lable at ScienceDirect

Water Research 109 (2017) 358e366

Contents lists avai

Water Research

journal homepage: www.elsevier .com/locate/watres

Roles of cation valance and exchange on the retention andcolloid-facilitated transport of functionalized multi-walled carbonnanotubes in a natural soil

Miaoyue Zhang a, d, Scott A. Bradford b, *, Jirka �Sim�unek c, Harry Vereecken a,Erwin Klumpp a

a Agrosphere Institute (IBG-3), Forschungszentrum Jülich GmbH, 52425 Jülich, Germanyb United States Department of Agriculture, Agricultural Research Service, U. S. Salinity Laboratory, Riverside, CA 92507, USAc Department of Environmental Sciences, University of California Riverside, Riverside, CA 92521, USAd Institute for Environmental Research (Biology V), RWTH Aachen University, Worringerweg 1, 52074 Aachen, Germany

a r t i c l e i n f o

Article history:Received 18 August 2016Received in revised form28 November 2016Accepted 28 November 2016Available online 29 November 2016

Keywords:Multi-walled carbon nanotubesSoilBreakthrough curvesRetention profilesCation exchangeSoil fractionation

* Corresponding author. United States DepartmenResearch Service, U. S. Salinity Laboratory, 450 W. Bi92507, USA.

E-mail address: scott.bradford@ars.usda.gov (S.A.

http://dx.doi.org/10.1016/j.watres.2016.11.0620043-1354/Published by Elsevier Ltd.

a b s t r a c t

Saturated soil column experiments were conducted to investigate the transport, retention, and releasebehavior of a low concentration (1 mg L�1) of functionalized 14C-labeled multi-walled carbon nanotubes(MWCNTs) in a natural soil under various solution chemistries. Breakthrough curves (BTCs) for MWCNTSexhibited greater amounts of retardation and retention with increasing solution ionic strength (IS) or inthe presence of Ca2þ in comparison to Kþ, and retention profiles (RPs) for MWCNTs were hyper-exponential in shape. These BTCs and RPs were well described using the advection-dispersion equa-tion with a term for time- and depth-dependent retention. Fitted values of the retention rate coefficientand the maximum retained concentration of MWCNTs were higher with increasing IS and in the pres-ence of Ca2þ in comparison to Kþ. Significant amounts of MWCNT and soil colloid release was observedwith a reduction of IS due to expansion of the electrical double layer, especially following cation ex-change (when Kþ displaced Ca2þ) that reduced the zeta potential of MWCNTs and the soil. Analysis ofMWCNT concentrations in different soil size fractions revealed that >23.6% of the retained MWCNT masswas associated with water-dispersible colloids (WDCs), even though this fraction was only a minorportion of the total soil mass (2.38%). More MWCNTs were retained on the WDC fraction in the presenceof Ca2þ than Kþ. These findings indicated that some of the released MWCNTs by IS reduction and cationexchange were associated with the released clay fraction, and suggests the potential for facilitatedtransport of MWCNT by WDCs.

Published by Elsevier Ltd.

1. Introduction

Carbon nanotubes (CNTs) are allotropes of carbon with acylindrical-shaped nanostructure (Iijima, 1991). They have beenutilized in numerous commercial applications (Gohardani et al.,2014) such as electrical cables and wires (Janas et al., 2014), solarcells (Guldi et al., 2005), hydrogen storage (Jones and Bekkedahl,1997), radar absorption (Lin et al., 2008) and considering as ab-sorbents for environmental remediation and water treatment

t of Agriculture, Agriculturalg Springs Road, Riverside, CA

Bradford).

(Camilli et al., 2014; Li et al., 2012; Mauter and Elimelech, 2008; Panand Xing, 2012; Zhang et al., 2010) due to their unique electric,chemical, and physical properties. The widespread commercial andpotential environmental applications will undoubtedly result intheir release into the subsurface. Current studies have investigatedthe transport behavior of CNTs in different porous media, which isaffected by various physical and chemical conditions includingionic strength (IS), water content, grain size, input concentrationand dissolved organic matter (Kasel et al., 2013a, 2013b; Tian et al.,2012; Yang et al., 2013). However, most of these studies wereconducted in model soil systems such as rigorously cleaned sand orglass beads, which do not reflect the full complexity and hetero-geneity of natural soils (e.g. complex grain size distribution andpore structure, and surface roughness and chemical heterogeneity)

M. Zhang et al. / Water Research 109 (2017) 358e366 359

(Cornelis et al., 2014). Kasel et al. (2013b) found limited transport offunctionalized multi-walled carbon nanotubes (MWCNTs) in anundisturbed, natural soil. However, information on CNTs in-teractions with soil colloids and the potential for colloid-facilitatedtransport of CNT is still limited. Furthermore, the detachment orrelease of retained engineered nanoparticles (ENPs) like CNTs fromthe solid phase is important for predicting the ultimate fate andtransport of ENPs in the subsurface, but has received little researchattention.

Previous studies have demonstrated that pore-water chemistryplays an important role in controlling the retention of ENPs in soils(Badawy et al., 2010; Gao et al., 2006; Tian et al., 2012; Wang et al.2014, 2015). Retention of ENPs is enhanced at higher ionic strength(IS) because the adhesive force increases with compression of theelectric double layer (EDL) and a decrease in the magnitude of thesurface potential (Elimelech et al., 2013; Israelachvili, 2011; Khilarand Fogler, 1998). Retention of ENPs is also higher in the presenceof similar concentrations of divalent than monovalent cations(Torkzaban et al., 2012; Yang et al., 2013) because the double layerthickness and magnitude of the surface potential decreases and theIS increases to a greater extent (Elimelech et al., 2013; Israelachvili,2011; Khilar and Fogler, 1998). In addition, adsorbed divalent cat-ions, such as Ca2þ, can enhance retention of ENPs by creatingnanoscale chemical heterogeneity on the solid surface that canneutralize or reverse the surface charge at specific locations(Grosberg et al., 2002) and/or produce a cation bridge betweennegatively charged sites on the surface of clays and ENPs(Torkzaban et al., 2012). The relative importance of solution IS andcation type on retention of ENPs is therefore expected to depend onthe presence of soil colloids, but these complexities have not yetbeen fully resolved for natural soils.

