Journal of Colloid and Interface Science 370 (2012) 1–10

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Effect of kaolinite, silica fines and pH on transport of polymer-modifiedzero valent iron nano-particles in heterogeneous porous media

Hye-Jin Kim a,b, Tanapon Phenrat a,c, Robert D. Tilton d,e, Gregory V. Lowry a,d,⇑a Center for Environmental Implications of Nanotechnology, Department of Civil & Environmental Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USAb Soil and Groundwater Research Division, Environmental Infrastructure Research Department, National Institute of Environmental Research, Incheon 404-708, South Koreac Department of Civil Engineering, Naresuan University, Phitsanulok 65000, Thailandd Center for Environmental Implications of Nanotechnology, Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USAe Center for Environmental Implications of Nanotechnology, Department of Biomedical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA

a r t i c l e i n f o

Article history:Received 20 June 2011Accepted 21 December 2011Available online 2 January 2012

Keywords:Environmental nanotechnologyNanoparticle heteroaggregationExposure assessmentNanoparticle transportSurface modified nanoparticleNanoparticle coatings

0021-9797/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.jcis.2011.12.059

⇑ Corresponding author at: Center for Environmentnology, Department of Civil & Environmental EnUniversity, Pittsburgh, PA 15213, USA. Fax: +1 412 26

E-mail address: glowry@cmu.edu (G.V. Lowry).

a b s t r a c t

Polymer coatings on nano-sized remediation agents and subsurface heterogeneity will affect their trans-port, likely in a pH-dependent manner. The effect of pH on the aggregation of polymer-coated nanoscalezerovalent iron (nZVI) and its deposition onto sand and clay (kaolinite) surfaces was studied. nZVI coat-ings included a high molecular weight (90 kg/mol) strong polyanion, poly(methacrylic acid)-b-(methymethacrylate)-b-(styrenesulfonate) (PMAA-PMMA-PSS) and a low molecular weight (2.5 kg/mol) weakpolyanion, polyaspartate. Aggregation and deposition increased with decreasing pH for both polyelectro-lytes. The extent was greater for the low MW polyaspartate coated nZVI. Enhanced deposition at lowerpH was indicated because the elutability of polyaspartate-modified hematite (which did not aggregate)also decreased at lower pH. The greater deposition onto clay minerals compared to similar sized silicafines is attributed to charge heterogeneity on clay mineral surfaces, which is sensitive to pH. Heteroag-gregation between kaolinite particles and nZVI over the pH range 6–8 confirmed this assertion. Excessunadsorbed polyelectrolyte in solution (100 mg/L) enhanced the transport of modified nZVI by minimiz-ing aggregation and deposition onto sand and clay. These results indicate that site physical and chemicalheterogeneity must be considered when designing an nZVI emplacement strategy.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction nZVI aggregation and deposition onto porous subsurface media

Nano-scale zerovalent iron (nZVI) and other nanoscale remedialagents (e.g. hydrophobic polymer beads and nano-hydroxyapatite)have been proposed as groundwater remediation agents. The oxi-dation of the zero valent iron (Fe0) is coupled with the reductionof chlorinated hydrocarbons to benign products or the reductionand immobilization of heavy metals. Their small size (5–70 nm)makes it possible to deliver these reactive materials [1–6] in situdirectly to the contamination source [7–9]. However, the transportof nZVI in laboratory scale columns is limited by aggregation andby deposition onto porous media, even under ideal flow conditionsin relatively homogeneous porous media [7,10,11]. Consistent withthis finding, field applications with bimetallic Fe/Pd nano-particles[4,12,13] have indicated that it is indeed difficult to emplace thenano-particles in the subsurface.

Adsorbed anionic polyelectrolyte surface coatings enhance nZVItransport by providing repulsive electrosteric forces that hinder

ll rights reserved.

al Implications of Nanotech-gineering, Carnegie Mellon8 7813.

which also usually has a net negative charge at near neutral pH,typical of most groundwater [10,14–17]. Still, varying pH and ionicconditions can limit the nZVI transport [16,18,19] by affecting thenZVI surface charge, which subsequently affects its aggregationand deposition to porous media. When subsurface media is physi-cally or surface chemically heterogeneous, as when mixtures of lar-ger and smaller sized particles co-exist or when clay minerals arepresent in interstitial spaces between larger grains, differentialinteractions with each subsurface component may determine thetransport of nZVI. Fig. 1 illustrates the possible mechanisms bywhich pH and the presence of subsurface media of varying size,mineralogy, and charge heterogeneity can affect nZVI transport.

Both pH and the surface polyanion type can affect nZVI aggrega-tion and transport. At pH < pH of the isoelectric point (pHiep) ofnZVI, aggregation may result from a change in the conformationof the adsorbed polyanion layer as the underlying particle surfacebecomes positively charged (Fig. 1a). The charge and associatedelectrosteric repulsion of an adsorbed weak polyanion would bemore sensitive to pH than a strong polyanion because the weakpolyanion has a higher pKa [20–25]. Polyaspartate is a weak poly-electrolyte with a pKa near pH 3.5–5 [22]. Its maximum deproto-nation occurs above pH 7 [23], so for pH < 7 the decreased

(c)

Deposition (sand) Enhanced Deposition (sand) + nZVI Aggregation

Flow

High pH Low pH Low pH + Clay

Flow Flow surface

modified

nZVI

sand

metal

oxides

Enhanced Deposition (sand + clay) + nZVI Aggregation

clay

Fig. 1. Conceptual model of physicochemical processes affecting aggregation and deposition of surface modified nZVI in heterogeneous porous media. (a) Decreasing pHpromotes aggregation of nZVI due to reduced net charge on polymer-modified nZVI and subsequent flattening of the adsorbed layer conformation. (b) Enhanced depositionoccurs on positively charged edge sites formed on kaolinite or metal oxide impurities on sand at low pH. (c) Particle filtration by deposition of aggregates formed at low pH,and by deposition and mechanical filtration by clay aggregates that form in the interstitial spaces at low pH.

2 H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10

charge in the polyaspartate layer will contribute to aggregation be-cause of the compression of electric double layer. Conversely, astrong polyelectrolyte such as PSS with a pKa near pH 2 shouldbe less affected by low pH. Aggregation of nZVI to large enoughaggregates can enhance deposition onto porous media due to in-creased gravitational settling or mechanical filtration through nar-row pore throats [24,25].

Even in the absence of nZVI aggregation, low pH can also en-hance nZVI deposition onto porous media with metal oxide surfaceimpurities (e.g. Fe-oxides or Al-oxides) that have a relatively highisoelectric point (pHiep = pH 7–9.5) [26–32] compared to silica(pHiep 2–3) [28,33] or with clay minerals that can develop posi-tively charged edge sites at low pH (Fig. 1b). Metal oxide impuritieson mineral surfaces with relatively high pHiep provide positivelycharged favorable deposition sites that limit nanoparticle transportthrough porous media at low pH. Even a small amount of hematiteon quartz sand has been shown to increase colloid deposition by 2or 3 orders of magnitude compared to clean quartz sand surfacesdue to the positively charged hematite surface at neutral pH[29,31,33].

