Nanoparticle Aggregation: Challenges to Understanding Transport and Reactivity in the Environment

Home , DLVO theory, Double layer forces, Particle aggregation

SPECIALTECHNICAL SUBMISSIONS REPORTS

Ernest M. Hotze, Tanapon Phenrat, and Gregory V. Lowry* Carnegie Mellon University

s of the submission date of this review, approximately Unique forms of manufactured nanomaterials, nanoparticles, 4700 articles on nanoparticle (NP)-related topics have been and their suspensions are rapidly being created by manipulating A properties such as shape, size, structure, and chemical published in the calendar year 2009 alone (Th ompson Reuters, composition and through incorporation of surface coatings. 2009). Many of these papers discuss novel particles, and applica- Although these properties make nanomaterial development tions of nanomaterials along with their use in consumer products interesting for new applications, they also challenge the ability are expanding rapidly (Project on Emerging Nanotechnologies, of colloid science to understand nanoparticle aggregation in 2009). For example, nanomaterials are used for environmental the environment and the subsequent eff ects on nanomaterial transport and reactivity. Th is review briefl y covers aggregation clean-up and biomedical applications (Alivisatos, 2001; Liu et al., theory focusing on Derjaguin-Landau-Verwey-Overbeak 2005). Th is expansion in applications, and the incorporation of (DLVO)-based models most commonly used to describe “nano” ingredients into products that may be released during the the thermodynamic interactions between two particles in a life cycle of those products, will increase the occurrence of manu- suspension. A discussion of the challenges to DLVO posed by factured nanomaterials in the environment. For example, the the properties of nanomaterials follows, along with examples from the literature. Examples from the literature highlighting leaching of silver NPs used as antimicrobial agents in cloth was the importance of aggregation eff ects on transport and reactivity recently reported (Benn and Westerhoff , 2008). Unintentionally and risk of nanoparticles in the environment are discussed. or intentionally, manufactured NPs might come into contact with biological receptors. For this reason, the risk of nanomaterials to humans and to the environment garners attention from the scien- tifi c community, governmental agencies, and public stakeholders. Nanoparticless in engineered or environmental systems could be thought of as dispersions of primary particles. However, primary NPs tend to aggregate into clusters up to several microns in size (Brant et al., 2007; Chen and Elimelech, 2007; Jiang et al., 2009; Limbach et al., 2005; Phenrat et al., 2008). Th erefore, NP aggregation plays an important role in determining reactivity, toxicity, fate, transport, and risk in the environment. Th e critical role of aggregation in environmental implications had occasion- ally been ignored (Baveye and Laba, 2008), but many recent studies have begun applying colloid science principles, based around Derjaguin-Landau-Verwey-Overbeak (DLVO) theory, to understand NP aggregation under various conditions. Manufactured NPs challenge the limits of colloid science, however, due to their small size, variable shape, structure, composition, and potential presence of adsorbed or grafted organic macromolecules. Laboratory aggregation studies, which are covered in detail in this review, have demonstrated that these challenges, acting simulta- neously, will ultimately aff ect how NPs behave in the environ- Copyright © 2010 by the American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. All rights ment. Additionally, particles released to the environment undergo reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including pho- tocopying, recording, or any information storage and retrieval system, Center for Environmental Implications of NanoTechnology (CEINT) and Deps. of Civil without permission in writing from the publisher. & Environmental Engineering, Carnegie Mellon Univ., Pittsburgh, PA 15213-3890. Assigned to Associate Editor Dionysios Dionysiou. J. Environ. Qual. 39:1909–1924 (2010) doi:10.2134/jeq2009.0462

Published online 20 May 2010. Abbreviations: CT, carbon tetrachloride; Df, fractal dimension; DLVO, Derjaguin-Landau- Received 18 Nov. 2009. Verwey-Overbeak; EDL, electrostatic double layer; MWCNT, multi-walled carbon *Corresponding author ([email protected]). nanotubes; NOM, natural organic matter; NP, nanoparticle; NZVI, nano zero valent iron;

© ASA, CSSA, SSSA PZC, point of zero charge; ROS, reactive oxygen species; Velas, elastic-steric potential; VOSM, 5585 Guilford Rd., Madison, WI 53711 USA osmotic potential; XDLVO, extended Derjaguin-Landau-Verwey-Overbeak.

1909 chemical and physical transformations and encounter a multi- (e.g., NP–clay particle attachment). Real environmental sys- tude of solution conditions important for aggregation, includ- tems contain natural particles (e.g., clay particles) in numbers ing solution pH, dissolved ions, naturally occurring organic far greater than the number of manufactured NPs, so heteroag- matter, clays, and biocolloids. gregation is an important fate process for manufactured NPs. Th is review focuses on three questions: (i) Can basic colloid Th e probability that two particles attach is given by a “stick- science predict the behavior of nanomaterials in the environ- ing coeffi cient,” also known as the attachment effi ciency (α) ment? We provide a brief overview of colloid science. Several (i.e., when α = 1, sticking occurs 100% of the time; when α = excellent resources are available for additional information 0.5, sticking occurs 50% of the time). Particles attaching at fi rst on colloid science (Evans and Wennerstrom, 1999; Hiemenz contact (α = 1) form large dendritic aggregates, whereas par- and Rajagopalan, 1997; Buffl e et al., 1998; Elimelech and ticles sticking only after several collisions (α < 1) form denser O’Melia, 1990; O’Melia, 1980). (ii) Where are challenges to and less dendritic aggregates (Buffl e and Leppard, 1995). Th e existing theory posed by manufactured nanomaterials? We concept of fractal dimension is used to describe aggregate struc-

discuss the intrinsic NP properties that raise these challenges. ture. Fractal dimension (Df) can range from 1 (a line) to 3 (a (iii) How will these challenges aff ect studies of fate and trans- sphere). Th ese are limits that come from Euclidean dimension, port, reactivity, and risk to the environment? We describe where 1 is a straight line (e.g., x1), 2 is a fl at surface (e.g., x2), several case studies examining environmental implications of and 3 is a volume (e.g., x3). Fractal dimensions are those that nanomaterial aggregation. lie between Euclidean dimensions. Ideal fractal structures are self-similar throughout. Homoaggregation under controlled Colloid Science and Aggregation laboratory conditions produces aggregates of reasonably pre- Colloid science is defi ned as a branch of chemistry dealing with dictable fractal dimension (Barbot et al., 2010; Chakraborti colloids, heterogeneous systems consisting of a mechanical et al., 2003; Hansen et al., 1999), whereas heteroaggregation mixture of dispersion (colloids) and dispersant. When particles typically forms natural fractals (statistically self-similar over a are dispersed, sizes range between 1 nm and 1 μm in a continu- limited range of length scales), making aggregation state more ous medium. Above this size range, particles begin to sediment diffi cult to predict. Th e physical dimensions of the aggregates out of suspension. Several theories to explain aggregation of formed can aff ect the reactive surface area, reactivity, bioavail- submicrometer spherical particles have been developed and ability, and toxicity. Aggregate structure must therefore be experimentally studied for decades, the most common being determined and considered when interpreting fate, transport, DLVO (Derjaguin and Landau, 1941; Verwey et al., 1948) and toxicity data. (an alternative approach is statistical thermodynamics; see Hermansson [1999] and Koponen [2009]). By defi nition, the Derjaguin-Landau-Verwey-Overbeak and Extended size range of colloidal particles overlaps the size range of manu- Derjaguin-Landau-Verwey-Overbeak factured nanomaterials (i.e., <100 nm). Th erefore, theories in According to the classical DLVO theory of aggregation, the colloid science should be applicable to manufactured nanoma- sum of attractive and repulsive forces determines attachment terials, but, due to the novelty of nanomaterial properties, sev- (Derjaguin and Landau, 1941; Verwey et al., 1948). Th is eral studies have found that this is not always the case (He et theory maintains that only two forces dominate interactions al., 2008; Kozan et al., 2008; Phenrat et al., 2008). However, between particles: van der Waals (vdW) attractive and electro- colloid science is fundamental to understanding and develop- static double layer (EDL) forces. Table 1 lists an expression for ing theories for nanomaterials systems. Th is section, therefore, the vdW attraction experienced by two spheres in suspension provides a very brief review of the types of aggregation and (Stumm and Morgan, 1996). Expressions for vdW commonly DLVO theory. calculate the attraction between two spherical particles; however, due to unique NP shapes and compositions, expanded Types of Aggregation and Aggregate Structure theoretical approaches may be needed (Gatica et al., 2005). Aggregation of colloidal particles occurs when physical pro- An EDL has a charge that refl ects not only the surface that it cesses bring particle surfaces in contact with each other and surrounds but also the solution chemistry of the surrounding short-range thermodynamic interactions allow for particle– media. Ionic strength, which is a measure of the amount of particle attachment to occur. For particles <100 nm in size, ions present, controls the extent of the radius of the diff use Brownian diff usion controls the long-range forces between layer from the surface. Low ionic strength means the EDL ion individual nanoparticles, causing collisions between particles. cloud extends far out from the particle; high ionic strength When contact occurs, it can result in attachment or repul- conditions compress the EDL. Table 1 lists an expression for sion. Short-range thermodynamic interactions that control the EDL experienced by two charged spheres in suspension this can be understood in the context of DLVO theory. Th ere (Stumm and Morgan, 1996). are two types of aggregation relevant to manufactured NPs in Classical DLVO simplifi es thermodynamic surface inter- the environment: homoaggregation and heteroaggregation. actions and predicts the probability of two particles stick- Homoaggregation refers to aggregation of two similar par- ing together by simply summing van der Waals and electric ticles (i.e., NP–NP attachment). Th is is observed in homo- double-layer potentials to determine if forces are net attractive geneous suspensions of particles that are typically studied in (−VT) or net repulsive (+VT). For example, in Fig. 1, van der the laboratory to make useful correlations with DLVO theory. Waals and electric double-layer potentials are plotted as a func- Heteroaggregation refers to aggregation of dissimilar particles tion of separation distance between the particles. Summing

1910 Journal of Environmental Quality • Volume 39 • November–December 2010 Table 1. Derjaguin-Landau-Verwey-Overbeak and extended Derjaguin-Landau-Verwey-Overbeak forces in aggregation. Force X/DLVO† Equations‡§ Origin References van der Waals DLVO ⎛⎞interactions of Stumm and Morgan VAaa2222sas()4 + attraction VDW =−⎜⎟ + +ln electrons in particles (1996) ⎜⎟()+ 22 kT64 kT⎝⎠ s a s ()22as++() as

Electrostatic DLVO +δ 2 2 charged surfaces Delgado et al. VnkTze64πΨ()a ⎡⎤⎛⎞−κ()s −2 δ double layer EDL = ⎢⎥tanh⎜⎟d exp (2007), Stumm and kTκ+() s24 a⎣⎦⎝⎠ kT Morgan (1996)

1 ⎡⎤N 2 22 ⎢⎥∑eznii κ=⎢⎥i ⎢⎥ååkT ⎢⎥rs 0 ⎣⎦

Magnetic XDLVO −πμ 23 aligning electron Phenrat et al. (2007) = 8 oSMa spins attraction VM ⎛⎞s 3 92⎜⎟+ ⎝⎠a

Hydrophobic (Lewis XDLVO V ⎛⎞ss− entropic penalty of Hoek and Agarwal acid-base) AB =ΔG AB exp⎜⎟0 separating hydrogen (2006), Wu et al. s0 λ surface area ⎝⎠ bonds in water (1999)

