Hexavalent chromium removal by chitosan modified

0 downloads 0 Views 2MB Size Report
small peak at $891 cm. А1 corresponds to the saccharide structure of chitosan (Fig. 3) (Darder et al., 2003; Yuan et al., 2010; Paluszkiewicz et al., 2011; Silva et ...
Available online at www.sciencedirect.com

ScienceDirect Geochimica et Cosmochimica Acta 210 (2017) 25–41 www.elsevier.com/locate/gca

Hexavalent chromium removal by chitosan modified-bioreduced nontronite Rajesh Singh a,1, Hailiang Dong a,b,⇑, Qiang Zeng b, Li Zhang a, Karthikeyan Rengasamy c b

a Department of Geology and Environmental Earth Science, Miami University, Oxford, OH 45056, United States State Key Laboratory of Biogeology and Environmental Geology, China University of Geosciences, Beijing 100083, China c Department of Biology, Washington University in St. Louis, MO 63130, United States

Received 10 June 2016; accepted in revised form 19 April 2017; Available online 26 April 2017

Abstract Recent efforts have focused on structural Fe(II) in chemically or biologically reduced clay minerals to immobilize Cr(VI) from aqueous solution, but the coulombic repulsion between the negatively charged clay surface and the polyanionic form of Cr(VI), e.g., dichromate, can hinder the effectiveness of this process. The purpose of this study was to investigate the efficiency and mechanism of Cr(VI) removal by a charge-reversed nontronite (NAu-2), an Fe-rich smectite. Chitosan, a linear polysaccharide derived from chitin found in soil and groundwater, was used to reverse the charge of NAu-2. Intercalation of chitosan into NAu-2 interlayer increased the basal d-spacing of NAu-2 from 1.23 nm to 1.83 nm and zeta potential from 27.17 to +34.13 mV, with the amount of increase depending on chitosan/NAu-2 ratio. Structural Fe(III) in chitosan-exchanged NAu-2 was then biologically reduced by an iron-reducing bacterium Shewanella putrefaciens CN32 in bicarbonate buffer with lactate as the sole electron donor, with and without electron shuttle, AQDS. Without AQDS, the extent of Fe(III) reduction increased from the lowest (9%) for the chitosan-free NAu-2 to the highest (12%) for the highest chitosan loaded NAu-2 (3:1 ratio). This enhancement of Fe(III) reduction was likely due to the attachment of negatively charged bacterial cells to charge-reversed (e.g., positively charged) NAu-2 surfaces, facilitating the electron transfer between cells and structural Fe (III). With AQDS, Fe(III) reduction extent doubled relative to those without AQDS, but the enhancement effect was similar across all chitosan loadings, suggesting that AQDS was more important than chitosan in enhancing Fe(III) bioreduction. Chitosan-exchanged, biologically reduced NAu-2 was then utilized for removing Cr(VI) in batch experiments with three consecutive spikes of 50 mM Cr. With the first Cr spike, the rate of Cr(VI) removal by charged-reversed NAu-2 that was bioreduced without and with AQDS was 1.5 and 6 mmol g1 h1, respectively. However, the capacity of these clays to remove Cr(VI) was progressively exhausted upon addition of subsequent Cr spikes. X-ray photoelectron spectroscopy (XPS) revealed that the reduction product of Cr(VI) by chitosan-exchanged-bioreduced NAu-2 was Cr(III), possibly in the form of Cr(OH)3. In summary, our results demonstrated that the combined effects of sorption and redox reactions by charge-reversed bioreduced nontronite may offer a feasible in-situ approach for remediating Cr(VI) polluted soil and groundwater. Ó 2017 Elsevier Ltd. All rights reserved. Keywords: Charge-reversed nontronite; Structural Fe(II); Chitosan; Shewanella putrefaciens; Cr(VI) removal

⇑ Corresponding author at: Department of Geology and Environmental Earth Science, Miami University, Oxford, OH 45056, United States. Fax: +1 513 529 1542. E-mail address: [email protected] (H. Dong). 1 Current address: Department of Biology, Washington University in St. Louis, MO 63130, United States.

http://dx.doi.org/10.1016/j.gca.2017.04.030 0016-7037/Ó 2017 Elsevier Ltd. All rights reserved.

26

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

1. INTRODUCTION Release of chromium into the environment as a consequence of growing industrialization has been a major concern worldwide (McNeill et al., 2012). Although trivalent chromium [Cr(III)] at a trace level is an essential nutrient for plant and animal metabolism (glucose metabolism, amino- and nucleic acid synthesis), high levels of chromium, especially hexavalent chromium [Cr(VI)], can cause serious diseases such as nausea, skin ulcerations, and lung cancer (Richard and Bourg, 1991). Despite these health hazards associated with chromium exposure, its application in steelworks, electroplating, leather tanning, and production of dyes and pigments, is still extensive, which often causes major environmental problems (Dhal et al., 2013). Among the range of oxidation states (2 to +6) of chromium, [Cr(VI)] is considered highly toxic, owing to its weak adsorption on mineral surfaces and strong oxidizing nature. This toxicity of Cr(VI) is in contrast to the less soluble Cr (III), which readily forms insoluble oxides and hydroxides at slightly acidic to alkaline pH conditions (Cervantes et al., 2001; Motzer and Engineers, 2004; McNeill et al., 2012). Due to the associated toxicity of chromium to biological systems, the United States environmental protection agency (US EPA) has set a maximum contamination level (MCL) of total chromium in drinking water to be 0.1 mg L1 (Sutton, 2010). In order to address the growing problem of chromium contamination in the environment, a considerable number of studies have used chemical, physical and biological methods to remediate this contaminant. Among these techniques, adsorption has been recognized as a popular method due to its simplicity of operation, cost effectiveness, high efficiency, easy recovery, regeneration capacity, and sludge-free operation (Zhang et al., 2016). However, adsorptive removal does not change the toxicity of Cr (VI), and this process may be partially or wholly reversible. Furthermore, high efficiency is usually achieved only at low pH, which makes it difficult to implement in-situ except for acid mine drainage areas. To overcome these problems, researchers have used structural Fe(II) in iron-bearing clay minerals such as montmorillonite, nontronite, illite, vermiculite, and kaolinite to reduce soluble and toxic Cr(VI) to insoluble and less toxic Cr(III) (Gan et al., 1996; Taylor et al., 2000; Zhuang et al., 2012; Bishop et al., 2014). Because clay minerals are ubiquitous in natural environments, redox processes involving structural Fe(II) may, therefore, play an important role in environmental remediation of heavy metals such as chromium. However, one difficulty with such approach is the inevitable coulombic repulsion between negatively charged clay surface and dichromate polyanion, which may impede the removal efficiency of anionic Cr(VI) (Taylor et al., 2000). One possible solution to this problem, especially for application in engineered system, is through charge reversal of clay surface by certain surfactants. Indeed, a number of studies have used various polymers to reverse the charge of clay minerals for various environmental applications, mostly via sorptive removal of various pollutants, including heavy metals, radionuclides, and inorganic anions (Krishna et al., 2000;

