Experimental Study of Tio2 Nanoparticle Adhesion to Silica and Fe(III) Oxide-Coated Silica Surfaces

Chemical Geology 332–333 (2012) 148–156

Contents lists available at SciVerse ScienceDirect

Chemical Geology

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

Experimental study of TiO2 nanoparticle adhesion to silica and Fe(III) oxide-coated silica surfaces

Lindsay A. Seders Dietrich a,⁎, Manoranjan Sahu b,1, Pratim Biswas b, Jeremy B. Fein a a University of Notre Dame, Department of Civil & Environmental Engineering & Earth Sciences, Notre Dame, IN 46556, USA b Washington University in St. Louis, Department of Energy, Environmental and Chemical Engineering, St. Louis, MO 63130, USA article info abstract

Article history: With the rapid expansion of industrial nanotechnology applications, engineered nanoparticles are being in- Received 8 June 2012 troduced into the environment before the controls on their fate and mobility are fully understood. In this Received in revised form 7 September 2012 study, we measured the adhesion of TiO2 nanoparticles onto silica and Fe(III) oxide-coated silica surfaces Accepted 14 September 2012 as a function of pH, nanoparticle concentration, and nanoparticle size. Batch TiO adhesion experiments Available online 4 October 2012 2 were conducted at pH 3–8 and 0.01 M NaClO4 with TiO2 concentrations ranging from 10 to 200 mg/L. Editor: B. Sherwood Lollar Three TiO2 size fractions, each containing a range of particle sizes, had initial average diameters of 16, 26, and 50 nm. Silica grains, both uncoated and coated with Fe(III) oxide, were used as the geosorbents. The ex-

Keywords: tent of TiO2 nanoparticle adhesion increased with increasing nanoparticle concentration, and pH exerted a Titanium dioxide nanoparticle strong effect on the adhesion behavior of the nanoparticles onto the uncoated silica particles. At and below

Adhesion pH 5, TiO2 nanoparticle adhesion increased with increasing pH; at pH 6 and above, adhesion occurred inde- Silica pendently of pH. In general, the differences in adhesion between the three nanoparticle sizes at a given pH Iron oxide coating were not large. Within a given size fraction, preferential adhesion of the larger TiO2 particles was suggested

below pH 6, and preferential adhesion of the smaller TiO2 particles was suggested at and above pH 6. Exper-

iments with the Fe-coated silica grains were conducted only with the 26 nm TiO2 nanoparticles, and, except at pH 6 where we observed signiﬁcantly enhanced adhesion to the Fe-coated silica relative to the uncoated silica, the extents of nanoparticle adhesion onto the two geosorbents were the same within experimental uncertainty. The similarity in adhesion behaviors onto solids with such different surface chemistries suggests

1. Introduction potentially detrimental in environmental systems where unwanted in- hibition of bacterial growth has been shown to occur (Adams et al.,

In recent years, the use of engineered nanoparticles has expanded 2006). In addition, exposure to nano-sized TiO2 particles may be toxic rapidly into a variety of industries, and many consumer products rang- (Long et al., 2006; Nel et al., 2006; Wiesner et al., 2006; Limbach et al., ing from shampoo and cosmetics to tires and tennis racquets now con- 2007; Wang et al., 2007; Simon-Deckers et al., 2008; Brunet et al., tain nanoparticles (Wiesner et al., 2006). Titanium dioxide (TiO2) 2009). However, despite their widespread industrial use and potential nanoparticles are especially common in a number of products and are release into the environment, the controls on the fate and mobility of being introduced into aquatic and subsurface geologic systems either TiO2 nanoparticles in the subsurface have not been well characterized. inadvertently, such as through the weathering of exterior paint (Kaegi Both agglomeration and surface adhesion can affect the mobility et al., 2008), or intentionally as a tool for environmental remediation of engineered nanoparticles in geologic systems. Most previous re- (e.g., Mattigod et al., 2005; Pena et al., 2005; Theron et al., 2008; search has focused on agglomeration processes, demonstrating the