Osmotic XDLVO ⎧⎫V concentration of Fritz et al. (2002), repulsion 20ds≤⇒⎨⎬OSM = ions between two Ortega-Vinuesa et ⎩⎭kT particles (Fig. 4) al. (1996), Phenrat et al. (2008), ⎪⎪⎧⎫π ⎛⎞⎛⎞2 Romero-Cano ≤< ⇒VaOSM =41 φ2 −χ − s ds2 d⎨⎬p ⎜⎟⎜⎟ d et al. (2001) ⎩⎭⎪⎪kT v1 ⎝⎠⎝⎠22

⎪⎪⎧⎫π ⎛⎞⎛⎞ ⎛⎞ <⇒VaOSM =41 φ22 −χ s −− 1 s sd⎨⎬p ⎜⎟ d⎜⎟ln ⎜⎟ ⎪⎪⎩⎭kT v1 ⎝⎠224⎝⎠ d ⎝⎠ d

Elastic-steric XDLVO ⎧⎫V molecules on (Fritz et al., 2002; repulsion ds≤⇒⎨⎬elas =0 particle surfaces Ortega-Vinuesa et ⎩⎭kT resist loss of entropy al., 1996; Phenrat due to compaction et al., 2008; Romero- ⎧ ⎧⎫⎡⎤2 ⎪⎪⎪Va⎛⎞2π ss⎛⎞3 − sd Cano et al., 2001) ds>⇒⎨⎨⎬elas =⎜⎟ φ d2 ρ ln ⎢⎥⎜⎟ kT Mpp d⎢⎥ d ⎝⎠2 ⎩⎪⎪⎪⎝⎠W ⎩⎭⎣⎦ ⎛⎞3 − sd ⎛⎞s ⎪⎫ −++6ln⎜⎟ 3⎜⎟ 1 ⎬ ⎝⎠2 ⎝⎠d ⎭⎪

Bridging XDLVO NA surface molecules Chen and Elimelech attraction bridge to other (2007), Domingos particles et al. (2009)

† DLVO, Derjaguin-Landau-Verwey-Overbeak; EDLVO, extended Derjaguin-Landau-Verwey-Overbeak. Δ AB ‡ A, Hamaker Constant; a, average particle radius; χ, Flory-Huggins solvency parameter; d, thickness of “brush” polymer layer (~δ); Gs :free energy of −1 −1 0 acid base interaction between particles at distance s0; δ, Stern layer thickness; ε, relative permittivity of water (=78.5 C V m ); εo, permittivity of a vacuum (=8.854 × 10−12); e, elementary charge of an electron; κ, inverse Debye length; k, Boltzmann constant; λ, decay length for acid–base interactions (=0.6

nm); Ms, saturation magnetization; μo, vacuum permeability; MW, molecular weight of polyelectrolyte; n, number concentration of ion pairs; ρp, density

of polymer; Ψd diff use potential (~zeta potential); s, distance between interacting surfaces; s0, minimum equilibrium separation distance (=0.158 nm); T,

absolute temperature; φp, volume fraction of polymer in “brush”; v1, volume of solvent molecule; z, electrolyte valence, § Refer to referenced publications for details of equation derivations and constraints. these curves (VT) demonstrates that particles can have a net gated, whereas particles in the secondary well are reversibly attraction in a primary or secondary minimum (well). Particles aggregated (i.e., re-entrainment is possible if shear forces are in the primary well are considered to be irreversibly aggre- exerted) (Hahn et al., 2004). Figure 1 also shows something

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1911 theories in colloid science, DLVO and XDLVO. Among these challenges are shape, size, structure, composition, and adsorbed or grafted organic macromolecules (e.g., coatings or NOM) (Fig. 2). Although particles not classifi ed as “nano” (i.e., <100 nm) also present some of these challenges, manufactured NPs are often designed with the intention of amplifying these properties. Examining some early studies demonstrating these challenges can inform our discussion in the previ- ous section on implications of NP aggregation on their transport and reactivity in the environment. Particle Size Th e small size of most NPs confl icts with a funda- Fig. 1. van der Waals force (dashed line), electrostatic double layer (EDL) force (gray mental assumption of DLVO. Equations describ- line), total Derjaguin-Landau-Verwey-Overbeak (DLVO) forces, and elastic force plotted ing EDL are mathematically incorrect if this together to fi nd total potential as a function of separation distance. XDLVO, extended Derjaguin-Landau-Verwey-Overbeak. V/KT, potential energy divided by Boltzmann’s assumption does not hold (Delgado et al., 2007). constant and absolute temperature. Th is is because when a particle reaches a small enough size, its surface curvature is too substantial that is commonly the case with manufactured NPs having an to assume it is fl at (i.e., when the assumption of organic coating: Classical DLVO forces alone are not suffi cient κ*a >>1 no longer applies). Th erefore, this is one of the pri- to accurately predict aggregation behavior. Steric repulsion mary challenges when considering NP aggregation in a DLVO

forces (VT + VELAS) resulting from adsorbed polymer or poly- framework. Aside from this, vdW and EDL forces (Table 1) electrolyte coatings or natural organic matter (NOM) makes it scale similarly with particle size (Evans and Wennerstrom, so these particles may only have a net attraction in a second- 1999; Hiemenz and Rajagopalan, 1997; Israelachvili, 1991), ary minimum. Coated NPs may therefore aggregate reversibly so moving from microscale to nanoscale particles should theo- (Graf et al., 2009), and aggregation reversibility has signifi cant retically aff ect a change on vdW and EDL forces to a similar consequences regarding the fate, transport, bioavailability, and degree. However, as a particle decreases in size, a greater per- eff ects of NPs in the environment. Th ese additional forces are centage of its atoms exist on the surface. Electronic structure, collectively known as extended DLVO (XDLVO). surface charge behavior, and surface reactivity can be altered as

In addition to steric repulsion forces (VT + VELAS) due to NP a result. Recent studies suggest that redistribution and chang- coatings, other XDLVO forces have been invoked to match ing coordination environment of charge and atoms at the sur- experimental data for various types of aggregating primary faces of NPs strongly infl uences their reactivity (Hochella et al., particles (Chen and Elimelech, 2007; Fritz et al., 2002; Hoek 2008). He et al. (2008) reported that as the particle size became and Agarwal, 2006; Ortega-Vinuesa et al., 1996; Phenrat et smaller, the particles became more susceptible to surface charge al., 2007; Phenrat et al., 2008; Romero-Cano et al., 2001; Wu titration. Th ey found an inverse correlation between particle et al., 1999). Th is means that additional short-range forces are size and the measured point of zero charge (PZC); 12, 32, and operating on an equivalent magnitude as vdW attraction and 65 nm hematite particles had PZCs of 7.8, 8.2, and 8.8 pH EDL. Th ese forces can include bridging (Chen and Elimelech, units, respectively (He et al., 2008). Th ese hematite NPs were 2007), osmotic (Fritz et al., 2002; Ortega-Vinuesa et al., 1996; synthesized using the same method so they would be identi- Phenrat et al., 2008; Romero-Cano et al., 2001), steric (Fritz cal in every aspect except for size. However, for these particles, et al., 2002; Ortega-Vinuesa et al., 1996; Phenrat et al., 2008; at pH 7 the smaller particles are less charged than larger par- Romero-Cano et al., 2001), hydrophobic Lewis acid-base (Wu ticles, leading to lower EDL repulsion forces and subsequently et al., 1999) (Hoek and Agarwal, 2006), and magnetic forces greater aggregation tendency. Indeed—given the same ionic (Phenrat et al., 2007). Table 1 summarizes these forces. Th e rel- strength—smaller hematites have faster aggregation rates than evant force type depends on the system tested, and frequently larger ones (He et al., 2008). Th is suggests that size polydisper- more than one XDLVO force is important. Moreover, forces sity in manufactured NPs aff ects their aggregation behavior. are not completely independent of one another. In the case of Polydispersion is the reality in large-scale applications. For nanomaterials, understanding aggregation using a DLVO- or example, reactive nano-scale zerovalent iron (NZVI) particles XDLVO-based theory presents many challenges. used for environmental remediation are highly polydisperse (Liu et al., 2005; Phenrat et al., 2007; Phenrat et al., 2009a). Nanoparticle Challenges to Derjaguin- Unlike van der Waals attraction, long-range magnetic attrac- Landau-Verwey-Overbeak tion between two NZVI particles increases with particle radius to the sixth power (Table 1). Polydispersity and particle size Colloid science demonstrates that several forces with diff er- therefore greatly aff ected the aggregation of NZVI having parent natures and origins govern the kinetics and extent of particle sizes ranging from 15 to 260 nm, with aggregation being ticle aggregation. Th is section elaborates how manufactured proportional to the fraction of large particles in the dispersion NPs present a combination of challenges to two predominant (Phenrat et al., 2009a).

1912 Journal of Environmental Quality • Volume 39 • November–December 2010 Fig. 2. Derjaguin-Landau-Verwey-Overbeak (DLVO) frames particle aggregation by modeling two solid spheres approaching in suspension. The sum of van der Waals (VDW) force and electrostatic double layer (EDL) force determines if the particles attach or repulse. Nanoparticles present challenges to DLVO frameworks because of shape, structure, composition, size, and macromolecule considerations. Not drawn to scale.

Another example of a size eff ects is “halo” stabilization, Particles with a high Hamaker constant have greater aggre- where nano sized materials can control electrostatic repulsion gation tendency compared with particles with a low Hamaker between two approaching micrometer-sized particles, hinder- constant at the same solution and surface chemistry. Because ing their fl occulation by keeping them separated beyond the the origin of van der Waals attraction is permanent or induced range of DLVO attractive forces (Fig. 3). Tohver et al. (2001) dipoles, a bulk material with electronic or molecular structure found that adding 6-nm zirconia NPs to a dispersion of unsta- that favors generation of permanent or induced dipoles nor- ble submicron silica particles (0.6 μm in diameter) stabilized mally has a large Hamaker constant value. For example, the the mixture (Tohver et al., 2001). Th e same phenomenon was Hamaker constants of gold, silver, and polystyrene are 45.3, observed for bimodal magnetorheological fl uids consisting of 39.8, and approximately 9.8 × 10−20 J, respectively (Hiemenz micrometer sized magnetite particles (1.45 μm in diameter) and Rajagopalan, 1997). van der Waals attraction forces are and magnetite NPs (8 nm diameter) (Viota et al., 2007). Size approximately fi ve times stronger for gold particles in compari- and resultant polydispersity, therefore, present a large challenge son with polystyrene (see equation in Table 1). In addition to to understanding how NP dispersions behave because these controlling van der Waals attraction, some materials can origi- parameters can enhance or diminish aggregation. nate forces that challenge the DLVO framework, including magnetic and hydrophobic interactions that promote aggrega- Chemical Composition tion. For example, ferromagnetic and ferrimagnetic materials, Colloid science captures the eff ect of chemical composition on such as zerovalent iron and magnetite, can cause long-range aggregation through the Hamaker constant, which governs van magnetic attraction without an applied magnetic fi eld (Table der Waals attraction; saturation magnetization, which control 1) (Elena et al., 2009; Phenrat et al., 2007). Th is magnetic magnetic attraction; hydrophobicity, which controls hydro- force contributes to abnormally rapid aggregation. Moreover, phobic interaction (Wu et al., 1999); and surface charge, which carbon-based nanomaterials, such as C60 or CNTs, have hydro- governs electrostatic double layer (EDL) interaction (Table 1) phobic surfaces in aqueous environments. Interaction of all- (Evans and Wennerstrom, 1999). carbon surfaces with water molecules involves a signifi cant entropic penalty driving an aggregation process that might