Bleiman and Mishael, 2010; Su et al., 2012; Pentra´k et al., 2014), organic pollutants (Churchman, 2002), and dyes (Crini and Badot, 2008; Auta and Hameed, 2014). In addition, other aluminosilicates such as Fe(II)-modified zeolites (Kiser and Manning, 2010; Lv et al., 2014) and surfactantmodified zeolites (Li and Bowman, 1997; Li et al., 1999; Leyva-Ramos et al., 2008; Swarnkar et al., 2011; Song et al., 2015; Szala et al., 2015; Ren et al., 2016) have been successfully utilized to remove Cr(VI) from aqueous solution either by sorption or reduction mechanism. Among various surfactants used to reverse the surface charge of clay minerals, synthetic and natural polymers such as polydiallyldimethylammonium chloride (PolyDADMAC) (Su et al., 2012) and chitosan (Pentra´k et al., 2014) have been recently used. In these studies, the structural Fe(III) in charge-reversed smectites was first chemically reduced by sodium dithionite to Fe(II) and the resulting Fe(II) was subsequently used for successful removal of nitrate from aqueous solution. However, both synthetic polymers and chemical reductants may not be readily available in nature. In such case, biological reduction of structural Fe(III) in natural biopolymer-modified clay minerals may be more relevant, however, the potential of such materials for Cr(VI) removal is unknown. Moreover, due in part to abundance of microorganisms in nature, relatively low cost and environmentally friendliness, biological reduction of charge-reversed clay minerals may be achieved in-situ and may offer a more feasible approach to remediate polluted soil and groundwater. Chitosan, a type of biopolymer (linear polysaccharide, poly-(D) glucosamine), is synthesized by living organisms (White et al., 1979; Synowiecki and Al-Khateeb, 1997; Pochanavanich and Suntornsuk, 2002; Wan Ngah et al., 2011), and is the most abundant natural biopolymer in the environment after cellulose (Rinaudo, 2006). These compounds also have applications in agriculture. Fragments from chitin and chitosan are known to have eliciting activities leading to a variety of defense mechanisms in host plants in response to microbial infections (Hadrami et al., 2010; Hafdani and Sadeghinia, 2011). As a result, interest has been growing in amending soils with chitin and chitosan to reduce the negative impacts of diseases on crop yield and quality (Hadrami et al., 2010; Hafdani and Sadeghinia, 2011; Sharp, 2013). These activities could make chitosan readily available in soil and groundwater. Chitin and chitosan-derivatives have gained wide attention as effective sorbents due to their low-cost and high contents of amino and hydroxyl functional groups, which may serve as sites for sorbing various aqueous pollutants (Sag˘ and Aktay, 2002; Lito et al., 2012; Wan Ngah et al., 2012). Because these compounds possess unique properties (e.g., biodegradability, bioactivity, biocompatibility, and nontoxicity), these biosorbents are widely used for heavy metal removal (Zhang et al., 2016). However, chitosanmodified clay minerals have not been used for removing anionic pollutants such as dichromate. In addition, chitosan may be used as a model compound to understand how the surface charge of clay minerals can be reversed to enhance removal of other heavy metals. After solubilization of chitosan in acidic solution, its ionized amino groups

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

can strongly interact with either negatively charged ions by electrostatic attraction (Kumar, 2000; Rinaudo, 2006) or positively charged cations owing to its nitrogen atoms holding free electron doublets (Wang and Chen, 2014). Therefore, the positively charged amino groups of acid solubilized chitosan can compensate for the net negative charge of nontronite (NAu-2) (Dong, 2012). When chitosan is present in a sufficient amount, the overall charge of the clay mineral may even become positive. The objective of this research was to investigate the kinetics and mechanisms (sorption vs. reduction) of Cr (VI) removal by chitosan-modified, biologically reduced nontronite (NAu-2). We hypothesize that after intercalation of chitosan into clay mineral interlayer, the surface charge of clay minerals can be reversed from negative to positive so that heavy metal removal is maximized. In this study, a laboratory based experiment was performed by using various proportions of chitosan to nontronite (NAu-2) at 37 °C and pH 7 to address the following questions: (i) how do different loadings of chitosan impact bioreduction extent and rate of structural Fe(III) in NAu2 by Shewanella putrefaciens CN32? (ii) how are the rates and extents of Cr(VI) removal by chitosan-modified, bioreduced NAu-2 affected by different loadings of chitosan? (iii) what mechanisms are involved in Cr(VI) removal by chitosan-modified-bioreduced NAu-2? Wet chemical methods, zeta potential measurements, X-ray diffraction (XRD), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FTIR), and X-ray photoelectron spectroscopy (XPS) were used to characterize the chitosan-modified NAu-2, to investigate the kinetics of Cr (VI) removal by structural Fe(II), and to determine the oxidation state of reduced Cr. Our results demonstrated enhancement of Cr(VI) removal from aqueous solution by chitosan-modified, bioreduced NAu-2 relative to original NAu-2. This study addresses the reactivity of biologically reduced organo-clays towards heavy metal remediation and expands our current understanding of clay-natural biopolymer-microbe-metal interactions in natural environment. 2. MATERIALS AND METHODS 2.1. Clay mineral preparation Nontronite (NAu-2), an iron rich smectite end member, was purchased from the Source Clays Repository of the Clay Minerals Society (West Lafayette, IN). The material has the formula of M0.72(Si7.55 Al0.16Fe0.29)(Al0.34Fe3.54Mg0.05)O20(OH)4 (M represents the interlayer cation), where the total Fe content is 23.4%, with 0.6% in the form of Fe(II) (Jaisi et al., 2005). To mimic typical clay size range in natural environment, bulk NAu-2 was fractionated to obtain 0.02–0.5 mm size fraction by manually crushing bulk NAu-2 followed by soaking in 0.1 N NaCl solution and sonication for 8 h in a water bath to completely disperse aggregated particles. The 0.02–0.5 mm size fraction was then separated from other particle sizes by repeated centrifugation and re-suspension in DI water.

27

Chloride anion was removed by repeated washing in doubly distilled water and its complete removal was tested with AgNO3. Our previous study (Jaisi et al., 2005) verified that this size fraction is free of any other mineral impurities. The fractionated clay material was then dried in an oven at 60 °C for 24 h and re-dispersed in sodium bicarbonate solution (2.5 mM) with a stock concentration of 5 g L1 for later use. 2.2. Intercalation of chitosan into NAu-2 Chitosan, a natural biopolymer with average molecular weight of 100,000–300,000 Da was purchased from Acros Organics, a division of Thermo Fisher Scientific (Pittsburgh, Pennsylvania, USA). Chitosan stock solution was first prepared by dissolving chitosan in acetic acid (2% v/ v, pH  3). Desired chitosan to NAu-2 ratios were obtained by mixing different amounts of chitosan stock solution with NAu-2 stock suspension. First, different amounts of chitosan solution, e.g., 0, 0.4, 1, 4, 8 mL (from a 50 g L1 stock), and 10 mL (from a 60 g L1 stock) were added into 50 mL polycarbonate centrifuge tubes and then diluted to 10 mL with DI water (18 M X cm). To each of these tubes, 40 mL of 5 g L1 NAu-2 stock suspension was added. The final chitosan to NAu-2 weight ratio of 0:1 (no chitosan), 0.1:1, 0.25:1, 1:1, 2:1 and 3:1 was achieved in separate bottles. The samples were then mixed by shaking in a horizontal shaker at 130 rpm. Although previous studies have shown an equilibration time of 2 h between chitosan and clay (Radian and Mishael, 2008; Su et al., 2012; Pentra´k et al., 2014), the samples in this experiment were equilibrated overnight with agitation. The pH of the resulting chitosan-clay suspension increased to 4–6 depending on specific chitosan loading. Previous studies have shown that a pH range of 1.5–2.0 was required to dissolve nontronite (Bickmore et al., 2001), and at pH 3 or above, nontronite dissolution was limited (Grybos et al., 2010). Thus, at pH of 4–6, nontronite dissolution was not expected. After equilibration, samples were washed with DI water for 5 times to remove excess chitosan and acetic acid, with the final pH maintained at 7 using 1 N NaOH. Chitosan-modified NAu-2 samples were then freeze-dried. Unmodified NAu2 was dried overnight in an oven at 60 °C. Dried NAu-2 samples were then stored air-tight in polycarbonate tubes for later use. Because of different amounts of chitosan loading to NAu-2, the initial amount of Fe(III) in various samples was variable (Table 1). 2.3. Bacterial growth Shewanella putrefaciens CN32 is one of the extensively studied microorganisms that has the capability to carry out dissimilatory Fe(III) reduction in clay minerals (Liu et al., 2001; Dong et al., 2003, 2009; Lovley et al., 2004; Luan et al., 2015). The strain was routinely cultured aerobically in tryptic soy broth (TSB) (30 g/L) at 37 °C. Growth was monitored by DAPI (40 ,6-diamidino-2-phenylindole) fluorescent staining using an Olympus AX-70 multimode light microscope.

