Science of the Total Environment 461–462 (2013) 19–27

Contents lists available at SciVerse ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Characterisation of Fe-oxide nanoparticles coated with humic acid andSuwannee River natural organic matter

Laura Chekli a,b, Sherub Phuntsho a, Maitreyee Roy c, Ho Kyong Shon a,b,⁎a School of Civil and Environmental Engineering, University of Technology, Sydney, Post Box 129, Broadway, NSW 2007, Australiab CRC CARE, PO Box 486, Salisbury, SA 5106, Australiac National Measurement Institute, PO Box 264, Lindfield, NSW 2070, Australia

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• HA and SRNOM were applied as surfacemodifiers to stabilise Fe2O3 NPs.

• Preferential adsorption for highmolecu-lar weight compounds was observed.

• Formation of small aggregateswith time,easily disaggregates with HA-coatedFe2O3 NPs.

• SRNOM-coated Fe2O3 NPs formedstronger aggregates than HA-coatedFe2O3 NPs.

⁎ Corresponding author at: School of Civil and Enviro2629; fax: +61 2 9514 2633.

E-mail address: Hokyong.Shon-1@uts.edu.au (H.K. S

0048-9697/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.scitotenv.2013.04.083

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 March 2013Received in revised form 27 April 2013Accepted 27 April 2013Available online xxxx

Editor: Damia Barcelo

Keywords:Dissolved organic matterIron oxide nanoparticlesFlFFFAggregationDisaggregation

Iron oxide nanoparticles are becoming increasingly popular for various applications including the treatmentof contaminated soil and groundwater; however, their mobility and reactivity in the subsurface environmentare significantly affected by their tendency to aggregate. One solution to overcome this issue is to coat thenanoparticles with dissolved organic matter (DOM). The advantages of DOM over conventional surfacemodifiers are that DOM is naturally abundant in the environment, inexpensive, non-toxic and readilyadsorbed onto the surface of metal oxide nanoparticles.In this study, humic acid (HA) and Suwannee River natural organic matter (SRNOM) were tested and com-pared as surface modifiers for Fe2O3 nanoparticles (NPs). The DOM-coated Fe2O3 NPs were characterisedby various analytical methods including: flow field-flow fractionation (FlFFF), high performance size exclu-sion chromatography (HPSEC) and Fourier transform infrared spectroscopy (FTIR). The stability of the coatedNPs was then evaluated by assessing their aggregation and disaggregation behaviour over time.Results showed that both HA and SRNOMwere rapidly and readily adsorbed on the surface of Fe2O3 NPs, pro-viding electrosteric stabilisation over a wide range of pH. HPSEC results showed that the higher molecularweight components of DOM were preferentially adsorbed onto the surface of Fe2O3. As SRNOM consists ofmacromolecules with a higher molecular weight than HA, the measured size of the SRNOM-coated Fe2O3

NPs was 30% larger than the HA-coated Fe2O3 NPs. FTIR results indicated the occurrence of hydrogen bondingarising from electrostatic interaction between the DOM and Fe2O3 NPs. Finally, a stability study showed that

nmental Engineering, University of Technology, Sydney, Post Box 129, Broadway, NSW 2007, Australia. Tel.: +61 2 9514

hon).

rights reserved.

20 L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

after 14 days, small agglomerates and aggregates were formed. The HA-coated Fe2O3 NPs formed agglomer-ates which were easily disaggregated using a vortex mixer, with the coated NPs returning to their initialsize. However, SRNOM-coated Fe2O3 NPs were only partially disaggregated using the same method, whichindicates that these aggregates have a more compact structure.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Anthropogenic contamination of water, soils and sediments withmanufactured nanoparticles, organic compounds and hazardous ele-ments is a worldwide problem stemming from many anthropogenicactivities, such as ore mining, ore processing and burning fossil fuels(Ribeiro et al., 2010; Silva et al., 2010). Over recent decades, someMNPs have attracted increasing attention due to their potential efficacyin the treatment of contaminated soil and groundwater (Crane andScott, 2012; Oliveira et al., 2012a, 2012b; Quispe et al., 2012).

Due to their low cost, highly reactive surface sites and high in-situreactivity, the most widely studied MNPs for soil and groundwaterremediation are nanoscale zero-valent iron (nZVI) nanoparticles(Wang and Zhang, 1997; Elliott and Zhang, 2001; Zhang, 2003).Nanoscale zero-valent iron particles have been shown to exhibithigh reactivity in remediating aquifers contaminated by nonaqueousphase liquids, hazardous element ions, and many other hazardouscompounds (Elliott and Zhang, 2001; Cundy et al., 2008; Geng et al.,2009). Delivering the nZVI to the contaminant source zone is essentialfor the success of in-situ remediation. However, many laboratory andpilot-scale field studies have demonstrated that the mobility and reac-tivity of iron-based nanoparticles are substantially limited in naturalporous systems such as soils and groundwater aquifers (Schrick et al.,2004; Quinn et al., 2005; He and Zhao, 2007; Saleh et al., 2007).

Aggregation is considered to be the primary cause of this reducedmobility and reactivity, and is the result of many factors includingsolution pH, ionic strength, and the presence of organic matter (Ponderet al., 2000; Saleh et al., 2005). To overcome this limitation, surfacemodification using charged polymers, polyelectrolytes or surfactantsis nowwidely used to disperse nanoparticles in environmentalmatricesof soil and water (Zhang et al., 1998; Schrick et al., 2004; Saleh et al.,2005; He et al., 2007; Saleh et al., 2007; Hajdú et al., 2009; Phenratet al., 2009; Sirk et al., 2009; Cirtiu et al., 2011). These modificationscan theoretically provide both electrostatic and steric (so-calledelectrosteric) stabilisation to prevent particles from aggregating andcan also reduce the propensity for surface attachment (Saleh et al.,2008). Although these different surface coatings can enhance nanopar-ticle stability, unfortunately, they can also be expensive, have toxiceffects on the environment, and alter the interaction of MNPs withcontaminants (Tiraferri et al., 2008).

Natural surface coating by the adsorption of dissolved organicmatter (DOM), such as humic and fulvic acids, on the surface ofnanoparticles has also been studied as an alternative “green” surfacecoating, and has been demonstrated to enhance nanoparticle stabilitythrough electrosteric stabilisation (Mylon et al., 2004; Illes andTombácz, 2006; Hu et al., 2010). The advantage of DOM over conven-tional surface modifiers is that DOM is naturally abundant in the envi-ronment, inexpensive, non-toxic, and has the ability to both adsorbonto metal oxide nanoparticles and complex with heavy metals (Liuet al., 2008; Dickson et al., 2012). A recent study by Zhang et al.(2013) has also demonstrated the capacity of humic acid (HA) coatediron oxide nanoparticles to remove organic dyes from wastewater.Finally, a study by Chen et al. (2011) demonstrated that DOM-coatednZVImay significantlymitigate bacterial toxicity due to the electrosterichindrance preventing direct contact.

