commercial titanium dioxide nanoparticles in both natural and synthetic water: comprehensive...

Published: October 20, 2011

r 2011 American Chemical Society 10045 dx.doi.org/10.1021/es2023225 | Environ. Sci. Technol. 2011, 45, 10045–10052

ARTICLE

pubs.acs.org/est

Commercial Titanium Dioxide Nanoparticles in Both Natural andSynthetic Water: Comprehensive Multidimensional Testing andPrediction of Aggregation BehaviorStephanie Ottofuelling, Frank Von Der Kammer,* and Thilo Hofmann*

Department of Environmental Geosciences, University of Vienna, Althanstrasse 14, Vienna 1090, Austria

bS Supporting Information

’ INTRODUCTION

Engineered nanomaterials and nanoparticles (ENPs) are alreadybeing emitted and can be found in the natural environment.1,2

The continuous release of ENPs into the environment raisesconcerns about their distribution, the adverse effects that theymay have on either individual organisms or entire ecosystems,3

and the cotransport of classical contaminants4 in surface waterand groundwater.5 Some TiO2 ENPs have already been shown tohave adverse effects on organisms in aquatic environments.6�8

The most important distribution pathway for TiO2 ENPs intothe environment is thought to be through wastewater treatmentplants (WWTPs).9 Such emissions will then impact the aquaticenvironment, except for the fraction that is retained in sludgewhich, if later used in agriculture, may impact on agricultural soilsystems.2,10,11 It is generally accepted that a detailed understand-ing of the behavior on ENPs in aquatic environments is crucialif a comprehensive assessment is to be made of their final distri-bution, the most important sinks, and the associated risks.12

We therefore propose that studies on the dispersion stabilityof ENPs under conditions that mimic real-world aquatic chem-istries should be an essential part of their general characteriza-tion. The results of such studies would enable the transport andfate of ENPs within natural aquatic systems to be predicted andthe results of toxicological tests to be accurately evaluated andcompared.12 The possible effects resulting from the behavior ofnanoparticles within the environment could thus be identified

and potentially minimized by modifying the ENPs before theyare able to cause any harm to humans or to the environment.

Recent studies have made use of various analytical techniquesand experiments to describe and understand the behavior ofENPs in synthetic growthmedia,13 natural waters,7 wastewater,14

and well-controlled synthetic waters.15 Results have indicatedthat the behavior, transport, and fate of nanoparticles in aqueoussystems are controlled both by their surface properties and by thechemistry of the aqueous system,8,16�18 but also that the investi-gated systems are highly complex and not yet understood indetail. Factors such as electrolyte concentration, the valence ofions countering the surface charge, the pH of the system, and thepresence of organic matter will cause either aggregation or stabi-lization of ENPs.16 For example, the presence of Ca2+ could dimi-nish or even cancel out the stabilizing effects of natural organicmatter.19Water constituents such as NOMor polyvalent ions canenhance, decrease, neutralize, or even reverse the surface chargeof certain nanoparticles through specific interactions (e.g., ad-sorption).18 Expanding or compressing the electrostatic doublelayer by decreasing or increasing the ionic strength (IS) will alsoalter the stability.20 In addition, specific adsorption of ions (e.g.,Ca2+, SO4

2�) may induce nanoparticle aggregation;15,21 divalent

Received: August 1, 2011Accepted: October 20, 2011Revised: October 18, 2011

ABSTRACT: Engineered nanoparticles (ENPs) from industrial applications and consumer productsare already being released into the environment. Their distribution within the environment is, amongother factors, determined by the dispersion state and aggregation behavior of the nanoparticles and, inturn, directly affects the exposure of aquatic organisms to EPNs. The aggregation behavior (or colloidalstability) of these particles is controlled by the water chemistry and, to a large extent, by the surfacechemistry of the particles. This paper presents results from extensive colloidal stability tests oncommercially relevant titanium dioxide nanoparticles (Evonik P25) in well-controlled synthetic waterscovering a wide range of pH values and water chemistries, and also in standard synthetic (EPA) watersand natural waters. The results demonstrate in detail the dependency of TiO2 aggregation on the ionicstrength of the solution, the presence of relevant monovalent and divalent ions, the presence andcopresence of natural organic matter (NOM), and of course the pH of the solution. Specificinteractions of both NOM and divalent ions with the TiO2 surfaces modify the chemistry of thesesurfaces resulting in unexpected behavior. Results from matrix testing in well-controlled batch systemsallow predictions to be made on the behavior in the broader natural environment. Our study provides the basis for a testing schemeand data treatment technique to extrapolate and eventually predict nanoparticle behavior in a wide variety of natural waters.

10046 dx.doi.org/10.1021/es2023225 |Environ. Sci. Technol. 2011, 45, 10045–10052

Environmental Science & Technology ARTICLE

ions screen the surface charge more efficiently than monovalentions,22 which is reflected in much lower critical coagulationconcentrations (roughly 100 times lower) for divalent ions thanfor monovalent ions.17 The surface charge of oxide nanoparticlesis controlled by the protonation or deprotonation of surface hy-droxyl groups and the specific adsorption of, for example, carbo-nate, sulfate, Ca2+, or NOM. It is, for example, well-known thatthe surface charge of TiO2 varies with the pH, and the point ofzero net proton charge can range between 4.8 and 5.9.23 Aggre-gation of particles is promoted at the pznpc24 due to the absenceof electrostatic repulsion.