28

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

Table 1 Bioreduction initial rates and final extents of chitosan-exchanged NAu-2 by Shewanella putrefaciens CN32 with and without AQDS. NAu-2 samples

Unmodified w/o cells Unmodified w/cells 0.1:1 w/cells 0.25:1 w/cells 1:1 w/cells 2:1 w/cells 3:1 w/cells

Bioreduction w/o AQDS

Bioreduction w/AQDS 1

Initial Fe(III), mM Extent (%)

Rate (mmol g

24.36 ± 0.87 30.01 ± 1.63 26.06 ± 1.21 26.87 ± 0.42 24.91 ± 1.51 23.62 ± 0.74 23.33 ± 1.31

1.57 ± 0.05 0.72 ± 0.34 2.10 ± 0.48 2.11 ± 0.81 3.62 ± 0.40 4.79 ± 0.60 4.03 ± 0.20

0.22 ± 0.11 9.28 ± 0.14 11.56 ± 0.16 10.35 ± 0.22 10.83 ± 0.06 11.54 ± 0.19 12.46 ± 0.71

2.4. Bioreduction of Fe(III) in chitosan-modified NAu-2 Dried NAu-2 samples were re-dispersed in 2.5 g L1 bicarbonate buffer in 20 mL glass vials with a final conc. of 5 g L1 (working volume of 12 mL). After sealing with thick butyl rubber stoppers and aluminum crimps, the vials were purged (15 min each in solution and headspace) with mixed N2:CO2 (80:20) with two needles pierced through the stoppers to remove any oxygen. One needle served as inlet and the other as vent. The vials were then autoclaved for an hour at 121 °C. After the vials were cooled down to room temperature, cells in the exponential phase (prewashed with 30 mM bicarbonate buffer for 3 times) were injected into the vials inside an anaerobic glove box (with an atmosphere of 95% N2 and 5% H2; Coy Laboratory Products, Grass Lake, MI, USA). The final cell concentration was 1.7  107 cells mL1 as determined by DAPI counting. Filter sterilized lactate (final conc. 0.2 mM) was added as the sole electron donor. In another set of experiment, 0.1 mM of anthraquinone 2, 6-disulfonate (AQDS) was supplied as an electron shuttle. Bioreduction experiments were conducted for a total of 14 days at 37 °C. All experiments were run in duplicate with abiotic controls (without cells) and an average was obtained for timecourse Fe(II) concentration. Once the total Fe(II) concentration leveled off, bioreduction experiment was stopped by means of pasteurization to eliminate any further cellular activity (Jaisi et al., 2009; Zhao et al., 2013; Bishop et al., 2014). The initial rate of bioreduction was calculated for the first 24 h (Eq. (1)), because after this period, Fe(II) sorption would affect the rate. The extent of Fe(III) bioreduction (Eq. (2)) was calculated for the whole reduction period (two weeks). Initial Reduction rate ¼

Reduction extent ¼

ðFeðIIÞfinal  FeðIIÞinitial Þ within the first 24 h 24 h ð1Þ

ðFeðIIÞtotal  FeðIIÞinitial Þ  100%: FeðIIIÞtotal

ð2Þ

Prior to subsequent Cr(VI) reduction experiments, aliquots of bioreduced NAu-2 samples were washed four times with anoxic and autoclaved DI water in order to remove any residual lactate, AQDS, cell debris, or aqueous Fe(II) possibly produced from reductive dissolution of NAu-2. Therefore, any contribution of aqueous Fe(II) towards Cr

1

h ) Initial Fe(III) (mM) Extent (%) 24.36 ± 0.87 26.57 ± 1.84 25.48 ± 0.30 25.85 ± 0.89 24.69 ± 1.13 24.95 ± 0.07 25.14 ± 1.33

0.22 ± 0.09 24.62 ± 0.27 23.76 ± 0.28 24.03 ± 0.54 23.30 ± 0.30 25.88 ± 0.13 23.25 ± 0.45

Rate (mmol g1 h1) 3.14 ± 0.00 48.11 ± 0.00 48.11 ± 0.00 22.01 ± 0.00 16.80 ± 0.01 17.19 ± 0.00 15.70 ± 0.00

(VI) reduction should be negligible. All sample manipulations were performed inside an anaerobic glove box. 2.5. Cr(VI) reduction by chitosan-modified-bioreduced NAu-2 Cr(VI) reduction experiments were performed by utilizing structural Fe(II) in chitosan-modified-bioreduced NAu2 as the sole electron donor and Cr(VI) as the sole electron acceptor. Chitosan-unmodified NAu-2 (e.g., regular bioreduced NAu-2) was used as a control. A stock solution of Cr(VI) (500 mmol L1) was prepared by dissolving potassium dichromate (K2Cr2O7) (Sigma-Aldrich) in presterilized DI water and degassed with N2 gas. Stock suspension of chitosan-modified-bioreduced NAu-2 and filtersterilized Cr(VI) stock solution were combined in various proportions in pre-sterilized bicarbonate buffer to achieve a final Cr(VI) concentration of 50 mmol L1 and Fe(II) concentration of 400 mmol L1 in 20 mL glass vials (working volume 12 mL). Because bioreduced samples contained different amounts of Fe(II) (because of different chitosan/ NAu-2 loading and AQDS effect), concentrations of NAu-2 samples were varied in order to achieve the identical Fe(II) concentration (400 mmol L1). Assuming the products of Cr(VI) reduction by structural Fe(II) are Cr(III) and Fe(III) (Brigatti et al., 2000; Taylor et al., 2000; Zhuang et al., 2012), this structural Fe(II) concentration was in excess relative to 50 mmol L1 Cr(VI) assuming a theoretical ratio of 3:1. Excess Fe(II) was used because that some of the Fe(II) in NAu-2 samples may not be fully reactive. Because bioreduction extent was higher in presence of AQDS than in its absence, lower sample concentrations of were required to achieve the same Fe(II) concentration (400 mM for AQDS-amended, bioreduced NAu-2 samples). The glass vials were then sealed with thick butyl rubber stoppers and capped with aluminum crimps. Changes of Cr(VI) and Fe(II) concentrations were measured over time. Upon exhaustion of the first Cr(VI) spike, a fresh spike of 50 mmol L1 of Cr(VI) was added to the reaction vials to determine the reduction capacity of structural Fe(II) in chitosan-modified-bioreduced NAu-2 samples. This procedure was repeated for three times until there was no more Cr(VI) removal upon addition of another Cr(VI) spike. All sample manipulations were performed inside an anaerobic glove box. Measurements were taken in duplicate and an average was obtained for each reading.

