratory and 464 supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy 465 under Contract No. DE-AC02-05CH11231. 466 467 Supporting Information 468 The Supporting Information provides detailed descriptions of experimental protocols (for Fe-EC 469 experiments, ζ-potential measurements, and fluorescent microscopy), the bacteria surface complexation 470 model, and the reactivity of strong oxidants produced in Fe-EC. Supporting figures and tables referenced 471 in the text are also included. 472 20 473 474 475 476 477 478 479 480 481 Figure 1: E. coli attenuation with Fe-EC and FeCl3, with and without 8 mM HCO3-. Fe dosage was 0.5 mM in all experiments. Panel a shows E. coli log attenuations. The asterisk indicates that the detection limit for bacteria attenuation was reached for some of the replicate experiments. Panels b-e show fluorescent microscopy images of live (green)-dead (red) stained E. coli cells. The blue dashed line is the average attenuation in all FeCl3 experiments (with and without HCO3-) and represents removal (blue arrow). E. coli log attenuations are compared to this baseline to deduce approximate log inactivations (red arrows). All experiments were conducted at pH 7.0. In 0.1 mM HCO3- experiments, 2 mM NaCl were added for conductivity. 482 21 483 484 485 486 487 488 489 490 491 492 493 494 Figure 2: Effect of Ca and Mg on E. coli attenuation in Fe-EC, in single solute electrolytes (panels a, b and c) and in groundwater-like electrolytes containing 8 mM HCO3-, 1.2 mM Si and 0.4 mM P (panels d, e and f). Panels a and d: effect of increasing Ca/Mg concentrations on E. coli log attenuation with an Fe dosage of 0.5 mM. The asterisk indicates that the detection limit for bacteria attenuation was reached for some of the replicate experiments. Panels b and e: effect of increasing Ca/Mg concentrations on the ζpotential of EC precipitates and E. coli cells (data points for 0 mM Ca and 0 mM Mg overlap on panel b). Panels c and f: E. coli attenuation as a function of complexed bacterial surface groups (combination of Figures 2a and S3, and 2d and S3 respectively). The dotted red lines highlight the inverse correlation between E. coli attenuation and the complexation of bacterial functional groups. All experiments were conducted at pH 7.0. Experiments with no Ca/Mg (panel a) were conducted in 2 mM NaCl for conductivity. 495 22 496 497 498 499 500 501 502 503 504 505 Figure 3: Effect of P and Si on E. coli attenuation by Fe-EC with an Fe dosage of 0.5 mM in single solute electrolytes (0.4 mM P or Si in 2mM NaCl background for conductivity; panels a,b and c) and groundwater-like electrolytes containing 8 mM HCO3-,1.2 mM Si and bivalent cations (panels, d, e and f). a) Effect of Si and P on E. coli attenuation. Asterisks indicate that the detection limit for bacteria attenuation was reached for some of the replicate experiments. b) Effect of Si (open symbols) and P (solid symbols) on the ζ-potential of EC precipitates and E. coli cells. c) Effect of P and Si on the removal of citrate (a proxy for carboxyl moieties) by Fe-EC. d) Effect of P on E. coli attenuation at different levels of Ca/Mg. e) Effect of P on the ζ-potential of EC precipitates. f) Effect of P on the ζ-potential of E. coli cells. All experiments were conducted at pH 7.0. 