In the present study, Fe2O3 nanoparticles (NPs) coated with DOMwere characterised using modern analytical methods: flow field-flowfractionation (FlFFF), high performance size exclusion chromatography(HPSEC) and Fourier transform infrared spectroscopy (FTIR), in order to

understand with greater confidence the interaction between DOM andFe2O3 NPs. The use of a multi-method approach for the characterisationof NPs has already been demonstrated in previous studies (Domingoset al., 2009; Cerqueira et al., 2011, 2012). Several characteristics wereinvestigated in terms of surface charge, size, adsorption capacity andchemical bonds. The aggregation and disaggregation behaviour of thecoated NPs was also investigated with FlFFF to assess their stabilityover time. Disaggregation is as important as aggregation for predictingthe fate and behaviour of NPs once released into the environment(Christian et al., 2008), however, there are only few studies availableon the disaggregation of NPs (Baalousha, 2009) and this is mainly dueto analytical challenges. The use of FlFFF to study the aggregation anddisaggregation behaviour of nanoparticles presents several advantagesover conventional size-measurement techniques. In particular, com-pared to dynamic light scattering which only measures an averageparticle size, FlFFF is a fractionation method and separation of the sam-ple allows accurate determination of the particle size distributionwhichis very useful for aggregation/disaggregation studies.

In addition, Fe2O3 NPs can be considered as representative of nZVI,since nZVI particles have been shown to have substantial shells ofiron oxide (Phenrat et al., 2007). Fe2O3 NPs demonstrate many similarproperties to nZVI when used to treat contaminated soil and ground-water, and thereby may be used as a model system for understandingaggregation behaviour (He et al., 2008).

2. Materials and methods

2.1. Chemicals and reagents

Commercially available Fe2O3 NPs (α-Fe2O3, average particle size30 nm, BET 50–245 m2/g, 20 wt.% dispersed in water at pH 4) andhumic acid (HA) (technical grade) were supplied by Sigma-AldrichAustralia. SRNOM was obtained from the International Humic Sub-stances Society (IHSS, St. Paul, USA). HA and SRNOM were employedas the DOM sources.

2.2. Sample preparation

Fe2O3 NPs were suspended in ultrapure water with a resistivity of18 MΩ cm (MilliQ, Millipore, USA) to obtain a final concentration of2 g/L at pH 4 ± 0.1. Solution pH was adjusted using 0.1 M HCl and0.1 M NaOH solutions and left for 24 h to equilibrate, after which thepH was re-measured and adjusted if necessary for all experiments.No buffers were used in this study, as their ionic strength may alterthe surface chemistry of the Fe2O3 NPs, enhancing their aggregation(Baalousha, 2009).

HA and SRNOMwere dissolved in ultrapurewater to obtain solutionswith a concentration of 500 mg/L. These were then filtered through a0.45 μm filter using vacuum suction. The filtrate was retained as thestock solution, and stored at 4 °C prior to experimental use. The totalorganic content (TOC) of the DOM solutions (dilution 1:10 of thestock solutions) was measured as 19.1 mgC/L and 18.9 mgC/L, for HAand SRNOM respectively, using a TOC analyser (TOC-VCPH, TNM-1,Shimadzu, Japan).

DOM-coated Fe2O3 NPs were prepared by mixing (i.e. using a mag-netic stirrer) 10 mL of concentrated Fe2O3 NPs (i.e. 2 g/L) with 10 mLof DOM stock solution and diluted with ultrapure water to obtain solu-tions with Fe2O3 NP concentrations of 200 mg/L and DOM concentra-tion of 50 mg/L. While stirring, the solutions were constantly kept at

Table 1Summary of the different FlFFF operating conditions used for particle size determination.

Sample Channel flow(mL/min)

Cross flow(mL/min)

Mobile phase

Fe2O3 NPs — 200 mg/L 1 0.5 Ultrapure water at pH 4HA and SRNOM — 50 mg/L Ultrapure water at pH 7HA-coated Fe2O3 NPs Ultrapure water at pH 7SRNOM-coated Fe2O3 NPs 0.15 Ultrapure water at pH 7

21L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

pH 4, as a previous study demonstrated that this is a favourable pH forDOM adsorption (Illés and Tombácz, 2004). Solutions were stirred for24 h, and samples were taken at different time intervals and measuredby the different analyticalmethods. Thefinal solutionswere then storedfor 14 days at ambient temperature for the stability study.

2.3. Characterisation of DOM-coated Fe2O3 NPs

2.3.1. Adsorption experimentsThe Fe2O3 NPs (200 mg/L) were equilibrated with HA and SRNOM

solutions (50 mg/L) for 24 h at ambient temperature as described inthe previous section. Samples were taken at different time intervals(i.e. 1 min, 2 min, 5 min, 10 min, 20 min, 30 min, 1 h, 2 h, 5 h, 10 hand 24 h) and then centrifuged for 10 min at 3500 rpm (Model2040, Centurion Scientific Ltd, UK) to separate the solution from thesolid particles. UV absorbance of the supernatant was then measuredvia a TOC analyser (TOC-VCPH, TNM-1, Shimadzu, Japan). The amountof DOM adsorbed on the surface of Fe2O3 NPs was calculated from thefollowing equation:

q tð Þ ¼ C0−Ctð Þ Vm

ð1Þ

where, C0 and Ct (mg/L) are the initial and concentration at time t ofDOM in solution, V (L) is the solution volume and m (g) is the mass ofthe Fe2O3 NPs.

Adsorption data were fitted to the pseudo first and pseudosecond-order kinetic models using linearised parameter estimations.The pseudo-second order kinetic model showed good correlationswhile the pseudo first-order model showed significantly low fit(R2 b 0.5). Therefore, in this study, the pseudo second-order kineticmodel was employed for data analysis. The linear form of thismodel can be described as (Febrianto et al., 2009):

tq tð Þ ¼

1kq2

eþ 1qe

ð2Þ

where, k (g/(mg·min)) is the rate of the pseudo second-order and qe(mg/g) is the amount of DOM adsorbed on the surface of Fe2O3 NPsat equilibrium. The k and qe values were calculated from the slopeand intercept of the y-axis obtained after plotting t/q(t) against trespectively.