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1913 Complex Crystal Structures Complex NP crystal structures can present challenges to colloid science because changes in the crystal structure can alter Hamaker constants and surface charges in ways not explain- able by theory. Additionally, defects in the crystal lattice can alter surface charge (Evans and Wennerstrom, 1999; Israelachvili, 1991). Th e crystal structure of materials with the same atomic composition can play a large role in

determining surface charge. Nano-sized TiO2, for example, has three diff erent polymorphs: rutile, anatase, and brookite, each with a unique crystal structure. Th e zeta potential is approxi- Fig. 3. Without nanoparticles present, microscale particles aggregate because net forces are mately −20 mV (French et al., 2009) when attractive. Nanoparticles (NPs) create a “halo” eff ect that increases electrostatic double layer repulsive forces and causes net forces between microscale particles to be unattractive. anatase and brookite structures are predominant and is approximately −35 mV (Lebrette et be understood through the use of XDLVO frameworks that al., 2004) when the rutile phase is predominant model hydrophobic Lewis acid–base interactions (Wu et al., (both measured at pH 7.5). Th is is consistent with fi ndings 1999). Th ese types of XDLVO frameworks have yet to be that the dielectric constant can be altered by annealing pure developed for carbon-based nanomaterials. anatase-phase TiO2 to the rutile phase (Kim et al., 2005). Th e Unique NP composition and morphology (e.g., core-shell Hamaker constant is a function of dielectric constant (Tabor NPs) and environmental transformations of NPs will chal- and Winterton, 1969); therefore, the Hamaker constant also lenge DLVO explanations of aggregation. Particles with a high can depend on the crystal structure of NPs. Th is is important surface potential have lower aggregation tendencies compared to consider for NP systems containing mixtures of phases or with particles with a low surface potential given the same where crystal structures are altered for applications, such as solution chemistry. Nanoparticle chemical composition alters photocatalysis with TiO2. surface potential by dissociation or ion adsorption by surface atoms (Evans and Wennerstrom, 1999; Israelachvili, 1991). Shape Surface charge of uncoated NPs is therefore governed in part In DLVO modeling, one of the primary assumptions is that by the type of atoms at the surface of the particles. However, particles are spherical. Th is assumption is reasonable for ideal the core of core-shell particles, which are common, can infl u- latex particles and some ideal colloidal systems. Manufactured ence the charge of the surface atoms. For example, uncoated NPs, however, come in a variety of nonspherical shapes. 0 Fe /Fe3O4 core-shell NPs have a zeta potential near −32 mV (in Triangles, icosahedrons, ellipses, rhombohedrons, spindle, −1 0 1 mmol L NaHCO3 at pH 7.5), but as the Fe core of the par- rod, and tube shapes are possible (AgNPs [Moon et al., 2009], ticles oxidizes to magnetite, the surface charge becomes more Fe2O3NPs [Wang et al., 2009], and Ag2S [Ma et al., 2007]), like that of magnetite (Phenrat et al., 2007). Gold NPs (syn- thereby complicating traditional DLVO and XDLVO. thesized via citrate reduction) have a charge of around −55 mV Both vdW and EDL forces are aff ected by changes in (Kim et al., 2009), presumably due to citrate acid adsorbed on shape. Several researchers have investigated these changes the surface. Multiwall carbon nanotubes (MWCNTs) treated (Bhattacharjee and Elimelech, 1997; Montgomery et al., 2000; with a mixture of sulfuric acid and nitric acid (3:1) together Vold, 1954). At a separation distance smaller than the mean with ultrasonication have zeta potential around −55 mV (Kim diameter (∼v1/3, where v is the volume of particles), the attrac- and Sigmund, 2004) due to carboxylation of the surface of the tion between anisometric particles, such as plates, rectangu- MWCNTs (Smith et al., 2009). Brant et al., Cheng et al., and lar rods, and cylinders, is larger than for spherical particles of Chen et al. have shown that stable fullerene clusters with nega- equal volume because a greater number of atoms are in close tive charge of approximately −40 to −80 mV (from pH 2 to proximity (Vold, 1954). Th e attraction between spheres, rods 12) were formed from mixing fullerene clusters in water for a (and cylinders), and platelets varies as s−1, s−2, and s−3, respec- period of several weeks (Brant et al., 2005; Chen and Elimelech, tively, where s is separation distance (Vold, 1954). For cylindri- 2009; Cheng et al., 2004). Th e nature of these charges is under cal particles, parallel orientation is energetically favored over a investigation. Th ese examples demonstrate that core-shell perpendicular orientation at small distances. For rectangular structures, the presence of coatings, and chemical transforma- rods or platelets, an orientation with the largest faces oppo- tions impart charges to NP surfaces. Th ese charge sources, or site each other is the most favorable energetically. Electrostatic combinations thereof, are not accounted for in colloid science. double-layer forces of cylindrical particles are theoretically a Understanding the fate of NPs in the environment therefore function of particle orientation (Vold, 1954). Th ese results requires a fundamental comprehension of the types of trans- imply that shape can theoretically control aggregation under formations that the NPs may undergo during manufacturing, some favored orientations. For example, Kozan et al. (2008) under commercial use, and after release into the environment studied the aggregation of tungsten trioxide (WO3) nanowire (as discussed later in this review). suspensions of diff erent types (uneven, single, or bundled in

1914 Journal of Environmental Quality • Volume 39 • November–December 2010 diameter) and dimensions (diameter of 40–200 nm and nomi- Surfactants and Macromolecular Surface Coatings nal lengths of 2, 4, 6, and 10 μm) in various polar solvents Nanomaterials are commonly manufactured with surface and compared the aggregate morphology of the nanowires with coatings (e.g., surfactants, polymers, and polyelectrolytes) to that of spherical WO NPs (40 nm in diameter) (Kozan et al., 3 enhance dispersion stability or to provide specifi c functionality 2008). Th e aggregates of spherical WO NPs were more com- 3 (He et al., 2007; Hezinger et al., 2008; Hydutsky et al., 2007; pact and more spherical in shape (D ~ 2.6 and R ~3.8 μm) f g Mayya et al., 2003; Phenrat et al., 2008; Saleh et al., 2008; than aggregates of nanowires with diameter of 200 nm (D ~ f Saleh et al., 2005; Zhang et al., 2002). Table 2 summarizes 2.3 and R ~2.2 μm). Moreover, aggregation of the nanow- g some examples of surface-coated NPs and their applications. ires for 7 d in a quiescent environment resulted in no signifi - Adsorbed or covalently bound surfactants prevent aggregation cant R increase but an increase in D (Kozan et al., 2008). g f and enhance dispersion stability of NPs by increasing surface Th is reorganization is related to the theoretical prediction that charge and electrostatic repulsion or by reducing interfacial anisometric particles are likely to aggregate under a secondary energy between particle and solvent (Rosen, 2004). Coulombic minimum (Vold, 1954). Th e degree of secondary minimum attractions (Rosen, 2004; Vaisman et al., 2006; Wang et al., aggregation is dependent on shape and is most favorable for 2004), hydrophobic interactions (Vaisman et al., 2006), and platelets, less favorable for rods and cylinders, and least favor- surfactant concentration (Vaisman et al., 2006) aff ect the able for spherical NPs (Vold, 1954). Loose aggregates formed adsorbed surfactant mass and layer conformation and hence in the secondary minimum might gradually restructure into the ability of a surfactant to stabilize a NP against aggrega- denser aggregates in the primary minimum. Nanoparticless tion. Table 3 summarizes the properties of surfactants that designed with a myriad of unique shapes (Cai and Sandhage, aff ect their ability to stabilize NPs against aggregation. Low- 2005; Kawano and Imai, 2008; Sanchez-Iglesias et al., 2009) molecular-weight surfactants reversibly adsorb on particle sur- therefore challenge colloid science to understand their aggrega- faces (Rosen, 2004), so, without a strong interaction between tion behavior in the environment because of the model com- the surfactant and NP surface, the surfactant may desorb or be plexity of understanding these shapes in a DLVO framework. displaced by higher-molecular-weight natural polymers such as Although nonconventional theories, such as surface element NOM when the particles are released into the environment. integration (Bhattacharjee and Elimelech, 1997), may account Th erefore, changes in the stability of surfactant-coated NPs are for interfacial forces in irregular shapes (e.g., ellipsoids), these expected on release to the environment. theories are diffi cult to apply to the unique (e.g., helical) and Polymers are organic macromolecules consisting of repeti- diverse types of nanomaterial shapes. tions of smaller monomer chemical units (Fleer et al., 1993).

Table 2. Examples of surface coated nanoparticles and their applications. Particle type Surface modifi er Purpose References Quantum dots amphiphilic alkyl-modifi ed alter particle hydrophilicity and modify quantum Luccardini et al. (2006) polyacrylic acid dot interactions with other biological compounds Titanium dioxide polyacrylic acid stabilize dispersions and modify surface Kanehira et al. (2008) chemistry allowing attachment of an antibody for a substrate selective photocatalytic system Fe0/Fe-oxide poly(styrene sulfonate), polyaspartate, inhibition of aggregation, decreasing adhesion He et al. (2009), Kanel et al. (2008), carboxymethyl cellulose, or triblock to solid surfaces, increasing mobility in the Phenrat et al. (2008), Phenrat et copolymers subsurface, and targeting specifi c pollutants al. (2009c), Phenrat et al. (2009a), Saleh et al. (2005) Carbon nanotubes various kinds of anionic, cationic, individually stabilized, single-walled carbon Moore et al. (2003) nonionic surfactants and polymers nanotubes Silver daxad 19 (sodium salt of a high- stabilize dispersions silver nanoparticles Sondi et al. (2003) molecular-weight naphthalene sulfonate formaldehyde condensate)

Table 3. Examples of surfactant properties and their relationship to particle stabilization. Property Findings References Surfactant A linear relationship exists between the surfactant chain length and the logarithm of Dederichs et al. (2009) chain length the dispersion concentration (cdc), where cdc is defi ned as the lowest concentration of a surfactant necessary to disperse hydrophobic particles. Molecular weight Stabilization of CNT† with nonionic surfactants increased with molecular weight (e.g., Brij Moore et al. (2003) 78 [MW = 1198 g mole−1] stabilized 4.3% of CNT, while Brij 700 [MW = 4670 g mole−1] could stabilize 6.4% of CNT) Type of ionic At similar molecular weight, cationic surfactants such as dodecyltrimethylammonium Moore et al. (2003) head groups bromide (MW = 308 g mole−1) and cetyltrimethylammonium bromide (MW = 364 g mole−1) were slightly more eff ective in stabilizing CNT than anionic sodium dodecylbenzenesulfonate (MW = 348 g mole−1). Self-assembly Sodium dodecyl sulfate, anionic surfactant, forms cylindrical micelles as helices or double O’Connell et al. (2002), Richard structure helices and hemicellar structures on CNT. Random adsorption of surfactants on CNT et al. (2003), Yurekli et al. (2004) hypothesized to enhance stabilization of the CNT dispersions.

† CNT, carbon nanotube.