A reduction in solution IS and cation exchange (monovalentdisplacing divalent cations) decreases the adhesive force andthereby induces release of clay particles and ENPs (Bradford andKim, 2010; Grolimund and Borkovec, 2006; Liang et al., 2013;Torkzaban et al., 2013). The migration of soil colloids can facilitatethe transport of pollutants and/or nanoparticles (Bradford andTorkzaban, 2008; Grolimund and Borkovec, 2005; Liang et al.,2013; Yan et al., 2016; Zhu et al., 2014). For example, Liang et al.(2013) provided experimental evidence that release of soil col-loids with IS reduction and cation exchange can facilitate thetransport of silver nanoparticles.Water-dispersible colloids (WDCs)are indicators for mobile soil colloids (de Jonge et al., 2004); e.g.,WDCs are particles from the soil clay fraction that are less than2 mm in size that are easily dispersible from soil with aqueous so-lution (Jiang et al., 2012, 2014). Knowledge of the interaction andassociation between ENPs and WDCs is therefore important forbetter understanding the fate of ENPs in soil. The effects of per-turbations in solution chemistry on the release and colloid-facilitated transport of MWCNTs in soil have not yet been studied.

The objective of this study is to better understand roles of soilcolloids, solution IS, and cation type on the transport, retention andrelease behavior of functionalized MWCNTs (1 mg L�1) in a naturalsoil. Breakthrough curves and retention profiles for MWCNTs weredetermined in column experiments, and a numerical model wasemployed to simulate their fate. The retained concentration ofMWCNTs was subsequently determined for the different soil sizefractions (e.g., sand, silt, and WDCs). Other experiments wereconducted to study the release of soil colloids and MWCNTs withperturbations in solution chemistry (e.g., IS reductions and cationexchange). Results provide valuable insight on the roles of cationtype, chemical perturbations, and soil colloids on the retention,release and colloid-facilitated transport of MWCNTs. This knowl-edge can be useful for environmental applications and risk man-agement of MWCNTs.

2. Materials and methods

2.1. Soil and MWCNT

Soil samples were collected from the upper 30 cm of an agri-cultural field site in Germany (Kaldenkirchen-Hülst, Germany),sieved to a fraction < 2 mm, and air dried. The soil was classified asa loamy sand with 4.9% clay (<2 mm), 26.7% silt (2e63 mm), and68.5% sand (63e2000 mm). It had a median grain size (d50) of120 mm, a total organic carbon content of 1.1%, a cation exchangecapacity of 7.8 cmolc kg�1, a pH value of 5.9 (0.01M CaCl2), a specificsurface area of 1.7 m2 g�1, and 0.8% iron. The clay faction wascomposed of illite, montmorillonite, and kaolinite minerals (Kaselet al., 2013b; Liang et al., 2013).

Radioactively (14C) labeled and unlabeled functionalizedMWCNTs (Bayer Technology Services GmbH, Leverkusen, Germany)were boiled with 70% nitric acid (Sigma-Aldrich Chemie GmbH,Steinheim, Germany) for 4 h to create additional oxygen-containingfunctional groups (e.g., carboxylic groups) on their surfaces (Kaselet al., 2013a). The functionalization and characterization ofMWCNTs were previously described by Kasel et al. (2013a, 2013b).In brief, a transmission electron microscope was used to charac-terize morphological properties of MWCNTs, X-ray photoelectronspectroscopy was used to identify oxygen containing functionalgroups on MWCNTs, and inductively coupled plasma-mass spec-trometry was used to determine the amount of metal catalystsbefore and after acid treatment of MWCNTs. The MWCNTs have amedian diameter of 10e15 nm and a median length of200e1000 nm (Pauluhn, 2010).

Suspensions of MWCNT at selected IS were prepared in deion-ized water with KCl and CaCl2. These suspensions were dispersedby ultrasonicating the stock suspension for 15 min at 65 W, andthen repeating this process 10 min before injection into soil col-umns. This low energy of sonication does not damage the MWCNTs(Li et al., 2012).

The hydrodynamic radius measured by dynamic light scattering(DLS) does not reflect the real geometric particle diameter for non-spherical particles like MWCNTs (Hassell€ov et al., 2008; Pecora,2000). Nevertheless, it can be used to study the aggregationbehavior of functionalized MWCNTs suspensions. The hydrody-namic radius of unlabeled functionalized MWCNTs suspended inKCl and CaCl2 at IS ¼ 10 mM were measured 0, 1, and 4 h aftersuspension preparation by DLS using a Zetasizer Nano (MalvernZetaSizer 4). The surface charge characteristics and electrophoreticmobility values of unlabeled functionalized MWCNT (pH z 5.4)and crushed soil (pH z 6) in selected electrolyte solutions werealso determined using the Zetasizer Nano apparatus.

2.2. Transport and retention experiments

Saturated MWCNT transport experiments were conducted atdifferent IS (1, 4, and 10 mM KCl) and different cation type at thesame IS (1 mM KCl and CaCl2, molar concentration 0.33 mM L�1

CaCl2) by using stainless steel columns (3 cm inner diameter and12 cm length) that were wet-packed with soil. Each side of thecolumn was fitted with a stainless steel plate (1 mm opening) anddouble PTFE mesh (100 mm openings) to support the soil and touniformly distribute the flow. The columns were wet packed byincrementally filling the column with soil and Milli-Q water, andthen gently tapping the columns with a rubber mallet to ensurecomplete water saturation. Steady-state flow was achieved using aperistaltic pump, with the flow direction from the column bottomto the top. The saturated hydraulic conductivity and bulk density ofthe packed soil column was approximately 0.73 cm min�1 and1.5 g cm�3, respectively.

M. Zhang et al. / Water Research 109 (2017) 358e366360

The soil in the column was equilibrated before initiating thetransport experiment by injecting approximately 30 pore volumes(PVs) of a selected background electrolyte solution at a slow, con-stant Darcy velocity of 0.18 cmmin�1. Conservative tracer (KBr) andMWCNT transport experiments were then conducted sequentiallyat the same IS and cation type as the equilibrium phase, but at ahigher, constant Darcy velocity of 0.71e0.73 cm min�1. The con-servative tracer experiment consisted of injecting a 2.1 PVs pulse(90 mL) of KBr solution, followed by continued eluting with thesame bromide-free electrolyte solution for another 5.5 PVs. Theeffluent concentrations of bromide were determined using a high-performance liquid chromatograph (STH 585, Dionex, Sunnyvale,CA, USA) equipped with a UV detector (UV2075, Jasco, Essex, UK).Transport experiments for MWCNTs (input concentration,1 mg L�1) were conducted in a similar manner as the conservativetracer; e.g., injection of a 2.1 PVs pulse of MWCNT suspension,followed by elution with the same particle-free electrolyte solutionfor another 5.5 PVs. The 14C-labeled effluent concentrations ofMWCNT were determined using a liquid scintillation counter (LSC)(PerkinElmer, Rodgau, USA). After recovery of the breakthroughcurve (BTC), the MWCNT retention profile (RP) was determined.Soil samples were excavated from a column in 0.5e1 cm in-crements, dried, crushed, combusted using a biological oxidizer at900 �C (OX 500, R.J. Harvey Instrumentation Corporation, Tappan,NY, USA), and retained concentrations of MWCNTs in the soil weredetermined by LSC measurement. A summary of the experimentalconditions is provided in Table 1.