Deposition of negatively charged engineered nanomaterialsonto positively charged clay minerals may decrease the transport

of nanomaterials in the subsurface. Clay minerals such as kaoliniteare ubiquitous in the subsurface [34–37]. Clays minerals can haveboth a permanent negative charge on the basal plane because ofisomorphic substitution of minerals (Si, Al) for lower positive va-lent ions and pH dependent edge charge due to surface hydroxylgroups (Al-OH, Si-OH) created from broken bonds which can pro-tonate or deprotonate depending on the pH in the solution[34,38]. The reported pHpzc-edge of kaolinite edge sites ranges frompH 5.8 to 7.5 [34,39–41]. The positively charged edge sites on claymay serve as deposition sites for negatively charged nZVI (Fig. 1b)at low pH, thereby decreasing transport through porous media.Heteroaggregation between positively charged iron oxide colloidsand negatively charged clay particles has been reported[38,42,43]. Fang et al. (2009) observed that retention of nano-sizedTiO2 (without surface modification) was higher in saturated soilcolumns with higher clay content [44]. However, polymer-coatednZVI nano-particles are expected to behave differently from previ-ously studied iron oxide and metal oxide particle due to the anio-nic polymer coating.

Real subsurface materials are not only chemically heteroge-neous but also physically heterogeneous. Fine particles (less than2 lm), including clay minerals, are ubiquitous inorganic natural

H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10 3

colloids in the subsurface, making up from less than 1% to tens ofpercent of subsurface materials [45,46]. The presence of fine parti-cles in pore spaces between larger sized media grains may furtherlimit nanoparticle transport by mechanical filtration. This is partic-ularly true for many clay minerals that exhibit ‘‘edge to face hete-roaggregation’’ to form large aggregates which can fill pore throats(Fig. 1c) [34,38,43]. This occurs predominantly at low pH where thepositive edge sites are attracted to the negative face sites of theclay particles. This process will be more significant if the nano-par-ticles also aggregate to larger sized particles.

Currently, we lack a fundamental understanding of how pH af-fects the aggregation, deposition, and transport of polyelectrolyte-modified nano-particles in heterogeneous subsurface media. Suchunderstanding is needed to design appropriate nanoparticle poly-meric modifiers for use in natural environments that are physicallyand chemically heterogeneous. The objectives of this study are todetermine the effects of pH and presence of excess free polymerin solution on (1) polyanion-modified nZVI aggregation, and depo-sition onto sand and clay minerals, and on (2) polyanion-modifiednZVI transport through physically heterogeneous porous mediacontaining large and small sized silica sand, or containing silicasand and clay minerals.

2. Materials and methods

2.1. Iron based nano-particles

One form of nZVI, Reactive Nano Iron Particles (RNIP), with Fe0

cores and Fe3O4 shells were provided by Toda Kogyo Corp. (Onoda,Japan). RNIP has no polymer coating and consists of polydisperseparticles with a primary particle size ranging from 5 to 70 nm witha median radius of �20 nm. The unmodified particles were storedanaerobically in water at pH 10.6 at �300 g/L. The synthesis andproperties of these particles have been described previously[5,10,47]. RNIP modified with adsorbed polyaspartate (MRNIP) witha molecular weight of 2–3 kg/mol (PAP2.5K) was also supplied byToda Kogyo Corp. (Onoda, Japan). Hematite nano-particles (a-Fe2O3) with an average radius of 20 nm were obtained from Nano-structured and Amorphous Materials Inc. (Los Alamos, NM). H2 gen-eration measurements [5] indicated that the Fe0 contents of thestock slurry of RNIP and MRNIP were 40% (w/w) and 37% (w/w)respectively when the experiments reported here were conducted.

2.2. Polyanion surface modifiers

Two polyanionic surface modifiers were used including a lowmolecular weight, weak polyanion, sodium polyaspartate, a syn-thetic homopolypeptide (PAP2.5K), and a higher molecular weighttriblock copolymer (90 kg/mol) containing a strong polyanion, poly(methacrylic acid)-block-poly (methyl methacrylate)-block-poly(styrene sulfonate)-block (p(MAA)42-p(MMA)26-p(SS)462). The PSSblock is the strong polyanion, and the PMAA and PMMA blocksserve to anchor the block copolymer to the iron oxide surface. Sub-scripts denote the degree of polymerization of each block. Polyas-partate was either supplied by Toda Kogyo (Onoda, Japan) pre-adsorbed to RNIP (MRNIP), or by Lanxess as sodium polyaspartate(2–3 kg/mol). The latter was used to modify hematite nano-parti-cles. The p(MAA)42-p(MMA)26-p(SS)462 triblock copolymer wassynthesized using atom transfer radical polymerization (ATRP) aspreviously described [7,9,15,16].

2.3. Porous matrix

Silica sand with a 300 lm average diameter (d50) and a specificgravity of 2.65 g/cm3 was used as a model porous matrix (Agsco

Corp., Hasbrouck Heights, NJ). The Fe2O3 and Al2O3 weight% inthe sand reported by the manufacturer were 0.14% and 0.49%respectively. The grain size distribution and other physical proper-ties are provided as Supplementary Material in Fig. S1. Acid-washed sand was prepared using a previously reported procedureto remove metal oxide impurities from the sand [30]. The proce-dure included reaction with sodium dithionite (0.1 M) for 2 h fol-lowed by immersion in a chromic and sulfuric acid mixture(Fisher Scientific, Pittsburgh PA) overnight to remove metal oxideimpurities. Then, the sand was soaked in hydrogen peroxide (5%H2O2) for 3 h to remove organic contaminants, followed by acidwashing in sulfuric/chromic acid overnight. Finally, the sand wasimmersed in hydrochloric acid (12 N HCl) overnight and thenrinsed repeatedly in DI water until the pH of the water sand mix-ture remained constant at 7.

Kaolin clay (kaolinite) with an average diameter of 1.3 lm waspurchased from ACROS Organics. The grain size distribution isshown in Supplementary Material Fig. S2. Silica fines (MINUSIL5) with a similar size (d50 = 1.7 lm) as kaolinite were supplied byUS silica. The size distribution as reported by the manufacturer isgiven in Supplementary Material Fig. S3.