To understand the behavior and characteristics of ENPs inaquatic systems researchers have conducted experiments in bothcomplex natural7,14 and synthetic waters.25,26 Although resultsfrom natural waters may be more realistic they are often unableto provide information on the processes involved due to thecomplexity of the water chemistry. Results from synthetic waters,with their reduced complexity, can help in understanding thebasic principles but are not necessarily representative of naturalsystems.27 In addition, quantitative descriptors can be deducedfrom dynamic studies28,29 that, depending on the experimentalapproach used, determine doublet formation, particle growthrate constants, or sedimentation rates.14,24

The objective of this study was to overcome the currentlimitations by investigating the effects that typical surface waterconstituents have on the stability of TiO2 nanoparticles, and tolink these synthetic results to the real-world behavior of theparticles. We performed static stability tests and selected NaCl,CaCl2, Na2SO4, and CaSO4 as typical, environmentally relevant,1:1, 2:1, 1:2 ,and 2:2 electrolytes. The effect of NOM and theeffect of CaCl2 in the presence of small but realistic concentra-tions of NOM were also investigated. Concentrations rangedfrom 10 μmol L�1 to 500 mmol L�1 and pH values ranged from4 to 8.

We hypothesized that the contour-type stability plots ob-tained would reveal zones of TiO2 nanoparticle stability and inst-ability, which would then enable the colloidal stability of TiO2

under real conditions to be estimated. This was investigated byfirst testing TiO2 stability in some typical surface waters, in tech-nical waters/effluents, and in synthetic test media with knownhydrochemical compositions. By measuring their water chemis-tries we were then able to relate the real waters to an appropriateset of synthetic waters. Comparing the results from the realsettings with those from the synthetic batch experiments revealeda good general agreement. The results may be used in exposureassessment and modeling.30

’MATERIALS AND METHODS

Nanoparticles.The TiO2 nanoparticles were obtained from abatch of Evonik P25 (produced by Evonik Industries, Germany)for theOrganization for EconomicCooperation andDevelopment(OECD) testing program. A detailed description of the TiO2

material is provided in the Supporting Information (Table S1).Stability Tests.We chose 9 different pH values and 7 different

electrolyte concentrations per single test matrix. To produceeach test matrix we suspended TiO2 in Milli-Q-water, addedthe required amounts of NaCl, CaCl2, Na2SO4, or NOM, andtitrated with HCl or NaHCO3 to the chosen pH values.We followed the titration and sampling protocol described inv.d. Kammer et al.15 For tests with the natural water and theEPA water a stock suspension of 1250 mg L�1 TiO2 was first

ultrasonicated for 30 min. One mL of this suspension was thenadded to 49 mL of natural or EPA water to achieve a TiO2

concentration of 25 mg L�1.Natural and Synthetic Water Samples. Seven natural water

samples were collected from a variety of sources: groundwaterfrom Hoersching (HOE), lake water from Lunz (LUNZ), tapwater from a household tap in Vienna (TAP), water from a peatbog at Tanner Moor (TAN), wastewater inflow and outflowfrom a wastewater treatment plant (WWTP) in Vienna, (WWIand WWO), all of which are in Austria, and seawater from Nor-mandy, France (FRA). Synthetic test water was prepared follow-ing the U.S. Environmental Protection Agency protocol (EPA-821-R-02-013: see Supporting Information, Table S2) and isreferred to in this paper as “EPA water”. All water samples werefiltered through a 0.2-μm cellulose acetate membrane (Whatman,Austria) and stored at 4 �C in the dark. The water chemistry andthe procedure for nanoparticle analyses are described in theSupporting Information.

’RESULTS AND DISCUSSION

Behavior and Characteristics of TiO2 in the Test Matrices.In our investigations we stabilized the pH by adding bicarbonate(as NaHCO3) to establish a pH-buffered environment typical ofmost natural systems The influence that NaHCO3

� on its ownhas on the stability of TiO2 nanoparticles has been previouslytested (Figure S2). At a pH either above or below the IEP(pH 5.0) TiO2 particles (residual concentration 13 mg L

�1 or 52%of the initial concentration, whichwas 25mgL�1 in all experiments)had a diameter of ∼300 nm. At the IEP the particles remaining inthe supernatant (maximum 20% of the initial concentration) werelarge aggregates greater than 2.5 μm in diameter (Figure S3).As can be deduced from the residual TiO2 concentrations in

the supernatant, the stability of TiO2 in the presence of sodiumchloride is influenced in different ways by the NaCl concentra-tion and the pH (Figure 1a and b). At a NaCl concentrationbelow 5 mM the system is predominantly regulated by the pHand the magnitude of the resulting zeta potential. The TiO2