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

2.6. Cr(VI) sorption by chitosan-exchanged-unreduced NAu-2 Unlike Cr(VI) reduction by structural Fe(II) in unmodified NAu-2 (Bishop et al., 2014), charge reversal of NAu-2 may result in significant Cr(VI) removal from aqueous solution by surface sorption through electrostatic attraction. To determine the relative amounts of Cr(VI) removal by sorption vs. reduction, a separate set of experiments were conducted using chitosan-modified but unreduced NAu-2. Since this NAu-2 was not bioreduced any Cr(VI) removal from aqueous solution should be due to sorption. A previous study (Bishop et al., 2014) has shown that the minor amount of structural Fe(II) initially present in unreduced NAu-2 (0.6%) does not reduce Cr(VI). To achieve a meaningful comparison, identical clay concentrations to those used in the Cr(VI) reduction experiments were used (Table 2) in bicarbonate buffer in 20 mL glass vials (working volume of 12 mL). The vials were sealed with thick butyl rubber stoppers and capped with aluminum crimps, followed by purging with N2/CO2 (80:20). After autoclaving at 121 °C for one hour, the cooled glass vials were passed inside an anaerobic glove box and filter-sterilized Cr(VI) stock solution was added into each vial to achieve a final concentration of 50 mmol L1. Cr(VI) concentration in aqueous solution was measured over time for a total of 100 h. All experiments were conducted in duplicates and an average was obtained for each reading. Combining the above two experiments (reduction + sorption and sorption only), the amount of Cr(VI) removal by reduction was then calculated by subtracting the amount of Cr(VI) removed by chitosan-modifiedunreduced NAu-2 [sorption only] from the amount of Cr (VI) removed by chitosan-modified-bioreduced NAu-2 [sorption + reduction]. 2.7. Analytical methods 2.7.1. Determination of total Fe(II) and Fe(III) by the 1,10phenanthroline method The Fe(II) and Fe(III) concentrations in all NAu-2 samples were determined spectrophotometrically (GENESYS 10 vis spectrophotometer, Thermal-Fisher, United States) using the 1,10-phenanthroline method (Amonette and Templeton, 1998). Because of the possibility of continuous reaction between residual Fe(II) in NAu-2 and aqueous Cr (VI) during the boiling step of the method (Bishop et al.,

29

2014), unreacted Cr(VI) in aqueous solution was first separated from NAu-2 by centrifugation (13,000 g for 10 min). Aqueous Cr(VI) concentration was measured in the supernatant, and Fe(II) concentration was measured in the clay pellet using the 1,10-phenanthroline method. Separate measurements without Cr(VI) indicated that aqueous Fe2+ concentration was negligible and should not contribute to any Cr(VI) reduction. 2.7.2. Cr(VI) measurement Aqueous Cr(VI) concentration was measured with the 1,5-diphenyl carbazide (DPC) colorimetric method (Urone, 1955). At each time point, 0.3 mL Cr-clay suspension was taken from the reaction vials and centrifuged at 13,000 g for 10 min. A small amount of supernatant (0.2 mL) was transferred to a 15 mL falcon tube containing 9.8 mL of 0.2 N H2SO4, and 0.5 mL of the diphenyl carbazide reagent. Concentration of Cr(VI) was then measured spectrophotometrically at the wavelength of 540 nm using a GENESYS 10 vis spectrophotometer. 2.7.3. X-ray diffraction (XRD) Effects of different chitosan loadings on the basal dspacing of NAu-2 were determined with XRD. Samples were prepared by smearing clay slurries onto petrographic slides and dried at 30 °C for 2 days. XRD patterns were collected with a Scintag X-ray powder diffractometer (CuKa radiation, k = 1.5418) with a fixed slit scintillation detector, and power of 1400 W (40 kV, 35 mA). XRD slides were scanned in 0.01 two-theta steps with a count time of 3 s per step and a scanning range of 2–12° two-theta. 2.7.4. Transmission electron microscopy (TEM) In addition to XRD, TEM was also employed to visually observe the effects of chitosan intercalation on the basal d-spacing of NAu-2 using JEOL JEM-2100 LaB6 TEM/ STEM at 200 keV accelerating voltage. This technique was further used to reveal the spatial relationship between NAu-2 particles and reduced Cr in Cr-reacted samples. Cr-unreacted samples were prepared by 50 dilution, pipetted onto 300 mesh copper grids, and dried overnight under room temperature, whereas Cr-reacted samples were washed with pre-sterilized, anoxic DI water for 3 times before pipetting onto copper grids and dried overnight inside an anaerobic glove box. For the Cr-unreacted samples, unmodified NAu-2 and a few modified ones (0.1:1, 1:1 and 3:1 chitosan/NAu-2 loadings) were used as repre-

Table 2 Cr(VI) sorption capacity of AQDS amended and un-amended unreduced NAu-2. Chitosan-modified, unreduced NAu-2 samples (Chitosan/NAu-2 ratios)