2.3.2. Surface charge, average hydrodynamic diameter and particle sizedistribution

Dynamic light scattering (DLS) was used to evaluate the hydrody-namic diameter of the particles in the sample. DLS technique measuresthe diffusion coefficient of particles by correlating thefluctuations of thescattered light intensity over time. These fluctuations come from theBrownian motion of the particles and from the fact that neighbouringparticles can have constructive or destructive interference of thescattered light intensity in a certain direction (Hassellöv et al., 2008).Detailed information on these methods, including their physical princi-ples, mathematical models, and limitations can be found elsewhere(Filella et al., 1997). A Zetasizer (ZEN3600; λ = 633 nm; MalvernInstruments, UK) was used to determine zeta potential, by electropho-retic mobility analysis, and Z-average hydrodynamic diameter, bydynamic light scattering (DLS), of the samples. Zeta potential as a func-tion of pH was measured in the range pH 3–10 for the Fe2O3 NPs(200 mg/L), HA (50 mg/L) and SRNOM (50 mg/L). At the end of theadsorption experiment (i.e. after 24 h stirring), solutions of DOM-coated Fe2O3 NPs were pH adjusted (i.e. from pH 3 to pH 10) and zetapotential measurements were carried out at different pH.

The average hydrodynamic diameter of the DOM-coated Fe2O3

NPs was measured at 1 h, 2 h, 5 h and 10 h time intervals. Three

aliquots were measured per sample, to obtain the reported valuesand associated standard deviations.

FlFFFwas used to assess the distribution of hydrodynamic diametersin the samples. FlFFF is a chromatography-like separation techniquebased on laminar flow (so-called channel flow) in a very thin(i.e. ~250 μm) channel with a cross flow applied perpendicular to thechannel flow. The channel flow has a parabolic velocity profile (i.e. themaximum velocity is at the centre of the channel). The cross flow forcesthe particles to move toward a membrane at the channel wall, fromwhere they canmoveback into the channel as a result of diffusion forcesin the normal elution mode (i.e. for particles smaller than 1 μm). Thesmallest particles, having the highest diffusion coefficient, will migratefarther into the channel at higher flow rates and will thus elute first.The theory and principles of FlFFF can be found in the literature(Giddings, 2000; Phuntsho et al., 2011). The FlFFF used for particlesize determination was an asymmetrical AF2000 Focus (PostnovaAnalytics, Germany) with channel dimensions: length, 29.8 cm (tip totip); width, 2 cm; thickness, 0.25 mm. The detection system was anultra-violet visible (UV–vis) detector (λ = 254 nm; SPD20A, Shimadzu,Japan). The software AF2000 Control (v1.1.0.23, Postnova Analytics,Germany), was used to control the FlFFF system. A regenerated cellulosemembrane (Z-AF4-MEM-612-10KD, Postnova Analytics, Germany)withamolecularweight cut-off of 10 kDawas used as a channel wall. Sodiumazide (0.1 mMNaN3) was used as bactericide in the mobile phase for allexperiments. The sample volumes were 20.8 μL (i.e. size of the sampleloop) and were injected using a 50 μL syringe (Rheodyne Corporation,USA); at least three independent replicates were run per sample andthe data were averaged. In general, good agreement of replicates wasobserved (i.e. peak heights differing by less than 5% and peak maximadiffering by less than 2%). The FlFFF operating conditions aresummarised in Table 1.

Measurements were conducted after 1 h, 2 h, 5 h and 10 h ofstirring time using ultrapure water at pH 7 for the mobile phase for allsamples except the bare Fe2O3 NPs which were run in ultrapure waterat pH 4 due to their instability at pH 7. This pH (i.e. pH 7) was chosenfor the DOM-coated Fe2O3 NPs as it falls within the range of groundwa-ter pH (i.e. about 5.5 to 8.5) (Chi and Amy, 2004) and these coatednanoparticles are likely to be used for the purpose of soil and ground-water remediation. FlFFF measurements were made on the samesamples at the same time points as for DLS experiments to ensuredata comparability.

Latex beads of 22 nm, 58 nm and 100 nm (Postnova Analytics,Germany) were used to create calibration curves from which hydro-dynamic diameters of the particles were determined. These curvescorrelated the retention time to particle size. Calibration curves wereestablished for all mobile phases and conditions (change in cross flowor channel flow) used in this study and regularly (i.e. once a week)re-drawn to check the accuracy of data.

2.3.3. Extended stability of DOM-coated Fe2O3 NPs and disaggregationstudy

The stability of DOM-coated Fe2O3 NPs was assessed by measuringtheir size distribution by FlFFF after 14 days, without any perturbation,following first measurements. The instrument conditions for the FlFFFwere the same as those described in Section 2.3.2.

Fig. 1. Zeta potential profiles of Fe2O3NPs, DOM-coated Fe2O3NPs, HA and SRNOM.

22 L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

After measuring the size of the aggregates formed during this14-day period, a vortex mixer (VELP Scientifica, 1 min, 3000 rpm)was used to induce disaggregation to assess the stability of theformed aggregates and agglomerates. An alternate FlFFF system wasused for this study. This system consisted of an Agilent 1200 high per-formance liquid chromatography unit (Agilent technologies, USA) andan Eclipse 3+ FlFFF (Wyatt Technology, USA; channel dimensions:length, 26.55 cm (tip to tip); thickness, 0.35 mm). A regenerated cellu-losemembrane (Millipore PLGC,Wyatt Technology, USA)with amolec-ular weight cut-off of 10 kDa was used as the channel wall. The on-linedetection system for eluted particles comprised a UV–vis absorbancediode array detector (DAD1200, Agilent Technologies, USA) with aspectral range from 190 nm to 950 nm, and a dynamic light scatteringdetector (λ = 658 nm; scattering angle: 149°; Dawn Helios II, WyattTechnology, USA). The delivery flow to the FlFFF was software con-trolled (ChemStation v.B.04.02 SP1, Agilent Technologies, USA). Astrasoftware (v.6.0.2,Wyatt Technology, USA)was used for data acquisitionand data processing. The mobile phase was ultrapure water at pH 7 forall experiments. Operating conditions were channel flow of 1 mL/minand cross flow of 0.5 mL/min.