2.3. Release experiments

Additional experiments were performed to study the releasebehavior of MWCNTs with IS reduction and cation exchange at aconstant Darcy velocity of 0.71e0.73 cm min�1. The initial deposi-tion phase (step A) was conducted using MWCNTs (1 mg L�1) in10 mM KCl (experiment I), 1 mM CaCl2 (experiment II), and 10 mMCaCl2 (experiment III) solutions in an analogous fashion toMWCNTs transport experiments discussed above (Section 2.2).Release experiments II and III were then initiated by changing theeluting solution chemistry in the following sequence: Milli-Q water(step B); KCl at the same IS as in step A (step C); Milli-Q water (stepD); 100 mM KCl (step E); and Milli-Q water (step F). Each solutionchemistry stepwas conducted for 7.6 PVs. Experiment I consisted ofonly steps A and B. The 14C-labeled effluent and soil concentrationsof MWCNT were measured in the same manner as in transportexperiments. Cation exchange and soil colloids were quantifiedduring the release experiments by measuring the effluent con-centrations of K, Ca, Fe, and Al by adding 5 mL of aqua regia for

Table 1Experimental conditions, hydraulic parameters and mass balance information for all col

IS [mM] Co [mg L�1] q [cm min�1] Disp. [cm] 4

Fig. 2 1, Kþ 1 0.72 0.604 0.4, Kþ 1 0.72 0.364 0.10, Kþ 1 0.71 0.417 0.

Fig. 3 1, Kþ 1 0.72 0.604 0.1, Ca2þ 1 0.72 0.711 0.

q [cm min�1] 4 ISA [mM] MA [%] MB H2O [%] ISC [mM] M

Fig. 4 I, 0.71 0.47 10, Kþ 4.6 42.2II, 0.72 0.49 1, Ca2þ 4.2 7.7 1, Kþ 0.III, 0.73 0.48 10, Ca2þ 0.03 0.3 10, Kþ 0.

Fig. 1: ionic strength effect; Fig. 2: cation type effect; Fig. 3: release of MWCNT by ionicvelocity; IS, ionic strength; 4,porosity; Disp.is the estimated longitudinal dispersivity;respectively; MA � MF are the mass percentages recovered from effluent (Steps A�F) inM*eff, M*soil, M*total are mass percentages of the replicate experiments recovered from efflshown in Figs. 1 and 2.

overnight digesting, diluting, and using inductively coupled opticalemission spectrometry (ICP- OES). Because radioactive samples(14C) are not allowed for ICP-OES measurement, the release ex-periments were repeated using unlabeled functionalized MWCNTsfor this analysis. The effluent from the release experiment was alsoanalyzed by DLS for soil colloids. A conservative tracer (KBr)experimentwas repeated (Section 2.2) after a release experiment toassess potential changes in the hydrodynamic properties of thecolumn.

2.4. Soil fractionation

Soil fractionation experiments were conducted to furtherelucidate the interactions between 14C-labeled functionalizedMWCNTs and the soil. The initial deposition phase was conductedsimilar to the MWCNT transport experiments discussed above(Section 2.2) at the same IS (1 mM KCl or CaCl2). After recovery ofthe MWCNT breakthrough curve, all the soil was excavated fromthe column, dried, and then the soil particle-size was fractionatedfollowing the method of S�equaris and Lewandowski (2003). Inbrief, 100 g of dried soil was added to a 1 L Duran bottle (Schott,Mainz, Germany) containing 200 mL of the same electrolyte solu-tion as the transport experiments (1 mM KCl or CaCl2) and hori-zontally shaken at 150 rpm for 6 h. An additional 600 mL of thesame electrolyte solution (1 mM KCl or CaCl2) solution was thenadded to this bottle and mixed before sedimentation. The pipettemethod was then used to separate the soil into three size classesbased on their sedimentation time according to Stoke's law (aparticle density of 2.65 g cm�3 was assumed): (i) >20 mm (sandsize) after 6 min; (ii) 2e20 mm (silt size) after 12 h; and (iii) <2 mm(water-dispersible colloids, WDCs) after 12 h. Triplicate 50 mLsamples of the WDC suspension were further fractionated bycentrifuging at 2525 � g for 4 min. Stoke's Law calculations indi-cated that the resulting supernatant consisted of electrolyte solu-tion and the WDC fraction that was <0.45 mm, whereas the pelletwas the WDC fraction that was 0.45e2 mm. Concentrations ofMWCNTs that were associated with the sand, silt, 0.45e2 mm, and<0.45 mm fractions were determined by LSC in a similar manner tothe transport experiments. Concentrations of MWCNTs in the col-umn effluent and residual soil samples from columns that were notused for soil fractionation were measured to ensure mass balance.All the experiments were replicated and exhibited similar results.

2.5. Numerical modeling

The one-dimensional MWCNT transport was described by theHYDRUS-1D code (�Sim�unek et al., 2016) using the advection-

umn experiments.

Meff [%] M*eff [%] Msoil [%] M*soil [%] Mtotal [%] M*total [%]

5 45.8 47.9 48.8 44.2 94.6 92.151 13.8 11.7 77.2 85.9 91.0 97.65 4 2.9 90.3 90.5 94.3 93.45 45.8 47.9 48.8 44.2 94.6 92.15 4 3.6 91.8 93.2 95.8 96.8

C [%] MD H2O [%] ISE [mM] ME [%] MF H2O [%] Msoil [%] Mtotal [%]

45.1 91.96 5.3 100, Kþ 0.4 21.7 50.6 90.502 11 100, Kþ 0.6 11.6 64.8 88.4

strength reduction and cation exchange; Co, MWCNT input concentration; q, DarcyMeff, Msoil, and Mtotal are mass percentages recovered from effluent, soil, and total,release experiments.uent, soil, and total, respectively. The data from the replicate experiments were not

M. Zhang et al. / Water Research 109 (2017) 358e366 361

dispersion equation with a kinetic retention site (Bradford et al.,2003; Kasel et al., 2013a, 2013b):

qvCvt

þ rvSvt

¼ qDv2Cvx2

� qvCvx

(1)

where q is volumetric water content [L3L�3], r is the bulk density ofthe soil [ML�3; where L and M denote units of length and mass,respectively], C is MWCNT concentration in the effluent [NL�3;where N denotes number], t is the time [T; where T denotes units oftime], x is the spatial coordinate [L], D is the hydrodynamicdispersion coefficient [L2T�1], q is the Darcy velocity [LT�1], and S[NM�1] is the solid phase concentrations of MWCNTs. The solidphase mass balance is given as:

rvSvt

¼ qjkswC � rkrsS (2)

where ksw [T�1] is the first-order retention rate coefficients, krs[T�1] is the first-order release rate coefficients, and j [�] aredimensionless functions that account for time- and depth-dependent retention. The parameter j is equal to:

j ¼�1� S

Smax

��d50 þ xd50

��b

(3)

where Smax [NM�1] is the maximum solid phase concentration, b[�] is a fitting parameter which controls the shape of the retentionprofile. The value of b ¼ 0.765 was selected based on reported re-sults for MWCNTs (Kasel et al., 2013a, 2013b).