506 507 508 509 510 511 512 513 23 514 515 516 517 518 Figure 4: Log attenuation of three different bacterial strains by Fe-EC, at an Fe dosage of 0.5 mM. All experiments were conducted at pH 7.5 in SBGW (8.2 mM HCO3-, 2.7 mM Ca, 2.0 mM Mg, 1.3 mM Si, 0.15 mM P, and 6.3 µM As(III)). The log attenuation of E. coli K12 in SBGW shown here has also been reported elsewhere (Delaire et al., 2015). 519 24 520 521 Alt, E., Leipold, F., Milatovic, D., Lehmann, G., Heinz, S., Schömig, A., 1999. Hydrogen peroxide for prevention of bacterial growth on polymer biomaterials. Ann. Thorac. Surg. 68, 2123–2128. doi:10.1016/S0003-4975(99)00832-2 522 523 524 Amrose, S.E., Bandaru, S.R.S., Delaire, C., van Genuchten, C.M., Dutta, A., DebSarkar, A., Orr, C., Roy, J., Das, A., Gadgil, A.J., 2014. Electro-chemical arsenic remediation: field trials in West Bengal. Sci. Total Environ. 488-489, 539–46. doi:10.1016/j.scitotenv.2013.11.074 525 526 Appenzeller, B.M.R., Duval, Y.B., Thomas, F., Block, J.-C., 2002. Influence of Phosphate on Bacterial Adhesion onto Iron Oxyhydroxide in Drinking Water. Environ. Sci. Technol. 36, 646–652. doi:10.1021/es010155m 527 528 Arai, Y., Sparks, D.L., 2001. ATR–FTIR Spectroscopic Investigation on Phosphate Adsorption Mechanisms at the Ferrihydrite– Water Interface. J. Colloid Interface Sci. 241, 317–326. doi:10.1006/jcis.2001.7773 529 530 Augusto, O., Miyamoto, S., 2011. Oxygen Radicals and Related Species | Ohara Augusto - Academia.edu, in: Pantopoulos, H.M. (Ed.), Principles of Free Radical Biomedicine. Volume 1. Nova Science Publishers, Inc. 531 532 533 Barrera-Dı́az, C., Ureña-Nuñez, F., Campos, E., Palomar-Pardavé, M., Romero-Romo, M., 2003. A combined electrochemicalirradiation treatment of highly colored and polluted industrial wastewater. Radiat. Phys. Chem. 67, 657–663. doi:10.1016/S0969-806X(02)00497-8 534 535 Beveridge, T.J., Koval, S.F., 1981. Binding of metals to cell envelopes of Escherichia coli K-12. Appl. Environ. Microbiol. 42, 325–35. 536 537 Borrok, D., Turner, B.F., Fein, J.B., 2005. A universal surface complexation framework for modeling proton binding onto bacterial surfaces in geologic settings. Am. J. Sci. 305, 826–853. doi:10.2475/ajs.305.6-8.826 538 539 Chan, C.S., Fakra, S.C., Edwards, D.C., Emerson, D., Banfield, J.F., 2009. Iron oxyhydroxide mineralization on microbial extracellular polysaccharides. Geochim. Cosmochim. Acta 73, 3807–3818. 540 541 542 Chassé, A.W., Ohno, T., Higgins, S.R., Amirbahman, A., Yildirim, N., Parr, T.B., 2015. Chemical Force Spectroscopy Evidence Supporting the Layer-by-Layer Model of Organic Matter Binding to Iron (oxy)Hydroxide Mineral Surfaces. Environ. Sci. Technol. 49, 9733–41. doi:10.1021/acs.est.5b01877 543 544 Chen, G., Walker, S.L., 2012. Fecal indicator bacteria transport and deposition in saturated and unsaturated porous media. Environ. Sci. Technol. 46, 8782–90. doi:10.1021/es301378q 545 546 547 Delaire, C., van Genuchten, C.M., Nelson, K.L., Amrose, S.E., Gadgil, A.J., 2015. Escherichia coli Attenuation by Fe Electrocoagulation in Synthetic Bengal Groundwater: Effect of pH and Natural Organic Matter. Environ. Sci. Technol. 49, 9945–53. doi:10.1021/acs.est.5b01696 548 549 550 Elzinga, E.J., Huang, J.