2.3.4. Characterisation of binding properties between Fe2O3 NPs andDOM

2.3.4.1. HPSEC analysis. The size distributions of DOM solutions beforeand after adsorption on Fe2O3NPswere determined byHPSEC. HPSEC isa low resolution chromatography technique which separates particleson the basis of molecular hydrodynamic size. In an HPSEC column, thesmaller molecules are trapped in the pores of the gel. The larger mole-cules simply pass by the pores as they are too large to enter it. Therefore,the larger molecules will elute quicker than the smaller ones (Mori andBarth, 1999). HPSEC used in this study (Shimadzu, Japan) consisted of aglycol-functionalised silica gel column (Protein-Pak 125, Waters, USA)with fluorescence detector (RF-10A, Shimadzu, Japan). Standard

Fig. 2. FlFFF fractograms of HA-coated and SRNOM-coa

polystyrene sulfonates (PSS: 210, 1800, 4600, 8000, and 18,000 Da,Polymer Standards Service, Germany) were used to calibrate the equip-ment. Details of the measurement methodology are given elsewhere(Shon et al., 2004). Aflow rate of 0.7 mL/minwas used. All injection vol-umes of the samples were 100 μL. Samples used in HPSEC experimentswere the same as for DLS and FlFFF experiments to ensure datacomparability.

2.3.4.2. FTIR analysis. Chemical bonding information on metal-oxygen,hydroxyl, and other functional groups was obtained with FTIR spec-troscopy using the IRAffinity-1 (FTIR-8400S, Shimadzu, Japan). Infra-red spectra were recorded on ZnSe through plate (PIKE technologies,USA). Each spectrum is the sum of 25 scans at a resolution of 2 cm−1.

ted Fe2O3NPs after different mixing time at pH 4.

Fig. 3. Adsorption kinetics of HA and SRNOM on Fe2O3NPs at pH 4 (A) experimental re-sults and (B) pseudo second-order kinetic model.

23L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

Samples of bare and coated Fe2O3 NPs (after 24 h stirring) werecompletely dried before performing measurements. For the coatedNPs, samples were first centrifuged and the supernatant discardedto remove the excess of DOM.

3. Results and discussion

3.1. Surface charge of DOM-coated Fe2O3 NPs

Fig. 1 shows the zeta potential profiles of Fe2O3NPs alone (200 mg/L),DOM-coated Fe2O3 NPs and DOM alone (50 mg/L), as a function of pH,ranging from 3 to 10. The zeta potential profile of Fe2O3 NPs showedthat the nanoparticles were highly positively charged at low pH values(i.e. pH 3–5). The zeta potential decreased as pH increased from 5 to 9and became highly negative above pH 10 with a point of zero charge(PZC) at around pH 7. This value is within the range of PZC valuesfound in the literature for Fe2O3 NPs (Tombácz et al., 2004; Illes andTombácz, 2006; Baalousha et al., 2008; Baalousha, 2009; Hu et al.,2010).

The zeta potential profiles of HA and SRNOM indicate that they arenegatively charged over the whole pH range. This is due to the fact thatDOMmacromolecules carry several negatively charged functional groups,including carboxylic and phenolic groups (Hajdú et al., 2009; Hu et al.,2010; Dickson et al., 2012). The zeta potential profiles of DOM-coatedFe2O3 NPs also remain negative across the whole pH range tested andare quite similar to the zeta potential profiles of the DOM. This indicatesthat both HA and SRNOM cover the surface of the bare Fe2O3 NPs, pro-viding electrostatic stabilisation over a wide range of pH.

3.2. Particle size and size distribution determination by FlFFF and DLS

The hydrodynamic sizes of the bare and DOM-coated nanoparticleswere firstly determined by FlFFF (Fig. 2). Measurements were made atpH 7 for the DOM-coated Fe2O3 NPs to assess their stability undergroundwater conditions (i.e. groundwater pH is usually within therange 5.5–8.5 (Chi and Amy, 2004)). Moreover, at this pH, the DOM-coated Fe2O3 NPs are highly negatively charged (as displayed in Fig. 1)which should theoretically enhance electrostatic stabilisation comparedto pH 4.

The FlFFF data allow a direct comparison between the bare andcoated particles. By comparing the fractograms of the bare and coatedFe2O3 NPs (Fig. 2), two key observations can be identified. Firstly,there was a significant increase in the void peak UV signal (i.e. from0.06 a.u. for the bare nanoparticles to about 0.35 a.u. and 0.75 a.u.for all HA-coated and SRNOM-coated nanoparticles respectively).This may be caused by unadsorbed DOM macromolecules which areconsiderably smaller than the Fe2O3 NPs. The applied cross flow wastoo low to retain the unadsorbed DOM, and so the elution ofunretained DOM was indicated by the larger void peak. When thestirring time increased from 1 h to 10 h (i.e. labelled (c) to (f) inFig. 2), a decrease in the void peak signal was observed on thefractograms of both DOM-coated Fe2O3 NPs. This decrease in thevoid peak signal can be explained by the increasing adsorption ofDOM onto the surface of the Fe2O3 NPs resulting in decreasinglyfewer free DOM species in the solution and therefore less elutedDOM in the void peak.

The second observation identified in the FlFFF data is a small shifttoward smaller sizes of the peak maxima (by comparing fractograms(c) with fractograms (d) to (f) for both coated nanoparticles), probablyindicating the formation of more stable coated particles. The shiftis more important for the HA-coated Fe2O3 NPs (i.e. peak maximadecreasing by 11.1% against 4.8% for the SRNOM-coated Fe2O3 NPs),whichmay indicate that the SRNOM-coated Fe2O3NPs reach equilibriummore rapidly (which was confirmed by the adsorption experimentsdisplayed on Fig. 3A and B). The broadening of the peak by comparingthe fractograms of the bare (i.e. fractograms (a)) and coated Fe2O3 NPs

(i.e. fractograms (c) to (f)) may be caused by the coated particles havingdifferent size and conformation. It should also be noted that the peaks inthe fractograms of HA-coated Fe2O3 NPs are broader than those ofSRNOM-coated Fe2O3 NPs. This may indicate that SRNOM-coated Fe2O3

NPs are more stable since the size distribution of the coated particles isnarrower, indicating less aggregation. On the fractograms of both coatedFe2O3 NPs, a second peak was observed between the void peak and theelution peak. This can be attributed to the formation of small aggregatesof DOM macromolecules. In fact, the coated-particles were prepared atpH 4 and at this pH, both HA and SRNOM are less negatively chargedthan at higher pH (as displayed in Fig. 1) which could promote theiraggregation.

These FlFFF results were compared with those from DLS. In general,the sizes measured by DLS (Table 2) were larger than FlFFF. DLS isknown to be extremely sensitive to larger particles and a very smallnumber of large particles (e.g. due to the formation of aggregates), caninduce a substantial shift toward larger sizes (Domingos et al., 2009).Moreover, DLS measurements were made at pH 4, pH at which thecoated nanoparticles are less negatively charged which might promotethe formation of some aggregates.