ENPs are relatively stable, with TiO2 concentrations of morethan 20% of the initial concentration at pH values above andbelow the IEP (Figure 1a). As expected, at the IEP the system isdestabilized and particles aggregated and were lost from thesupernatant. At NaCl concentrations above about 5 mM theinfluence of pH is diminished, particles are aggregated to formentities larger than 1.5 μm, and the concentration in the super-natant is reduced (C/C0 10%). The zeta potential plot revealsthat the IEP shifts from pH 5 to pH 6.5 with increasing NaCl(from 0.1 to 500 mmol L�1), indicating that the NaCl doesnot act as a completely indifferent electrolyte but that a weakinteraction of the Na+ with the TiO2 surface is shifting the zetapotential to more positive values at elevated Na+ concentrations(Figure 1b). This interaction could be the result of adsorption onthemineral surface or of neutralization of negative patches, whichproduce elevated positive potentials at intermediate NaCl con-centrations (5� 100 mM). This is accompanied by an increasedTiO2 stability at 10 mM NaCl concentration; it is evident fromconcentration and zeta potential data that this stability increaseoriginates from Na+ interaction and deviates from the expectedplot of a noninteracting electrolyte.The effects of a positively charged divalent cation and a

monovalent anion on TiO2 stability were investigated usingcalcium chloride (Figure 1c and d). In this test matrix we used



concentrations ranging from 0.01 to 10 mmol L�1 due to anexpected lower critical coagulation concentration (CCC) for thedivalent cation. Typical calcium concentrations in the naturalenvironment range from 1.0 mg L�1 in rainwater and 40 mg L�1

in surface water, up to 430 mg L�1 in seawater (Table S3). Themarked shift of the IEP to a higher pH with increasing CaCl2concentration indicates a specific interaction (adsorption) of Ca2+ ions to the titania surface. Increasing the CaCl2 concentrationresulted in a charge reversal at pH values above 6.5 (Figure 1d).At the lowest CaCl2 concentration the stability follows a trendwith pH similar to that followed by NaCl. With greater than0.1 mM CaCl2 and a pH above 5 the TiO2 particles are generallyaggregated (TiO2 particles >1000 nm, Figure S4). However anextended zone of relative stability can be identified at a pH belowand above 5 with up to 40% of stable TiO2 in the supernatant(Figure 1c and d). At a pH below 5 and for Ca2+ concentrationsranging from 0.05 to 5 mmol L�1 the residual TiO2 concentra-tions are greater than 20% (Figure 1c) and the particle diametersare less than 1000 nm (Figure S4). However, at concentrationsgreater than 5 mM the particles again aggregate (>1000 nmdiameter) and there is an accompanying decrease in TiO2 inthe supernatant (<20%). These findings are in agreement withthose of v.d. Kammer et al.15 and Domingos et al.25 but covera broader range of parameters, and the addition of HCO3

� stabi-lizes the pH nicely over the experimental period of 15 h (data notshown)while not changing the overall conclusion compared to thenonbuffered systems. It should be noted that the adsorption of adivalent cation increases the positive surface charge (representedby the zeta potential) and stabilizes the particles, since at pH lowerthan the isoelectric point of TiO2 the counterion here is mono-valent chloride. The enhanced positive surface charge leads to astable dispersed system, which would not usually be expectedunder these conditions.

Sulfate in the form of sodium sulfate was used in concentra-tions ranging from 0.01 to 10 mmol L�1 as a 1:2 electrolyte andtomimic a typically occurring divalent anion (Figure 1e and f). Incontrast to the monovalent Na+ and the divalent Ca2+ cations,the presence of the divalent SO4

2� anion strongly promotesaggregation. Under all tested conditions, about 80% of the initialTiO2 was removed from the supernatant (Figure 1e). The onlyexception to this overall picture occurs at high pH and low SO4

2�

concentrations. Interestingly, the stability plot does not correlateas well with the zeta potential plot as it does in the other cases. Itseems evident that the SO4

2� ion interacts specifically with thetitania surface, resulting in low to elevated negative zeta poten-tials at all tested pH values (Figure 1f) with the IEP at about pH 4.The zeta potential of TiO2 in SO4

2� suspension is almost com-pletely independent of the Na2SO4 concentration, and is mainlycontrolled by the pH. The shift in negative zeta potential (�30mV)toward a minimum (�5 mV) from high to low pH is congruentwith an increase in particle diameter from 1000 to 1600 nm(Figures 1f and S4). A comparison with the NaCl matrix(Figure 1a and b) illustrates the different influences that thechloride and sulfate anions have on TiO2 stability. A charge neu-tralization may be induced at certain concentrations, as has previ-ously been reported for anatase particles in SO4

2� suspension.31

Despite the high negative zeta potential observed, the TiO2 par-ticles were aggregated implying the influence of forces other thanelectrostatic forces such as, for example, the bridging effectsthat have been demonstrated for other nanoparticles.19,21

The combined effects of divalent cations and anions (2:2 elec-trolyte) on the TiO2 stability is represented by the CaSO4 matrixin Figure 1g and h. In this set the concentration of CaSO4 rangesfrom 0.01 to 5 mmol L�1. Only at a pH of less than 5 and CaSO4

concentrations below 0.1 mM are the TiO2 particles less aggre-gated, with diameters of less than 1000 nm and residual C/C0

Figure 1. Contour plots of particle stability expressed as the residual concentration in supernatant after a 15 h aggregation and sedimentation period, asa function of pH and electrolyte concentration (upper row). The particle zeta potential as a function of pH and electrolyte concentration is shown in thelower row.