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1915 Th ey can be synthetic (e.g., polyethylene glycol) or naturally occurring (e.g., biomacromolecules, such as proteins and poly- saccharides) (Brewer et al., 2005; Jain et al., 2005; Liu et al., 2006). Th ey also are surface active agents but are generally of higher molecular weight than surfactants discussed previously. Unlike surfactants, the repeating monomer units can adsorb to NPs at multiple places along the polymer chain, making polymer adsorption stronger than surfactant adsorption and relatively irreversible. For example, Kim et al. (2009) modi- ﬁ ed NZVI by physisorption of various types of strong and weak polyelectrolytes and found that <30% by mass of polymer desorbed from the NPs over a period of 4 mo (Kim et al., 2009). Polyelectrolytes are charged polymers. Th eir charge is derived from ionizable groups built into the polymer structure − − − (e.g., from -COO , -SO4 , or -SO3 groups) (Fleer et al., 1993). Adsorbed polymers and polyelectrolytes provide electrosteric repulsions that prevent NP aggregation. Th ese repulsive forces

have osmotic (VOSM) and elastic-steric (Velas) components due to their charge and large molecular weight, respectively (Fig. 4). Th e magnitude of the electrosteric repulsion is governed by the surface excess (adsorbed mass of polymers per unit surface area of a NP), polymer molecular weight, polymer density, adsorbed layer thickness, and solution composition (see expressions in Table 1 and examples in Tables 4 and 5) (Fritz et al., 2002; Ortega-Vinuesa et al., 1996; Phenrat et al., 2008; Romero-Cano et al., 2001). Fig. 4. Osmotic plus elastic repulsive forces (VOSM + Velas) of polyelectro- Surfactant–NP and polymer–NP interactions are extremely lytes are dependent on surface layer thickness. complex and often kinetically controlled rather than being in thermodynamic equilibrium. Th e science describing or summarizes some of the environmental interactions of NPs predicting adsorbed mass and adsorbed layer conformation that aff ect or are aff ected by aggregation and are discussed next. is immature, and the adsorbed mass and layer conformation depends on NP properties and solution conditions (pH, ionic Eff ect of Solution Chemical Composition on strength, ionic composition). Th erefore, predicting the eff ects Aggregation: pH and Ionic Solutes of adsorbed surfactants, polymers, or polyelectrolytes on NP pH and dissolved ionic solutes play crucial roles on aggregation behavior in the environment is challenging. of NPs in engineered and natural systems. Th ese parameters can be controlled in laboratory settings but may vary spatially and Eff ect of Nanoparticle Aggregation on temporally in natural systems. Spatial heterogeneity of miner- Applications and Implications als in the subsurface alters the concentration of ionic species present in diff erent systems (e.g., groundwater versus surface for the Environment water). Variation in aquatic environment pH stems from the Understanding NP aggregation under well controlled solution same phenomenon. A better understanding of how pH and dis- conditions in the laboratory is diffi cult because NP properties solved ionic solutes aff ect the behavior of organic macromol- present challenges for colloid science. However, ultimately we ecule–coated NPs in the environment is needed. Surface charge would like to comprehend the fate, transport, reactivity, and titration and EDL screening are the two primary ways that risks associated with manufactured NPs released into the envi- pH and ionic solutes promote NP aggregation. Most NP sur- ronment. How do these intrinsic challenges in understanding faces have surface functional groups (e.g., hydroxide and oxide aggregation aff ect our understanding of NP fate, transport, and groups) that are titratable by H+ or OH−. Excess of H+ (low eff ects in more complicated environmental settings? Figure 5 pH) normally results in a positively charged particle surface,

Table 4. Examples of polymer properties and their relationship to particle stabilization. Polymer property Findings References Adsorbed layer thickness and surface excess Correlate to dispersion stability of polymer modifi ed Fe0–Fe Phenrat et al. (2008) oxide nanoparticles. Molecular weight Dispersion stability of carbon nanotubes increased as the Moore et al. (2003) molecular weight of Pluronic increased.

Condition of polymer adsorption: heat treatment Adsorption of polyacrylic acid on TiO2 enhanced at 150°C Kanehira et al. (2008) versus 20°C. This resulted in the enhanced dispersion stability.

Condition of polymer adsorption: solvent property More polyacrylic acid adsorbed on TiO2 in dimethylformamide Kanehira et al. (2008) than in water.

1916 Journal of Environmental Quality • Volume 39 • November–December 2010 Table 5. Examples of particle stabilization and destabilization by natural organic matter and biomolecules. System Findings References

TiO2 (positively charged) Disaggregation of TiO2, resulting in a smaller particle size (3 nm) than the Domingos et al. (2009) −1 at pH 2 to 6 with 1 mg L FA† original bare TiO2 obtained from the manufacturer (10 nm).

TiO2 (negatively charged) Flocculation of TiO2 due to bridging FA was observed. Aggregation decreased Domingos et al. (2009) pH > PZC with 1 mg L−1 FA with increasing ionic concentration. High ionic concentration screened the

negative charges of FA and TiO2, resulting in greater FA adsorption onto TiO2 promoting stabilization, not bridging.

−1 −1 C60 or CNT with HA (CaCl2 Humic acid as low as 1 mg L increased dispersion stability of C60 up to 5 mg L . Chappell et al. (2009), concentration <40 mmol L−1) Humic acid concentrations increased the dispersion stability of CNTs from 5 to Chen and Elimelech (2007) 300 mg L−1.

C60 with HA (CaCl2 concentration Unadsorbed HA aggregated through intermolecular bridging via calcium Chen and Elimelech (2007) −1 >40 mmol L ) complexation. Aggregates subsequently bridged C60, resulting in fl occulation. AuNPs with BSA Gold nanoparticles coated with BSA were stable as colloids at the PZC of bare Brewer et al. (2005) AuNPs, suggesting steric stabilization due to adsorbed BSA layers. AuNPs with lysozymes Enhanced aggregation of lysozyme–AuNP assemblies at physiological pH Zhang et al. (2009) at very low lysozyme concentrations (16 nmol L−1). Au nanoparticles were found embedded in the lysozyme matrix.

† AuNP, gold nanoparticle; BSA, bovine serum albumin; CNT, carbon nanotubes; FA, fulvic acid; HA, humic acid; PZC, point of zero charge.

Fig. 5. Hypothesized eff ects of the environment on nanoparticle aggregation and how aggregation state could alter nanoparticle (NP) environmental interactions. (A) pH and ions. Increasing acidity can lead to charge neutralization at the point of zero charge. Increasing ionic strength reduces the size of the electrostatic double layer and repulsive forces. Certain ions (e.g., Ca2+) can bridge functional groups on the surface of nanoparticles. (B) Heteroaggregation. Macromolecules present in the environment coat nanomaterials leading to a steric eff ect. Larger macromolecules trap NPs in a mesh or gel causing stabilization or destabilization. Nanoparticles also associate with other particles such as clays or biocolloids. Nanoparticles fl ow through porous media depending on their aggregation state. (C) Biological interactions. Cells activate phagocytosis mechanisms depending on if a particle is aggregated or not. Nanoparticles containing metals aggregate at the surface of organisms and release metal ions at varying rates. (D) Transformations. Oxidants or sunlight degrade particle coatings or the particles themselves. Aggregated photoactive NPs only have the outside particles active.

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1917 whereas excess of OH− (high pH) normally yields a negatively tion, but this information is useful for understanding the infl u- charged surface. For each particular system, the pH at which ence of various parameters on aggregation and deposition. + − the H and OH concentration causes suspended particles to Nanoparticle Interaction with Naturally Occurring Organic Matter, have a neutral charge is called the PZC. As pH moves toward Biomolecules, and Biomacromolecules this point, the EDL repulsion decreases, and aggregation is promoted by vdW attraction (Fig. 5A). Charge reversal (i.e., from As with synthetic surface coatings used on NPs (e.g., polyelec- negatively charged to positively charged particles) is also possi- trolytes), naturally occurring organic macromolecules (e.g., ble. Th is change would dramatically aff ect NP behavior because NOM) in the environment can signifi cantly alter the aggre- most surfaces in the environment are negatively charged at near gation behavior of NPs. Even without fully defi ned structure, neutral pH. As for the ionic species eff ect, high ionic concen- due to its macromolecular nature, NOM is expected to prevent tration decreases the Debye length (κ−1) (see Table 1 for calcu- aggregation and deposition presumably due to electrosteric sta- lation of EDL using DLVO), resulting in the decrease of the bilization (Amirbahman and Olson, 1993; Amirbahman and extent of EDL repulsion (Fig. 5A). Th e ionic concentration Olson, 1995). Natural organic matter consists of mainly fulvic where the repulsive energy barrier is completely screened and and humic substances. Fulvic compounds typically make up 40 rapid aggregation occurs is known as the critical coagulation to 80% of NOM compounds, and, due to their small size (∼1 concentration. Table 6 summarizes factors governing nanoma- kDa), they can coat the surfaces of particles. Fulvic compounds terial critical coagulation concentration and dispersion sensitiv- alter the surface charge and have been shown to stabilize with ity toward electrolytes. In many cases, surface titration and EDL an eff ect similar to surfactants (Domingos et al., 2009). In fact, screening eff ects occur in NP dispersions as they occur for large NOM may stabilize MWCNTs better than surfactants com- colloidal particles. However, there are notable exceptions where monly used for laboratory suspensions of MWCNTs (Hyung et al., 2007). However, enhanced dispersion stability due to elec- the pHPZC depends on the NP size. Table 7 summarizes parameters altering nanoparticle PZC. trosteric forces and destabilization due to bridging have been experimentally observed, and it is not often possible to predict Heteroaggregation and Deposition these interactions. Natural organic matter attaches to the sur- Th e environment is a complex system with multiple primary face of particles in a variety of ways. For example, irreversible molecules and particles as well as solution compositions. adsorption onto the surfaces of iron oxide via ligand exchange Straightforward DLVO theory is diffi cult to apply to these sys- between carboxyl/hydroxyl functional groups of humic acid tems, and typically XDLVO forces are brought into models to and iron oxide surfaces (Gu et al., 1994) or hydrophobic inter- improve results (Wu et al., 1999). At best, DLVO + XDLVO actions with carbon based nanomaterials (Hyung et al., 2007) provides qualitative information about aggregation and deposi- have been reported. Whether or not attachment results in stabi-

Table 6. Factors governing nanomaterial critical coagulation concentration and dispersion sensitivity toward electrolyte concentration (ionic strength). Parameter Findings References

Type of ionic Regardless of particle type, CCC† for divalent ions (e.g., CaCl2 or MgCl2) Chen and Elimelech (2007), Evans and species is lower than for monovalent ions (e.g., NaCl). Wennerstrom (1999), Saleh et al. (2008)

Particle size Values of CCC increase with increasing particle size (e.g., CCC values of Fe2O3 with He et al. (2008) diameter of 12, 32, and 65 nm have CCCs of 45, 54, and 70 mmol L−1 NaCl, respectively). Type of particles The CCC of carbon nanomaterials such as fullerene appears to be higher than that of Chen and Elimelech (2007), −1 metal oxide such as Fe2O3 [e.g., CCC of C60 [dp = 60 nm] is around 120 mmol L NaCl, He et al. (2008) −1 whereas that of Fe2O3 [dp = 65 nm] is 70 mmol L NaCl). This is presumably due to stronger van der Waals attractions of Fe2O3 in comparison to C60, making the stability of metal oxide nanoparticles more sensitive to screening eff ects.