Despite differing in absolute values, sizemeasurements by FlFFF andDLS show similar trends. Both the hydrodynamic diameter (from FlFFF)and z-average hydrodynamic diameter (from DLS) of the coated parti-cles slightly decreased with increasing stirring time. Also, both FlFFFand DLS results indicate that the size of the SRNOM-coated NPs is largerthan those of HA-coated NPs. This is consistent with SRNOM having alarger molecular weight than HA (as shown on the HPSEC chromato-grams in Fig. 4).

Table 2Z-average hydrodynamic diameter of bare Fe2O3 NPs, HA-coated Fe2O3 NPs and SRNOM-coated Fe2O3 NPs as determined by DLS (at pH 4 for bare Fe2O3 NPs and at pH 7 forDOM-coated Fe2O3 NPs).

Z-average hydrodynamic diameter (nm)

Bare Fe2O3 NPs 63 ± 4

Mixing time HA-coated Fe2O3 NPs SRNOM-coated Fe2O3 NPs

1 h 95 ± 4 127 ± 52 h 91 ± 3 124 ± 45 h 90 ± 3 124 ± 310 h 89 ± 2 122 ± 3

24 L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

3.3. DOM adsorption to Fe2O3 NPs

The adsorption kinetics of both HA and SRNOM to Fe2O3 NPs(Fig. 3A) show a steep initial slope before reaching a plateau at equi-librium, implying a high affinity of binding sites for both HA andSRNOM at pH 4 (Kang and Xing, 2008). Illés and Tombácz (2004) dem-onstrated that the adsorption of DOM is favourable under acidic condi-tions where negatively charged functional groups of DOM are attractedby the positively charged Fe2O3 NPs.

The adsorption data were then fitted with the pseudo second-order kinetic model as shown in Fig. 3B. The results indicated that thecorrelation coefficient (i.e. R2) for both DOM was higher than 0.999and the calculated equilibrium adsorption capacity (i.e. qe) was consis-tent with the experimental results for both HA and SRNOM. This sug-gested that kinetic data are well described with pseudo second-orderkinetic model, indicating that the rate-limiting step may be chemicalsorption (Wu et al., 2001).

Finally, the adsorption of SRNOM onto Fe2O3 NPs was faster thanthe adsorption of HA (Fig. 3A and B), as equilibrium was reachedafter only 60 min compared with 120 min for HA and as indicated by

Fig. 4. HPSEC chromatograms of HA-coated Fe2O3NPs and SRN

the higher value of the constant k related to the adsorption rate whichis consistent with the FlFFF data.

3.4. Characterisation of Fe2O3 NP-bound DOM.

DOM is a mixture of heterogeneous components having differentmolecular weight and chemical composition. The polydispersity ofDOM is supposed to be responsible for the adsorption of a preferentialsize fraction of DOM (Gu et al., 1995). It is thus interesting to investi-gate which fraction of DOM is preferentially adsorbed onto the Fe2O3

NP surface. A possible approach is to compare the size distribution ofthe original DOM with the DOM-coated particles using HPSEC.

Fig. 4 displays the HPSEC chromatograms of HA, SRNOM, and bothHA-coated and SRNOM-coated Fe2O3 NPs. The chromatograms of bothHA and SRNOM have multiple peaks, indicating the polydispersity oftheDOM. Themolecularweights of SRNOMrange from250 Da (organicacids) to 20,000 Da (high molecular weight compounds (HMW), suchas colloids) against 250 Da to 1000 Da for HA, with the highest fractionat 850–1000 Da (humic substances) for both. This is in accordancewiththe general feature of DOM (Shon et al., 2005). The peak at a retentiontime of 10 min, which only appeared in the SRNOM chromatograms,suggests that SRNOM does possess higher molecular weight than HA.The peak at 250 Da, present in both chromatograms, has a higher inten-sity for HA indicating that HA has a greater amount of low-molecularweight components. The peak at 36,000 Da appearing only on theHPSEC chromatograms of the DOM-coated NPs can be attributed tothe coated Fe2O3 NPs. Finally, the peak at 250,000 Da is probably relatedto the formation of small aggregates of coated particles and has higherintensity in the HA chromatograms. This can be related to the broaderpeaks in the FlFFF fractograms (Fig. 2).

By comparing the chromatograms between DOM and DOM-coatednanoparticles, the intensity of the peaks ranging from 850 Da to20,000 Da (i.e. humic substances and HMW molecules) decreasedsignificantly, while the peak at 250 Da (i.e. organic acids) remained

OM-coated Fe2O3NPs after different mixing time at pH 4.

Fig. 6. FlFFF fractograms of (A) HA-coated Fe2O3NPs and (B) SRNOM-coated Fe2O3NPsafter 2 weeks. (N.B.: 1 h, 2 h, 5 h and 10 h denote the mixing time originally used toprepare the different samples).

25L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

relatively high. As the measured fluorescence response is proportionalto the DOM concentration, the concentrations of both HMW com-pounds and humic substances showed a significant decrease duringadsorption. This demonstrated the preferential adsorption of both highmolecular weight DOM and low molecular weight humic substanceson the metal oxide surface. However, the smallest molecular weightcompounds in the range of 250 Da were not adsorbed onto Fe2O3 NPs.Previous studies also indicated the preferential adsorption of highmolecular weight DOM onto metal oxide surfaces (McKnight et al.,1992; Vermeer and Koopal, 1998; Zhou et al., 2000).

To further investigate the interactions between DOM and Fe2O3

NPs, FTIR spectra of bare and coated Fe2O3 NPs were used to identifythe physico-chemical binding mechanisms (Fig. 5).

No apparent peaks can be assigned in the bare Fe2O3 NP spectra.The O\H stretch observed at about 3400 cm−1 may be due to thepresence of small amount of water in the sample during analysis.The spectra of both DOM-coated Fe2O3 NPs show a C_O stretch atapproximately 1600 cm−1 which may indicate the carboxylate anioninteracting with the iron oxide surface, since the C_O stretches infree carboxylic acid would be above 1700 cm−1 (Yantasee et al.,2007). As the peak of the C_O stretches in the SRNOM-coated Fe2O3

NP spectrum have higher intensity (i.e. compared to the spectrum ofHA-coated Fe2O3 NPs), it may be concluded that this type of bond ismore pronounced between SRNOM and Fe2O3 NPs than between HAand Fe2O3 NPs. The broad O\H stretch peak present in bothDOM-coated Fe2O3 NPs spectra could indicate the occurrence of hydro-gen bonding resulting from the interaction between the positivelycharged Fe-OH+ and the negatively charged DOM in acidic conditions(Lin et al., 2010). In fact, many studies have demonstrated that theadsorption of DOM on the surface of Fe2O3 NPs is mainly governed byCoulombic interactions via ligand-exchange reactions below the pH ofPZC (Filius et al., 2000; Illés and Tombácz, 2004).