-H., Chorover, J., Kretzschmar, R., 2012. ATR-FTIR spectroscopy study of the influence of pH and contact time on the adhesion of Shewanella putrefaciens bacterial cells to the surface of hematite. Environ. Sci. Technol. 46, 12848–55. doi:10.1021/es303318y 551 552 Felix, A., Pitt, R.M., 1935. Virulence and Immunogenic Activities of B. typhosus in Relation to its Antigenic Constituents. J. Hyg. (Lond). 35, 428–36. 553 554 Filius, J.D., Lumsdon, D.G., Meeussen, J.C.L., Hiemstra, T., Van Riemsdijk, W.H., 2000. Adsorption of fulvic acid on goethite. Geochim. Cosmochim. Acta 64, 51–60. doi:10.1016/S0016-7037(99)00176-3 555 556 Ghernaout, D., Badis, A., Kellil, A., Ghernaout, B., 2008. Application of electrocoagulation in Escherichia coli culture and two surface waters. Desalination 219, 118–125. doi:10.1016/j.desal.2007.05.010 557 558 559 Hamid, R.D., Swedlund, P.J., Song, Y., Miskelly, G.M., 2011. Ionic strength effects on silicic acid (H4SiO4) sorption and oligomerization on an iron oxide surface: an interesting interplay between electrostatic and chemical forces. Langmuir 27, 12930–7. doi:10.1021/la201775c 560 561 Holt, P.K., Barton, G.W., Mitchell, C.A., 2005. The future for electrocoagulation as a localised water treatment technology. Chemosphere 59, 355–67. doi:10.1016/j.chemosphere.2004.10.023 562 563 Hug, S.J., Leupin, O., 2003. Iron-catalyzed oxidation of arsenic(III) by oxygen and by hydrogen peroxide: pH-dependent formation of oxidants in the Fenton reaction. Environ. Sci. Technol. 37, 2734–42. 564 565 Ikawa, S., Kitano, K., Hamaguchi, S., 2010. Effects of pH on Bacterial Inactivation in Aqueous Solutions due to LowTemperature Atmospheric Pressure Plasma Application. Plasma Process. Polym. 7, 33–42. doi:10.1002/ppap.200900090 566 567 Jacobsen, F., Holcman, J., Sehested, K., 1998. Reactions of the ferryl ion with some compounds found in cloud water. Int. J. Chem. Kinet. 30, 215–221. doi:10.1002/(SICI)1097-4601(1998)30:3<215::AID-KIN7>3.0.CO;2-V 25 568 569 570 571 Jacobson, K.H., Gunsolus, I.L., Kuech, T.R., Troiano, J.M., Melby, E.S., Lohse, S.E., Hu, D., Chrisler, W.B., Murphy, C.J., Orr, G., Geiger, F.M., Haynes, C.L., Pedersen, J.A., 2015. Lipopolysaccharide Density and Structure Govern the Extent and Distance of Nanoparticle Interaction with Actual and Model Bacterial Outer Membranes. Environ. Sci. Technol. 49, 10642–10650. doi:10.1021/acs.est.5b01841 572 573 Johnson, K.J., Szymanowski, J.E.S., Borrok, D., Huynh, T.Q., Fein, J.B., 2007. Proton and metal adsorption onto bacterial consortia: Stability constants for metal–bacterial surface complexes. Chem. Geol. 239, 13–26. 574 575 576 Kanematsu, M., Young, T.M., Fukushi, K., Green, P.G., Darby, J.L., 2013. Arsenic(III, V) adsorption on a goethite-based adsorbent in the presence of major co-existing ions: Modeling competitive adsorption consistent with spectroscopic and molecular evidence. Geochim. Cosmochim. Acta 106, 404–428. doi:10.1016/j.gca.2012.09.055 577 578 Keenan, C.R., Sedlak, D.L., 2008. Factors Affecting the Yield of Oxidants from the Reaction of Nanoparticulate Zero-Valent Iron and Oxygen. Environ. Sci. Technol. 42, 1262–1267. doi:10.1021/es7025664 579 580 Kelly, S.D., Hesterberg, D., Ravel, B., 2008. Analysis of soils and minerals using X-ray absorption spectroscopy, in: Methods of Soil Analysis. Part 5. Mineralogical Methods. 