2.4. Particle preparation for aggregation and transport studies

MRNIP was provided with PAP2.5K pre-adsorbed in concen-trated slurry that contains excess unadsorbed polyaspartate. Themass ratio of polymer to MRNIP was 1:6. This stock slurry was di-luted to 6 g/L with 1 mM NaHCO3 solution (pH = 8.3). Triblockcopolymer modified RNIP (Triblock-RNIP) was prepared using thefollowing procedure. The stock slurry of (unmodified) RNIP wassonicated with an ultrasonic probe (550 Sonic Dismembrator, Fish-er Scientific) for 30 min to break agglomerates formed during stor-age. Triblock copolymer (1 g/L) and RNIP (6 g/L) were mixed andequilibrated by end-over-end rotation at 30 rpm for 2 days. Forboth MRNIP and Triblock-RNIP, suspensions were settled for5 min to exclude very large aggregates. The upper 80 vol.% of thesupernatant was taken and ultracentrifuged (Sorvall, OTD65B) at27,000 rpm (68,400 g) for 80 min to separate the nano-particlesfrom the unadsorbed polymer solution. The supernatant was dis-carded and the particles were re-suspended in 1 mM sodium bicar-bonate (NaHCO3) solution (pH 8.3) with sonication (550 SonicDismembrator, Fisher Scientific). This washing process was re-peated three times to completely remove free polyelectrolyte fromthe suspension [15]. For consistency, and because oxidation of nZVIcan affect its aggregation and transport [48], the unmodified RNIPsuspension used the same rinsing procedure as for the surfacemodified RNIP. After rinsing, the unmodified or modified RNIPwas suspended in 1 mM NaHCO3 solution to be used as stock forheteroaggregation and transport studies. The concentrations ofunmodified and PAP modified and, triblock copolymer modifiedRNIP stock suspensions were 1.78 g/L, 3.42 g/L and 3.23 g/L,respectively.

With a comparable surface chemistry to MRNIP, polyaspartate-modified hematite particles (PAP-hematite) that were stableagainst aggregation were used to distinguish effects of aggregationfrom effects of deposition on RNIP transport. PAP-hematite wasprepared by adsorbing PAP2.5K at pH 5 in 1 mM NaHCO3 + 10 mMNaCl solution. It should be noted that at pH = 5 some or all of theadded bicarbonate may leave the system through volatilization ofCO2(g). The isoelectric point of hematite occurs at pHiep = 7–9.5[49] so a pH of 5 was used to maximize the adsorbed mass ofPAP to ensure good stabilization against aggregation (Fig. S4).The excess unadsorbed polymer was removed with same pH andionic strength solution using the same washing protocol as de-scribed above to provide a stock suspension (10.68 g/L) of PAP-hematite.

4 H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10

2.5. Characterization of surface modified nano-particles and porousmedia

The electrophoretic mobility (EPM) of polyelectrolyte-modifiedRNIP or hematite was measured at pH 6 and 8. The electrophoreticmobility (EPM) of MRNIP, triblock-RNIP, and PAP-hematite was mea-sured on dilute suspensions (�15 mg/L) in aqueous NaCl solutions(from 5 mM to 60 mM NaCl) containing 1 mM NaHCO3 using a Mal-vern Zetasizer 3000 (Southborough, MA). While loss of CO2(g) fromthe system may occur at pH,7, the pH of these suspensions measuredbefore and after measurements remained at the target pH.

Aggregation of surface modified nano-particles was determinedat pH 6 and 8 and in 10 mM NaCl + 1 mM NaHCO3 by dynamic lightscattering (DLS) (Malvern Zetasizer 3000, Southborough, MA) at300 mg/L, which is the same concentration used in the columntransport studies. The aggregate size of kaolinite clay at pH 4 and 8was also measured by DLS at 200 mg/L in 10 mM NaCl + 1 mM NaH-CO3 background electrolyte. It should be noted that at pH = 4 theadded bicarbonate likely leaves through volatilization of CO2(g),but the pH was well buffered by the acidity of the system. The CON-TIN algorithm was used to convert intensity autocorrelation func-tions to intensity-weighted particle hydrodynamic diameterdistributions assuming that the particles are spherical. Samples forboth electrophoretic mobility and size measurement were sonicatedin a water bath (Bransonic 2510) for 5 min just prior to the measure-ment to break any weak agglomerates formed during storage.

The streaming potential of the porous medium (sand and acidwashed sand) was measured as a function of pH using an Electro Ki-netic Analyzer (Anton Paar GmbH, Graz, Austria) equipped with acylindrical cell to house the granular porous medium. Sand or acidtreated sand was wet-packed into the cylindrical cell (1.5 cm insidediameter � 3 cm length) in a solution having the same ionic compo-sition and pH as the solution used during the corresponding columnexperiments. Before the start of each measurement, the cell wasequilibrated by circulating the solution in alternate directions for20 min [50]. Streaming potentials were converted to zeta potentialsusing the Helmholtz–Smoluchowski equation.

To further demonstrate the existence of pH dependent chemicalcharge heterogeneity on the sand surface, PAP2.5K adsorption ontothe sand was determined by quantifying adsorbed PAP mass at pH6 and 8 in 1 mM NaHCO3 + 10 mM NaCl. Adsorption to the sandsurface was measured by solution depletion in 20 mL slurries con-taining 300 g/L sand at different initial concentrations of PAP2.5K.The sand slurry and the stock polymer solution were prepared sep-arately and equilibrated at either pH 6 or 8 for 3 days. The pH wasadjusted every 12 h until there was no change in the pH, and thenthe sand and PAP were mixed. In this process the added carbonatehad likely escaped from the system for the pH = 6 condition. After2 days, the PAP2.5K concentration in the supernatant was mea-sured using a total organic carbon analyzer (OI Analytical) as de-scribed in Kim et al. [15].

The EPM of kaolinite clay was determined between pH 5 and 8using a Malvern Zetasizer 3000 (Southborough, MA). Dilute sus-pensions (�15 mg/L) prepared in 1 mM NaHCO3 plus 10 mM NaClwere analyzed. Some carbonate had likely volatilized from the sys-tem at pH = 5. The measured EPM were converted to the apparentzeta potential using the Smoluchowski approximation. Clay sus-pensions were sonicated in a water bath (Bransonic 2510) for5 min just prior to EPM measurement, and pH was adjusted byadding NaOH (0.1 or 0.01 N) or HCl (0.1 or 0.01 N) as needed.

2.6. Heteroaggregation of surface modified RNIP and clay particles

Heteroaggregation of uncoated or modified RNIP and kaoliniteparticles was determined by monitoring the optical density changeof colloidal suspensions using UV–vis spectrophotometry at

508 nm (Varian Palo Alto, CA) for 1 day. A suspension of kaoliniteclay (6 g/L) was dispersed in 1 mM NaHCO3 solution and was al-lowed to settle quiescently for 1 day to remove the largest aggre-gates and to obtain a clay suspension that is highly stable againstaggregation. The upper 80 vol.% of the supernatant clay suspensionwas used as stock for the heteroaggregation study. A kaolinite sus-pension at 800 mg/L and a suspension of kaolinite (800 mg/L) plusRNIP (200 mg/L) were prepared in 10 mM NaCl plus 1 mM NaH-CO3. The pH was adjusted to pH 6, 7 or 8 by adding HCl or NaOH(0.1 N or 0.01 N). Samples were sonicated in a water bath (Branson2510) for 5 min before the measurement. pH measured after themeasurements indicated that the pH had not changed over thetime frame of the measurement. All the samples were preparedin triplicate and the reported values are the average and one stan-dard deviation of the three measurements. By comparing the opti-cal density-based sedimentation curves for kaolinite alone and forthe mixture of RNIP and kaolinite at different pH, the extent ofheteroaggregation between RNIP and kaolinite is determined. Toassess the ability of excess unadsorbed polymer to minimize hete-roaggregation, the measurements were repeated in the presence of270 mg/L of excess unadsorbed PAP2.5K in solution. The tempera-ture was controlled at 25 �C over the time of measurement.