3.5. Stability of DOM-coated Fe2O3 NPs

The stability of the coated particles was assessed by measuring theirsize 14 days after the preparation of the “fresh coated nanoparticles”(i.e. 14 days after the 24 hour mixing time) without any modifications.Fig. 6 shows the FlFFF fractograms of both DOM-coated Fe2O3 NPs. Thesize of the HA-coated Fe2O3 NPs was approximately 200 nm for all thesamples, while SRNOM-coated Fe2O3 NPs were larger at 250 nm forall samples. These values are greater than the size obtained with“fresh samples” (i.e. around 60–70 nm for HA-coated Fe2O3 NPs andaround 100–105 nm for SRNOM-coated Fe2O3 NPs), indicating the for-mation of some aggregateswith time. This also suggests that the stirringtime has no influence on the stability of the coated nanoparticles.Increasing the stirring time after 1 h did not improve the stability ofthe coating by providing increased steric stability, which would havecaused entropically unfavourable conditions and prevented the coated

Fig. 5. FTIR spectra of bare Fe2O3NPs, HA-coated Fe2O3NPs and SRNOM-coated Fe2O3NPs.

nanoparticles from aggregation (Tiller and O'Melia, 1993; Illés andTombácz, 2004; Silva et al., 2012a, 2012b, 2012c).

The stability of the formed aggregateswas then assessed by studyingthe effect of vortexmixing on the disaggregation of the coated nanopar-ticle aggregates. Studying the stability of the formed aggregates iscrucial for application in soil and groundwater remediation. In fact, forthis application, nanoparticles are usually applied directly on-site via in-jection (Cundy et al., 2008). We demonstrated that after a short periodof time (i.e. few days), the coated-particles aggregated slightly whichcould decrease theirmobility once injected in the subsurface. Therefore,finding simple and rapid methods to disaggregate and stabilise the pre-pared coated-particles prior to their injection on-site is essential.

For HA-coated NPs (Fig. 7A), the effect of vortexmixingwas that thesize of the aggregated samples decreased from 200 nm to 70 nmi.e. back to the initial size of the HA-coated sample, after 1 min of vortexmixing. This indicates that theHA-coated Fe2O3 NPswere agglomeratedrather than aggregated and they were only held byweak van derWaalsforces (Jiang et al., 2009). The results also suggest that the bonding ofHA-Fe2O3 NPs is strong, otherwise vortex mixing the sample wouldhave also broken the bonds between HA and the surface of Fe2O3 NPs.

For SRNOM-coated Fe2O3 NPs (Fig. 7B), the same conditions wereused and the results showed the presence of 2 peaks in the fractogramof the coated nanoparticles after vortex mixing. This indicates that onlya fraction (about 50%) of the samplewas disaggregated (i.e. same size as

Fig. 7. Effect of vortex on the disaggregation of (A) HA-coated Fe2O3NPs and (B)SRNOM-coated Fe2O3NPs.

26 L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

the original samples) but the rest remained unchanged. This could beexplained by the structure of the aggregateswhichmay have a substan-tial influence on the disaggregation of nanoparticles (Christian et al.,2008). In fact, the aggregate structure (i.e. the conformation and poros-ity) can vary significantly with the concentration and type of DOM. Arecent study by Baalousha et al. (2008) demonstrated that, in theabsence of HA, Fe2O3 NPs formed open and porous aggregates, whereasin the presence of HA, they formed compact aggregates whichwere dif-ficult to disaggregate without applying any exterior mechanical forces.SRNOM has a more complex structure than HA, with more HMW mol-ecules; therefore the structure of the formed aggregates may be evenmore complex, making the disaggregation process more difficult.

4. Conclusions

DOM-coated Fe2O3 NPs are now used for various applications suchas the removal of heavy metals and harmful organic compounds inwater or wastewater treatment. It is thus important to fully under-stand the interactions between DOM and Fe2O3 NPs to be able topredict their fate and behaviour once applied in the environment. Inthis study, DOM-coated Fe2O3 NPs were characterised with a rangeof techniques. Fe2O3 NPs proved to be a particularly good adsorbentfor both HA and SRNOM under acidic conditions. The main interactionarises from a ligand exchange reaction between oppositely chargedparticles; however, carboxylate anions interacting with the iron oxidesurfacewas also important between SRNOMand Fe2O3NPs. Preferentialadsorption for highermolecular sized components was observedwhichcould potentially lead not only to electrostatic stabilisation but also to

steric stabilisation by causing entropically unfavourable conditionswhen the coated particles come closer to one another.

Finally, an aggregation and disaggregation study revealed thatafter 14 days, small aggregates were formed but they remained in thenanosize range. HA-coated Fe2O3 NPs formed agglomerates whichwere easily disaggregated using a vortex mixer and returned to theirinitial state. The SRNOM-coated Fe2O3 NPs formed more stable aggre-gates, where only a fraction of the coated nanoparticleswere recovered.Future studies should focus on the fate of these coated NPs in real appli-cations such as wastewater treatment in order to assess their stabilityand behaviour in such conditions over a longer time period.

Acknowledgements

This research was funded by the Cooperative Research Centre forContamination Assessment and Remediation of the Environment(CRC CARE).

References

Baalousha M. Aggregation and disaggregation of iron oxide nanoparticles: influence ofparticle concentration, pH and natural organic matter. Sci Total Environ2009;407(6):2093–101.

Baalousha M, Manciulea A, Cumberland S, Kendall K, Lead JR. Aggregation and surfaceproperties of iron oxide nanoparticles: influence of pH and natural organic matter.Environ Toxicol Chem 2008;27(9):1875–82.

Cerqueira B, Vega F, Serra C, Silva L, Andrade M. Time of flight secondary ion mass spec-trometry and high-resolution transmission electron microscopy/energy dispersivespectroscopy: a preliminary study of the distribution of Cu2+ and Cu2+/Pb2+ on aBt horizon surfaces. J Hazard Mater 2011;195:422–31.

Cerqueira B, Vega FA, Silva LF, Andrade L. Effects of vegetation on chemical and miner-alogical characteristics of soils developed on a decantation bank from a coppermine. Sci Total Environ 2012;421:220–9.

Chen J, Xiu Z, Lowry GV, Alvarez PJJ. Effect of natural organic matter on toxicity and re-activity of nano-scale zero-valent iron. Water Res 2011;45(5):1995–2001.