Oyama et al., 2009). TiO2 nanoparticles exhibit antibacterial properties, prevalence of TiO2 nanoparticle agglomerates under a wide range which may be beneﬁcial in pharmaceutical applications but are of aqueous conditions and with nanoparticles of many sizes (Lecoanet et al., 2004; Dunphy Guzman et al., 2006; Ridley et al., ⁎ Corresponding author at: Department of Civil and Environmental Engineering, 2006; Choy et al., 2008; Domingos et al., 2009; Fatisson et al., 2009; Southern Methodist University, P.O. Box 750340, Dallas, TX 75275-0340, USA. Tel.: +1 French et al., 2009; Jiang et al., 2009; Keller et al., 2010; Petosa et 214 768 1991; fax: +1 214 768 2164. al., 2010; Ottofuelling et al., 2011). In general, nanoparticle agglom- E-mail addresses: [email protected] (L.A. Seders Dietrich), [email protected] eration increases with increasing ionic strength, and agglomeration (M. Sahu), [email protected] (P. Biswas), [email protected] (J.B. Fein). 1 Present address: Advanced Energy Technology Initiative, Prairie Research Institute, also increases as the suspension pH approaches the zero point of University of Illinois at Urbana—Champaign, Champaign, IL 61820, USA. charge (pHzpc) of the nanoparticles (e.g., French et al., 2009; Jiang

0009-2541/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chemgeo.2012.09.043 L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156 149 et al., 2009). Agglomerate size also increases as the pH approaches Brieﬂy, the nanoparticle sizes reported here were based on the equiva- the pHzpc (Dunphy Guzman et al., 2006; Fatisson et al., 2009). The lent diameter of the particles as determined by their measured surface presence of other dissolved components, such as cations, organic area. The same particles were also measured by microscopic methods matter, or surfactants, can also affect the agglomeration behavior of by counting more than 100 particles from representative images. Infor-

TiO2 nanoparticles (Tkachenko et al., 2006; Domingos et al., 2009; mation on the average particle size and standard deviation is given in French et al., 2009; Joo et al., 2009; Keller et al., 2010; Ottofuelling more detail in Jiang et al. (2008). The nanoparticles were observed to et al., 2011). be spherical, and each size fraction contained a range of particle sizes Nanoparticle agglomeration behavior can affect or be affected by in- (Jiang et al., 2007, 2008; Sahu et al., 2011). The average diameters teractions with geosorbents (Dunphy Guzman et al., 2006; Choy et al., (dry) of the three nanoparticle size fractions used in these experiments 2008; Fang et al., 2009; Fatisson et al., 2009; Joo et al., 2009; Solovitch were 16 nm, 26 nm, and 50 nm. The hydrodynamic diameter of the et al., 2010), significantly influencing particle transport. Dunphy particles (i.e., agglomerate size) is known to change with ionic strength Guzman et al. (2006) found that larger agglomerates had higher inter- and pH (Jiang et al., 2009; Sahu et al., 2011), and agglomeration was action energies, which led to an increase in nanoparticle deposition; expected under our experimental conditions. The crystallinity and however, TiO2 nanoparticles and their agglomerates were still found phase of the material were determined using X-ray diffractometry to be highly mobile under most conditions. Fatisson et al. (2009) (XRD) with a Rigaku D-MAX/A9 diffractometer and Cu Kα radiation found that the deposition rates of TiO2 nanoparticles onto silica differed (λ=1.5418 Å). The crystalline phase of the material was anatase as de- depending upon changes in pH and ionic strength due to the effect of termined from the XRD pattern. The resulting nanoparticles were in these parameters on particle agglomeration. Nanoparticle agglomera- powder (dry) form and were used without washing. Specific surface tion and deposition were also observed by Fang et al. (2009) to depend areas determined previously for TiO2 nanoparticles of similar size and upon the characteristics of the soil with which the nanoparticles were in manufactured using the same process were 95.8, 61.5, and 31.5 m2/g contact. As is true of most particles, high ionic strength, pH, and zeta po- for the 16, 26, and 50 nm particles, respectively (Jiang et al., 2007, tential affected TiO2 settling, while increasing the dissolved organic car- 2008, 2009). The isoelectric points of nanoparticles manufactured bon concentration and clay content helped to stabilize the nanoparticles using the same procedure described above were shown previously to in suspension. In some cases, the formation of large agglomerates with- decrease with increasing nanoparticle size from pH 5.8 at 16 nm to in soil columns completely prevented their passage (Fang et al., 2009). pH 5.2 at 26 nm to pH 5.0 at 53 nm (Suttiponparnit et al., 2010). Using sand columns, Choy et al. (2008) found almost complete retention of TiO2 nanoparticles, although the roles of straining and adhesion 2.2. Geosorbents could not be differentiated. Joo et al. (2009) observed similar retention of TiO2 nanoparticles in silica columns, but in the presence of an The geosorbents used in these experiments were amorphous sil- adsorbed polymer, nanoparticle transport was enhanced. Thus, TiO2 ica gel (30–60 mesh) purchased from Sigma Aldrich (St. Louis, MO), nanoparticles remain disagglomerated and mobile in certain soils and as well as this same amorphous silica onto which an Fe(III) oxide under certain aqueous conditions, but they agglomerate and become coating was applied following the procedure of Ams et al. (2004) immobile due to straining and/or adhesion in others. This change in mo- (described in more detail in the Supplementary materials). The silica bility as a function of agglomeration state is true of most particles over a grains to which the Fe(III) oxide coating was applied will hereafter range of compositions and sizes. be referred to as simply Fe-coated. Amorphous silica was selected There have been few studies to examine the adhesion behavior of as the geosorbent because it represents a common mineral surface