AQDS amended NAu-2

AQDS un-amended NAu-2

Clay conc., mg/mL

Sorption capacity, mmol Cr (VI)/mg NAu-2

Clay conc., mg/mL

Sorption capacity, mmol Cr (VI)/mg NAu-2

0:1 0.1:1 0.25:1 1:1 2:1 3:1

2.09 0.18 0.17 0.18 0.17 0.18

0.00 0.03 0.03 0.03 0.06 0.06

2.09 0.36 0.34 0.33 0.33 0.31

0.00 0.01 0.02 0.03 0.03 0.05

30

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

sentatives. For the Cr-reacted samples, the 1:1 chitosan: NAu-2 loading was used as a representative. TEM images were recorded using a Gatan Orius SC200D camera attached on a Gatan 863 Tridiem GIF Post-Column Energy Filter (Gatan Image Filter) and chemical composition was determined with EDS (Bruker AXS Microanalysis Quantax 200 with 4030 SDD detector). 2.7.5. Zeta potential Because the surface charge of chitosan-modified NAu-2 samples was found to play an important role in reducing nitrate (Pentra´k et al., 2014) and possibly Cr(VI), it was necessary to measure their surface charge. Freeze-dried, chitosan-modified NAu-2 samples were dispersed in DI water and stirred for 48 h. After complete dispersion, samples were transferred to an electrophoretic cell, and zeta potential was measured with a zeta meter 4.0 system. In addition, zeta potentials of chitosan-modified-bioreduced NAu-2 samples were also measured because bioreduction of structural Fe(III) in NAu-2 may result in more negative surface charge of the clay (Stucki and Kostka, 2006). 2.7.6. Fourier Transform Infrared Spectroscopy (FTIR) The interaction between chitosan and NAu-2 samples was investigated with FTIR (Perkin-Elmer Frontier) using an attenuated total reflection (ATR) unit. Approximately 10 milligrams of freeze-dried sample were evenly placed on a ZnSe crystal and pressed tight with an adjustable sample bar. Sample was analyzed immediately with fifty scans per spectrum over the 650–4000 cm1 range with a spectral resolution of 4 cm1. 2.7.7. X-ray photoelectron spectroscopy (XPS) XPS is a widely used technique for studying surface chemistry of various materials (Biesinger et al., 2004; Payne et al., 2011). In order to determine the valence state and the corresponding phase of chromium after the reaction with structural Fe(II) in clay (after third Cr spike), a representative sample (e.g., 0.1:1 chitosan:clay loading at the end of the 3rd Cr(VI) spike), was analyzed with XPS. The sample was washed with anaerobic ddH2O inside an anaerobic glove box and mounted over silicon wafer. After overnight drying inside the anaerobic glove box, the sample was analyzed with PHI VersaProbeII Scanning XPS Microprobe equipped with a monochromated Al Ka X-ray source. The XPS spectra were fitted assuming Gaussian– Lorentzian distribution for each peak with a Shirley background model and with binding energies referenced to the C 1s (Binding energy 284.8 eV). 3. RESULTS 3.1. Characterization of chitosan-exchanged NAu-2 Intercalation of chitosan into NAu-2 resulted in a progressive increase in the basal d-spacing of NAu-2, from 1.23 nm for the unmodified to 1.83 nm for a chitosan: NAu-2 ratio of 3:1 (Fig. 1a). However, TEM observations revealed smaller d-spacing, e.g., from 1.04 nm to 1.67 nm (Fig. 1b to e), likely due to NAu-2 layer collapse inside

the high vacuum of TEM column. Intercalation of chitosan into NAu-2 resulted in a progressive increase in zeta potential from 27.2 to +34.1 mV when the chitosan to NAu-2 ratio increased from 0:1 to 3:1 (Fig. 2). A full FTIR characterization of pure NAu-2 and chitosan endmembers was necessary in order to understand their interactions. The mid-infrared region (MIR) of the unmodified NAu-2 showed absorption peaks of OH and SiAO groups that are common for Fe-bearing smectites. The broad band between 3100 and 3750 cm1 was attributed to the OAH stretching of hydroxyl group in the octahedral layer (Anirudhan and Ramachandran, 2006). The band at 3570 cm1) is characteristic of the nontronite structure, in which Fe-Fe-OH grouping dominates the octahedral sheet (Madejova´, 2003). Broad bands at 3405 cm1 (HAOAH stretching) and 1625 cm1 (HAOAH bending) indicated the presence of adsorbed water on NAu-2 samples (Fig. 3) (Madejova´, 2003; Silva et al., 2012; Leita˜o et al., 2015). The band at 975 cm1 was assigned to the stretching vibration of SiAO, which is typically influenced by the presence of octahedrally coordinated iron atoms (Leita˜o et al., 2015). The bands between 650 and 900 cm1 could be attributed to octahedral bending deformations of RAOAH groups, in which R can be either Fe or Mg atoms (Silva et al., 2012; Leita˜o et al., 2015). Pure chitosan showed a broad band near 3346 cm1 (Fig. 3) attributed to OH and NAH stretching vibrations, aliphatic CAH stretching vibrations at 2927 and 2877 cm1 (Madejova´ et al., 2009, 2011; Pentra´k et al., 2014), carbonyl stretching modes of secondary amides at 1645 cm1, and bending modes of protonated amide (NAH) group near 1585 cm1 (Fig. 3) (Brugnerotto et al., 2001; Marchessault et al., 2006; Pentra´k et al., 2014). A group of bending vibrations of CAH associated with amide and CAC skeletal groups were clearly recognizable at 1421 cm1 and 1371 cm1, respectively. The bands at 1318 and 1254 cm1 represent symmetric stretching vibrations of the CH3 groups of tertiary amide and CAOAH bending, respectively (Paulino et al., 2006; Abugoch et al., 2011; Pentra´k et al., 2014). Further, the stretching vibrations at 1055 cm1 and 1033 cm1 were attributed to coupled CAOAN bending and CAO bending modes. The band located near 1150 cm1 is related to asymmetric vibrations of CO in oxygen bridge resulting from deacetylation of chitosan (Silva et al., 2012). The small peak at 891 cm1 corresponds to the saccharide structure of chitosan (Fig. 3) (Darder et al., 2003; Yuan et al., 2010; Paluszkiewicz et al., 2011; Silva et al., 2012). The mid-infrared region (MIR) of the chitosan-modified NAu-2 samples clearly showed the interaction between chitosan and NAu-2. The HAOAH stretching vibration of NAu-2 showed a progressive shift to a lower wave number after its interaction with different chitosan loadings (electronic annex, Table EA-1). A shift of the amide vibration band was expected when the NH+ 3 group interact electrostatically with the negatively charged sites of NAu-2 (Darder et al., 2003, 2005; Han et al., 2010; Silva et al., 2012). Indeed, the vibrational band at 1585 cm1 shifted to 1520 cm1 upon interaction with NAu-2, and with increased chitosan loading, this peak shifted to higher wave

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

31

Fig. 1. (a) XRD patterns of chitosan-exchanged NAu-2. A progressive increase in the basal d-spacing (0 0 1) from 1.23 nm to 1.469 nm, 1.623 nm, 1.738 nm, 1.801, and 1.828 nm is likely due to the insertion of chitosan into the interlayer to various extents. (b, c, d, e) HRTEM images showing the basal d-spacing of NAu-2 samples (b = unmodified, c = 0.1:1, d = 1:1 and e = 3:1). Smaller d-spacing values (by 0.2 nm) relative to XRD-determined values is likely due to layer collapse in the high vacuum of the TEM column.

numbers and increased in intensity (Fig. 3 and Table EA-1). The secondary amide band at 1645 cm1 of chitosan is overlapped with the HAOAH bending vibration band at 1625 cm1 of the adsorbed water molecules and therefore was not diagnostic of the NAu-2-chitosan interaction. The bands at 2877 and 2927 cm1 were indicative of CAH stretching from alkanes of intercalated chitosan in NAu-2 (Madejova´ et al., 2009, 2011; Pa´lkova´ et al., 2011). Similarly, a prominent shoulder in the 1050– 1200 cm1 region in chitosan-modified clay indicated ACAOACA vibrations in glycosidic linkages of chitosan (Madejova´ et al., 2009; Bleiman and Mishael, 2010; Kumirska et al., 2010; Su et al., 2012), again indicating intercalation of chitosan into NAu-2. 3.2. Microbial reduction of structural Fe(III) in chitosanmodified NAu-2 and subsequent physicochemical changes The effect of chitosan on Fe(III) bioreduction was investigated by comparing changes of initial rate and final extent across different chitosan loadings. Abiotic controls did not show any significant bioreduction without or with AQDS (Fig. 4a and b). In contrast, Shewanella putrefaciens CN32 reduced structural Fe(III) in all NAu-2 samples with varying initial rates and extents. In absence of AQDS, the

Fig. 2. Zeta potential of NAu-2 exchanged with various amounts of chitosan. A negative zeta potential of the unmodified NAu-2 (27.17 mV) progressively became more positive with increased loadings of chitosan. Bioreduction of chitosan-modified NAu-2 samples with AQDS exhibited less negative charge despite their higher reduction extents. Averages of two measurements from duplicate experimental tubes were reported.