Particle shape The CCC of carbon nanotubes is lower than C60 particles. Van der Waals attraction is Chen and Elimelech (2007), greater and EDL repulsion is lower for rod-like shapes than for spherical shapes. Saleh et al. (2008) Surface modifi er Due to steric repulsions polymeric surface modifi cation decreases the sensitivity Kanehira et al. (2008)

of aggregation induced by ionic species. NZVI (dp = 162 nm) modifi ed by guar-gum −1 prevented aggregation at ionic concentrations up to 0.5 mmol L . TiO2 modifi ed by PAA prevented aggregation in ionic concentrations up to 0.5 mol L−1 NaCl.

† CCC, critical coagulation concentration; EDL, electrostatic double layer; NZVI, nano zero valent iron; PAA, polyacrylic acid.

Table 7. Parameters altering nanoparticle point of zero charge. Parameter Findings References Chemical composition PZCs of metal NPs,† metal oxides, and carbon nanomaterials have a wide range Brant et al. (2005), Kim et al. (2009), and need to be measured for each material and preparation method. Kosmulski (2009), Sirk et al. 2009 Surface modifi cation Surface modifi cation using anionic polyelectrolytes shifts PZC toward lower Kim et al. (2009), Sirk et al. (2009) values (e.g., ~3 pH units for Fe0/Fe-oxide nanoparticles). Particle size PZCs for hematite nanoparticles of diameter 12, 32, and 65 nm are ~7.5, 7.8, He et al. (2009) and 8.5, respectively. Particle transformation Aged Fe0/Fe oxide nanoparticles have lower PZC than fresh NPs by ~3 pH Kim et al. (2009) units, presumably due to surface oxidation and hydrolysis.

† NP, nanoparticle; PZC, point of zero charge.

1918 Journal of Environmental Quality • Volume 39 • November–December 2010 lization or destabilization depends on several factors, including sion stability of engineered NPs via heteroaggregation (Fig. type of NOM, NOM concentration, and solution chemistry 5B). Heteroaggregation with bacteria can prevent aggregation (Chen and Elimelech, 2007; Chen et al., 2006; Diegoli et al., or promote disaggregation of NPs. Cerium oxide NPs were 2008; Domingos et al., 2009). Stabilization usually results from reported to attach the surface of bacteria at the maximum −2 NOM forming a charged stabilizing layer on the outside of the adsorption capacity of 15 mg of CeO2 m due to electrostatic particle. Destabilization, on the other hand, results from par- attraction (Th ill et al., 2006). Holden et al. (2009c) showed ticles being bridged by larger NOM molecules, such as rigid that TiO2 NPs were disaggregated in the presence of micro- biopolymers (Fig. 5B, left). Th is occurs for relatively large-sized organisms (unpublished data). Heteroaggregation of NZVI NOM (~10–100 kDa) and can dominate interactions between and bacteria also increases dispersion stability when com- nanoparticles (Buffl e et al., 1998). Large-molecular-weight bio- pared with dispersions containing no bacteria (Li et al., 2010). molecules and biomacromolecules, including proteins, poly- Heteroaggregation with bacteria, NOM, or mobile colloids peptides, and amino acids (McKeon and Love, 2008; Moreau (e.g., clay particles) could enhance NP stability, but heteroag- et al., 2007; Vamunu et al., 2008), aff ect aggregation of NPs in gregation to large enough particles could also destabilize NP aquatic environments similarly. Table 5 summarizes reports of dispersions. A complete understanding of the impacts of het- NP stabilization and destabilization aff orded by the presence of eroaggregation on NP behavior requires more experimentation. NOM and biomacromolecules reported in literature. Eff ect of Aggregation on Deposition and Sedimentation Nanoparticle Interaction with Inorganic and Biocolloids Aggregation of NPs aff ects their fate and transport in ground Nanoparticles released to the environment can interact with and surface waters (Fig. 6). In the subsurface, aggregation aff ects inorganic colloids (e.g., clays, aluminoscilicates and iron oxy- deposition processes; in surface waters, aggregation drives sedi- hydroxides) and biocolloids (e.g., bacteria), altering the disper- mentation processes. Homoaggregation of NPs mostly leads to

Fig. 6. Nanoparticles are hypothesized to transport diff erently in the environment depending on their homo- or heteroaggregation state. Homoaggregated particles are more likely to attach to surfaces and be removed in porous media. Homoaggregation can also lead to the formation of aggregates that are large enough to be settled out of surface waters by gravitational forces. Heteroaggregation may lead to surface particle coatings that enhance transport in porous media. Heteroaggregation can also lead to stabilization or destabilization eff ects in surface waters (e.g., with bacteria or natural organic matter).

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1919 high deposition rates in porous media because of increased col- or XDLVO theory and is an important area of environmental lisions between NP clusters and aquifer materials via sedimenta- nanotechnology research. tion and interception as opposed to relying only on Brownian motion. For example, NZVI injected into the subsurface at high Eff ect of Aggregation on Nanoparticle Reactivity particle concentrations (1–10 g L−1) aggregate rapidly, and their and Eff ect of Reactivity on Aggregation State mobility in the subsurface suff ers as a result (Kanel et al., 2008; Aggregation can alter NP reactivity, and NP aggregation state Phenrat et al., 2007; Saleh et al., 2007) (i.e., they do not trans- can be altered by chemical or photochemical environmen- port far from their injection point). tal processes (Fig. 5D). Several types of NPs are photoactive Th ere are numerous studies that suggest prevention of (e.g., C , quantum Dots, and TiO ) and can produce reactive aggregation improves transport in porous media. Polymeric 60 2 oxygen species (ROS). A relationship between Df and photoac- surface modifi cation has been used to decrease NZVI aggrega- tivity indicates that the structure of the aggregates formed has a tion and therefore deposition (Kanel et al., 2008; Kim et al., role in determining ROS production and subsequent reactivity. 2009; Phenrat et al., 2009c; Phenrat et al., 2008; Saleh et al., Fullerene (C ) is a photoactive carbon-based class of NP that 2008). He et al. (2009) and Tiraferri and Sethi (2009) observed 60 forms crystalline aggregates in water with high Df (i.e., closely enhanced mobility of NZVI modifi ed with adsorbed carboxy- packed aggregates). Th is aggregation promoted triplet–triplet methyl cellulose and guar gum (He et al., 2009; Tiraferri and annihilation and self-quenching due to close contact between Sethi, 2009). Phenrat et al. (2009a) demonstrated that NZVI cages and severely reduced the quantum yield of C60 (Hotze transport was limited specifi cally by NP aggregation during et al., unpublished data). Fullerol (hydroxylated fullerene), transport in porous media by comparing the transport of non- which has a lower theoretical quantum yield than C60 produced aggregating hematite NPs and aggregating NZVI of the same greater amounts of ROS than C under identical solution con- size and surface chemistry. When NP deposition is interpreted 60 ditions because fullerol formed aggregates with lower Df (less by classical fi ltration theory (i.e., assuming no aggregation closely packed fractal aggregates), which decreased the amount during particle transport in porous media), the eff ect of aggre- of triplet–triplet annihilation and increased ROS yield relative gation is neglected and can lead to error in the reported attach- to C60 (Hotze et al., unpublished data). ment coeffi cient. Th is makes comparisons between diff erent Aggregation has been shown to decrease the rate of dissolu- experiments and systems diffi cult. tion of lead sulfi de (PbS) NPs (Liu et al., 2008). Th e dissolu- In contrast to homoaggregation, heteroaggregation with tion rate of 240 nm aggregates of PbS NPs (4.7 × 10−10 mol m−2 mobile, natural colloidal particles (e.g., clay) and with natu- s−1) was an order of magnitude lower than for monodisperse ral particles that have low density (e.g., microorganisms) can 14-nm primary PbS NPs (4.4 × 10−9 mol m−2 s−1) and also enhance transport of NPs in porous media by decreasing the lower than for micrometer-sized particles of the same materi- particle collision rate. For example, the presence of 20 mg als. In addition to aggregation decreasing available surface area −1 L NOM enhanced NZVI transportability in porous media for dissolution, they attributed the retarded dissolution to inhi- (Johnson et al., 2009). Clay particles were also used as NZVI bition in the highly confi ned space between the particles in supports to enhance NZVI mobility (Hydutsky et al., 2007), the aggregates. In confi ned space, water molecules arrange dif- which indicates that heteroaggregation between clay and NZVI ferently than in the bulk, increasing viscosity at the interface. enhances NZVI mobility. Enhanced viscosity in confi ned space decreases the diff usion In surface waters, homoaggregation can enhance sedimen- rate of water to the surface and therefore the dissolution rate. tation if aggregates or agglomerates reach a size where gravity Th ey also hypothesized that the overlap of the diff usion layer forces acting on the particles becomes signifi cant compared between two adjacent particles in an aggregate can reduce the with Brownian diff usion (O’Melia, 1980). Nanoparticles with concentration gradient (i.e., the driving force for dissolution) low dispersion stability therefore tend to accumulate in the from NP surfaces to bulk solution. sediment near their source (Fig. 6). Heteroaggregation, with Another example of aggregation decreasing NP reactivity low-density natural colloids, may facilitate stabilization or dis- is the dechlorination of carbon tetrachloride (CT) by 9-nm- aggregation of NPs, which would increase their residence times diameter magnetite NPs (Vikesland et al., 2007). Th e CT in water bodies (Fig. 6). Stabilizing or destabilizing NPs against dechlorination rate decreased as the degree of aggregation aggregation can have far-reaching eff ects. For example, cerium of the magnetite NPs increased. By promoting aggregation oxide NPs stabilized with surfactant remained dispersed as col- through addition of diff erent concentrations of monovalent loids and were not eff ectively removed by an activated sludge or divalent cations, an inverse relationship between aggre- system (Limbach et al., 2008). Additionally, the removal effi - gate size (measured by DLS) and the CT dechlorination rate ciency of functionalized versus nonfunctionalized fullerene was established. NPs by alum coagulation was compared. Increased steric sta- Although aggregation can aff ect NP reactivity, NP reactiv- bilization by surface functionalization prevented aggregation ity can also aff ect aggregation. Clusters of C60 also undergo and attraction with alum fl ocs (Hyung and Kim, 2009). In photochemical oxidation and subsequent disaggregation in contrast, surface coatings, such as Tween 20 on SiO NPs, 2 water (Hou and Jafvert, 2009; Lee et al., 2009). For exam- have been shown to enhance their fl occulation and removal ple, Hou and Jafvert (2009) reported that the photochemical during water treatment, whereas uncoated SiO2 NPs were not transformation of aqueous C60 clusters in sunlight decreased removed (Jarvie et al., 2009). Th e ability to predict the aggre- mean aggregate size from 500 to 160 nm and produced gation eff ects of adsorbed or grafted organic macromolecular unidentifi ed soluble reaction products after 65 d of exposure coatings on NP surfaces is not possible with current DLVO

1920 Journal of Environmental Quality • Volume 39 • November–December 2010 to natural sunlight (Hou and Jafvert, 2009). Lee et al. (2009) ity. For example, 20-nm NPs were found to deposit mostly in observed similar transformations under UV light exposure the alveolar region, whereas 5- to 10-nm particles deposited