TiO2 nanoparticles onto well-characterized, uniform geosorbents, so type, and its high surface area coupled with its settling properties quantitative models of nanoparticle transport behavior in geologic sys- make it an ideal material for these experiments. The pHzpc of the tems are poorly constrained. Furthermore, in transport studies such as uncoated silica grains is ~pH 2 and that of the Fe-coated grains is those conducted by Dunphy Guzman et al. (2006), Choy et al. (2008), ~pH 7–8(Ams et al., 2004). The BET surface area of the uncoated and Fang et al. (2009),andJoo et al. (2009),itisdifficult to differentiate be- Fe-coated silica grains was analyzed using a Micromeritics ASAP 2020 tween nanoparticle adhesion and straining of large nanoparticle ag- surface area analyzer (Micromeritics Instrument Corp., Norcross, GA). glomerates in order to understand the relative effects of electrostatic The silica grains were rinsed with DI water to remove fines and allowed controls versus agglomeration controls on TiO2 transport. In this to air dry. The Fe-coated grains had already been rinsed well and were study, we determined the influence of pH, nanoparticle size, and nano- analyzed without further rinsing. Approximately 0.3 g of each sample particle concentration on the adhesion of TiO2 nanoparticles to silica was outgassed at 30 °C for 20 h. Surface areas were measured in tripli- grains. One nanoparticle size fraction was also selected for adhesion cate, yielding averages of 703.3±8.3 m2/g for the uncoated grains and experiments with iron-coated silica grains in order to determine the 528.4±3.7 m2/g for the Fe-coated grains. These surface areas suggest effect of geosorbent mineral surface composition on particle adhesion that the porosity of the grains is high. Images of the uncoated and behavior. Fe-coated silica grains (Figs. S1 and S2 in Supplementary materials) were also taken using an environmental scanning electron microscope 2. Materials and methods (ESEM; Carl Zeiss LEO EVO 50) and showed considerable surface rough- ness but no porosity on the scale of the nanoparticle agglomerates 2.1. Nanoparticles present in these experiments. Thus, the pores in the silica grains are likely small enough so that the nanoparticles cannot access them.

TiO2 nanoparticles of the same crystalline phase and of different sizes were synthesized in a premix flame aerosol reactor by control- 2.3. Batch experiments ling the temperature and residence time history of the particles in the high temperature combustion zone (Jiang et al., 2007). Titanium Batch adhesion experiments were conducted at a fixed ionic tetra-isopropoxide (TTIP, 97%, Aldrich) was used as the precursor for strength in 0.01 M NaClO4,atpH3–8. Nanoparticle concentrations synthesizing the TiO2 particles. The volumetric flow rates of N2/TTIP, O2, were analyzed using UV/vis absorbance due to the limitations of and CH4 were precisely controlled by mass flow controllers. Nanoparti- ICP-OES, DLS, and GF-AAS (described in more detail in the Supplemen- cle size and morphology were characterized by field emission scanning tary materials). As described in Section 2.4, the UV/vis method yielded or (scanning) transmission electron microscopy (FE-SEM, model: JEOL excellent calibration curves and provided additional qualitative infor- 7001LVF; FE-(S)TEM, model: JEOL 2100F). Details on the methodology mation unavailable from more traditional methods. In order to mini- used to determine particle size are provided in Jiang et al. (2008). mize pH drift during the experiments, the uncoated silica grains were 150 L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156