concentrations greater than 20% (Figure 1g). Above both of thesethresholds TiO2 is aggregated, with average particle diameters ofup to 1600 nm (Figure S4) and concentrations less than 20% ofthe initial concentration (Figure 1g). The zeta potential increasesfrom�20 mV at pH 8 to zero for pH less than 5. With increasingCaSO4 (1 mM) the IEP shifts toward pH 6 (Figure 1h). How-ever, the impact of Ca2+ adsorption to TiO2 at pH 4 is still visiblethrough a slightly positive zeta potential at intermediate CaSO4

concentrations. Figure 1 shows the effect of two interactingdivalent but counter charged ions with the TiO2 surface. WhileSO4

2� “alone” (with the almost indifferent sodium as a counter-ion) shifted the zeta potentials toward more negative values atall concentrations and all tested pH values, and calcium “alone”(with the almost noninteracting chloride as a counterion) shiftedthe zeta potential toward more positive values at elevated con-centrations and low pH values, a balancing of both these effectscan be observed. Calcium and sulfate both result in a low magni-tude zeta potential and a corresponding colloidal instability ofTiO2 particles, as expressed in low concentrations (Figure 1eand g) and large hydrodynamic diameters (Figure S4).NOMadsorbs to the surface ofmetal oxides andmay reduce or

neutralize an existing positive surface charge, or even (at sufficientconcentrations) induce a reversal from a positive to a negative sur-face charge. The effects of NOM on the stability of nanoparticlesare well-known,14�16 but depending on theNOMconcentration,the surface chemistry, and the surrounding cations, the presenceof NOM does not necessarily always have a unidirectional stabi-lizing effect. NOM concentrations between 0.5 and 100 mg L�1

were used to construct the NOMmatrix, which are equivalent to0.2 to 41.1 mg L�1 of DOC (Figure 2). At DOC concentrations

greater than 0.4 mg L�1 the organic matter throughout stabilizesthe TiO2 nanoparticles in suspension, which is consistent withresults obtained for metal-oxide particles in other studies, e.g., byZhang et al.,32 who tested the stability of TiO2 at various DOCconcentrations and observed the stability over 2 h with 0.4 mgL�1 DOC. Despite the effect of HCO3

� in our study, we ob-served aggregation of the TiO2 particles (∼1000 nm diameter) atless than 0.2 mg L�1 DOC and a pH below the IEP of the pristinematerial (Figure S4). Above this DOC concentration and at apH above the IEP the TiO2 particles were covered with organicmatter, resulting in zeta potentials of less than �20, a high resi-dual concentration (C/C0 > 50%) (Figure 2), and particle dia-meters of about 200 nm (Figure S4).So far we illustrated the effects of weakly and strongly inter-

acting, as well as of surface potential determining, inorganic ions,and the stabilizing effect of even small amounts of NOM. Wefurther aimed to investigate the effect of a combination of bothNOM and electrolytes. Because Ca2+ interacts with NOM andTiO2 and acts as a strong coagulant we chose CaCl2 as the addi-tional reagent in multicomponent testing.25,26 We set the back-ground NOM concentration to a realistic surface water concen-tration of 1 mg L�1 DOC and repeated the CaCl2 matrix withelectrolyte concentrations ranging from 0.01 to 10 mmol L�1

over a pH range from 4 to 8 (Figure 2c and d). In this system therelatively low NOM concentration of 1 mg L�1 DOC radicallychanges the behavior of the TiO2. At CaCl2 concentrations below1 mmol L�1 elevated concentrations of colloidally stable TiO2

were observed (g50% of the initial concentrations) (Figure 2c).These match the concentrations found in the NOM matrix at1 mg L�1 DOC (Figure 2a) and are higher than those of any pure

Figure 2. Contour plots of particle stability, expressed as the residual concentration in supernatant after a 15 h aggregation and sedimentation period.The left panel shows results for various concentrations of NOM and the right panel shows results for various concentrations of calcium chloride. Theupper row of the central panel shows results for the same CaCl2 concentrations in the presence of 1 mg L�1 of DOC, while the lower row shows thecorresponding zeta potentials.



inorganic matrix in this test (Figure 1a to h). Apart from thisremarkable effect of 1 mg L�1 DOC at CaCl2 concentrationsbelow 1 mmol L�1, it is the CaCl2 concentration rather thanthe pH that further controls the stability of TiO2 under theseconditions.In contrast to the pure NOM matrix (Figure 2b), it is evident

that the magnitude of the negative zeta potential is reduced whenthere is CaCl2 present (Figure 2d).The corresponding reduction in colloidal stability can be

related to the reduction in electrostatic stabilization, or to com-pression of the NOMmolecules due to complexation of the Ca2+

ions by the carboxylic groups of the NOM and associated loss ofrepulsive negative charges. Here it is not possible to distinguishbetween the effects of compression of the double layer thicknessand the reduction of NOM charge density through complexationof Ca2+ by the NOM carboxyl groups. No indications of stericstabilization by the NOM have been observed. However, atCaCl2 concentrations greater than 1 mmol L�1 a slight decreasein zeta potential is observed with increasing pH, together with aneven more marked increase in aggregate size (Figure 2d andFigure S4). This is remarkable since protonation of the NOM,accompanied by reduction in the negative surface potential of thesurface adsorbedNOM, had been expected at low pH values. Theobserved behavior may be caused by the complexation of Ca2+ -by carboxyl functional groups within the NOM structure,33 or bythe combined effect of a reduction in NOM adsorption to theTiO2 followed by adsorption of Ca2+ to negatively charged sur-face functional groups. Although the nanoparticles were coatedwith NOM the aggregation was enhanced in the presence ofmore than 1mmol L�1 CaCl2, and only 20% to almost 60% of theinitial TiO2 concentration remained colloidally stable (Figure 2c).Above 0.5 mM CaCl2 we observed an increase in particle dia-meter from 260 to 900 nm for pH less than 6, and strong aggre-gation occurred where the pH was greater than 6 with particlediameters of 1.7 μm (Figure S4) concomitant with a slightlylower negative zeta potential (Figure 2d).Behavior and Characteristics of TiO2 in Natural and EPA