(Lee et al., 2009). Photochemical transformations of C60 in tracheobronchial region, and particles smaller than 10 nm also took place during heteroaggregation with NOM (Li et accumulate mostly in the upper respiratory tract (Heyder et al., 2009) and enhanced the dispersion stability of NOM- al., 1986; Oberdorster et al., 2007; Swift et al., 1994). Th is coated C60 in comparison to nonilluminated samples (Li et indicates that aggregation can alter exposure locations or al., 2009). Enhanced dispersion appeared to be due to a sur- exposure pathways. Nanoparticle aggregation was also shown face erosion or dissolution-recrystallization process catalyzed to aff ect the mode of cellular uptake of NPs and subsequent by sunlight, rather than a simple breakage of C60 from clus- biological responses. For example, phagocytosis by macro- ters (Li et al., 2009). Th erefore, NPs designed to be reactive phages and giant cells is a mechanism used by cells to clear under photoilluminated conditions (e.g., C60 or TiO2) could particles of a few microns or less (Oberdorster et al., 2007). have their aggregation properties altered by long-term sun- Phenrat et al. (2009b) found that aggregation aff ected uptake light exposure. of bare and polymer-modifi ed NZVI particles by mammalian Chemical oxidation could act in a similar manner by alter- glial cells. Although the primary particle sizes of bare NZVI ing surface coatings or the underlying NP so that the particles particles (20–100 nm) are smaller than the optimal size range gain or lose stability against aggregation (Fig. 5D). For exam- (1–2 μm) for phagocytosis, the NZVI loosely aggregated to ple, ozone reacts with C60 (Fortner et al., 2007) and CNTs, micrometer-sized clusters in the physiological buff er, trigger- breaking the carbon cage structure and altering the stability of ing phagocytosis and subsequent uptake. Biological eff ects, the NPs against aggregation in aqueous suspension. Th is ozo- therefore, vary depending on whether or not a particle is nation process has been demonstrated to render the oxidized aggregated, among other confounding factors (e.g., endotox-

C60 more photoactive than its unoxidized parent aggregates ins, adsorption of biomacromolecules, etc.). Given this fact, (Cho et al., 2009). Additionally, surfactants and coatings are understanding aggregation in biological systems is essential susceptible to these same types of oxidation processes possi- for elucidating mechanisms of toxicity. bly decreasing the stability of NPs in the environment due to loss of the stabilizing coating. A better understanding of the Conclusions biological and abiotic reactions aff ecting NP stability against Th e ability to manipulate atoms allows scientists to create aggregation is needed to predict their fate in the environment. NPs with signifi cantly diff erent reactivity than their micro- sized analogs (He et al., 2008; Kanehira et al., 2008). Eff ects of Aggregation on Biological Impacts Reports on NP behavior and eff ects to date indicate that from Nanoparticles the aggregation behavior of the NPs will signifi cantly aff ect Aggregation of NPs can alter their biological eff ects by aff ecting their fate, transport, reactivity, bioavailability, and eff ects in ion release from the surface, if and where particles collect inside environmental systems. Fortunately, the science of aggrega- an organism, and the reactive surface area of NPs (Fig. 5C). tion is well developed. Unfortunately, the translation of that Calculation of metal fl ux at cell surfaces is important for deter- science into the regime of nanomaterials remains a challenge mining the bioavailability and ecotoxicity of nanomaterials. Th is due their unique physical and chemical properties (e.g., size, is especially true in the case of silver NPs where silver ion release shape, composition, and reactivity). Th e presence of organic plays an important role in the inhibition of bacterial growth macromolecular coatings on NPs including those engineered (Taurozzi et al., 2008; Ju-Nam and Lead, 2008). Ion release as part of the NP, or those that are acquired in the environ- may also prove to be important for determining the impacts of ment (e.g., adsorption of NOM), further complicate analysis quantum dots and iron NPs, among others. Heteroaggregation of NP behavior by traditional colloid science. Nanomaterial can confound understanding of metal fl ux because the metal aggregation studies, therefore, aim to take into account can also bind to surfaces and internal sites of naturally occur- how these types of novel surface properties will alter what ring colloids of various sizes and compositions (Buffl e et al., is already understood about aggregation science. Ultimately, 2007). Solution chemistry (e.g., ionic strength and the amount this will guide scientists’ grasp of how nanomaterials and and type of NOM) controls the fractal dimension of NP aggre- NPs will occur in and aff ect environmental systems. Much work remains to gain a full understanding of NP aggregation gates, with lower Df (1.3–2.0) formed in solutions with high and how to apply these principles to create better applica- NOM, and higher Df (2.3–2.7) formed in solutions without signifi cant NOM present (Zhang et al., 2007). Aggregate struc- tions and lower impacts of nanomaterials. ture plays an important role in how metal ions diff use inside of the fl ocs where the kinetics of metal association and dissociation Acknowledgments from binding sites is slowed by orders of magnitude compared Th is material is based on work supported by the National Science with rates expected for free ions in solution (Buffl e et al., 2007). Foundation (NSF) and the Environmental Protection Agency (EPA) Understanding mechanisms by which aggregate structure aff ects under NSF Cooperative Agreement EF-0830093, Center for the Environmental Implications of NanoTechnology (CEINT). Any the dissolution and release of ions and resulting ecotoxicity (e.g., opinions, fi ndings, conclusions or recommendations expressed in this from silver NPs) will improve insights into the potential impacts material are those of the author(s) and do not necessarily refl ect the of such particles in the environment. views of the NSF or the EPA. Th is work has not been subjected to Aggregation alters the NP size distribution, in turn aff ect- EPA review, and no offi cial endorsement should be inferred. ing the fate of NPs in biological systems as well as their toxic-

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1921 References 43:1282–1286. Elena, D.V., M. Coisson, C. Appino, F. Vinai, and R. Sethi. 2009. Magnetic Alivisatos, A. 2001. Less is more in medicine: Sophisticated forms of nanotech- characterization and interaction modeling of zerovalent iron nanopar- nology will fi nd some of their fi rst real-world applications in biomedical ticles for the remediation of contaminated aquifers. J. Nanosci. Nano- research, disease diagnosis and, possibly, therapy. Sci. Am. 285:66–73. technol. 9:3210–3218. Amirbahman, A., and T.M. Olson. 1993. Transport of humic matter-coated Elimelech, M., and C. O’Melia. 1990. Eff ect of particle size on collision ef- hematite in packed beds. Environ. Sci. Technol. 27:2807–2813. fi ciency in the deposition of Brownian particles with electrostatic energy Amirbahman, A., and T.M. Olson. 1995. Th e role of surface conformations in barriers. Langmuir 6:1153–1163. the deposition kinetics of humic matter-coated colloids in porous media. Evans, D.F., and H. Wennerstrom. 1999. Th e colloidal domain: Where phys- Colloids Surf. A Physicochem. Eng. Asp. 95:249–259. ics, chemistry, biology, and technology meet. Wiley-VCH, New York. Barbot, E., P. Dussouillez, J.-Y. Bottero, and P. Moulin. 2010. Coagulation of Fleer, G.J., M.A. Cohen Stuart, J.M.H.M. Scheutjens, T. Cosgrove, and B. bentonite suspension by polyelectrolytes or ferric chloride: Floc breakage Vincent. 1993. Polymers at interfaces. Chapman & Hall, London, UK. and reformation. Chem. Eng. J. 156:83–91. Fortner, J.D., D.-I. Kim, A.M. Boyd, J.C. Falkner, S. Moran, V.L. Colvin, J.B. Baveye, P., and M. Laba. 2008. Aggregation and toxicology of titanium dioxide Hughes, and J.-H. Kim. 2007. Reaction of water-stable C60 aggregates nanoparticles. Environ. Health Perspect. 116:A152–A152. with ozone. Environ. Sci. Technol. 41:7497–7502. Benn, T.M., and P. Westerhoff . 2008. Nanoparticle silver released into wa- French, R.A., A.R. Jacobson, B. Kim, S.L. Isley, R.L. Penn, and P.C. Baveye. ter from commercially available sock fabrics. Environ. Sci. Technol. 2009. Infl uence of ionic strength, pH, and cation valence on aggrega- 42:4133–4139. tion kinetics of titanium dioxide nanoparticles. Environ. Sci. Technol. Bhattacharjee, S., and M. Elimelech. 1997. Surface element integration: A 43:1354–1359. novel technique for evaluation of DLVO interaction between a particle Fritz, G., V. Schadler, N. Willenbacher, and N. Wagner. 2002. Electrosteric and a fl at plate. J. Colloid Interface Sci. 193:273–285. stabilization of colloidal dispersions. Langmuir 18:6381–6390. Brant, J., H. Lecoanet, M. Hotze, and M. Wiesner. 2005. Comparison of Gatica, S., M. Cole, and D. Velegol. 2005. Designing van der Waals forces electrokinetic properties of colloidal fullerenes (nC60) formed using two between nanocolloids. Nano Lett. 5:169–173. procedures. Environ. Sci. Technol. 39:6343–6351. Graf, P., A. Mantion, A. Foelske, A. Shkilnyy, A. Masic, A.E. Th uenemann, Brant, J.A., J. Labille, C.O. Robichaud, and M. Wiesner. 2007. Fullerol cluster and A. Taubert. 2009. Peptide-coated silver nanoparticles: Synthesis, formation in aqueous solutions: Implications for environmental release. surface chemistry, and pH-triggered, reversible assembly into particle as- J. Colloid Interface Sci. 314:281–288. semblies. Chem. Eur. J. 15:5831–5844. Brewer, S.H., W.R. Glomm, M.C. Johnson, M.K. Knag, and S. Franzen. Gu, B., J. Schmitt, Z. Chen, L. Liang, and J.F. McCarthy. 1994. Adsorption 2005. Probing BSA binding to citrate-coated gold nanoparticles and sur- and desorption of natural organic matter on iron oxide: Mechanisms and faces. Langmuir 21:9303–9307. models. Environ. Sci. Technol. 28:38–46. Buffl e, J., and G. Leppard. 1995. Characterization of aquatic colloids and mac- Hahn, M., D. Abadzic, and C. O’Melia. 2004. Aquasols: On the role of sec- romolecules: 1. Structure and behavior of colloidal material. Environ. ondary minima. Environ. Sci. Technol. 38:5915–5924. Sci. Technol. 29:2169–2175. Hansen, P., M. Malmsten, B. Bergenstahl, and L. Bergstrom. 1999. Ortho- Buffl e, J., K.J. Wilkinson, S. Stoll, M. Filella, and S. Zhang. 1998. A gener- kinetic aggregation in two dimensions of monodisperse and bidisperse alized description of aquatic colloidal interactions: Th e three-colloidal colloidal systems. J. Colloid Interface Sci. 220:269–280. component approach. Environ. Sci. Technol. 32:2887–2899. He, F., D. Zhao, J. Liu, and C.B. Roberts. 2007. Stabilization of Fe-Pd Buffl e, J., Z. Zhang, and K. Startchev. 2007. Metal fl ux and dynamic spe- nanoparticles with sodium carboxymethyl cellulose for enhanced trans- ciation at (bio)interfaces. Part I: Critical evaluation and compilation of port and dechlorination of trichloroethylene in soil and groundwater. physicochemical parameters for complexes with simple ligands and ful- Ind. Eng. Chem. Res. 46:29–34. vic/humic substances. Environ. Sci. Technol. 41:7609–7620. He, F., M. Zhang, T. Qian, and D. Zhao. 2009. Transport of carboxymethyl Cai, Y., and K.H. Sandhage. 2005. Zn2SiO4-coated microparticles with bio- cellulose stabilized iron nanoparticles in porous media: Column experi- logically-controlled 3D shapes. Phys Status Solidi A 202:R105–R107. ments and modeling. J. Colloid Interface Sci. 334:96–102. Chakraborti, R., K. Gardner, J. Atkinson, and J. Van Benschoten. 2003. Chang- He, Y.T., J. Wan, and T. Tokunaga. 2008. Kinetic stability of hematite es in fractal dimension during aggregation. Water Res. 37:873–883. nanoparticles: Th e eff ect of particle sizes. J. Nanopart. Res. 10:321–332. Chappell, M.A., A.J. George, K.M. Dontsova, B.E. Porter, C.L. Price, P. Zhou, Hermansson, M. 1999. Th e DLVO theory in microbial adhesion. Colloids E. Morikawa, A.J. Kennedy, and J.A. Steevens. 2009. Surfactive stabiliza- Surf. B Biointerfaces 14:105–119. tion of multi-walled carbon nanotube dispersion with dissolved humic Heyder, J., J. Gebhart, G. Rudolf, and C. Schiller. 1986. Deposition of par- substances. Environ. Pollut. 157:1081–1087. ticles in the human respiratory tract in the size range 0, 005–15 μm. J. Chen, K., and M. Elimelech. 2007. Infl uence of humic acid on the aggrega- Aerosol Sci. 17:811–825. tion kinetics of fullerene (C ) nanoparticles in monovalent and divalent 60 Hezinger, A.F.E., J. Tessmar, and A. Göpferich. 2008. Polymer coating of electrolyte solutions. J. Colloid Interface Sci. 309:126–134. quantum dots: A powerful tool toward diagnostics and sensorics. Eur. J. Chen, K.L., and M. Elimelech. 2009. Relating colloidal stability of fullerene Pharm. Biopharm. 68:138–152. (C ) nanoparticles to nanoparticle charge and electrokinetic properties. 60 Hiemenz, P.C., and R. Rajagopalan. 1997. Principles of colloid and surface Environ. Sci. Technol. 43:7270–7276. chemistry. CRC, New York. Chen, K.-L., S.E. Mylon, and M. Elimelech. 2006. Aggregation kinetics of Hochella, M.F., S.K. Lower, P.A. Maurice, R.L. Penn, N. Sahai, D.L. Sparks, alginate-coated hematite nanoparticles in monovalent and divalent elec- and B.S. Twining. 2008. Nanominerals, mineral nanoparticles, and earth trolyte. Environ. Sci. Technol. 40:1516–1523. systems. Science 21:1631–1635. Cheng, X.K., A.T. Kan, and M.B. Tomson. 2004. Naphthalene adsorption and Hoek, E., and G. Agarwal. 2006. Extended DLVO interactions between spher- desorption from aqueous C fullerene. J. Chem. Eng. Data 49:675–683. 60 ical particles and rough surfaces. J. Colloid Interface Sci. 298:50–58. Cho, M., J.D. Fortner, J.B. Hughes, and J.-H. Kim. 2009. Escherichia coli Hotze, E.M., J. Labille, P. Alvarez, and M.R. Wiesner. 2008. Mechanisms of inactivation by water-soluble, ozonated C derivative: Kinetics and 60 photochemistry and reactive oxygen production by fullerene suspensions mechanisms. Environ. Sci. Technol. 43:7410–7415. in water. Environ. Sci. Technol. 42:4175–4180. Dederichs, T., M. Möller, and O. Weichold. 2009. Colloidal stability of hy- Hou, W.-C., and C.T. Jafvert. 2009. Photochemical transformation of aqueous drophobic nanoparticles in ionic surfactant solutions: Defi nition of the C clusters in sunlight. Environ. Sci. Technol. 43:362–367. critical dispersion concentration. Langmuir 25:2007–2012. 60 Hydutsky, B.W., E.J. Mack, B.B. Beckerman, J.M. Skluzacek, and T.E. Mal- Delgado, A.V., F. Gonzalez-Caballero, R.J. Hunter, L.K. Koopal, and J. Lykl- louk. 2007. Optimization of nano- and microiron transport through ema. 2007. Measurement and interpretation of electrokinetic phenom- sand columns using polyelectrolyte mixtures. Environ. Sci. Technol. ena. J. Colloid Interface Sci. 309:194–224. 41:6418–6424. Derjaguin, B.V., and L. Landau. 1941. Th eory of the stability of strongly Hyung, H., J. Fortner, J. Hughes, and J. Kim. 2007. Natural organic matter charged lyophobic sols and of the adhesion of strongly charged particles stabilizes carbon nanotubes in the aqueous phase. Environ. Sci. Technol. in solutions of electrolytes. Acta Phys. Chim. URSS 14:633–662. 41:179–184. Diegoli, S., A.L. Manciulea, S. Begum, I.P. Jones, J.R. Lead, and J.A. Preece. Hyung, H., and J.-H. Kim. 2009. Dispersion of C60 in natural water and re- 2008. Interaction between manufactured gold nanoparticles and natu- moval by conventional drinking water treatment processes. Water Res. rally occurring organic macromolecules. Sci. Total Environ. 402:51–61. 43:2463–2470. Domingos, R.F., N. Tufenkji, and K.J. Wilkinson. 2009. Aggregation of tita- Israelachvili, J.N. 1991. Intermolecular & surface forces. Academic Press, nium dioxide nanoparticles: Role of a fulvic acid. Environ. Sci. Technol.