HYDRUS-1D includes a provision to determine model parame-ters by inverse optimization to experimental data. Values of q andthe dispersivity (lÞ were determined by fitting to the conservativetracer BTC, whereas retention model parameters (ksw, krs, and Smax)were obtained by optimization to BTCs and RPs for MWCNTs.

3. Results and discussion

3.1. MWCNT suspension stability

The hydrodynamic radius of MWCNTs in KCl and CaCl2 solutionsat an IS ¼ 10 mM was measured 0, 1, and 4 h after suspensionpreparation by DLS. The hydrodynamic radius was always withinthe measurement error, and did not show a systematic trend withthe cation type or time. The MWCNT suspensions were thereforeconsidered to be stable during the injection phase (approx. 18 min)of all transport and release experiments discussed below.

3.2. Zeta potential

Fig. 1 presents plots of measured zeta potentials (Fig. 1a) andelectrophoretic mobility (Fig. 1b) values for MWCNTs and soil as afunction of IS and cation valence (Kþ and Ca2þ). Both MWCNTs andsoil exhibited a net negative charge for the considered solutionchemistry conditions. This result indicates that net electrostaticinteractions between the MWCNTs and soil were repulsive. Thezeta potential was generally similar in magnitude for MWCNTs andsoil under the same solution chemistry condition. Furthermore,increasing the solution IS and cation valence tended to decrease themagnitude of both zeta potentials. These trends are consistent withpublished literature (Elimelech et al., 2013; Liang et al., 2013;McNew and LeBoeuf, 2016).

3.3. Transport and retention of MWCNTs

Fig. 2 presents observed and simulated BTCs and RPs for func-tionalized MWCNTs when the IS ¼ 1, 4, and 10 mM KCl. The BTCs(Fig. 2a) are plotted as the normalized effluent concentration (C/Co;where Co is the influent suspension concentration) versus porevolumes, whereas the RPs are plotted as normalized solid phaseconcentration (S/Co) as a function of distance from the column inlet.The experimental conditions, hydraulic parameters (q and l), andmass balance information for these experiments are presented inTable 1. Simulations provided an excellent description of BTCs andRPs with a Pearson's correlation coefficient (R2) > 0.95. Table 2provides a summary of fitted retention model parameters.

The total mass balance (Mtotal) of column experiments in Fig. 2was >91% (Table 1). The mass percentage recovered from theeffluent (Meff) strongly decreased from 45.8% to 4.0% as the ISincreased from 1 to 10 mM (KCl), resulting in a corresponding in-crease in the solid phase mass percentage (Msoil) from 48.8% to90.3%. Fitted values of ksw and Smax=Co also increased with IS,whereas krs decreased with IS. These trends are attributable to anincrease in the adhesive force for MWCNTs on soil with IS due tocompression of the double layer thickness and a decrease in themagnitude of the zeta potential (Fig. 1), and is consistent with otherstudies (Jaisi et al., 2008; Tian et al., 2012; Yang et al., 2013). Itshould also be mentioned that the influence of nanoscale hetero-geneities on colloid retention increases with an increase in IS(Bradford and Torkzaban, 2013).

BTCs for MWCNTs (Fig. 2a) exhibited increasing breakthroughconcentrations over time due to blocking; e.g., filling of a limitednumber of retention sites with continued MWCNT injection. Thebreakthrough time for MWCNTs was also delayed in comparisonwith the tracer (data not shown). This occurs because ksw wassufficiently high to produce complete retention until availableretention sites fill enough to induce breakthrough (e.g., Leij et al.,2015). This delay in breakthrough increases with IS because alarger value of Smax=Co takes longer to fill. Other studies that haveinvestigated the effect of IS on CNTs transport in clean sands orglass beads have not observed this result (Jaisi et al., 2008; Tianet al., 2012; Yang et al., 2013), but it has been previous demon-strated for other nanoparticles (e.g.,Sasidharan et al., 2014). Thisapparent discrepancy is due to the strong dependency of blockingon ksw, Smax, Co, and the input pulse duration (Leij et al., 2015). Inparticular, delay in the breakthrough will not occur if ksw; Co, andthe input pulse duration is too low, or if Smax is too high.

The RPs for MWCNTs (Fig. 2b) exhibited a hyper-exponentialshape that was well described using the model with a depth-dependent retention function (Eq. (3)). This indicates that agreater retention rate occurred near the column inlet then theoutlet. Hyper-exponential RPs have previously been observed forMWCNTs (Kasel et al., 2013a, Kasel et al. 2013b). Liang et al. (2013)also observed hyper-exponential RPs for spherical silver nano-particles (diameter of 15 nm) in this same soil. A variety of potentialexplanations for hyper-exponential RPs have been identified in theliterature, including: straining (Bradford et al., 2002, 2003;Bradford and Bettahar, 2006), colloid aggregation (Chen andElimelech, 2006, 2007), chemical heterogeneity on the soil andcolloid (Tong and Johnson, 2007; Tufenkji and Elimelech, 2005),and system hydrodynamics (Bradford et al., 2009; Li et al., 2005). Itis difficult to ascertain the potential contribution of each factor tothe observed RP. However, MWCNTs retention still resulted in ahyper-exponential shape even under low IS conditions (1 mM KCl)that should minimize the contribution of colloid aggregation andchemical heterogeneity. Results for theoretical calculations thatconsider forces and torques that act on colloids near heterogeneoussurfaces, and comparison of retention in batch and column studies

Fig. 1. Zeta potentials (a) and electrophoretic mobility (b) values of MWCNTs (pH z 5.4) and soil (pH z 6) as a function of the solution (KCl and CaCl2) ionic strength.

Fig. 2. Effect of ionic strength on the transport and retention of MWCNTs in soil: observed and fitted breakthrough curves (a) and retention profiles (b) of MWCNTs under 1, 4, and10 mM KCl, respectively.

Table 2Fitted model parameters.