Waters.By testing the behavior of P25 TiO2 nanoparticles undervarious hydrochemical conditions we have been able to demon-strate the usefulness of this generic approach to understand thebasics of the material's behavior in natural waters. To facilitatea comparison between the stability of nanoparticles in naturalwater and the synthetic results of the matrix testing, a range ofdifferent, naturally occurring water types were selected, includingmarine, fresh, and wastewater. The chemical characteristics of thechosen waters are summarized in the Supporting Information(Table S4). The pH did not vary much among sources except forthe slightly acidic peat bogwater, which occurred over granite andhad a pH of 5.2. The ionic strength (IS: 0.5 to 146 meq L�1) andthe DOC content (<0.5 to 68 mg L�1), however, varied signi-ficantly among samples. Three synthetic waters proposed by theEPA34 were chosen as potential test media to cover different pHvalues (pH 6�8) and hardnesses (soft to hard water). The IS ofthe EPA waters ranged from 0.3 to 9.4 meq L�1.Stability of TiO2 in Natural Waters, Waste Waters, and

Synthetic EPA Waters. The stability of TiO2 in natural water isstrongly influenced by the presence of DOC and the concentra-tions of monovalent and divalent ions (Table 1 and Table S4).The concentration of TiO2 in the supernatant of the naturalwater samples decreased proportionally with decreasing DOCcontent. The organic matter in the low mineralized peat bog-water (TAN) stabilized the TiO2 nanoparticles most effectivelyT

able1.

Aggregation

Stateof

TiO

2Nanop

articles

inNaturalWater

andSyntheticEPAWater

a

ground

water

lake

water

tapwater

peatbogw

ater

seawater

wastewaterinflow

wastewateroutflow

EPAvery

soft

EPAmoderatelyhard

EPAvery

hard

unit

HOE

LUNZ

TAP

TAN

FRA

WWI

WWO

VS

MH

VH

particlediam

eter

nm1160

((154)

1180

((54)

1170

((118)

219((

10)

1340

((92.4)

750((

55.1)

1210

((91.5)

1190

((165)

1480

((90)

1520

((96)

zetapotential

mV

�1((

0.6)

�15((

0.6)

�9((

0.3)

�26((

0.2)

+2((

2.2)

�17((

1.18)

�14((

0.8)

�20((

1.6)

�16((

0.6)

�10((

0.2)

colloidalstablefractio

nmgL�

12((

0.6)

2((

0.1)

2((

0.2)

24((

1.2)

1((

0.1)

15((

2.2)

9((

1.5)

8((

0.7)

2(

0.1)

1((

0.4)

aThe

particlediam

eter

(z-average),zetapotential,andconcentration,asaverages

of3replicates

(n=3)

and(1SD

,wereanalyzed

inthesupernatantafter15

hof

sedimentatio

n.All0.2μm

filtered.



(24 mg L�1 TiO2 in the supernatant). Together with the highTiO2 concentration we found small particle sizes (TAN: 219 nm)and a high negative surface charge (TAN: �26 mV), as has alsobeen shown to be the case in other studies for carbon nano-tubes,35 goethite,36 and other metal oxide nanoparticles.32

Compared to the TAN sample the 0.2-μm filtered WWTPinflow had a higher DOC content (68 mg L�1) but a lower levelof TiO2 stability. The 15 mg L�1 of TiO2 particles remaining inthe supernatant had a diameter of 750 nm and a zeta potentialof �17 mV. The IS and, in particular, the presence of divalentions (WWI: Ca2+ 48 mg L�1 and SO4

2� 45 mg L�1), promoteaggregation, as has been demonstrated in the matrix tests and canbe seen in Figure 1, even in the presence of NOM (Figure 2).Some studies have suggested destabilization due to bridgingeffects or charge neutralization in the presence of alginate andCa2+.19,28 Our results are in agreement with those of Zhanget al.,32 who found a threshold of 4 meq L�1 Ca2+ above whichaggregation of NOM-coated TiO2 nanoparticles was induceddue to a reduction of the energy barrier. In our study we foundincreased aggregation for samples with Ca2+ exceeding 2 meq L�1