Chi F-H, Amy GL. Kinetic study on the sorption of dissolved natural organic matter ontodifferent aquifer materials: the effects of hydrophobicity and functional groups.J Colloid Interface Sci 2004;274(2):380–91.

Christian P, Von der Kammer F, Baalousha M, Hofmann T. Nanoparticles: structure,properties, preparation and behaviour in environmental media. Ecotoxicology2008;17(5):326–43.

Cirtiu CM, Raychoudhury T, Ghoshal S, Moores A. Systematic comparison of the size,surface characteristics and colloidal stability of zero valent iron nanoparticlespre- and post-grafted with common polymers. Colloids Surf A Physicochem EngAsp 2011;390(1–3):95-104.

Crane RA, Scott TB. Nanoscale zero-valent iron: future prospects for an emerging watertreatment technology. J Hazard Mater 2012;211-212:112–25.

Cundy AB, Hopkinson L, Whitby RLD. Use of iron-based technologies in contaminatedland and groundwater remediation: a review. Sci Total Environ 2008;400(1–3):42–51.

Dickson D, Liu G, Li C, Tachiev G, Cai Y. Dispersion and stability of bare hematitenanoparticles: effect of dispersion tools, nanoparticle concentration, humic acidand ionic strength. Sci Total Environ 2012;419:170–7.

Domingos RF, Baalousha MA, Ju-Nam Y, Reid MM, Tufenkji N, Lead JR, et al. Character-izing manufactured nanoparticles in the environment: multimethod determina-tion of particle sizes. Environ Sci Technol 2009;43(19):7277–84.

Elliott DW, Zhang WX. Field assessment of nanoscale bimetallic particles for ground-water treatment. Environ Sci Technol 2001;35(24):4922–6.

Febrianto J, Kosasih AN, Sunarso J, Ju Y-H, Indraswati N, Ismadji S. Equilibrium and ki-netic studies in adsorption of heavy metals using biosorbent: a summary of recentstudies. J Hazard Mater 2009;162(2):616–45.

Filella M, Zhang J, Newman ME, Buffle J. Analytical applications of photon correlationspectroscopy for size distribution measurements of natural colloidal suspensions:capabilities and limitations. Colloids Surf A Physicochem Eng Asp 1997;120(1):27–46.

Filius JD, Lumsdon DG, Meeussen JCL, Hiemstra T, Van Riemsdijk WH. Adsorption offulvic acid on goethite. Geochim Cosmochim Acta 2000;64(1):51–60.

Geng B, Jin Z, Li T, Qi X. Preparation of chitosan-stabilized Fe0 nanoparticles for removalof hexavalent chromium in water. Sci Total Environ 2009;407(18):4994–5000.

Giddings JC. Field flow fractionation handbook. Chapter 1: the field-flow fractionationfamily: underlying principles. Wiley-interscience; 2000.

Gu B, Schmitt J, Chen Z, Liang L, McCarthy JF. Adsorption and desorption of different or-ganic matter fractions on iron oxide. Geochim Cosmochim Acta 1995;59(2):219–29.

Hajdú A, Illés E, Tombácz E, Borbáth I. Surface charging, polyanionic coating and colloidstability of magnetite nanoparticles. Colloids Surf A Physicochem Eng Asp2009;347(1–3):104–8.

Hassellöv M, Readman JW, Ranville JF, Tiede K. Nanoparticle analysis and characterizationmethodologies in environmental risk assessment of engineerednanoparticles. Ecotox-icology 2008;17(5):344–61.

27L. Chekli et al. / Science of the Total Environment 461–462 (2013) 19–27

He F, Zhao D. Manipulating the size and dispersibility of zerovalent iron nanoparticlesby use of carboxymethyl cellulose stabilizers. Environ Sci Technol 2007;41(17):6216–21.

He F, Zhao D, Liu J, Roberts CB. Stabilization of Fe–Pd nanoparticles with sodiumcarboxymethyl cellulose for enhanced transport and dechlorination of trichloro-ethylene in soil and groundwater. Ind Eng Chem Res 2007;46(1):29–34.

He YT, Wan J, He YT, Wan J, Tokunaga T. Kinetic stability of hematite nanoparticles: theeffect of particle sizes. J Nanopart Res 2008;10(2):321–32.

Hu J-D, Zevi Y, Kou X-M, Xiao J, Wang X-J, Jin Y. Effect of dissolved organic matter onthe stability of magnetite nanoparticles under different pH and ionic strength con-ditions. Sci Total Environ 2010;408(16):3477–89.

Illés E, Tombácz E. The role of variable surface charge and surface complexation in theadsorption of humic acid on magnetite. Colloids Surf A Physicochem Eng Asp2004;230(1–3):99-109.

Illes E, Tombácz E. The effect of humic acid adsorption on pH-dependent surface chargingand aggregation of magnetite nanoparticles. J Colloid Interface Sci 2006;295(1):115–23.

Jiang J, Oberdörster G, Biswas P. Characterization of size, surface charge, and agglomerationstate of nanoparticle dispersions for toxicological studies. J Nanopart Res 2009;11(1):77–89.

Kang S, Xing B. Humic acid fractionation upon sequential adsorption onto goethite.Langmuir 2008;24(6):2525–31.

Lin YH, Tseng HH, Wey MY, Lin MD. Characteristics of two types of stabilized nanozero-valent iron and transport in porous media. Sci Total Environ 2010;408(10):2260–7.

Liu J, Zhao Z, Jiang G. Coating Fe3O4 magnetic nanoparticles with humic acid for highefficient removal of heavy metals in water. Environ Sci Technol 2008;42(18):6949–54.

McKnight DM, Bencala KE, Zellweger GW, Aiken GR, Feder GL, Thorn KA. Sorption ofdissolved organic carbon by hydrous aluminum and iron oxides occurring at theconfluence of Deer Creek with the Snake River, Summit County, Colorado. EnvironSci Technol 1992;26(7):1388–96.

Mori S, Barth HG. Size exclusion chromatography. Springer Verlag; 1999.Mylon SE, Chen KL, Elimelech M. Influence of natural organic matter and ionic compo-

sition on the kinetics and structure of hematite colloid aggregation: implications toiron depletion in estuaries. Langmuir 2004;20(21):9000–6.

Oliveira ML, Ward CR, French D, Hower JC, Querol X, Silva LF. Mineralogy and leachingcharacteristics of beneficiated coal products from Santa Catarina, Brazil. Int J CoalGeol 2012a;94:314–25.

Oliveira ML, Ward CR, Izquierdo M, Sampaio CH, de Brum IA, Kautzmann RM, et al.Chemical composition and minerals in pyrite ash of an abandoned sulphuric acidproduction plant. Sci Total Environ 2012b;430:34–47.