581 582 Kim, J.Y., Park, H.-J., Lee, C., Nelson, K.L., Sedlak, D.L., Yoon, J., 2010. Inactivation of Escherichia coli by nanoparticulate zerovalent iron and ferrous ion. Appl. Environ. Microbiol. 76, 7668–70. doi:10.1128/AEM.01009-10 583 584 Li, L., Li, J., Shao, C., Zhang, K., Yu, S., Gao, N., Deng, Y., Yin, D., 2014. Arsenic removal in synthetic ground water using iron electrolysis. Sep. Purif. Technol. 122, 225–230. doi:10.1016/j.seppur.2013.11.012 585 586 Li, L., van Genuchten, C.M., Addy, S.E.A., Yao, J., Gao, N., Gadgil, A.J., 2012. Modeling As(III) oxidation and removal with iron electrocoagulation in groundwater. Environ. Sci. Technol. 46, 12038–45. doi:10.1021/es302456b 587 Madigan, M.T., Martinko, J.M., Parker, J., 2000. Brock biology of microorganisms. Prentice Hall, Upper Saddle River NJ. 588 589 Mazel, D., Dychinco, B., Webb, V.A., Davies, J., 2000. Antibiotic resistance in the ECOR collection: integrons and identification of a novel aad gene. Antimicrob. Agents Chemother. 44, 1568–74. 590 591 McBride, M.B., Kung, K.-H., 1991. Adsorption of phenol and substituted phenols by iron oxides. Environ. Toxicol. Chem. 10, 441–448. doi:10.1002/etc.5620100403 592 593 Medinas, D.B., Cerchiaro, G., Trindade, D.F., Augusto, O., 2007. The carbonate radical and related oxidants derived from bicarbonate buffer. IUBMB Life 59, 255–62. doi:10.1080/15216540701230511 594 595 596 Miot, J., Benzerara, K., Obst, M., Kappler, A., Hegler, F., Schädler, S., Bouchez, C., Guyot, F., Morin, G., 2009. Extracellular iron biomineralization by photoautotrophic iron-oxidizing bacteria. Appl. Environ. Microbiol. 75, 5586–91. doi:10.1128/AEM.00490-09 597 598 599 Mohanty, S.K., Torkelson, A.A., Dodd, H., Nelson, K.L., Boehm, A.B., 2013. Engineering solutions to improve the removal of fecal indicator bacteria by bioinfiltration systems during intermittent flow of stormwater. Environ. Sci. Technol. 47, 10791–8. doi:10.1021/es305136b 600 601 Neta, P., Huie, R.E., Ross, A.B., 1988. Rate Constants for Reactions of Inorganic Radicals in Aqueous Solution. J. Phys. Chem. Ref. Data 17, 1027. doi:10.1063/1.555808 602 603 Ngwenya, B.T., Sutherland, I.W., Kennedy, L., 2003. Comparison of the acid-base behaviour and metal adsorption characteristics of a gram-negative bacterium with other strains. Appl. Geochemistry 18, 527–538. 604 605 Norén, K., Loring, J.S., Persson, P., 2008. Adsorption of alpha amino acids at the water/goethite interface. J. Colloid Interface Sci. 319, 416–28. doi:10.1016/j.jcis.2007.11.046 606 607 Parikh, S.J., Chorover, J., 2006. ATR-FTIR spectroscopy reveals bond formation during bacterial adhesion to iron oxide. Langmuir 22, 8492–500. doi:10.1021/la061359p 608 609 610 Parikh, S.J., Mukome, F.N.D., Zhang, X., 2014. ATR-FTIR spectroscopic evidence for biomolecular phosphorus and carboxyl groups facilitating bacterial adhesion to iron oxides. Colloids Surf. B. Biointerfaces 119, 38–46. doi:10.1016/j.colsurfb.2014.04.022 611 612 613 Roberts, L.C., Hug, S.J., Ruettimann, T., Billah, M.M., Khan, A.W., Rahman, M.T., 2004. Arsenic Removal with Iron(II) and Iron(III) in Waters with High Silicate and Phosphate Concentrations. Environ. Sci. Technol. 