32

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

Fig. 3. MIR spectra in the 4000–650 cm1 range for a. chitosan only, b. NAu-2 only, c. chitosan:clay (0.1:1), d. chitosan:clay (0.25:1), e. chitosan:clay (1:1), f. chitosan:clay (2:1), and g. chitosan:clay (3:1).

Fig. 4. Time course production of Fe(II) by Shewanella putrefaciens CN32 without AQDS (a) and with AQDS (b) of different chitosanmodified NAu-2 as measured by the 1, 10-phenantholine method. In contrast to the no-AQDS experiment, which showed a progressive increase in reduction extent with an increase in chitosan loading, the AQDS experiment showed similar reduction extents regardless of the amount of chitosan added. Error bars represent standard deviation from averages of two measurements from duplicate experimental tubes.

extent of reduction increased progressively from 9.2% for the unmodified NAu-2 to 12.5% for the highest chitosan loaded NAu-2 (3:1) (Table 1 and Fig. 4a). However, with AQDS, the extent of reduction doubled relative to that

without AQDS (23% to 25%), but the amount of chitosan loading in NAu-2 did not make any difference (Table 1 and Fig. 4b). Likewise, the presence of AQDS doubled initial rates of bioreduction. Fe(III) bioreduction resulted in a

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

33

Fig. 5. Cr(VI) removal and the corresponding structural Fe(II) oxidation by chitosan-modified-bioreduced NAu-2. A complete Cr removal was achieved by all bioreduced NAu-2 samples during the first Cr spike, and the rate of Cr(VI) removal was higher by bioreduced, AQDSamended samples than bioreduced, AQDS-unamended samples. During the 2nd Cr spike, both rate and extent became distinctly different among different chitosan loadings. (a) The Cr removal rate followed the order of 3:1 > 2:1 > 1:1 > 0.25:1 > 0.1:1 > unmodified-bioreduced NAu-2 by the bioreduced, AQDS-unamended NAu-2 samples; (b) The Cr removal order by the bioreduced, AQDS-amended NAu-2 samples followed: unmodified-bioreduced NAu-2 > 3:1 > 2:1 > 1:1 > 0.25:1 > 1:1. Cr removal was insignificant after the 3rd Cr spike. (c) The rate of Fe(II) oxidation in bioreduced NAu-2 followed the order of 0.1:1 > 0.25:1 > 1:1 > 2:1 > 3:1 > unmodified-bioreduced NAu-2; and (d) the rate of Fe(II) oxidation in bioreduced, AQDS-amended NAu-2 samples followed the order of unmodified-bioreduced NAu2 > 0.1:1 > 0.25:1 > 1:1 > 2:1 > 3:1. Error bars represent standard deviation from averages of two measurements from duplicate experimental tubes.

decrease in zeta potential, and the amount of decrease was more in the absence of AQDS than in its presence (Fig. 2). XRD patterns showed no secondary mineral formation as a result of bioreduction (shown in the electronic annex, Fig. EA-1). 3.3. Cr(VI) removal by chitosan-modified-bioreduced NAu-2 Different chitosan loadings resulted in different rates and extents of Cr(VI) removal. The first spike of aqueous Cr (VI) (50 mM) was completely removed within 100 h by all bioreduced NAu-2 samples. Because of rapid reduction, any difference in Cr(VI) removal rate and extent was not apparent among different chitosan loadings. Bioreduced NAu-2 samples that were bioreduced without AQDS showed a similar rate of 1.5 mmol g1 h1. However, during the reduction of the 2nd and the 3rd Cr(VI) spikes a difference showed up, where the NAu-2 sample with the

highest chitosan loading exhibited the highest Cr(VI) removal rate and extent, and the one without or little chitosan exhibited the lowest (Fig. 5a). This trend was consistent with the measured zeta potential, i.e., the more positive sample exhibited the highest rate and extent of Cr(VI) removal. Multiple spiking of Cr(VI) gradually decreased the reactivity of chitosan-modified-bioreduced NAu-2, ultimately leading to a complete stop of Cr(VI) removal, despite that some amounts of Fe(II) still remained in the samples after the 3rd Cr spike (Fig. 5c and d). Bioreduced NAu-2 samples that were bioreduced with AQDS showed a higher rate of Cr(VI) removal (6 mmol g1 h1, Fig. 5b). Again, because of rapid removal of the first Cr(VI) spike, any difference in the rate and extent was not apparent among different NAu-2 samples. During the 2nd and the 3rd Cr(VI) spikes, this difference became distinct (Fig. 5b). However, in comparison with the bioreduced, AQDS-unamended samples, the differ-

34

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

Fig. 6. Adsorption of Cr(VI) by chitosan-modified NAu-2 samples during the first 100 h, which corresponded to reduction of the first Cr(VI) spike. (a) Without AQDS; (b) with AQDS. The highest adsorption was achieved by the highest chitosan loaded NAu-2. The presence of AQDS did not make a difference to Cr(VI) adsorption. Error bars represent standard deviation from averages of two measurements from duplicate experimental tubes.

Fig. 7. Net Cr(VI) reduction by chitosan-modified-bioreduced NAu-2 during the first Cr spike (100 h). This net reduction was calculated by subtracting the amount of adsorbed Cr(VI) from the total removal (subtract Fig. 5a and b from Fig. 4a and b, respectively). (a) By bioreduced, AQDS-unamended NAu-2, (b) By bioreduced, AQDS-amended NAu-2. The data show that the highest Cr(VI) reduction was achieved by the least chitosan loaded NAu-2.

Fig. 8. The ratio of total Fe(II) oxidized over Cr(VI) reduced during the first Cr spike (100 h). The ratio was close to the theoretical ratio of 3:1 (dashed line). The chitosan-modified-bioreduced NAu-2 was produced without AQDS (a) and with AQDS (b). Higher ratios in some samples may be due to possible oxidation of Fe(II) during sample handing.

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

Fig. 9. Cr 2p XPS spectra of the 0.1:1 chitosan:NAu-2 bioreduced sample after reaction with the third Cr spike. The result shows 90% of Cr as Cr(III) and 10% as Cr(VI).

ence among these AQDS-amended samples was smaller. There was also a positive correlation between Cr(VI) removal extent/rate and chitosan loading. Surprisingly, the unmodified-bioreduced NAu-2 sample exhibited the highest rate and extent of Cr(VI) removal among this group (Fig. 5b), in contrast to the other bioreduced sample (e.g., without AQDS, Fig. 5a). Interestingly, the trend of Fe(II) oxidation did not correspond to that of Cr(VI) removal. For the NAu-2 samples that were bioreduced without AQDS, the relative order of Fe(II) oxidation (Fig. 5c) was almost opposite to the relative order of Cr(VI) removal (Fig. 5a). The highest Fe(II)