IS [mM] b Smax/Co [cm3g�1] S.E. Smax/Co ksw [min�1] S.E. ksw krs [min�1] S.E. krs R2

Fig. 2 1, Kþ 0.765 5.45 1.51 10.30 5.89E-01 6.80E-03 5.00E-04 0.9754, Kþ 0.765 5.98 NF 28.22 6.88E-01 5.70E-03 4.59E-04 0.95810, Kþ 0.765 6.10 4.65E-01 49.69 1.17Eþ00 4.56E-03 1.30E-03 0.991

Fig. 3 1, Kþ 0.765 5.45 1.51 10.30 5.89E-01 6.80E-03 5.00E-04 0.9751, Ca2þ 0.765 9.33 NF 44.54 1.39Eþ00 3.31E-03 1.34E-03 0.971

Fig. 1 ionic strength effect; Fig. 2: cation type effect; R2 reflects the correlation of observed and fitted data; NF - denotes not fitted. ksw, the first-order retention rate coefficient;krs, the first-order release rate coefficient; Smax/Co, normalized maximum solid phase concentration of deposited MWCNTs; S.E. e standard error.

M. Zhang et al. / Water Research 109 (2017) 358e366362

indicate that retention of MWCNTs is controlled by surface strain-ing near macroscopic roughness locations and grain-grain contactsunder low IS conditions (Bradford and Torkzaban, 2015; Zhanget al., 2016). Similarly, other studies have concluded that strainingplayed a dominant role in the retention of MWCNTs (Jaisi et al.,2008; Kasel et al., 2013a, 2013b; Wang et al., 2012).

The effect of cation type on the transport and retention ofMWCNTs in soil is presented in Fig. 3. In this case, MWCNTs weredeposited and eluted at the same IS ¼ 1 mM in the presence ofmonovalent Kþ or divalent Ca2þ cations. Fig. 3 presents observedand simulated BTCs and RPs. Table 2 summarizes the fitted reten-tion model parameters that provided an excellent description ofthis data (R2 > 0.97). Retention of MWCNTs (Table 1) and fittedvalues of ksw and Smax=Co (Table 2) were much higher in thepresence of Ca2þ than Kþ at the same IS. In comparison to Kþ, Ca2þ

produces a stronger adhesive force due to the smaller magnitude ofthe zeta potential (Fig. 1), localized neutralization and/or reversal ofsurface charge (Grosberg et al., 2002), and cation bridging

(Torkzaban et al., 2012) that all enhances retention of MWCNTs.Fitted values of Smax=Co clearly indicate that many more retentionsites were available in the presence of Ca2þ than Kþ (an increase of171%). This large increase in the value of Smax=Co in the presenceCa2þ also influenced the blocking behavior in the MWCNT BTC. Inparticular, a greater delay in the breakthrough time and a slowerrate of increase in the BTC with time was observed (Fig. 3b). Similarto RPs in the presence of Kþ (Fig. 2b), the RP in the presence of Ca2þ

also exhibited a hyper-exponential shape. This observation sug-gests that similar retention mechanisms were operative in thepresence of both monovalent and divalent cations.

3.4. Release of MWCNTs

Very little release of MWCNTs was observed under steady-statesolution chemistry conditions (Figs. 2 and 3, and Table 2).Conversely, significant amounts of colloid and nanoparticle releasehave been reported when transient solution chemistry reduce the

Fig. 3. Effect of cation type on the transport and retention of MWCNTs in soil: observed and fitted breakthrough curves (a) and retention profiles (b) of MWCNTs under 1 mM KCland CaCl2, respectively.

M. Zhang et al. / Water Research 109 (2017) 358e366 363

adhesive force (e.g., Liang et al., 2013). Additional experiments weretherefore conducted to better understand the effect of IS reductionand cation exchange on the release of MWCNTs in soil. Step A ofFig. 4a presents BTCs when MWCNTs were retained and eluted inthe presence of IS¼ 10mMKCl (experiment I), and CaCl2 solution atIS ¼ 1 mM (experiment III) and IS ¼ 10 mM (experiment III).Transport of MWCNTs was limited in Step A for all of these condi-tions. Indeed, effluent mass balance during step A (MA) onlyequaled 4.6, 4.2, and 0.03% for experiments I, II, and III, respectively.

Step B in Fig. 4a presents the subsequent release curves whenretained MWCNTS were eluted by Milli-Q water. Release was muchmore pronounced when the MWCNTs were retained (step A) in thepresence of monovalent Kþ (42.2%) than divalent Ca2þ (<7.7%)cation. This result indicates that the adhesive interaction wasstronger and less reversible in the presence of Ca2þ than Kþ.Furthermore, release of MWCNTs was less pronounced when theywere retained in IS ¼ 10 mM CaCl2 (0.3%) than IS ¼ 1 mM CaCl2(7.7%) solution, respectively. This implies that a stronger adhesiveforce continued to act on the MWCNTs when they were initiallyretained in the presence of a higher concentration of Ca2þ. A po-tential explanation for these observations is due to localized chargeneutralization/reversal or cation bridging by Ca2þ which createsstrong interactions with MWCNTs (Jaisi et al., 2008; Liang et al.,2013).

Experiments II and III were continued to further study therelease of MWCNTs that were retained in the presence of Ca2þ

using the following elution sequence: KCl at the same IS as in step A(step C); Milli-Q water (step D); 100 mM KCl (step E); and Milli-Qwater (step F). The effluent concentrations of MWCNTs duringeach of these steps are shown in Fig. 4a, with corresponding massbalance information given in Table 1 (denoted as MC-MF). Effluentconcentrations of Kþ and Ca2þ, and Al and Fe in a replicate releaseexperiment for MWCNT in IS ¼ 1 mM CaCl2 solution during step Aare shown in Fig. 4b and c, respectively. The RPs for MWCNTsfollowing completion of transient release experiments I, II, and IIIare shown in Fig. 4d.

MWCNT release always occurred when Milli-Q water wasinjected into the column during steps B, D, and F, but graduallyceased when KCl was injected during steps C and E (Fig. 4a). Theamount of MWCNTs that was released with Milli-Q water stronglydepended on the concentrations of injected Ca2þ during step A andKþ during steps C and E. Release of MWCNTs that occurred with ISreduction during steps D and F were also influenced by the amountof cation exchange that occurred during steps C and E. Specifically,greater amounts of Ca2þ ions were exchanged back into theaqueous phase when a higher concentration of Kþ was injected(Fig. 4b). This exchange process reduced the strength of the

adhesive interaction, such that greater amounts of MWCNTs werereleased with a subsequent reduction in solution IS. Consequently,MD was less when the IS equaled 1 mM (MD ¼ 5.3%) than 10 mM(MD ¼ 11%) KCl during step C. The value ofMF was influenced by: (i)the initial concentration of Ca2þ during deposition (step A); (ii) theamount of cation exchange (steps C and E); and (iii) the previousamounts of MWCNT that were released (steps B and D). In this case,the value of MF equaled 21.1 and 11.6% when the IS equaled 1 and10 mM CaCl2, respectively, during step A. A similar trend wasobserved for the total recoveredmass of MWCNTs during steps A-F,and this indicates that increased strength of the adhesive force at ahigher Ca2þ concentration (step A) was the dominantconsideration.