1922 Journal of Environmental Quality • Volume 39 • November–December 2010 New York. synthesis and formation mechanism of nanoparticles-assembled Ag2S Jain, T.K., M.A. Morales, S.K. Sahoo, D.L. Leslie-Pelecky, and V. Labhaset- nanorods and nanotubes. J. Cryst. Growth 304:163–168. war. 2005. Iron oxide nanoparticles for sustained delivery of anticancer Mayya, K.S., B. Schoeler, and F. Caruso. 2003. Preparation and organization agents. Mol. Pharm. 2:194–205. of nanoscale polyelectrolyte-coated gold nanoparticles. Adv. Funct. Ma- Jarvie, H., H. Al-Obaidi, S. King, M. Bowes, M. Lawrence, A. Drake, M. ter. 13:183–188. Green, P. Dobson. 2009. Fate of silica nanoparticles in simulated pri- McKeon, K.D., and B.J. Love. 2008. Th e presence of adsorbed proteins on mary wastewater treatment. Environ. Sci. Technol. 43:8622–8628. particles increases aggregated particle sedimentation, as measured by a Jiang, J., G. Oberdorster, and P. Biswas. 2009. Characterization of size, surface light scattering technique. J. Adhes. 84:664–674. charge, and agglomeration state of nanoparticle dispersions for toxico- Montgomery, S., M. Franchek, and V. Goldschmidt. 2000. Analytical disper- logical studies. J. Nanopart. Res. 11:77–89. sion force calculations for nontraditional geometries. J. Colloid Interface Johnson, R.L., G.O. Johnson, J.T. Nurmi, and P.G. Tratnyek. 2009. Natural Sci. 227:567–584. organic matter enhanced mobility of nano zerovalent iron. Environ. Sci. Moon, S.Y., T. Kusunose, and T. Sekino. 2009. CTAB-assisted synthesis of Technol. 43:5455–5460. size- and shape-controlled gold nanoparticles in SDS aqueous solution. Ju-Nam, Y., and J.R. Lead. 2008. Manufactured nanoparticles: An overview of Mater. Lett. 63:2038–2040. their chemistry, interactions and potential environmental implications. Moore, V.C., M.S. Strano, E.H. Haroz, R.H. Hauge, R.E. Smalley, J. Schmidt, Sci. Total Environ. 400:396–414. and Y. Talmon. 2003. Individually suspended single-walled carbon Kanehira, K., T. Banzai, C. Ogino, N. Shimizu, Y. Kubota, and S. Sonezaki. nanotubes in various surfactants. Nano Lett. 3:1379–1382.