35

oxidation rate was observed for the sample with the lowest chitosan loading, where the Cr(VI) removal rate was the slowest. For the samples that were bioreduced with AQDS, the relationship between Fe(II) oxidation and Cr(VI) removal (compare Fig. 5b and d) was similar to the one for the bioreduced, AQDS-unamended samples, but unexpectedly the unmodified-bioreduced NAu-2 sample exhibited the highest Fe(II) oxidation rate and extent (Fig. 5d). To determine the relative contribution of sorption vs. reduction, a separate set of experiments was conducted using chitosan-modified but unreduced NAu-2 samples. The results revealed a positive correlation between the amount of Cr(VI) sorptive removal and the amount of chitosan loading (Fig. 6a and b). By the end of 100 h, the Cr (VI) sorption capacity was in the range of 0.01–0.05 and 0.03–0.06 mmol Cr(VI) per mg of chitosan-modified, unreduced NAu-2 sample in the absence and presence of AQDS, respectively (Table 2). When the amount of Cr(VI) sorptive removal (Fig. 6) was subtracted from the total amount of Cr(VI) removal (Fig. 5a and b), the resulting Cr(VI) reduction data (Fig. 7a and b) were consistent with their corresponding Fe(II) oxidation (Fig. 5c and d). The calculated ratio of the amount of structural Fe(II) oxidized over the amount of Cr(VI) reduced was fairly close to the stoichiometric ratio of 3–1 for all NAu-2 samples (Fig. 8a and b), which was expected assuming that the reaction products of Cr(VI) reduction by structural Fe(II) were Cr(III) and Fe(III) (Brigatti et al., 2000; Taylor et al., 2000; Zhuang et al., 2012; Bishop et al., 2014). Surprisingly, these data revealed that the highest amount of Cr(VI) reduction was observed for the sample with the lowest chitosan loading, progressively followed by higher chitosan loadings. XPS result for the 0.1:1 (chitosan:clay) sample after the third Cr spike showed two distinct bands at binding

Fig. 10. A representative STEM image, elemental maps, and EDS composition of a 1:1 chitosan-modified, bioreduced NAu-2 sample after reaction with Cr(VI) for 500 h.

36

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

energies of 577.78 eV and 587.48 eV, corresponding to Cr 2p3/2 and Cr 2p1/2 orbitals, respectively (Fig. 9), suggesting trivalent chromium, most likely in the form of Cr(OH)3 (Moulder et al., 1992). Additionally, XPS data also showed that 90% of chromium was in the form of Cr(III) and the remaining 10% as Cr(VI). STEM/elemental maps coupled with EDS clearly showed the association of reduced Cr phase with typical NAu-2 composition, Si, Fe, Al, and O in Cr-reacted samples with a possible formation of secondary Fe-Cr mineral (Fig. 10). 4. DISCUSSION 4.1. Characterization of chitosan-modified NAu-2 clays The protonated amino groups produced after the dissolution of chitosan in acidic solution (Rinaudo, 2006) are easily attracted to negatively charged clay surface by coulombic forces or substituted into interlayer via cation exchange (Bleiman and Mishael, 2010). These two mechanisms can eventually replace the exchangeable Na+ in NAu-2 on both external surfaces and in the interlayer position (Alexander, 1999; Zhang et al., 2014). Our data showed that with the excess amino group relative to the original amount of Na+, the surface charge of NAu-2 was reversed from negative to positive (Fig. 2). A similar charge reversal was reported in the previous studies, where polycations such as chitosan or poly-DADMAC reversed the surface charge of Na+-montmorillonite SWy-2 (Bleiman and Mishael, 2010) and ferruginous smectite SWa-1 (Su et al., 2012; Pentra´k et al., 2014). The gradual increase of the basal d-spacing of NAu-2 with increased chitosan loading suggests the intercalation of chitosan into NAu-2 as a dominant mechanism. At a low chitosan/NAu-2 ratio (0.1:1–1:1) one polymer chain should have entered the NAu-2 interlayer; whereas at higher ratios (2:1 and 3:1) two polymer chains may have entered the interlayer. This result was consistent with the previous observations of chitosan intercalation in ferruginous smectite (SWa-1) where low chitosan to clay ratio (0.045:1) resulted in a slight expansion of basal spacing by 0.02 nm corresponding to a polycation monolayer in clay interlayer, whereas clusters or multi-layer stacks formed and basal spacing increased by as much as 0.55 nm when the chitosan to clay ratio increased to 0.6:1 (Pentra´k et al., 2014). Our XRD and TEM results supported this model, where layer expansion gradually increased but eventually leveled off around 1.80 nm (Fig. 1), apparently due to the formation of multi-layer stacking of chitosan into the NAu-2 interlayer. Other studies have shown that the basal spacing of chitosanintercalated SWy-2, a different type of smectite clay, increased in two steps; at low chitosan loadings (0.024:1– 0.154:1) the basal spacing increased to 1.46 nm (corresponding to monolayer intercalation) and at higher loading (>0.3:1), it increased to 2.18 nm (corresponding to multilayer intercalation) (Darder et al., 2005; Bleiman and Mishael, 2010). Thus, the basal spacing of our modified NAu-2 samples (Fig. 1) should reflect the degree of chitosan intercalation in NAu-2 interlayers. Because of different

layer charge and expandability of smectite samples, different loadings of chitosan may be required in their interlayer region to transition from mono-layer to multi-layer stacking. 4.2. Effects of chitosan intercalation on bioreduction of NAu2 The observed enhancement of Fe(III) bioreduction by chitosan in the absence of AQDS (Fig. 4a), suggests that the positive charge of chitosan-modified NAu-2 samples facilitated the attachment of negatively charged bacterial cells to chitosan-modified NAu-2 surface such that the electron transfer between the cells and Fe(III) in NAu-2 became favorable. However, the presence of chitosan in the interlayer of NAu-2 would also impede electron transfer (Zeng et al., 2016), but its effect may be less severe and may not have been manifested in the overall results. Our data suggest that the enhancement effect dominated over the inhibitory effect. In the presence of AQDS, chitosan did not show any enhancement effect in Fe(III) bioreduction (Fig. 4b), likely due to the electron shuttling function of AQDS. In this case, bacterial attachment to clay surface was not required and thus there was no effect of facilitation from chitosan. Our data suggest that AQDS also played a role in electron transfer. With AQDS, the rates and extents of bioreduction were much higher than those without AQDS (Table 1, Fig. 4a and b) and these differences can be explained by the electron transfer pathway from cells to the structural Fe(III) within NAu-2. In the absence of AQDS, a direct contact between the bacterium and the mineral dictates the electron transfer pathway parallel to the (0 0 1) layers of NAu-2 (Dong et al., 2009) and bioreduction may be limited to Fe(III) near NAu-2 particle edges (Zhao et al., 2015; Shi et al., 2016). In this case, the observed bioreduction enhancement by chitosan may be largely achieved via increased attachment of bacterial cells to clay surface. In the presence of AQDS, electron transfer pathway may be both parallel and perpendicular to the (0 0 1) layers (Dong et al., 2009). AQDS and its reduced form AH2DS might be able to access Fe(III) within the nontronite structure that microorganisms have no access to and thus electron transfer from cells to the octahedral Fe(III) in the NAu-2 structure is facilitated (Liu et al., 2011; Zhang et al., 2012, 2014; Bishop et al., 2014; Liu et al., 2016). The two electron transfer pathways [parallel and perpendicular to the (0 0 1) layers] should have enhanced both the rate and extent of bioreduction in this experiment. Indeed, recent studies (Newmann et al., 2013; Zhao et al., 2015; Shi et al., 2016) suggests that electron transfer can take place both perpendicular and parallel to the basal plane of clay minerals, which further confirm previous speculations (Komadel et al., 2006; Stucki and Kostka, 2006; Dong et al., 2009). 4.3. Cr(VI) removal by chitosan-modified-bioreduced NAu-2 Uptake of metallic anions by NAu-2 and chitosanmodified NAu-2 is the direct effect of electrostatic interac-