Fig. 4d presents the RPs for MWCNTs following completion ofthe release experiments. Similar to Figs. 2 and 3, RPs were stillhyper-exponential in shape. However, close inspection of Figs. 2b,3b and 4d reveals that retained MWCNTs in release experimentswere shifted from the top into deeper layers. This observationsupports the potential for continued slow remobilization ofMWCNTs in the subsurface due to the effects of IS reduction andcation exchange.

Bradford and Kim (2010) examined the release behavior of insitu kaolinite clay from sand due to cation exchange and IS reduc-tion, and observed similar trends as seen for the MWCNTs in Fig. 4.This observation suggests the potential for colloid-facilitatedtransport of MWCNTs in this study. Indeed, effluent samplesexhibited differences in sample turbidity due to release of soilcolloids, and DLS measurements indicated the presence of soilcolloids with hydrodynamic diameter ranged from 150 to 308 nm.To better understand the potential association of released MWCNTsand soil colloids, Fig. 4c presents plots of effluent concentrations ofAl, Fe, and MWCNTs during the release experiment. Note thatrelease behavior of Fe, Al, and MWCNTs closely follow each otherwith IS reduction and cation exchange. Liang et al. (2013) alsoobserved this same phenomenon for silver nanoparticles with Aland Fe. All of these observations strongly support the potential forcolloid-facilitated transport of MWCNTs during the release exper-iments. However, it should be mentioned that BTCs for the con-servative tracer before and after release experiments were nearlyidentical (data not shown). This observation suggests that changesin the soil hydraulic properties due to soil colloid release wereinconsequential during the release experiment.

3.5. Soil size fractionation

Soil fractionation was conducted following completion ofMWCNT transport experiments at IS ¼ 1 mM in KCl and CaCl2

Fig. 4. (a) Breakthrough and release behavior, and retention profiles (d) of MWCNTs insoil. Deposition (step A) occurred at an IS ¼ 10 mM using KCl for experiment I, and anIS ¼ 1 and 10 mM using CaCl2 for experiment II and III, respectively, whereas releasewas initiated by ionic strength reduction (steps B, D, and F, Milli-Q water) and cation

M. Zhang et al. / Water Research 109 (2017) 358e366364

solutions to further investigate the association between MWCNTsand soil colloids in the presence of different cation types. Fig. 5ashows a plot of the soil mass percentage of the different size frac-tions. Results indicate that the soil was composed of 2.38% WDCs(1.51% in the range of 0.45e2 mm and 0.87% less than 0.45 mm),11.5% silt (2e20 mm), and 86.2% sand (20e2000 mm). This sizefractionation is comparable to that obtained by (Kasel et al., 2013b),with small variations likely due to the use of different soil frac-tionation methods.

Fig. 5b presents a plot of the recovered mass percentages ofMWCNTs in the various soil size fractions when the IS ¼ 1 mM KCland CaCl2. The total mass recovery was calculated to be more than85%. Mass percentages of MWCNTs in the sand and silt fractionswere only 9.5 and 17.2%, respectively, even though they accountedfor 97.7% of the soil mass. Mass percentage of MWCNTs in the0.45e2 mm WDC fraction that accounted for 1.51% of the soil masswas 23.6%. The mass percentage of MWCNTs in the <0.45 mm WDCfraction and the so-called electrolyte phase (Jiang et al., 2014) was49.7%. This fraction includedMWCNTs associatedwith very fine soilcolloids and those released during the fractionation procedure.

Fig. 5c presents the retained concentration (S/Co) of MWCNTs indifferent soil fractions in the presence of KCl or CaCl2 when theIS ¼ 1 mM. The retained concentration of MWCNTs in each soilfraction was more pronounced in the presence of Ca2þ than Kþ,especially in the WDC fraction. This observation can be explainedby charge reversal/neutralization (Grosberg et al., 2002) and/orbridging complexation between soil grains and functionalizedMWCNTs in the presence of Ca2þ (Torkzaban et al., 2012). Similar toexperimental observations, bridging complexation is expected tobe more pronounced with the WDC fraction (Torkzaban et al.,2012).

The above information indicates that the WDC fractions wereenriched in MWCNTs in comparison with the sand and silt frac-tions, and that the association between WDCs and MWCNTs wasrelated to the cation valance. This strong association betweenWDCs and MWCNTs provides further evidence for colloid-facilitated transport of MWCNTs during the release experiments.

4. Conclusions

Findings in this study provide important insight on the roles ofsolution IS, cation valance, and soil colloids on the transport,retention, and remobilization of MWCNTs in soils. Experimentaland modeling results indicate that the mobility of MWCNTs washighly sensitivity to the IS and cation valence, with greater amountsof retention occurring at high IS and with divalent cations. BTCs forMWCNTs exhibited a delay in breakthrough that increased withsolution IS and cation valence due to blocking. RPs for MWCNTsshowed a hyper-exponential shape likely due to surface strainingprocesses. Significant amounts of MWCNTs could be released byperturbations in solution chemistry that reduced the adhesive force(e.g., IS reduction and cation exchange). This release behavior wasdemonstrated to depend on the concentration of Ca2þ duringMWCNT deposition and the release of soil colloids with a high

exchange (steps C and E) as summarized in Table 1. (b) Effluent concentrations of K andCa during steps A�F in experiment II. (c) Release of MWCNTs and naturally occurringminerals due to ionic strength reduction (steps B, D, and F, Milli-Q water) and cationexchange (steps C and E) in soil in experiment II. For experiment III, the range of stepA-F is not shown as the blue dotted line due to the different experimental conditionscompared with experiment II. The injection procedure in experiment II is step A(0e7.61 PV), step B (7.61e15.23 PV), step C (15.23e17.51 PV), step D (17.51e22.85 PV),step E (22.85e25.13 PV), and step F (25.13e30.47 PV).(For interpretation of the refer-ences to colour in this figure legend, the reader is referred to the web version of thisarticle.)

Fig. 5. (a) Mass percentage of each soil sized fraction: sand, silt, 0.45e2 mm WDCs and<0.45 mmWDCs; (b) Mass percentage of MWCNTs in sand, silt and 0.45e2 mmWDCs ofsoil. The <0.45 mm WDCs also contained the MWCNTs in the so-called electrolytephase (1 mM KCl and CaCl2); (c) Retained MWCNTs in sand, silt, 0.45e2 mm WDCs and<0.45 mm WDCs including electrolyte phase under different cation type.

M. Zhang et al. / Water Research 109 (2017) 358e366 365

association for MWCNTs. These results suggest the potential for soilcolloids to facilitated the transport of MWCNTs in the subsurfaceenvironment, and thereby pose a potential risk of groundwaterpollution, especially during rainfall or irrigation events that alterthe solution chemistry.