2.7. Transport in porous media

Column experiments were conducted to determine how pH,clay minerals, and free polymer in solution affect the transport ofsurface modified RNIP in heterogeneous porous media. To evaluatethe simultaneously acting effects of aggregation and deposition ontransport, RNIP transport was measured as a function of pH. Toevaluate the effect of pH on deposition only, independent of aggre-gation, transport was measured as a function of pH using PAP-hematite nano-particles that do not aggregate under the currentconditions. To isolate the effect of grain size on the transport, with-out heterogeneously charged clay minerals, silica fines with a sim-ilar size as the clay minerals were added to the sand.

Transport studies were performed using MRNIP, Triblock-RNIP,and non-aggregating PAP-hematite particles in stainless steel col-umns (15-cm length � 1.25 cm o.d.) as described previously [7].The porosity was 0.33 and the linear pore water velocity was main-tained at 3.2 � 10�4 m/s using a peristaltic pump. This velocity rep-resents that for an injection well or injection–extraction well pairused for emplacing nZVI in porous media. Either 2 wt.% of silicafines (D50 = 1.7 lm) or 2 wt.% of kaolinite (D50 = 1.36 lm) weremixed with sand (d50 = 300 lm) and a small amount of waterand dried in the oven at 40 �C to create a relatively uniform mix-ture of sand and silica fines or of sand and clay [51]. Column exper-iments were conducted as follows. After the columns were packed,at least 20 pore volumes (PV) of background solution containing10 mM NaCl plus 1 mM NaHCO3 at pH 6, 7 or 8 were elutedthrough the columns at 3.2 � 10�4 m/s until the effluent pHequaled the influent pH. A square pulse (1 PV) of nZVI or hematitesuspension at 300 mg/L in 10 mM NaCl and 1 mM NaHCO3 ad-justed to pH 6, 7 or 8 was introduced into the column at3.2 � 10�4 m/s, then the column was flushed with backgroundelectrolyte solution (particle free) for three additional pore vol-umes. The column effluent was collected for four pore volumes.The ability of excess polymer in solution to enhance the transportof nZVI through the column was determined by introducing onesquare pulse of MRNIP mixed with 100 mg/L of PAP. During injec-tion, the particle suspension was sonicated with an ultrasonicprobe to prevent aggregation prior to introduction into the column.An aliquot of the effluent was acid digested in concentrated HCl(trace metal grade) and analyzed for total Fe using flame atomicabsorption spectrometry (GBC at 248.3 nm within the range of0–10 mg/L Fe) as previously described [7,10,15]. The influent

H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10 5

uncoated or modified RNIP, or hematite particle concentration, wasmeasured in the same way. All transport experiments were con-ducted in duplicate. The reported values are the average and errorbars represent one standard deviation of the replicatemeasurements.

3. Results and discussion

3.1. pH affects charge and aggregation of surface modified RNIP

The electrophoretic mobility of the surface modified RNIP at pH6 and 8 are shown in Fig. 2. Although both the MRNIP and triblock-RNIP are negatively charged at pH 6 and 8, the magnitude of theEPM is lower at pH 6 compared to pH 8 even for the triblockcopolymer with the strong PSS polyanion. This is likely a resultof the positive charge on the underlying RNIP particle at pH 6(pHpzc = 6.3) compared to pH 8 [15]. The triblock-RNIP exhibiteda more negative electrophoretic mobility compared to MRNIP ateach pH tested, and the change in the magnitude for triblock-RNIPwas less than for MRNIP when pH was decreased from 8 to 6. Thegreater apparent charge and lower sensitivity to pH for the tri-block-RNIP compared to MRNIP is consistent with expectation forthe strong PSS polyanion as compared to the weak PAP polyanion.The decreased magnitude of EPM at low pH compared to the higherpH is consistent with decreased charge in the surface and adsorbedlayer which could decrease the repulsion between particles and inturn enhance aggregation. The magnitude of the EPM for PAP mod-ified hematite was greater than for either modified RNIP and wasindependent of pH for pH 6 to pH 8. The greater charge of hematitecompared to either of the RNIP’s is likely due to a greater adsorbedmass of PAP since the negatively charged PAP was adsorbed tohematite at pH = 5 where it is positively charged [52].

Low pH promoted the aggregation of surface modified RNIP. Themeasured aggregate size distributions at pH 6 and 8 for each nano-particle type are shown in Fig. 3. At pH 6, the aggregate sizes ofboth MRNIP and triblock-RNIP were larger than at pH 8. Notably,the aggregate sizes for both RNIP systems were >1.5 lm at pH 6.

The greater aggregation of surface modified RNIP at pH 6 com-pared to pH 8 is consistent with the decreased electrophoreticmobility at pH 6. At pH = 6 the underlying RNIP becomes positivelycharged (pHiep = 6.3) [15] so the interactions between the particlesurface and the negatively charged adsorbed polyelectrolyte be-come more favorable. This can result in a flatter adsorbed layerconformation compared with the higher pH where RNIP is nega-tively charged when there is no excess unadsorbed polymer in

NaCl concentration (mM)0 10 20 30 40 50 60

u e(1

0-8m

2 V-1s-1

)

-4.0

-3.5

-3.0

-2.5

-2.0

-1.5

-1.0

Fig. 2. Electrophoretic mobility of MRNIP, triblock-RNIP, and PAP-hematite vs. ionicstrength at pH 6 and pH 8. The symbols correspond to MRNIP at pH 6 (d) or pH 8(s), Triblock-RNIP at pH 6 (j) or pH 8 (h) and PAP-hematite particles at pH 6 (N) orpH 8 (4).

the solution. This flattening would decrease the electrosteric repul-sive force between particles and increase aggregation as was ob-served [20]. The effect of RNIP aggregation on its transport inporous media is discussed later.

3.2. pH affects deposition onto heterogeneous sand surfaces

Limited mobility of polyelectrolyte modified RNIP in sand col-umns can result from aggregation of the RNIP or RNIP depositiononto the sand surface, or both. A lower mobility of surface modifiedRNIP through the sand column was observed at pH 6 compared topH 8 (Fig. 4) for all of coated nano-particles used here. Data in Fig. 4are for sand that was not acid-treated to remove metal oxide impu-rities and contained no clay minerals or silica fines, and the nano-particles were all washed – there was no unadsorbed polyelectro-lyte in solution for these measurements. A clear trend of increasingmobility through the column with increasing pH is observed. FornZVI transport, the effect of aggregation (average particle size) onits mobility cannot be distinguished from the effect of pH on depo-sition because both factors vary at the same time. However, com-paring the transport of aggregating nZVI (MRNIP and triblock-RNIP with a DLS dh � 1.5 lm) with PAP-hematite (DLSdh � 200 nm) can be used to infer the impact of aggregation (par-ticle size) on transport.