(Table 1). For these samples the zeta potential was close to theIEP (HOE: �0.5, TAP: �8.7, FRA: +1.7, VH: �9.7 mV)resulting in nearly 90% aggregation and sedimentation (Table 1).The stability of TiO2 in test media was investigated in the very

soft (VS), moderately hard (MH), and very hard (VH) watertypes proposed by EPA.34 These waters were in equilibrium withCO2, their pH was stable at 6.4 (VS), 7.4 (MH), and 8.4 (VH),their IS ranged from 0.3 to 9.6 meq L�1, and their electricalconductivities ranged from 35 to 1046 μS cm�1. In the EPAwaters the TiO2 stability decreased with increasing pH and IS.The zeta potential was �20 mV (VS), �16 mV (MH), and�10mV (VH). The particle diameter of the remaining 30% (VS),10% (MH), and 5% (VH) of TiO2 in the supernatant rangedfrom 1190 nm (VS), to 1480 nm (MH), and 1520 nm (VH). Ourresults show that the colloidal stability of TiO2 nanoparticlesin natural water is governed by the presence of organic matter,resulting in higher particle stability, and the concentration ofbivalent ions, resulting in aggregation and lower particle stability.In addition, we have demonstrated that despite significantvariations in the water chemistry there are similarities in thebehavior of TiO2 between the various chosen water types (e.g.,groundwater and lake water). This suggests the presence of

stability-determining master parameters (such as pH and a smallset of ions).Predictive Potential of the Test Matrices. A comparison

with TiO2 stability in natural and near-natural waters and syn-thetic matrix tests demonstrates how the synthetic matrix testresults can be used to predict the behavior of TiO2 in naturalwaters and test media.The most important step in relating the synthetic matrix

results to natural waters is to identify which of the syntheticmatrices best represents the particular natural water under consi-deration. The natural waters tested, as well as the synthetic EPAwaters, were therefore classified by identifying the parametersthat dominated the water chemistry through a STIFF diagramplotting procedure (see explanation in SI and Figure S5). On thebasis of the matrix testing we assumed that the concentration ofdivalent ions, the pH values, and the DOC concentrations wouldbe the parameters with the strongest influence on aggregationbehavior. This enabled the relevant matrix testing plot to bechosen from the STIFF diagrams (Figures 1 and 2). Ground-water (HOE) was classified as a CaCl2 type and lake water(LUNZ), tap water (TAP), and EPA waters (VS, MH, VH) wereclassified as Ca2+/SO4

2� types. The organic-rich waters, suchas the peat bogwater (TAN) and the wastewater inflow (WWI)and outflow (WWO), were classified as Ca2+/DOC types. Theseawater (FRA) is not represented in the diagrams due to ionconcentrations that exceeded those of the test matrix by a factorgreater than 10. Each value of TiO2 concentration, particle dia-meter, and zeta potential was extracted from the synthetic matrixtests by means of interpolated data. The values from the naturalwaters and the EPA waters were then plotted against the corres-ponding value in the test matrix (Figure 3).The results from the test matrices characterized the natural

and EPA waters well, but in some cases the test matrices under-estimated (TAN) or overestimated (HOE,WWO) themeasuredvalues (Figure 3). We explain this by the considerable quantitiesof SO4

2� present in HOE and WWO, but not being representedin the respective matrix (Table S4), which promoted aggregation(Table 1).The results from the WWTP inflow (WWI) are well repre-

sented in the chosen test matrix which is indicated by the stabi-lizing effects of organic matter. We found the NOM surrogate tobe less effective in stabilizing the TiO2 particles than the natural

Figure 3. Correlating TiO2 concentration, zeta potential, and particle diameter in natural/synthetic water with results from the relevant synthetictest matrix. The appropriate synthetic test matrix was selected after categorization of the natural and test waters according to their water chemistry.STIFF diagrams (see Supporting Information) were used to support the decision making.



organic-rich water (TAN), and consequently the test matrixunderestimated the concentration of TiO2 in the supernatant.The adsorption, and hence the stabilizing capacity, of NOM cor-relates with their functional groups.17 We have observed differ-ences in TiO2 stabilization related to theNOM-type but verifyingthis observation has been outside the scope of this study.Surprisingly, the very soft EPA water (VS), with the lowest IS,

was not well represented by the test matrix. The TiO2 concen-tration in the EPA water was reduced by a factor of 2.3 comparedto the value from the corresponding test matrix (Ca2+/SO4

2�-matrix). The zeta potential measurements showed a very goodcorrelation.In general, the matrices systematically underestimate the zeta

potential for several samples (TAN, VS, MH, VH, and HOE).This is due to the existence of DOC in all of the natural watersamples, which is not represented in the purely inorganic matrixtests used to predict behavior in these waters. Conclusively, withincreasing NOM adsorption to the TiO2 surface zeta potentialsbecome more negative [e.g., ref 32] and therefore natural sam-ples (WWI, WWO) correlate better to the Ca2+/DOC matrix.Environmental Implications. The colloidal stability of TiO2

nanoparticles in an aquatic environment is strongly influenced bythe water chemistry. The process and state of aggregation is animportant factor affecting the fate and transport of ENPs in theaqueous environment, and may alter the exposure of microorg-anisms to ENP or even determine which organisms are exposedat all. It also is an important input parameter to support exposuremodeling. We have presented a generic testing scheme for thestability and aggregation of ENPs and have validated this schemethrough experiments with natural waters. The results from syn-thetic waters with a wide variety of compositions, ion concentra-tions, and pH values, have enabled us to identify zones of TiO2

nanoparticle stability and aggregation. By combining this infor-mation with results from tests in natural waters we have shownin the presented case that it is possible to predict the generalbehavior of TiO2 nanoparticles. Even though the underlyingmechanisms are not yet fully understood, the presented approachto predict nanoparticle behavior bridges the gap between detailed,single-parameter laboratory studies and real, natural situations. Asimilar approach is also likely to be successful if applied to othertypes of nanoparticles. Further work is needed to address theeffects of different types of NOM and the presence of otherparticulates.