Acknowledgments

The first author thanks the China Scholarship Council (CSC) forfinancial support. The authors would like to acknowledge StephanK€oppchen for bromide measurement, and Andrea Kubika andMartina Krause for TOC measurements. Thanks to Herbert Philippand ClaudiaWalraf for their technical assistance, and Xiaoqian Jiangand Wulf Amelung for the helpful discussion related to soilfractionation.

References

Badawy, A.M.E., Luxton, T.P., Silva, R.G., Scheckel, K.G., Suidan, M.T., Tolaymat, T.M.,2010. Impact of environmental conditions (pH, ionic strength, and electrolytetype) on the surface charge and aggregation of silver nanoparticles suspensions.Environ. Sci. Technol. 44 (4), 1260e1266.

Bradford, S.A., Bettahar, M., 2006. Concentration dependent transport of colloids insaturated porous media. J. Contam. Hydrol. 82 (1), 99e117.

Bradford, S.A., Kim, H., 2010. Implications of cation exchange on clay release andcolloid-facilitated transport in porous media. J. Environ. Qual. 39 (6), 2040.

Bradford, S.A., Kim, H.N., Haznedaroglu, B.Z., Torkzaban, S., Walker, S.L., 2009.Coupled factors influencing concentration-dependent colloid transport andretention in saturated porous media. Environ. Sci. Technol. 43 (18), 6996e7002.

Bradford, S.A., �Sim�unek, J., Bettahar, M., van Genuchten, M.T., Yates, S.R., 2003.Modeling colloid attachment, straining, and exclusion in saturated porousmedia. Environ. Sci. Technol. 37 (10), 2242e2250.

Bradford, S.A., Torkzaban, S., 2008. Colloid transport and retention in unsaturatedporous media: a review of interface-, collector-, and pore-scale processes andmodels. Vadose Zone J. 7 (2), 667e681.

Bradford, S.A., Torkzaban, S., 2013. Colloid interaction energies for physically andchemically heterogeneous porous media. Langmuir 29 (11), 3668e3676.

Bradford, S.A., Torkzaban, S., 2015. Determining parameters and mechanisms ofcolloid retention and release in porous media. Langmuir 31 (44), 12096e12105.

Bradford, S.A., Yates, S.R., Bettahar, M., �Sim�unek, J., 2002. Physical factors affectingthe transport and fate of colloids in saturated porous media. Water Resour. Res.38 (12), 63-1e63-12.

Camilli, L., Pisani, C., Gautron, E., Scarselli, M., Castrucci, P., D'Orazio, F.,Passacantando, M., Moscone, D., De Crescenzi, M., 2014. A three-dimensionalcarbon nanotube network for water treatment. Nanotechnology 25 (6), 065701.

Chen, K.L., Elimelech, M., 2006. Aggregation and deposition kinetics of fullerene(C60. Nanoparticles. Langmuir 22 (26), 10994e11001.

Chen, K.L., Elimelech, M., 2007. Influence of humic acid on the aggregation kineticsof fullerene (C60) nanoparticles in monovalent and divalent electrolyte solu-tions. J. Colloid Interface Sci. 309 (1), 126e134.

Cornelis, G., Hund-Rinke, K., Kuhlbusch, T., Van den Brink, N., Nickel, C., 2014. Fateand bioavailability of engineered nanoparticles in soils: a review. Crit. Rev.Environ. Sci. Technol. 44 (24), 2720e2764.

de Jonge, L.W., Moldrup, P., Rubæk, G.H., Schelde, K., Djurhuus, J., 2004. Particleleaching and particle-facilitated transport of phosphorus at field scale. VadoseZone J. 3 (2), 462e470.

Elimelech, M., Gregory, J., Jia, X., 2013. Particle Deposition and Aggregation: Mea-surement, Modelling and Simulation. Butterworth-Heinemann.

Gao, B., Saiers, J.E., Ryan, J., 2006. Pore-scale mechanisms of colloid deposition andmobilization during steady and transient flow through unsaturated granularmedia. Water Resour. Res. 42 (1).

Gohardani, O., Elola, M.C., Elizetxea, C., 2014. Potential and prospective imple-mentation of carbon nanotubes on next generation aircraft and space vehicles:a review of current and expected applications in aerospace sciences. Prog.Aerosp. Sci. 70, 42e68.

Grolimund, D., Borkovec, M., 2005. Colloid-facilitated transport of strongly sorbingcontaminants in natural porous media: mathematical modeling and laboratorycolumn experiments. Environ. Sci. Technol. 39 (17), 6378e6386.

Grolimund, D., Borkovec, M., 2006. Release of colloidal particles in natural porousmedia by monovalent and divalent cations. J. Contam. Hydrol. 87 (3), 155e175.

Grosberg, A.Y., Nguyen, T., Shklovskii, B., 2002. Colloquium: the physics of chargeinversion in chemical and biological systems. Rev. Mod. Phys. 74 (2), 329.

Guldi, D.M., Rahman, G., Prato, M., Jux, N., Qin, S., Ford, W., 2005. Single-wall carbonnanotubes as integrative building blocks for solar-energy conversion. Angew.Chem. 117 (13), 2051e2054.

Hassell€ov, M., Readman, J.W., Ranville, J.F., Tiede, K., 2008. Nanoparticle analysis andcharacterization methodologies in environmental risk assessment of engi-neered nanoparticles. Ecotoxicology 17 (5), 344e361.

Iijima, S., 1991. Helical microtubules of graphitic carbon. Nature 354 (6348), 56e58.

M. Zhang et al. / Water Research 109 (2017) 358e366366

Israelachvili, J.N., 2011. Intermolecular and Surface Forces, revised third ed. Aca-demic press.

Jaisi, D.P., Saleh, N.B., Blake, R.E., Elimelech, M., 2008. Transport of single-walledcarbon nanotubes in porous media: filtration mechanisms and reversibility.Environ. Sci. Technol. 42 (22), 8317e8323.

Janas, D., Herman, A.P., Boncel, S., Koziol, K.K., 2014. Iodine monochloride as apowerful enhancer of electrical conductivity of carbon nanotube wires. Carbon73, 225e233.

Jiang, C.-L., S�equaris, J.-M., Vereecken, H., Klumpp, E., 2012. Effects of inorganic andorganic anions on the stability of illite and quartz soil colloids in Na-, Ca-andmixed NaeCa systems. Colloids Surf. A Physicochem. Eng. Asp. 415, 134e141.

Jiang, C., S�equaris, J.-M., Wacha, A., B�ota, A., Vereecken, H., Klumpp, E., 2014. Effectof metal oxide on surface area and pore size of water-dispersible colloids fromthree German silt loam topsoils. Geoderma 235, 260e270.

Jones, A.D.K., Bekkedahl, T., 1997. Storage of hydrogen in single-walled carbonnanotubes. Nature 386, 377.