38, 307–315. doi:10.1021/es0343205 614 Schwertmann, U., Cornell, R.M., 2000. Iron Oxides in the Laboratory: Preparation and Characterization. Wiley. 615 Senn, A.-C., Kaegi, R., Hug, S.J., Hering, J.G., Mangold, S., Voegelin, A., 2015. Composition and structure of Fe(III)- 26 616 617 precipitates formed by Fe(II) oxidation in water at near-neutral pH: Interdependent effects of phosphate, silicate and Ca. Geochim. Cosmochim. Acta 162, 220–246. doi:10.1016/j.gca.2015.04.032 618 619 620 Stachowicz, M., Hiemstra, T., van Riemsdijk, W.H., 2008. Multi-competitive interaction of As(III) and As(V) oxyanions with Ca(2+), Mg(2+), PO(3-)(4), and CO(2-)(3) ions on goethite. J. Colloid Interface Sci. 320, 400–14. doi:10.1016/j.jcis.2008.01.007 621 622 STEC center, 2016. STEC center website [WWW Document]. URL http://shigatox.net/new/reference-strains/ecor.html (accessed 1.1.16). 623 624 Stevenson, G., Neal, B., Liu, D., Hobbs, M., Packer, N.H., Batley, M., Redmond, J.W., Lindquist, L., Reeves, P., 1994. Structure of the O antigen of Escherichia coli K-12 and the sequence of its rfb gene cluster. J. Bacteriol. 176, 4144–56. 625 626 Truesdail, S.., Lukasik, J., Farrah, S.., Shah, D.., Dickinson, R.., 1998. Analysis of Bacterial Deposition on Metal (Hydr)oxideCoated Sand Filter Media. J. Colloid Interface Sci. 203, 369–378. doi:10.1006/jcis.1998.5541 627 628 van Genuchten, C.M., Addy, S.E.A., Peña, J., Gadgil, A.J., 2012. Removing arsenic from synthetic groundwater with iron electrocoagulation: an Fe and As K-edge EXAFS study. Environ. Sci. Technol. 46, 986–94. doi:10.1021/es201913a 629 630 van Genuchten, C.M., Gadgil, A.J., Peña, J., 2014a. Fe(III) nucleation in the presence of bivalent cations and oxyanions leads to subnanoscale 7 Å polymers. Environ. Sci. Technol. 48, 11828–36. doi:10.1021/es503281a 631 632 633 van Genuchten, C.M., Peña, J., Amrose, S.E., Gadgil, A.J., 2014b. Structure of Fe(III) precipitates generated by the electrolytic dissolution of Fe(0) in the presence of groundwater ions. Geochim. Cosmochim. Acta 127, 285–304. doi:10.1016/j.gca.2013.11.044 634 635 636 Voegelin, A., Kaegi, R., Frommer, J., Vantelon, D., Hug, S.J., 2010. Effect of phosphate, silicate, and Ca on Fe(III)-precipitates formed in aerated Fe(II)- and As(III)-containing water studied by X-ray absorption spectroscopy. Geochim. Cosmochim. Acta 74, 164–186. doi:10.1016/j.gca.2009.09.020 637 638 Walker, S.L., Redman, J.A., Elimelech, M., 2004. Role of Cell Surface Lipopolysaccharides in Escherichia coli K12 adhesion and transport. Langmuir 20, 7736–46. doi:10.1021/la049511f 639 WHO, 2011. WHO | Guidelines for drinking-water quality, fourth edition. World Health Organization. 640 641 Yan, W., Wang, H., Jing, C., 2016. Adhesion of Shewanella oneidensis MR-1 to Goethite: A Two-Dimensional Correlation Spectroscopic Study. Environ. Sci. Technol. 50, 4343–9. doi:10.1021/acs.est.6b00066 642 643 27 Figure 1 Click here to download high resolution image Figure 2 Click here to download high resolution image Figure 3 Click here to download high resolution image Figure 4 Click here to download high resolution image Electronic Supplementary Material (for online publication only) Click here to download Electronic Supplementary Material (for online publication only): Supporting Information-Delaire et al.pdf