rinsed prior to use at least five times with 0.01 M NaClO4 that was ad- experimental control was then measured and adjusted, if necessary, justed to the pH of a given experiment. The last wash was left rotating using minute aliquots of concentrated HNO3 or NaOH. The acid/base ad- overnight before decanting. At values above pH 6, it was necessary to ditions did not significantly affect the ionic strength of the suspensions. readjust the pH with NaOH several times over the course of at least All systems were rotated for 3 h, with hourly pH measurements and 24 h in order to obtain a stable pH. The Fe-coating procedure involved readjustments. After the 3 hour reaction time, the final pH of each sus- rinsing the silica grains at least 20 times with ultrapure water, so the pension was measured. The final pH value of each suspension was with- coated silica grains were rinsed only twice more with 0.01 M NaClO4 in ±0.4 pH units of the desired experimental pH, with the majority at the experimental pH. The final wash solution was left to rotate for within ±0.2 pH units. The geosorbent grains settled as soon as the ex- about 90 min during which HNO3 or NaOH was added to stabilize the perimental suspensions were removed from the rotator, and the super- pH at the desired value for each experiment. The shorter equilibration natant from each experiment was transferred with a pipette to a 28 mL time was used to minimize the removal/dissolution of the Fe-coating. Teflon tube. Any settling of the nanoparticles due to agglomeration was However, in some cases, dissolution of the Fe coating was found to in- expected to be identical to the settling that would occur in the stan- terfere with the results (see Supplementary materials); thus, only dards under the same conditions. The experimental control suspensions data at pH 5 and above are reported here. Both the coated and uncoated were also transferred to Teflon tubes in the same manner as the exper- silica grains were added to the nanoparticle suspensions wet, and the imental suspensions in order to evaluate nanoparticle loss to the Teflon dry weight of the silica grains was determined by evaporating the liquid jars during the 3 hour reaction time. Standard suspensions were never from the used geosorbents after completion of the experiments. moved from the Teflon jars, which meant that any nanoparticles ad- Geosorbent-only controls were conducted as a part of each ex- hered to the container walls could be resuspended during the final periment. The silica grains settled from solution as soon as rotating ultrasonication step. stopped. In addition, at each pH of interest here, we measured the dissolved Si and/or Fe concentrations in a set of geosorbent-only 2.4. UV/vis analysis control experiments involving either the uncoated or the Fe-coated silica grains using ICP-OES (Perkin Elmer, Optima DV 2000). In all The following UV/vis method of determining nanoparticle concen- cases, the Si levels were below saturation with respect to amorphous trations was selected because more traditional methods were found to silica and were such that SiO2(aq) should be the dominant Si species have limitations in our experimental system. Complete, reproducible in solution. Furthermore, polymers did not form and were thus un- digestion of TiO2 suspensions was not possible, which prohibited the likely to affect nanoparticle agglomeration (Iler, 1979). Dissolved use of ICP-OES analysis. The UV/vis approach provided highly linear Fe levels were as follows: 0.061 ppm at pH 5, 0.24 ppm at pH 6, and reproducible calibration curves that made it possible to relate ab- 1.00 ppm at pH 7, and 1.09 ppm at pH 8. These values are higher than sorbance to concentration. Also, standards that exactly replicate the Fe(III) oxide solubility limits, which makes it likely that there was samples in every way except for exposure to the mineral surfaces some colloidal iron in suspension. It is unlikely that the pH adjustments were used. Not only does UV/vis offer rigorous quantitative analysis of necessary during the rinsing and equilibration of the silica grains TiO2 concentrations in suspension, but it also offers additional informa- stemmed from dissolution of the grains, as the presence of the neutral tion not available from other approaches. Changes in the UV/vis absor- aqueous silica species should not affect pH, and the dissolved Fe present bance profile from one set of experiments to another make it possible to in solution was not enough to affect pH. qualitatively infer changes in bulk nanoparticle or agglomerate size for For the experiments containing nanoparticles, a parent suspen- the particles in suspension. While this information is traditionally sion of 200 mg/L TiO2 nanoparticles in 0.01 M NaClO4 at each pH obtained using dynamic light scattering (DLS), the apparent particle condition studied was prepared in a Teflon container immediately size in our suspensions, as measured during DLS tests that we prior to conducting experiments in order to minimize nanoparticle conducted in order to attempt to verify the particle size of the experi- agglomeration and settling over time. However, nanoparticle ag- mental TiO2 suspensions, was found to increase over the 9 min required glomeration was unavoidable in these experiments. For this reason, for triplicate analyses. This is likely due to changes in agglomerate size great care was taken to treat all nanoparticle standards and nanopar- with time and potentially settling. A UV/vis scan using our approach ticle suspensions exposed to silica grains in exactly the same manner lasted for only 45 s, thereby limiting the time for such agglomeration so that the agglomeration that did occur would be consistent for each changes to occur. set of experimental conditions. All 200 mg/L TiO2 nanoparticle par- All UV/vis analyses were conducted with a Varian Cary 300 Bio UV/ ent suspensions were placed into an ultrasonic water bath (210 W, visible double-beam spectrophotometer. Prior to each UV/vis measure- 50/60 Hz) for 20 min to ensure suspension of the nanoparticles ment, all suspensions (samples and standards) were placed in an ultra- and to achieve a consistent degree of agglomeration in each experi- sonic water bath for 40 min and were then immediately scanned from ment (Jiang et al., 2009). For consistency, following ultrasonication, 800 to 350 nm. Experiments were very carefully conducted to keep thesame200mg/LTiO2 nanoparticle parent suspension was used the duration of rotation and ultrasonication as similar as possible for to make the nanoparticle concentration standards without exposure each sample and standard suspension. Using absorbance at 750 nm to geosorbents, the experimental suspensions to which geosorbent after background subtraction, linear calibration curves were created grains were added, and the experimental geosorbent-free controls from the standard suspensions at each nanoparticle size and pH and that were used to quantify loss to the labware. The nanoparticle con- were used to calculate the nanoparticle concentrations remaining in centrations in these suspensions ranged from 2 to 200 mg/L, and all the samples taken from the geosorbent-bearing experiments. Selection were made in 30 mL Teflon jars. All of these suspensions were then of 750 nm as the wavelength of interest is explained in detail in the placed into the ultrasonic water bath for an additional 20 min. Im- Supplementary materials. Briefly, absorbance at 750 nm showed good mediately following the second ultrasonication, the nanoparticle agreement between the absorbance profiles of the standards and the suspensions were analyzed by measuring light absorbance from geosorbent-bearing suspensions. All calibration curves had calculated 800 to 350 nm (described in more detail in Section 2.4). R2 values of 0.99 or above (discussed in more detail in the Supplemen- Experiments using the uncoated silica grains were conducted using tary materials, Fig. S6). The nanoparticle concentration adhered to the each of the three sizes of TiO2; the experiments that involved the geosorbent was calculated by difference between the initial and final Fe-coated silica grains were conducted using only the 26 nm TiO2 size nanoparticle concentrations. fraction. To each of the 10 mL experimental suspensions, 0.5 g of silica The validity of this analytical approach was verified by conducting grains or Fe-coated silica grains (dry weight) was added for a solid to several parallel experiments with nanoparticle concentrations in suspen- suspension ratio of 50 g/L. The pH of each standard, experiment, and sion analyzed using graphite furnace-atomic absorption spectrometry L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156 151