Kasel, D., Bradford, S.A., �Sim�unek, J., Heggen, M., Vereecken, H., Klumpp, E., 2013a.Transport and retention of multi-walled carbon nanotubes in saturated porousmedia: effects of input concentration and grain size. Water Res. 47 (2),933e944.

Kasel, D., Bradford, S.A., �Sim�unek, J., Pütz, T., Vereecken, H., Klumpp, E., 2013b.Limited transport of functionalized multi-walled carbon nanotubes in twonatural soils. Environ. Pollut. 180 (0), 152e158.

Khilar, K.C., Fogler, H.S., 1998. Migrations of Fines in Porous Media. Springer Science& Business Media.

Leij, F.J., Bradford, S.A., Wang, Y., Sciortino, A., 2015. Langmuirian blocking of irre-versible colloid retention: analytical solution, moments, and setback distance.J. Environ. Qual. 44 (5), 1473e1482.

Li, C., Sch€affer, A., S�equaris, J.-M., L�aszl�o, K., T�oth, A., Tomb�acz, E., Vereecken, H., Ji, R.,Klumpp, E., 2012. Surface-associated metal catalyst enhances the sorption ofperfluorooctanoic acid to multi-walled carbon nanotubes. J. Colloid InterfaceSci. 377 (1), 342e346.

Li, X., Zhang, P., Lin, C., Johnson, W.P., 2005. Role of hydrodynamic drag on micro-sphere deposition and Re-Entrainment in porous media under unfavorableconditions. Environ. Sci. Technol. 39 (11), 4012e4020.

Liang, Y., Bradford, S.A., �Sim�unek, J., Heggen, M., Vereecken, H., Klumpp, E., 2013.Retention and remobilization of stabilized silver nanoparticles in an undis-turbed loamy sand soil. Environ. Sci. Technol. 47 (21), 12229e12237.

Lin, H., Zhu, H., Guo, H., Yu, L., 2008. Microwave-absorbing properties of Co-Filledcarbon nanotubes. Mater. Res. Bull. 43 (10), 2697e2702.

Mauter, M.S., Elimelech, M., 2008. Environmental applications of carbon-basednanomaterials. Environ. Sci. Technol. 42 (16), 5843e5859.

McNew, C.P., LeBoeuf, E.J., 2016. Nc60 deposition kinetics: the complex contributionof humic acid, ion concentration, and valence. J. Colloid Interface Sci. 473,132e140.

Pan, B., Xing, B., 2012. Applications and implications of manufactured nanoparticlesin soils: a review. Eur. J. Soil Sci. 63 (4), 437e456.

Pauluhn, J., 2010. Multi-walled carbon nanotubes (Baytubes®): approach for deri-vation of occupational exposure limit. Regul. Toxicol. Pharmacol. 57 (1), 78e89.

Pecora, R., 2000. Dynamic light scattering measurement of nanometer particles inliquids. J. Nanoparticle Res. 2 (2), 123e131.

Sasidharan, S., Torkzaban, S., Bradford, S.A., Dillon, P.J., Cook, P.G., 2014. Coupledeffects of hydrodynamic and solution chemistry on long-term nanoparticletransport and deposition in saturated porous media. Colloids Surf. A Phys-icochem. Eng. Asp. 457, 169e179.

S�equaris, J.-M., Lewandowski, H., 2003. Physicochemical characterization of po-tential colloids from agricultural topsoils. Colloids Surf. A Physicochem. Eng.Asp. 217 (1), 93e99.

�Sim�unek, J., Genuchten, M.T.V., �Sejna, M., 2016. Recent developments and appli-cations of the hydrus computer software packages. Vadose Zone J. 15 (7), 25.http://dx.doi.org/10.2136/vzj2016.04.0033.

Tian, Y., Gao, B., Wu, L., Munoz-Carpena, R., Huang, Q., 2012. Effect of solutionchemistry on multi-walled carbon nanotube deposition and mobilization inclean porous media. J. Hazard. Mater. 231e232, 79e87.

Tong, M., Johnson, W.P., 2007. Colloid population heterogeneity drives hyper-exponential deviation from classic filtration theory. Environ. Sci. Technol. 41 (2),493e499.

Torkzaban, S., Bradford, S.A., Wan, J., Tokunaga, T., Masoudih, A., 2013. Release ofquantum dot nanoparticles in porous media: role of cation exchange and agingtime. Environ. Sci. Technol. 47 (20), 11528e11536.

Torkzaban, S., Wan, J., Tokunaga, T.K., Bradford, S.A., 2012. Impacts of bridgingcomplexation on the transport of surface-modified nanoparticles in saturatedsand. J. Contam. Hydrol. 136, 86e95.

Tufenkji, N., Elimelech, M., 2005. Spatial distributions of cryptosporidium oocysts inporous media: evidence for dual mode deposition. Environ. Sci. Technol. 39(10), 3620e3629.

Wang, D., Jaisi, D.P., Yan, J., Jin, Y., Zhou, D., 2015. Transport and retention ofpolyvinylpyrrolidone-coated silver nanoparticles in natural soils. Vadose Zone J.14 (7).

Wang, Y., Bradford, S.A., �Sim�unek, J., 2014. Physicochemical factors influencing thepreferential transport of in soils. Vadose Zone J. 13 (1).

Wang, Y., Kim, J.H., Baek, J.B., Miller, G.W., Pennell, K.D., 2012. Transport behavior offunctionalized multi-wall carbon nanotubes in water-saturated quartz sand as afunction of tube length. Water Res. 46 (14), 4521e4531.

Yan, J., Lazouskaya, V., Jin, Y., 2016. Soil colloid release affected by dissolved organicmatter and redox conditions. Vadose Zone J. 15 (3).

Yang, J., Bitter, J.L., Smith, B.A., Fairbrother, D.H., Ball, W.P., 2013. Transport ofoxidized multi-walled carbon nanotubes through silica based porous media:influences of aquatic chemistry, surface chemistry, and natural organic matter.Environ. Sci. Technol. 47 (24), 14034e14043.

Zhang, S., Shao, T., Kose, H.S., Karanfil, T., 2010. Adsorption of aromatic compoundsby carbonaceous adsorbents: a comparative study on granular activated carbon,activated carbon fiber, and carbon nanotubes. Environ. Sci. Technol. 44 (16),6377e6383.

Zhang, M., Bradford, S.A., Simunek, J., Vereecken, H., Klumpp, E., 2016. Do goethitesurfaces really control the transport and retention of multi-walled carbonnanotubes in chemically heterogeneous porous media? Environ. Sci. Technol.http://dx.doi.org/10.1021/acs.est.6b03285.

Zhu, Y., Ma, L.Q., Dong, X., Harris, W.G., Bonzongo, J., Han, F., 2014. Ionic strengthreduction and flow interruption enhanced colloid-facilitated Hg transport incontaminated soils. J. Hazard. Mater. 264, 286e292.