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

tion between the reactive amine and hydroxyl sites of chitosan and metal polyanions (Guibal, 2004; Boddu et al., 2008; Wan Ngah et al., 2011). For unmodified NAu-2 surface, Cr(VI) sorption to edge OH group may be possible via the ligand exchange mechanism, as observed for phosphate (Dobias and Stechemesser, 2005). On the other hand, for chitosan-modified NAu-2, adsorption may occur due to the electrostatic interaction between chromate anion and oppositely charged chitosan amino groups. At neutral pH, about 50% of total amine groups remain protonated giving the polymer a cationic behavior and consequently providing potential sites to attract metal polyanions (Guibal, 2004). Thus, Cr(VI) removal by modified chitosan-NAu-2 composites observed in our study should be a combined effect of both sorption by electrostatic attraction described above and reduction mechanisms. For the first Cr(VI) spike, the higher rate of Cr(VI) removal (6 mmol g1 h1) by chitosan-modifiedbioreduced NAu-2 samples (produced with AQDS) than the equivalent ones produced without AQDS (1.5 mmol g1 h1) could be directly related to their more positive zeta potentials (Fig. 2), because AQDS itself was removed by repeated washing. More positive zeta potentials should have contributed to the higher removal rate of Cr (VI) not only by sorption but also by bringing dichromate anions closer to NAu-2 surfaces to facilitate the electron transfer between Cr(VI) and structural Fe(II) in these samples promoting the reduction of Cr(VI) to Cr(III). For the 2nd and 3rd Cr(VI) spikes, the Cr(VI) removal was slower and incomplete (Fig. 5a and b), likely due to decrease of Fe(II) reactivity and available sorption sites. A non-linear decline in Cr(VI) reduction rate was also observed previously when Fe(II) in bioreduced NAu-2 samples was reacted with multiple spikes of Cr(VI) (Bishop et al., 2014). In contrast to the first spike, the rate and extent of Cr(VI) removal during the 2nd Cr(VI) spike were positively correlated with the measured zeta potentials of these samples, e.g., the sample with more positive zeta potential exhibited higher rate and extent of Cr(VI) removal. These data suggest sorption as the dominant mechanism for Cr(VI) removal during the 2nd Cr(VI) spike. Indeed, our Cr(VI) sorption experiment using chitosan-modified but unreduced NAu-2 suggests that sorption of Cr(VI) would likely have continued beyond 100 h (Fig. 6a and b). However, by the 3rd Cr(VI) spike, the sorption capacity and the Fe(II) reactivity were apparently both exhausted. The exceptionally high rate and extent of Cr(VI) removal by the unmodified-bioreduced NAu-2 sample (Fig. 5b) was unexpected. Possible presence of AQDS in this sample could have enhanced the reduction of Cr(VI) for the following reasons: (1) this behavior was not observed for the equivalent sample without AQDS; (2) the exceptionally high rate and extent of Cr(VI) removal only occurred during the 2nd spike, when sorptive removal of Cr(VI) was the predominant mechanism. Because of lack of chitosan coating on the surface of this unmodified NAu2 sample, any residual AQDS would have greatly facilitated electron transfer from Fe(II) to Cr(VI). In contrast, for all other samples, the presence of chitosan would have blocked

37

electron transfer, and any residual AQDS may not have had an effect. Our calculated Cr(VI) reduction data for the 100 h, after accounting for the sorption, showed an unexpected trend, i.e., the NAu-2 sample with the lowest chitosan/NAu-2 ratio (e.g., the least positive charge) was the most reactive in reducing Cr(VI), but the one with the highest chitosan loading was the least reactive. Similar results have been reported previously, where nitrate reduction rate was the highest by chemically reduced SWy-2 that was chargereversed by a small amount of chitosan (Pentra´k et al., 2014) or PDADMAC (Su et al., 2012). This difference could be explained by a change in chitosan configuration on NAu-2 particle surface with increasing chitosan loading. At a low chitosan loading, the surface and the interlayer region of NAu-2 particles should be populated with monolayer chitosan, which would attract dichromate anions and enhance the electron transfer between structural Fe(II) in NAu-2 and aqueous Cr(VI). At higher chitosan loadings, chitosan may have formed multi-layer stacks, covering the physical surface of NAu-2 particles and clogging up the interlayer space. Although these multi-layer stacks may create more sorption sites for dichromate anions, they could also impede the electron transfer between structural Fe(II) and Cr(VI), eventually diminishing Cr(VI) reduction. Considering the fact that the majority of biogenic Fe(II) in NAu-2 resides within clay mineral structures with a negligible amount of aqueous Fe(II) (Bishop et al., 2011, 2014), which should have been removed by repeated washings, the reactive Fe(II) species of these clay minerals must have been structural Fe(II). The ratio of the amount of Fe(II) oxidized to the amount of Cr(VI) reduced during the first 100 h was fairly close to the theoretical ratio of 3:1 (Fig. 8a and b), consistent with previous studies suggesting that Cr(VI) was reduced to Cr(III) (Eary and Rai, 1988; Patterson and Fendorf, 1997; Kiser and Manning, 2010; Bishop et al., 2014). The higher ratios (3.2–3.8) in some of the samples could possibly be due to the oxidation of Fe(II) during sample handling or manipulations. A close association of the reduced Cr(III) with clay minerals (Fig. 10) is also consistent with our previous data and suggests a distribution of reduced Cr within NAu-2 matrix. 4.4. Implication for Cr(VI) immobilization A number of studies (Krishna et al., 2000; Bleiman and Mishael, 2010; Su et al., 2012; Pentra´k et al., 2014) have used various polymers to reverse the charge of clay minerals for sorptive removal of pollutants but its high efficiency is only achieved at lower pH, making its application difficult to the environment, except for perhaps in acid mine drainage areas. To address this problem, researchers have used structural Fe(II) in iron-bearing clays and clay minerals to reduce Cr(VI) to Cr(III) (Gan et al., 1996; Taylor et al., 2000; Zhuang et al., 2012; Bishop et al., 2014). However, the inevitable coulombic repulsion between negatively charged clay surface and dichromate anions may hinder removal efficiency. In such a scenario, the use of chitosanmodified bioreduced nontronite offers a better opportunity

38

R. Singh et al. / Geochimica et Cosmochimica Acta 210 (2017) 25–41

for Cr(VI) removal by increasing the extent of both sorption and reduction of Cr(VI) to Cr(III). Relative to iron oxides, our approach of using natural polymer modified, biologically produced Fe(II) in clay minerals can be of advantage in several respects: (1) their widespread occurrence in nature; (2) In those environments where natural reductive capacity of structural Fe(II) in minerals may have been exhausted by prolonged exposure to heavy metals, structural Fe(III) in clays can be regenerated via addition of organic carbon, thus this approach represents a renewable source of reducing power in sediments and groundwater aquifers (Ernstsen et al., 2006). Our recent studies have shown that structural Fe in clay minerals can undergo multiple redox cycles without losing any reactivity (Yang et al., 2012; Zhao et al., 2015); (3) Our experimental data show that Cr(VI) at a concentration as high as 50 mM can be decreased to an undetectable level within days. Although sorption alone is not adequate to remove Cr(VI) with a lower removal efficiency than that by Fe oxides (Ajouyeda et al., 2010), a combination of sorption and reduction achieved 100% removal efficiency, even at neutral pH. With the reduction step, 100% removal by Fe oxides was only achieved at acidic pH (