Phenrat T, Kim HJ, Fagerlund F, Illangasekare T, Tilton RD, Lowry GV. Aggregation andsedimentation of aqueous nanoscale zerovalent iron dispersions. Environ SciTechnol 2007;41(1):284–90.

Phenrat T, Saleh N, Sirk K, Tilton RD, Lowry GV. Particle size distribution, concentration,and magnetic attraction affect transport of polymer-modified Fe0 nanoparticles insand columns. Environ Sci Technol 2009;43(13):5079–85.

Phuntsho S, Shon H, Vigneswaran S, Cho J. Assessing membrane fouling potential ofhumic acid using flow field-flow fractionation. J Membr Sci 2011;373(1–2):64–73.

Ponder SM, Darab JG, Mallouk TE. Remediation of Cr(VI) and Pb(II) aqueous solutionsusing supported, nanoscale zero-valent iron. Environ Sci Technol 2000;34(12):2564–9.

Quinn J, Geiger C, Clausen C, Brooks K, Coon C, O'Hara S, et al. Field demonstration ofDNAPL dehalogenation using emulsified zero-valent iron. Environ Sci Technol2005;39(5):1309–18.

Quispe D, Pérez-López R, Silva LF, Nieto JM. Changes in mobility of hazardous elementsduring coal combustion in Santa Catarina power plant (Brazil). Fuel 2012;94:495–503.

Ribeiro J, Flores D, Ward CR, Silva LF. Identification of nanominerals and nanoparticlesin burning coal waste piles from Portugal. Sci Total Environ 2010;408(23):6032–41.

Saleh N, Phenrat T, Sirk K, Dufour B, Ok J, Sarbu T, et al. Adsorbed triblock copolymersdeliver reactive iron nanoparticles to the oil/water interface. Nano Lett2005;5(12):2489–94.

Saleh N, Sirk K, Liu Y, Phenrat T, Dufour B, Matyjaszewski K, et al. Surface modificationsenhance nanoiron transport and NAPL targeting in saturated porous media. Envi-ron Eng Sci 2007;24(1):45–57.

Saleh N, Kim HJ, Phenrat T, Matyjaszewski K, Tilton RD, Lowry GV. Ionic strength andcomposition affect the mobility of surface-modified Fe0 nanoparticles inwater-saturated sand columns. Environ Sci Technol 2008;42(9):3349–55.

Schrick B, Hydutsky BW, Blough JL, Mallouk TE. Delivery vehicles for zerovalent metalnanoparticles in soil and groundwater. Chem Mater 2004;16(11):2187–93.

ShonH, Vigneswaran S, Kim IS, Cho J, NgoH. The effect of pretreatment to ultrafiltration ofbiologically treated sewage effluent: a detailed effluent organic matter (EfOM) char-acterization. Water Res 2004;38(7):1933–9.

Shon H, Vigneswaran S, Ngo H, Kim JH. Chemical coupling of photocatalysis with floc-culation and adsorption in the removal of organic matter. Water Res 2005;39(12):2549–58.

Silva LF, Hower JC, Izquierdo M, Querol X. Complex nanominerals and ultrafine parti-cles assemblages in phosphogypsum of the fertilizer industry and implicationson human exposure. Sci Total Environ 2010;408(21):5117–22.

Silva L, Sampaio C, Guedes A, Fdez-Ortiz de Vallejuelo S, Madariaga J. Multianalyticalapproaches to the characterisation of minerals associated with coals and the diag-nosis of their potential risk by using combined instrumental microspectroscopictechniques and thermodynamic speciation. Fuel 2012a;94:52–63.

Silva LF, DaBoit K, Sampaio CH, Jasper A, Andrade ML, Kostova IJ, et al. The occurrenceof hazardous volatile elements and nanoparticles in Bulgarian coal fly ashes andthe effect on human health exposure. Sci Total Environ 2012b;416:513–26.

Silva LF, Jasper A, Andrade ML, Sampaio CH, Dai S, Li X, et al. Applied investigation onthe interaction of hazardous elements binding on ultrafine and nanoparticles inChinese anthracite-derived fly ash. Sci Total Environ 2012c;419:250–64.

Sirk KM, Saleh NB, Phenrat T, Kim HJ, Dufour B, Ok J, et al. Effect of adsorbed polyelectro-lytes on nanoscale zero valent iron particle attachment to soil surfacemodels. EnvironSci Technol 2009;43(10):3803–8.

Tiller CL, O'Melia CR. Natural organic matter and colloidal stability: models andmeasurements. Colloids Surf A Physicochem Eng Asp 1993;73:89-102.

Tiraferri A, Chen KL, Sethi R, Elimelech M. Reduced aggregation and sedimentation ofzero-valent iron nanoparticles in the presence of guar gum. J Colloid Interface Sci2008;324(1–2):71–9.

Tombácz E, Libor Z, Illés E, Majzik A, Klumpp E. The role of reactive surface sites andcomplexation by humic acids in the interaction of clay mineral and iron oxide par-ticles. Org Geochem 2004;35(3):257–67.

Vermeer A, Koopal L. Adsorption of humic acids to mineral particles. 2. Polydispersityeffects with polyelectrolyte adsorption. Langmuir 1998;14(15):4210–6.

Wang C-B, Zhang W-X. Synthesizing nanoscale iron particles for rapid and completedechlorination of TCE and PCBs. Environ Sci Technol 1997;31(7):2154–6.

Wu F-C, Tseng R-L, Juang R-S. Enhanced abilities of highly swollen chitosan beads forcolor removal and tyrosinase immobilization. J Hazard Mater 2001;81(1):167–77.

Yantasee W, Warner CL, Sangvanich T, Addleman RS, Carter TG, Wiacek RJ, et al. Re-moval of heavy metals from aqueous systems with thiol functionalizedsuperparamagnetic nanoparticles. Environ Sci Technol 2007;41:5114–9.

Zhang W-X. Nanoscale iron particles for environmental remediation: an overview.J Nanopart Res 2003;5:323–32.

Zhang W-x, Wang C-B, Lien H-L. Treatment of chlorinated organic contaminants withnanoscale bimetallic particles. Catal Today 1998;40(4):387–95.

Zhang X, Zhang P, Wu Z, Zhang L, Zeng G, Zhou C. Adsorption of methylene blue ontohumic acid-coated Fe3O4 nanoparticles. Colloids Surf A Physicochem Eng Asp2013.

Zhou Q, Cabaniss SE, Maurice PA. Considerations in the use of high-pressure size exclu-sion chromatography (HPSEC) for determining molecular weights of aquatichumic substances. Water Res 2000;34(14):3505–14.