2008. Properties of TiO2–polyacrylic acid dispersions with potential for Moreau, J.W., P.K. Weber, M.C. Martin, B. Gilbert, I.D. Hutcheon, and J.F. molecular recognition. Colloids Surf. B Biointerfaces 64:10–15. Banfi eld. 2007. Extracellular proteins limit the dispersal of biogenic Kanel, S.R., R.R. Goswami, T.P. Clement, M.O. Barnett, and D. Zhao. nanoparticles. Science 316:1600–1603. 2008. Two dimensional transport characteristics of surface stabilized O’Connell, M.J., S.M. Bachilo, C.B. Huff man, V.C. Moore, M.S. Strano, E.H. zero-valent iron nanoparticles in porous media. Environ. Sci. Technol. Haroz, K.L. Rialon, P.J. Boul, W.H. Noon, C. Kittrell, J.P. Ma, R.H. 42:896–900. Hauge, R.B. Weisman, and R.E. Smalley. 2002. Band gap fl uorescence Kawano, T., and I. Hiroaki. 2008. A simple preparation technique for shape- from individual single-walled carbon nanotubes. Science 297:593–596. controlled zinc oxide nanoparticles: Formation of narrow size-distrib- O’Melia, C.R. 1980. ES&T features. Aquasols: Th e behavior of small particles uted nanorods using seeds in aqueous solutions. Colloid Surface A in aquatic systems. Environ. Sci. Technol. 14:1052–1060. 319:130–135. Oberdorster, G., V. Stone, and K. Donaldson. 2007. Toxicology of nanopar- Kim, B., and W.M. Sigmund. 2004. Functionalized multiwall carbon nano- ticles: A historical perspective. Nanotoxicology 1:2–25. tube/gold nanoparticle composites. Langmuir 20:8239–8242. Ortega-Vinuesa, J., A. Martín-Rodríguez, and R. Hidalgo-Alvarez. 1996. Col- Kim, H.-J., T. Phenrat, R.D. Tilton, and G.V. Lowry. 2009. Fe0 nanoparticles loidal stability of polymer colloids with diff erent interfacial properties: remain mobile in porous media after aging due to slow desorption of Mechanisms. J. Colloid Interface Sci. 184:259–267. polymeric surface modifi ers. Environ. Sci. Technol. 43:3824–3830. Phenrat, T., H.-J. Kim, F. Fagerlund, T. Illangasekare, R.D. Tilton, and G.V. Kim, J., D. Kim, H. Jung, and K. Hong. 2005. Infl uence of anatase-rutile Lowry. 2009a. Particle size distribution, concentration, and magnetic at- 0 phase transformation on dielectric properties of sol-gel derived TiO2 thin traction aff ects transport of polymer-modifi ed Fe nanoparticles in sand fi lms. Jpn. J. Appl. Phys. 1:6148–6151. columns. Environ. Sci. Technol. 43:5079–5085. Koponen, I.T. 2009. Cluster growth poised on the edge of break-up, II: From Phenrat, T., T. Long, G. Lowry, and B. Veronesi. 2009b. Partial oxidation (“ag- reaction kinetics to thermodynamics. Physica A 388:2659–2665. ing”) and surface modifi cation decrease the toxicity of nanosized zerova- Kosmulski, M. 2009. pH-dependent surface charging and points of zero charge. lent iron. Environ. Sci. Technol. 43:195–200. IV. Update and new approach. J. Colloid Interface Sci. 337:439–448. Phenrat, T., Y. Liu, R. Tilton, and G. Lowry. 2009c. Adsorbed polyelectrolyte Kozan, M., J. Th angala, R. Bogale, M.P. Menguc, and M.K. Sunkara. 2008. coatings decrease Fe0 nanoparticle reactivity with TCE in water: Con- ceptual model and mechanisms. Environ. Sci. Technol. 43:1507–1514. In-situ characterization of dispersion stability of WO3 nanoparticles and nanowires. J. Nanopart. Res. 10:599–612. Phenrat, T., N. Saleh, K. Sirk, H. Kim, R. Tilton, and G. Lowry. 2008. Stabilization of aqueous nanoscale zerovalent iron dispersions by an- Lebrette, S., C. Pagnoux, and P. Abélard. 2004. Stability of aqueous TiO2 suspensions: Infl uence of ethanol. J. Colloid Interface Sci. 280:400–408. ionic polyelectrolytes: Adsorbed anionic polyelectrolyte layer proper- Lee, J., M. Cho, J.D. Fortner, J.B. Hughes, and J.-H. Kim. 2009. Transforma- ties and their eff ect on aggregation and sedimentation. J. Nanopart. Res. 10:795–814. tion of aggregated C60 in the aqueous phase by UV irradiation. Environ. Sci. Technol. 43:4878–4883. Phenrat, T., N. Saleh, K. Sirk, R.D. Tilton, and G.V. Lowry. 2007. Aggrega- tion and sedimentation of aqueous nanoscale zerovalent iron dispersions. Li, Q., B. Xie, Y.S. Hwang, and Y. Xu. 2009. Kinetics of C60 fullerene dispersion in water enhanced by natural organic matter and sunlight. Environ. Environ. Sci. Technol. 41:284–290. Sci. Technol. 43:3574–3579. Project on Emerging Nanotechnologies. 2009. Nanotechnology consumer Li, Z., K. Greden, P. Alvarez, K. Gregory, and G.V. Lowry. 2010. Adsorbed products inventory. Available at http://www.nanotechproject.org (veri- polymer and NOM limits adhesion and toxicity of nano scale zero-valent fi ed 22 Apr. 2010). Project on Emerging Nanotechnologies, Washing- iron (NZVI) to E. coli. Environ. Sci. Technol. (in press). ton, DC. Limbach, L.K., R. Bereiter, E. Muller, R. Krebs, R. Glli, and W.J. Stark. 2008. Richard, C., F. Balavoine, P. Schultz, T.W. Ebbesen, and C. Mioskowski. 2003. Removal of oxide nanoparticles in a model wastewater treatment plant: Supramolecular self-assembly of lipid derivatives on carbon nanotubes. Infl uence of agglomeration and surfactants on clearing effi ciency. Envi- Science 300:775–778. ron. Sci. Technol. 42:5828–5833. Romero-Cano, M., A. Martin-Rodriguez, and F. de las Nieves. 2001. Elec- Limbach, L.K., Y.C. Li, R.N. Grass, T.J. Brunner, M.A. Hintermann, M. trosteric stabilization of polymer colloids with diff erent functionality. Muller, D. Gunther, and W.J. Stark. 2005. Oxide nanoparticle uptake Langmuir 17:3505–3511. in human lung fi broblasts: Eff ects of particle size, agglomeration, and Rosen, M.J. 2004. Surfactants and interfacial phenomena. 3rd ed. Wiley-In- diff usion at low concentrations. Environ. Sci. Technol. 39:9370–9376. terscience, New York. Liu, J., D.M. Aruguete, J.R. Jinschek, J.D. Rimstidt, and M.F. Hochella. Saleh, N., H.-J. Kim, T. Phenrat, K. Matyjaszewski, R.D. Tilton, and G.V. 2008. Th e non-oxidative dissolution of galena nanocrystals: Insights into Lowry. 2008. Ionic strength and composition aff ect the mobility of mineral dissolution rates as a function of grain size, shape, and aggrega- surface-modifi ed Fe0 nanoparticles in water-saturated sand columns. En- tion state. Geochim. Cosmochim. Acta 72:5984–5996. viron. Sci. Technol. 42:3349–3355. Liu, Y., S.A. Majetich, R.D. Tilton, D.S. Sholl, and G.V. Lowry. 2005. TCE Saleh, N., T. Phenrat, K. Sirk, B. Dufour, J. Ok, T. Sarbu, K. Matyiaszewski, dechlorination rates, pathways, and effi ciency of nanoscale iron particles R.D. Tilton, and G.V. Lowry. 2005. Adsorbed triblock copolymers de- with diff erent properties. Environ. Sci. Technol. 39:1338–1345. liver reactive iron nanoparticles to the oil/water interface. Nano Lett. Liu, Z., W. Cai, L. He, N. Nakayama, and K. Chen. 2006. In vivo biodistribu- 5:2489–2494. tion and highly effi cient tumour targeting of carbon nanotubes in mice. Saleh, N., K. Sirk, Y. Liu, T. Phenrat, B. Dufour, K. Matyjaszewski, R.D. Til- Nature Nanotechnol. 2:47–52. ton, and G.V. Lowry. 2007. Surface modifi cations enhance nanoiron Luccardini, C., C. Tribet, F. Vial, V. Marchi-Artzner, and M. Dahan. 2006. transport and DNAPL targeting in saturated porous media. Environ. Size, charge, and interactions with giant lipid vesicles of quantum dots Eng. Sci. 24:45–57. coated with an amphiphilic macromolecule. Langmuir 22:2304–2310. Sanchez-Iglesias, A., M. Grzelczak, B. Rodriguez-Gonzalez, R.A. Alvarez- Ma, D., X. Hu, H. Zhou, J. Zhang, and Y. Qian. 2007. Shape-controlled Puebla, L.M. Liz-Marzan, and N.A. Kotov. 2009. Gold colloids with unconventional angled shapes. Langmuir 25:11431–11435.

Hotze et al.: Nanoparticle Aggregation: Environmental Transport and Reactivity 1923 Sirk, K.M., N.B. Saleh, T. Phenrat, H.-J. Kim, B. Dufour, J. Ok, P.L. Golas, K. titanium dioxide nanoparticles. Int. J. Nanomedicine 3:69–74. Matyjaszewski, G.V. Lowry, and R.D. Tilton. 2009. Eff ect of adsorbed Verwey, E., J. Overbeek, and K. van Nes. 1948. Th e theory of the stability polyelectrolytes on nanoscale zero valent iron particle attachment to soil of liophobic colloids: Th e interaction of sol particles having an electric surface models. Environ. Sci. Technol. 43:3803–3808. double layer. Elsevier, Amsterdam. Smith, B., K. Wepasnick, K.E. Schrote, H.-H. Cho, W.P. Ball, and D.H. Fair- Vikesland, P.J., A.M. Heathcock, R.L. Rebodos, and K.E. Makus. 2007. Par- brother. 2009. Infl uence of surface oxides on the colloidal stability of ticle size and aggregation eff ects on magnetite reactivity toward carbon multi-walled carbon nanotubes: A structure-property relationship. Lang- tetrachloride. Environ. Sci. Technol. 41:5277–5283. muir 25:9767–9776. Viota, J.L., F. González-Caballero, J.D.G. Durán, and A.V. Delgado. 2007. Sondi, I., D.V. Goia, and E. Matijevic. 2003. Preparation of highly concen- Study of the colloidal stability of concentrated bimodal magnetic fl uids. trated stable dispersions of uniform silver nanoparticles. J. Colloid Inter- J. Colloid Interface Sci. 309:135–139. face Sci. 260:75–81 . Vold, M.J. 1954. Van der Waals’ attraction between anisometric particles. J. Stumm, W., and J.J. Morgan. 1996. Aquatic chemistry: Chemical equilibria Colloid Sci. 9:451–459. and rates in natural waters. 3rd ed. John Wiley & Sons, New York. Wang, G.-H., W.-C. Li, K.-M. Jia, B. Spliethoff , F. Schüth, and A.-H. Lu. Swift, D., Y. Cheng, Y. Su, and H. Yeh. 1994. Ultrafi ne aerosol deposition in 2009. Shape and size controlled -Fe2O3 nanoparticles as supports for the human nasal and oral passages. Ann. Occup. Hyg. 38:77–81. gold-catalysts: Synthesis and infl uence of support shape and size on cata- Tabor, D., and R. Winterton. 1969. Th e direct measurement of normal and lytic performance. Appl. Catal. A 364:42–47. retarded van der Waals forces. Proceed. R. Soc. London A. 312:435–450. Wang, W., B. Gu, L. Liang, and W.A. Hamilton. 2004. Adsorption and Taurozzi, J.S., H. Arul, V.Z. Bosak, A.F. Burban, T.C. Voice, M.L. Bruen- structural arrangement of cetyltrimethylammonium cations at the silica ing, and V.V. Tarabara. 2008. Eff ect of fi ller incorporation route on the nanoparticle-water interface. J. Phys. Chem. B 108:17477–17483. properties of polysulfone-silver nanocomposite membranes of diff erent Wu, W., R.F. Giese, and C.J. van Oss. 1999. Stability versus fl occulation of porosities. J. Membr. Sci. 325:58–68. particle suspensions in water-correlation with the extended DLVO ap- Th ill, A., O. Zeyons, O. Spalla, F. Chauvat, J. Rose, M. Auff an, and A.M. proach for aqueous systems, compared with classical DLVO theory. Col-

Flank. 2006. Cytotoxicity of CeO2 nanoparticles for Escherichia coli: loids Surf. B Biointerfaces 14:47–55. Physico-chemical insight of the cytotoxicity mechanism. Environ. Sci. Yurekli, K., C.A. Mitchell, and R. Krishnamoorti. 2004. Small-angle neutron Technol. 40:6151–6156. scattering from surfactant-assisted aqueous dispersions of carbon nano- Th ompson Reuters. 2009. Web of knowledge. Available at http://WebofKnowl- tubes. J. Am. Chem. Soc. 126:9902–9903. edge.com (verifi ed 22 Apr. 2009). Th ompson Reuters, Philadelphia, PA. Zhang, D., O. Neumann, H. Wang, V.M. Yuwono, A. Barhoumi, M. Per- Tiraferri, A., and R. Sethi. 2009. Enhanced transport of zerovalent iron ham, J.D. Hartgerink, P. Wittung-Stafshede, and N.J. Halas. 2009. Gold nanoparticles in saturated porous media by guar gum. J. Nanopart. Res. nanoparticles can induce the formation of protein-based aggregates at 11:635–645. physiological pH. Nano Lett. 9:666–671. Tohver, V., J.E. Smay, A. Braem, P.V. Braun, and J.A. Lewis. 2001. Nanopar- Zhang, Y., N. Kohler, and M. Zhang. 2002. Surface modifi cation of super- ticle halos: A new colloid stabilization mechanism. Proc. Natl. Acad. Sci. paramagnetic magnetite nanoparticles and their intracellular uptake. USA 98:8950–8954. Biomaterials 23:1553–1561. Vaisman, L., H.D. Wagner, and G. Marom. 2006. Th e role of surfac- Zhang, Z., J. Buffl e, and D. Alemani. 2007. Metal fl ux and dynamic speciation tants in dispersion of carbon nanotubes. Adv. Colloid Interface Sci. at (bio)interfaces. Part II: Evaluation and compilation of physicochemi- 128–130:37–46. cal parameters for complexes with particles and aggregates. Environ. Sci. Vamunu, C.I., P.J. Hol, Z.E. Allouni, S. Elsayed, and N.R. Gjerdet. 2008. Technol. 41:7621–7631. Formation of potential titanium antigens based on protein binding to

1924 Journal of Environmental Quality • Volume 39 • November–December 2010