’ASSOCIATED CONTENT

bS Supporting Information. Additional material and results.This material is available free of charge via the Internet at http://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected] or [email protected].

’ACKNOWLEDGMENT

We acknowledge the funding received by The AustrianMinistry for Agriculture, Forestry and the Environment, andthe European Chemical Industry Council (CEFIC) within theframework of the LRI Project on Detection, Fate and Uptake of

Engineered Nanoparticles in Aquatic Systems. We also are gratefulto the OECD’s WPMN, Evonik, and JRC for providing us withthe P25 OECD material (NM 105) used in this study.

’REFERENCES

(1) Kiser, M. A.; Westerhoff, P.; Benn, T.; Wang, Y.; P�erez-Rivera, J.;Hristovski, K. Titanium nanomaterial removal and release from waste-water treatment plants. Environ. Sci. Technol. 2009, 43 (17), 6757–6763.

(2) Kaegi, R.; Ulrich, A.; Sinnet, B.; Vonbank, R.; Wichser, A.;Zuleeg, S.; Simmler, H.; Brunner, S.; Vonmont, H.; Burkhardt, M.;Boller, M. Synthetic TiO2 nanoparticle emission from exterior facadesinto the aquatic environment. Environ. Pollut. 2008, 156 (2), 233–239.

(3) Handy, R. D.; Von Der Kammer, F.; Lead, J. R.; Hassell€ov, M.;Owen, R.; Crane, M. The ecotoxicology and chemistry of manufacturednanoparticles. Ecotoxicology 2008, 17 (4), 287–314.

(4) Hofmann, T.; von der Kammer, F. Estimating the relevance ofengineered nanoparticle facilitated transport of hydrophobic organiccontaminants in porousmedia.Environ. Pollut. 2009, 157 (4), 1117–1126.

(5) Lead, J. R.; Wilkinson, K. J. Aquatic colloids and nanoparticles:Current knowledge and future trends. Enviorn. Chem. 2006, 3, 159–171.

(6) Hartmann, N.; von der Kammer, F.; Hofmann, T.; Baalousha,M.; Ottofuelling, S.; Baun, A. Algal testing of titanium dioxide nano-particles� testing considerations, inhibitory effects and modification ofcadmium bioavailability. Toxicology 2009, 269 (2�3), 190–197.

(7) Gao, J.; Youn, S.; Hovsepyan, A.; Llaneza, V. n. L.; Wang, Y.;Bitton, G.; Bonzongo, J.-C. J. Dispersion and toxicity of selected manu-factured nanomaterials in natural river water samples: Effects of waterchemical composition. Environ. Sci. Technol. 2009, 43 (9), 3322–3328.

(8) Battin, T. J.; v. d. Kammer, F.; Weilhartner, A.; Ottofuelling, S.;Hofmann, T. Nanostructured TiO2: Transport behavior and effects onaquatic microbial communities under environmental conditions. Envir-on. Sci. Technol. 2009, 43 (21), 8098–8104.

(9) Gottschalk, F.; Scholz, R. W.; Nowack, B. Probabilistic materialflow modeling for assessing the environmental exposure to compounds:Methodology and an application to engineered nano-TiO2 particles.Environ. Model. Software 2010, 25 (3), 320–332.

(10) Fang, J.; Shan, X.-q.; Wen, B.; Lin, J.-m.; Owens, G. Stability oftitania nanoparticles in soil suspensions and transport in saturatedhomogeneous soil columns. Environ. Pollut. 2009, 125 (3), 1101–1109.

(11) Brar, S. K.; Verma, M.; Tyagi, R. D.; Surampalli, R. Y. En-gineered nanoparticles in wastewater and wastewater sludge - Evidenceand impacts. Waste Manage. 2010, 30 (3), 504–520.

(12) Alvarez, P. J. J.; Colvin, V.; Lead, J.; Stone, V. Research prioritiesto advance eco-responsible nanotechnology. ACS Nano 2009, 3 (7),1616–1619.

(13) Ji, Z.; Jin, X.; George, S.; Xia, T.;Meng, H.;Wang, X.; Suarez, E.;Zhang, H.; Hoek, E. M. V.; Godwin, H.; Nel, A. E.; Zink, J. I. Dispersionand stability optimization of TiO2 nanoparticles in cell culture media.Environ. Sci. Technol. 2010, 44 (19), 7309–7314.

(14) Keller, A. A.; Wang, H.; Zhou, D.; Lenihan, H. S.; Cherr, G.;Cardinale, B. J.; Miller, R.; Ji, Z. Stability and aggregation of metal oxidenanoparticles in natural aqueous matrices. Environ. Sci. Technol. 2010, 44(6), 1962–1967.

(15) v. d. Kammer, F.; Ottofuelling, S.; Hofmann, T. Assessment ofthe pysico-chemical behavior of engineered nanoparticles in aqauticenvironments using multi-dimensional parameter testing. Environ. Pol-lut. 2010, 158 (12), 3472–3481.