Elution of PAP-hematite (DLS hydrodynamic diameter�200 nm) was limited to less than 10% at pH 6 compared to 70%at pH 7 and 80% at pH 8, indicating that pH affects the depositionof nano-particles onto the sand grains independent of aggregation(Fig. 4). Elutability was highly limited at pH 6 indicating that depo-sition onto the sand surfaces (as opposed to mechanical filtration)is very important at this pH. At pH 7 and 8 elution of PAP-hematitewas rather high, indicating a relatively weak role of direct deposi-tion on transport. The deposition of MRNIP and triblock-RNIP wasgreater than PAP-hematite at all pH values evaluated, suggestingthat aggregation of MRNIP and triblock-RNIP increased depositionby mechanical filtration which decreased transport. The greatertransport of MRNIP and triblock-RNIP at pH = 8 compared topH = 6 and 7 even though the DLS dh sizes was similar at pH = 6and 8, suggests that enhanced deposition onto sand grains at lowerpH (perhaps due to the lower EPM at pH = 6 compared to pH = 8)has a greater effect on transport than pH induced aggregation ofMRNIP or triblock-RNIP. It cannot, however, be ruled out that thegreater mobility of PAP-hematite compared to MRNIP or triblock-RNIP through porous media is simply a result of the higher nega-tive surface charge of PAP-hematite compared to either modifiedRNIP. Regardless, it is clear that lower pH increases depositionand decreases transport of polymer-modified RNIP.

The greater deposition (and hence lower transport) at pH 6compared to pH 8 for all coated RNIP evaluated may also be dueto the presence of metal oxide (particularly Fe-oxide) impuritieson the sand surface. These impurities may be positively chargedat pH 6 and promote deposition of the negatively charged modifiedRNIP. To test this hypothesis, the transport of PAP-hematite nano-particles, in the absence of excess unadsorbed polyelectrolyte, wasmeasured using sand that was acid washed to remove the impuri-ties (Fig. 5). Transport of the nano-particles at pH 6 was signifi-cantly greater for the acid washed sand than for the untreatedsand, suggesting that the metal oxide impurities were indeed pro-viding favorable deposition sites. The zeta potential of the acidwashed sand (Fig. 6) was significantly more negative and closerto that expected for silica than for the untreated sand, further sug-gesting the presence of metal oxide impurities. Acid treatment ofthe sand also affected adsorption of PAP to the sand (Fig. 7). PAPadsorption to untreated sand at pH 6 was approximately one orderof magnitude greater than at pH 8, even though the zeta potentialof the untreated sand was nearly independent of pH. This suggests

20 40 60 80 200 400 600 8001000

5

10

15

20

25

30

35

500 750 2500 5000 75001000

0

20

40

60

80(a)

pH=8

MRNIP

pH=6

Inte

nsity

(%

)

Diameter (nm)

2.0 m1.1 m

PAP-hematite

Diameter (nm)500 750 25001000

0

20

40

60

80

1 μm 1.5 m

Diameter (nm)

Triblock-RNIP(b) (c)pH=8 pH=6 pH=6

pH=8

200 nm

200 nm

Fig. 3. Intensity-weighted aggregate size distribution measured by dynamic light scattering at a time of 3 min after ending sonication. (a) MRNIP, (b) triblock-RNIP and (c)PAP-hematite in 1 mM NaHCO3 and 10 mM NaCl at pH 6 and 8. The open circles (s) are for pH 8 and the closed circles (d) are for pH 6.

MRNIP triblock-RNIP PAP-hematite

Elut

ion

(%)

0

20

40

60

80

100pH 6pH 7pH 8

Fig. 4. Percent of mass eluted for MRNIP, triblock-RNIP, and PAP-hematite elutedthrough 15 cm columns filled with silica sand at pH 6, 7 and 8. The eluted mass oftriblock-RNIP was near zero at pH = 6. The initial RNIP or hematite particleconcentration was 300 mg/L, porosity of 0.33, linear pore water velocity was3.2 � 10�4 m/s and ionic strength was controlled at 10 mM NaCl plus 1 mMNaHCO3.

acid washed sand sand

Elut

ion

(%)

0

20

40

60

80

100

120pH6pH7pH8

PAP-hematite

Fig. 5. Percent of PAP-hematite mass eluted through 15 cm columns filled withsand and acid washed sand at pH = 6, pH = 7 and pH = 8 with porosity of 0.33. Theinitial hematite particle concentration was 300 mg/L, linear pore water velocity was3.2 � 10�4 m/s and ionic strength was controlled at 10 mM NaCl plus 1 mMNaHCO3.

pH6 7 8 9

-50

-40

-30

-20

-10

0

Zeta

pot

entia

l (m

V)

sand

acid washed sand

Fig. 6. Zeta potential of sand and acid washed sand at pH 6, 7 and 8. Zeta potentialswere calculated from measured streaming potentials. Experiments were conductedwith 10 mM NaCl plus 1 mM NaHCO3 and as a background electrolyte at 25 �C.

Caq (mg/L)0 50 100 150 200

Adso

rbed

mas

s (m

g/kg

)

0

10

20

30

40

50

Γmax = 68.05 mg/kg

pH 6

pH8 Γmax = 8.86 mg/kg

Fig. 7. Adsorption isotherm of PAP onto untreated sand at pH 6 (closed triangle N)and pH 8 (open triangle 4) in the presence of 10 mM NaCl + 1 mM NaHCO3. Thefitted line is based on the Langmuir isotherm for convenience.

6 H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10

that at the lower pH, anionic PAP may be adsorbing preferentiallyto positively charged impurities on the sand surface that have apHiep higher than pH 6. This may be an important considerationwhen extrapolating transport results from laboratory studies usingclean sands to real field soils where chemical heterogeneity such asadsorbed iron oxide are likely to be present and may affect trans-port considerably.

3.3. Heteroaggregation between clay and bare or modified RNIP insuspension

A heteroaggregation study was performed to determine if thereis pH dependent deposition of uncoated or modified RNIP ontokaolinite particles to support measurements of the effect of kaolin-ite on column transport discussed below. In each case, the aggrega-tion and sedimentation of a mixture of RNIP (bare or coated,

H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10 7

200 mg/L) and kaolinite (800 mg/L) was monitored and comparedto the homoaggregation and sedimentation of bare or modifiedRNIP or kaolinite at the same pH. Heteroaggregation is indicatedby a shift in the critical time for sedimentation to occur [10,14].Heteroaggregation between bare or surface modified RNIP andkaolinite occurred at pH 6 7, but not at pH 8 (Fig. 8 shows datafor bare RNIP; results for both uncoated and modified RNIP typesare summarized in Table 1). The pH where heteroaggregation

Time (min)

0.01

0.1

1T1

T2

kaolinite

RNIP + kaolinite

aggregation regime

sedimentation regime

pH=8

OD

508 /

initi

al O

D50

8

Time (min)

0 200 400 600 800 1000 1200 1400

0 200 400 600 800 1000 1200 14000.01

0.1

1

T1

T2kaolinite

ΔΤ : heteroaggregationpH=7

RNIP + kaolinite

OD

508 /

initi

al O

D50

8

Time (min)

0 50 100 150 200

OD

508 /

initi

al O

D50

8

0.1

1T1T2

kaolinite

pH=6

RNIP + kaolinite

(a)

(b)

(c)

Fig. 8. Heteroaggregation was evaluated by monitoring the sedimentation of bareRNIP or MRNIP (200 mg/L), kaolinite (800 mg/L), or a mixture of MRNIP andkaolinite at (a) pH 8, (b) pH 7, and (c) pH 6. Sedimentation was monitored bymeasuring optical density at 508 nm using UV–vis spectrophotometer in 1 mMNaHCO3 + 10 mM NaCl for 24 h. The black line is RNIP, the orange line is kaolinite,and the green line is the mixture of kaolinite and RNIP. T1 represents the criticalsedimentation time of kaolinite and T2 represents the critical sedimentation time ofthe mixture of RNIP and kaolinite. (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

occurs in this study falls into the reported range of pHpzc-clay edge

(from pH 5.8 to 7.3) and pHpzc-RNIP = 6.3 [34,38], suggesting thatuncoated and modified RNIP are depositing onto the positivelycharged edge sites of the kaolinite at pH < 7 or that the net repul-sive force between the RNIP and the kaolinite is decreasing due to alower net charge on both at the lower pH.

The homoaggregation measurements reveal other kaolinitebehaviors that can affect uncoated and coated RNIP transport insand amended with kaolinite. At pH < 7, the critical time tohomoaggregation for kaolinite was decreased relative to high pH.This is a result of ‘‘edge-to-face aggregation’’ among kaolinite par-ticles having both the permanent negatively charged face and pos-itively charged edge sites simultaneously in the structure. This isfurther supported by the overall decrease in the kaolinite zeta po-tential with decreasing pH due to the positive charge developmentat the edge sites of kaolinite (Fig. 9), as well as an increase in kao-linite aggregate size from that measured at pH 8 (�800 nm) to thatmeasured at low pH (�4 lm) (Fig. S5). Also, aggregation of kaolin-ite in porous media was observed at pH 6 6 (Fig. S6) which couldplug pore spaces and prevent transport via mechanical filtration.In summary, the heteroaggregation study suggests that at pH < 7there is enhanced deposition of coated or uncoated RNIP onto kao-linite clay particles. Also, enhanced mechanical filtration of RNIPdue to aggregation of the kaolinite in the pore spaces is possible.Either process may decrease the mobility of nZVI through subsur-face media as described next.

3.4. Effect of clay minerals or silica fines on the transport of surfacemodified RNIP

The effect of the presence of kaolinite or silica fines in sand col-umns on nZVI transport was investigated by comparing the trans-port of coated RNIP in sand columns amended with either 2 wt.%silica fines or 2 wt.% kaolinite (Figs. 10–12). The presence of eithersilica fines or kaolinite increases the available surface area for po-tential RNIP deposition and can decrease the effective pore size be-tween sand grains, which can in turn decrease transport bymechanical filtration. Even though the silica fines and the kaoliniteparticles were the same size (dp � 1.5 lm), the presence of 2 wt.%of one or the other of these particles had a different effect on trans-port due differences in their surface chemistry. Despite the in-creased surface area for deposition, the presence of silica finesincreased transport compared to unamended sand columns for tri-block-RNIP and for PAP-hematite (Figs. 11 and 12). Silica fines hadlittle effect on transport of MRNIP which was less stable againstaggregation and deposition compared to triblock-RNIP (Fig. 10).The greater transport for triblock-RNIP and for PAP-hematite inthe presence of silica fines is due in part due to the increase inaverage porewater velocity in the pore throats due to the presenceof fines [48], however, the increased velocity cannot explain all ofthe increase in transport. (Supporting Information Fig. S7). Thissuggests that the presence of silica fines (without surface impuri-ties) is also blocking access to favorable deposition sites on thesand surfaces.

In contrast to the silica fines, the presence of 2 wt.% kaolinite inporous media generally decreased the mobility of RNIP relative tothe silica fines at all pH conditions evaluated (Figs. 10 and 11). Thetriblock-RNIP was less affected than MRNIP, likely due to the great-er electrosteric repulsion [14,16] from the negatively charged kao-linite faces afforded by this polymer relative to the PAP. Enhanceddeposition onto kaolinite occurred even for the PAP-coated hema-tite (Fig. 12) which had the greatest overall surface charge (Fig. 2)and did not aggregate. Deposition was greatest at pH 6 where hete-roaggregation between uncoated or modified RNIP and the kaolin-ite occurs in suspension, and where the kaolinite tends to undergoface-to-edge aggregation and can block pore spaces.

Table 1Summary of the heteroaggregation study.

Critical sedimentation time (min) pH 8a pH 7a pH 6a pH 5a

Kaolinite 197 ± 6 200 120 ± 17 17 ± 1Kaolinite + uncoated RNIP 197 ± 6 108 ± 21 7 ± 1 4 ± 1Kaolinite + MRNIP 263 ± 21 203 ± 6 14 ± 1 9 ± 1Kaolinite + triblock-RNIP 220 ± 17 200 ± 10 13 ± 1 9Kaolinite + excess PAP (267 mg/L) 223 ± 21 173 ± 36 178 ± 21 180 ± 12Kaolinite + MRNIP + excess PAP (267 mg/L) 210 ± 12 200 193 ± 12 193 ± 12

a Values were measured from triplicate samples and the average values and ±standard deviations are reported.

pH4 5 6 7 8 9

-60

-55

-50

-45

-40

-35

-30

Zeta

pot

entia

l (m

V)

Fig. 9. Zeta potential of kaolinite at pH 5, 6, 7 and 8. Kaolinite concentration was15 mg/L and the ionic strength was controlled at 1 mM NaHCO3 + 10 mM NaCl.

sand 2% fine 2% kaolinite

Elut

ion

(%)

0

20

40

60

80

100pH6pH7pH8

MRNIP

Fig. 10. Percent of mass eluted for MRNIP (300 mg/L) through a 15 cm column filledwith silica sand, sand + 2 wt.% silica fines (1.7 lm), or sand + 2 wt.% clay (1.36 lm)at pH 6, 7 and 8. The porosity was 0.33, the linear pore water velocity was3.2 � 10�4 m/s, and ionic strength was controlled at 10 mM NaCl + 1 mM NaHCO3.

sand 2% fine 2% kaoliniteEl

utio

n(%

)0

20

40

60

80

100pH6pH7pH8

triblock-RNIP

Fig. 11. Percent of mass eluted for triblock-RNIP (300 mg/L) through a 15 cmcolumn filled with silica sand, sand + 2 wt.% silica fines (1.7 lm), or sand + 2 wt.%clay (1.36 lm) at pH 6, 7 and 8. The porosity was 0.33, the linear pore water velocitywas 3.2 � 10�4 m/s, and ionic strength was controlled at 10 mM NaCl + 1 mMNaHCO3.

sand 2% fine 2% kaolinite

Elut

ion

(%)

0

20

40

60

80

100pH6pH7pH8

PAP-hematite

Fig. 12. Percent of mass eluted for PAP-hematite (300 mg/L) through a 15 cmcolumn filled with silica sand, sand + 2 wt.% silica fines (1.7 lm), or sand + 2 wt.%clay (1.36 lm) at pH 6, 7 and 8. The porosity was 0.33, the linear pore water velocitywas 3.2 � 10�4 m/s, and ionic strength was controlled at 10 mM NaCl + 1 mMNaHCO3.