(16) Johnson, R. L.; Johnson, G. O. B.; Nurmi, J. T.; Tratnyek, P. G.Natural organic matter enhanced mobility of nano zerovalent iron.Environ. Sci. Technol. 2009, 43 (14), 5455–5460.

(17) Hyung, H.; Kim, J.-H. Natural organic matter (NOM) adsorp-tion to multi-walled carbon nanotubes: Effect of NOM characteristicsand water quality parameters. Environ. Sci. Technol. 2008, 42 (12), 4416–4421.

(18) Chen, K. L.; Elimelech, M. Relating colloidal stability of fulle-rene (C60) nanoparticles to nanoparticle charge and electrokineticproperties. Environ. Sci. Technol. 2009, 43 (19), 7270–7276.



(19) Chen, K. L.; Mylon, S. E.; Elimelech, M. Aggregation kinetics ofalginate-coated hematite nanoparticles in monovalent and divalentelectrolytes. Environ. Sci. Technol. 2006, 40 (5), 1516–1523.(20) Chen, K. L.; Elimelech, M. Aggregation and deposition kinetics

of fullerene (C60) nanoparticles. Langmuir 2006, 22, 10994–11001.(21) Fernandez-Nieves, A.; de las Nieves, F. J. The role of ζ potential

in the colloidal stability of different TiO2/electrolyte solution interfaces.Colloids Surf., A 1999, 148 (3), 231–243.(22) Hardy, W. B. A preliminary investigation of the conditions

which determine the stability of irreversible hydrosols. J. Phys. Chem.1899, 4 (4), 235–253.(23) Guzm�an, D. K. A.; Finnegan, M. P.; Banfield, J. F. Influence of

surface potential on aggregation and transport of titania nanoparticles.Environ. Sci. Technol. 2006, 40 (24), 7688–7693.(24) French, R. A.; Jacobson, A. R.; Kim, B.; Isley, S. L.; Penn, R. L.;

Baveye, P. C. Influence of ionic strength, pH, and cation valence onaggregation kinetics of titanium dioxide nanoparticles. Environ. Sci.Technol. 2009, 43 (5), 1354–1359.(25) Domingos, R. F.; Peyrot, C.; Wilkinson, K. J. Aggregation of

titanium dioxide nanoparticles: Role of calcium and phosphate. Environ.Chem. 2010, 7 (1), 61–66.(26) Domingos, R. F.; Tufenkji, N.; Wilkinson, K. J. Aggregation of

titanium dioxide nanoparticles: Role of a fulvic acid. Environ. Sci. Technol.2009, 43 (5), 1282–1286.(27) Kretzschmar, R.; Holthoff, H.; Sticher, H. Influence of pH

and humic acid on coagulation kinetics of kaolinite: A dynamic lightscattering study. J. Colloid Interface Sci. 1998, 202 (1), 95–103.(28) Chen, K. L.; Elimelech, M. Interaction of fullerene (C60)

nanoparticles with humic acid and alginate coated silica surfaces: Mea-surements, mechanisms, and environmental implications. Environ. Sci.Technol. 2008, 42 (20), 7607–7614.(29) Smith, B.; Wepasnick, K.; Schrote, K. E.; Bertele, A. R.; Ball,

W. P.; O’Melia, C.; Fairbrother, D. H. Colloidal properties of aqueoussuspensions of acid-treated, multi-walled carbon nanotubes. Environ. Sci.Technol. 2008, 43 (3), 819–825.(30) Mueller, N. C.; Nowack, B. Exposure modelling of engineered

nanoparticles in the environment. Environ. Sci. Technol. 2008, 42 (12),4447–4453.(31) Kosmulski, M.; Rosenholm, J. B. High ionic strength electro-

kinetics of anatase in the presence of multivalent inorganic ions. ColloidsSurf., A 2004, 248 (1�3), 121.(32) Zhang, Y.; Chen, Y.; Westerhoff, P.; Crittenden, J. Impact of

natural organic matter and divalent cations on the stability of aqueousnanoparticles. Water Res. 2009, 43 (17), 4249–4257.(33) Kim, J.; Shan, W.; Davies, S. H. R.; Baumann, M. J.; Masten,

S. J.; Tarabara, V. V. Interactions of aqueous NOMwith nanoscale TiO2:Implications for ceramic membrane filtration-ozonation hybrid process.Environ. Sci. Technol. 2009, 43 (14), 5488–5494.(34) EPA. Short-term Methods for Estimating the Chronic Toxicity of

Effluents and Receiving Waters to Freshwater Organisms; United StatesEnvironmental Protection Agency: Washington, DC, 2002; p 335.(35) Hyung, H.; Fortner, J. D.; Hughes, J. B.; Kim, J. H. Natural org-

anic matter stabilizes carbon nanotubes in the aqueous phase. Environ. Sci.Technol. 2007, 41 (1), 179–184.(36) Weng, L. P.; Van Riemsdijk, W. H.; Koopal, L. K.; Hiemstra, T.

Adsorption of humic substances on goethite: Comparison betweenhumic acids and fulvic acids. Environ. Sci. Technol. 2006, 40 (24), 7494–7500.

commercial titanium dioxide nanoparticles in both natural and synthetic water: comprehensive...

Documents