8 H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10

3.5. Effect of excess polymer on the transport of surface modified RNIP

Excess PAP in solution was shown to increase transport ofMRNIP and hematite in (untreated) sand columns (Fig. 13) byadsorbing to and modifying the sand surface (Fig. 7). Similarbehavior was observed for transport in the presence of silica finesand kaolinite in the presence of excess PAP in solution. Adding asmall amount of excess polymer (100 mg/L) to the injection solu-tion dramatically improved the mobility of MRNIP through all ofthree types of porous media used (sand, sand + 2 wt.% silica fines,and sand + 2 wt.% kaolinite clay) (Fig. 13a). The presence of excessunadsorbed polymer prevented heteroaggregation between kao-linite and MRNIP at all pH values evaluated (Table 1), as well asedge to face aggregation of kaolinite. These effects can all be attrib-uted to polyanionic PAP adsorbing to positively charged metal

oxide impurities on untreated sand and edge sites on kaolinite.These results suggest that excess PAP in the nZVI injection solutionmay inhibit nZVI deposition onto the positively charged sites ofkaolinite and decreases aggregation of clay in the pore spaces be-tween sand grains. Eliminating these two mechanisms for filteringnZVI results in significantly less deposition and greater transport. Itis noteworthy that 100 mg/L of excess unadsorbed PAP in solutionappears to have mobilized a fraction of the clay within the column,as judged by the visual appearance of the eluted suspension(Fig. 13b). This may have also enhanced transport of the MRNIP,if some MRNIP were adsorbed to, but shuttled by, mobilizedkaolinite.

sand 2% fine 2% clay

(b)(a)

Fig. 13. (a) Elution of 300 mg/L MRNIP in the presence of 100 mg/L excess unadsorbed polymer in solution (corresponds to a 0.333 mass ratio of unadsorbed PAP2.5K to RNIP)in silica sand and silica fine- and kaolinite-amended sand columns at pH = 6. Column length was 15 cm with porosity of 0.33. The linear pore water velocity was 3.2 � 10�4 m/s and ionic strength was 10 mM NaCl plus 1 mM NaHCO3. (b) Appearance of eluted MRNIP suspension with 100 mg/L excess polymer in solution at pH 6.

H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10 9

4. Conclusions and environmental implications

This study highlights the importance of commonly occurringsite specific hydrogeochemical factors such as physical and chem-ical heterogeneity of porous media and groundwater pH on subsur-face transport of surface modified nZVI. At pH 6, aggregation anddirect deposition of polyanion-modified RNIP limited its transport.At higher pH, aggregation continues to play a role in decreasedtransport, but the strong polyanion coatings better serve to de-crease direct deposition. Direct deposition to impurities with pHpzc

between 6 and 8 on the sand surface limited surface modified RNIPtransport. Even for seemingly favorable conditions for transport,(i.e. sandy materials with less than 1% metal oxide impurities),the presence of small percentages of clay minerals resulted in RNIPdeposition and severely limited surface modified RNIP transport atpH 6 to 8.

For the nZVI types and injection conditions evaluated here,transmission of nZVI through the 15 cm column was less than70% without addition of excess polymer. This suggests transportdistances of �1 m for 90% removal of particles, which could makenZVI emplacement over large contaminated regions impracticalwithout addition of excess polymer. One way to improve nZVItransport may be to co-inject small amounts of excess unadsorbedpolymer in solution [13]. Excess polymer in the injection can min-imize surface modified nZVI deposition onto sand or clay surfacesby adsorbing to positively charged sites in the subsurface media,making the surface charge of porous media uniformly negativeand increasing the electrosteric repulsions between nZVI and themedia. In mildly acidic soils, and where the porous medium con-tains a high portion of surface charge heterogeneity due to metaloxide or clay minerals, using excess free polymer in the injectionsolution can increase surface modified nZVI mobility through sub-surface media. The results are based on an injected nZVI concentra-tion of 0.3 g/L, which is much lower than is typically applied in thefield (2–10 g/L). While aggregation and enhanced deposition ontofavorable deposition sites on the porous media can both limittransport of nZVI, the higher nZVI concentrations used in practicewill promote aggregation, potentially making physical strainingthe dominant factor limiting transport [25,48]. Ionic strength andcomposition (held constant in this study), and adsorbed natural or-ganic matter on the coated nZVI and the sand and clay surfaces willalso affect the deposition and transport behavior of nZVI. All ofthese variables must be evaluated to better predict coated nZVItransport behavior in natural porous media.

Acknowledgments

This research was funded in part by the US EPA (R833326), theNSF (BES-068646 and EF-0830093), the Department of Defense

through the Strategic Environmental Research and DevelopmentProgram (W912HQ-06-C-0038). We specially thank Sharon L.Walker of the University of California at Riverside and HyunjungN. Kim of the Chonbuk National University in Korea for streamingpotential analysis.

Appendix A. Supplementary material

Additional information about the grain size distributions ofsand, silica fines, kaolinite clay, adsorption isotherm of polyaspar-tate 2.5 K on hematite nano-particles, pH dependent aggregate sizeof kaolinite clay and microscopic observations of MRNIP aggrega-tion and deposition onto kaolinite clay in a sand–clay mixed mediain a micro-fluidic cell. Supplementary data associated with thisarticle can be found, in the online version, at doi:10.1016/j.jcis.2011.12.059.

References

[1] E. Dalla Vecchia, M. Luna, R. Sethi, Environ. Sci. Technol. 43 (2009) 8942.[2] F. He, D.Y. Zhao, Environ. Sci. Technol. 41 (2007) 6216.[3] F. He, D.Y. Zhao, J.C. Liu, C.B. Roberts, Ind. Eng. Chem. Res. 46 (2007) 29.[4] D.W.W. Keith, W. Henn, Remediation 16 (2006) 57.[5] Y.Q. Liu, H. Choi, D. Dionysiou, G.V. Lowry, Chem. Mater. 17 (2005) 5315.[6] D. Wang, L. Chu, M. Paradelo, W. Peijnenburg, Y. Wang, D. Zhou, J. Colloid

Interf. Sci. 360 (2011) 398.[7] N. Saleh, K. Sirk, Y.Q. Liu, T. Phenrat, B. Dufour, K. Matyjaszewski, R.D. Tilton,

G.V. Lowry, Environ. Eng. Sci. 24 (2007) 45.[8] P.G. Tratnyek, R.L. Johnson, Nano Today 1 (2006) 44.[9] N. Saleh, T. Phenrat, K. Sirk, B. Dufour, J. Ok, T. Sarbu, K. Matyiaszewski, R.D.

Tilton, G.V. Lowry, Nano Lett. 5 (2005) 2489.[10] T. Phenrat, N. Saleh, K. Sirk, R.D. Tilton, G.V. Lowry, Environ. Sci. Technol. 41

(2007) 284.[11] N.D. Berge, C.A. Ramsburg, Environ. Sci. Technol. 43 (2009) 5060.[12] D.W. Elliott, W.X. Zhang, Environ. Sci. Technol. 35 (2001) 4922.[13] R.L. Johnson, G.O. Johnson, J.T. Nurmi, P.G. Tratnyek, Environ. Sci. Technol. 43

(2009) 5455.[14] T. Phenrat, N. Saleh, K. Sirk, H.J. Kim, R.D. Tilton, G.V. Lowry, J. Nanopart. Res.

10 (2008) 795.[15] H.-J. Kim, T. Phenrat, R.D. Tilton, G.V. Lowry, Environ. Sci. Technol 43 (2009)

3824.[16] N. Saleh, H.-J. Kim, T. Phenrat, K. Matyjaszewski, R.D. Tilton, G.V. Lowry,

Environ. Sci. Technol. 42 (2008) 3349.[17] T. Phenrat, J.E. Song, C.M. Cisneros, D.P. Schoenfelder, R.D. Tilton, G.V. Lowry,

Environ. Sci. Technol. 44 (2010) 4531.[18] A. Majzik, E. Tombacz, Org. Geochem. 38 (2007) 1330.[19] A. Majzik, E. Tombacz, Org. Geochem. 38 (2007) 1319.[20] C.W. Hoogendam, A. de Keizer, M.A.C. Stuart, B.H. Bijsterbosch, J.G. Batelaan,

P.M. van der Horst, Langmuir 14 (1998) 3825.[21] J.K. Wolterink, L.K. Koopal, M.A.C. Stuart, W.H. Van Riemsdijk, Colloids Surf. A

291 (2006) 13.[22] A. Katchalsky, N. Shavit, H. Eisenberg, J. Polym. Sci. 13 (1954) 69.[23] Y.T. Wu, C. Grant, Langmuir 18 (2002) 6813.[24] T. Phenrat, A. Cihan, H.J. Kim, M. Mital, T. Illangasekare, G.V. Lowry, Environ.

Sci. Technol. 44 (2010) 9086.[25] T. Phenrat, H.J. Kim, F. Fagerlund, T. Illangasekare, G.V. Lowry, J. Contam.

Hydrol. 118 (2010) 152.[26] G.M. Litton, T.M. Olson, J. Colloid Interf. Sci. 165 (1994) 522.

10 H.-J. Kim et al. / Journal of Colloid and Interface Science 370 (2012) 1–10

[27] N. Tufenkji, M. Elimelech, Langmuir 21 (2005) 841.[28] G.M. Litton, T.M. Olson, Environ. Sci. Technol. 27 (1993) 185.[29] J.Y. Chen, C.H. Ko, S. Bhattacharjee, M. Elimelech, Colloids Surf. A 191 (2001) 3.[30] M. Elimelech, M. Nagai, C.H. Ko, J.N. Ryan, Environ. Sci. Technol. 34 (2000)

2143.[31] L.F. Song, P.R. Johnson, M. Elimelech, Environ. Sci. Technol. 28 (1994) 1164.[32] J.C. Joo, C.D. Shackelford, K.F. Reardon, J. Colloid Interf. Sci. 317 (2008) 424.[33] P.R. Johnson, N. Sun, M. Elimelech, Environ. Sci. Technol. 30 (1996) 3284.[34] E. Tombacz, M. Szekeres, Appl. Clay Sci. 34 (2006) 105.[35] V. Ravisangar, K.E. Dennett, T.W. Sturm, A. Amirtharajah, Environ. Eng. 127

(2001) 531.[36] H. Van Olphen, An introduction to clay colloid chemistry, Wiley, New York,

1977.[37] K. Itami, H. Fujitani, Colloids Surf. A 265 (2005) 55.[38] E. Tombacz, Z. Libor, E. Illes, A. Majzik, E. Klumpp, Org. Geochem. 35 (2004)

257.[39] T.M. Herrington, A.Q. Clarke, J.C. Watts, Colloids and Surf 68 (1992) 161.[40] R. Kretzschmar, H. Holthoff, H. Sticher, J. Colloid Interf. Sci. 202 (1998) 95.[41] E. Tombacz, G. Filipcsei, M. Szekeres, Z. Gingl, Colloids Surf. A 151 (1999) 233.

[42] C. Galindo-Gonzalez, J. de Vicente, M.M. Ramos-Tejada, M.T. Lopez-Lopez, F.Gonzalez-Caballero, J.D.G. Duran, Langmuir 21 (2005) 4410.

[43] E. Tombacz, C. Csanaky, E. Illes, Colloid. Polym. Sci. 279 (2001) 484.[44] J. Fang, X.Q. Shan, B. Wen, J.M. Lin, G. Owens, Environ. Pollut. 157 (2009) 1101.[45] J.N. Ryan, P.M. Gschwend, Environ. Sci. Technol. 28 (1994) 1717.[46] S.B. Roy, D.A. Dzombak, Colloids Surf. A 107 (1996) 245.[47] J.T. Nurmi, P.G. Tratnyek, V. Sarathy, D.R. Baer, J.E. Amonette, K. Pecher, C.M.

Wang, J.C. Linehan, D.W. Matson, R.L. Penn, M.D. Driessen, Environ. Sci.Technol. 39 (2005) 1221.

[48] T. Phenrat, H.-J. Kim, T. Fagerlund, T. Illangasekare, R.D. Tilton, G.V. Lowry,Environ. Sci. Technol. 43 (2009) 5079.

[49] U.S. Rochelle, M. Cornell, The Iron Oxides, second edition., Oxford UniversityPress, New York, 2000.

[50] H.N. Kim, S.A. Bradford, S.L. Walker, Environ. Sci. Technol. 43 (2009) 4340.[51] N.B. Saleh. An assessment of novel polymeric coatings to enhance transport

and in situ targeting of nanoiron for remediation of non-aqueous phase liquids(NAPLs). Carnegie Mellon University, Pittsburgh, 2007.

[52] Y.T. He, J.M. Wan, T. Tokunaga, J. Nanopart. Res. 10 (2008) 321.