0.8 (GF-AAS; Perkin Elmer AAnalyst 800). The two analytical methods 50 nm yielded consistent nanoparticle concentrations in suspension over the 0.7 26 nm range of concentrations studied here. Thus, UV/vis was found to yield re- 0.6 16 nm sults that were consistent with those obtained with GF-AAS but with the 0.5 beneﬁt of providing additional information, as discussed below. The results of these control experiments are described in more detail in the Sup- 0.4 plementary materials. 0.3 absorbance 0.2 2.5. Adhesion isotherm calculations 0.1 The Langmuir isotherm approach was used to model the experimen- 0 tal nanoparticle adhesion data. The Langmuir approach is based on the 350 400 450 500 550 600 650 700 750 800 assumption of a ﬁnite supply of reaction sites (Stumm and Morgan, wavelength (nm) 1996), and the Langmuir isotherm equation can be expressed as: Fig. 1. Particle size dependence of the 50 mg/L TiO2 standards at pH 4. x bCN ¼ max ð1Þ m 1 þ bC

x fi where m is the mass of the sorbate (mg TiO2)dividedbythemassofthe the 16 nm standards had a similar shape to the pro les of the 50 nm sorbent (g dry silica grains), b is a constant (L/mg), and C is the concen- samples except below approximately 400 nm (Fig. 2b). Therefore, the tration of sorbate remaining in suspension (mg TiO2/L) after equilibra- fact that the UV/vis absorbance profiles for the 50 nm suspensions be- tion. Nmax is the maximum possible adhesion by the sorbent (mg/g) came less steep during interaction with the silica grains suggests the under the experimental conditions, and the Langmuir constant, KL,is possibility that the larger nanoparticles within these suspensions pref- fi de ned as: KL =bNmax. At low sorbate concentrations, KL approaches erentially adhered to the silica surface, leaving the smaller TiO2 in the linear distribution coefficient Kd and a linear relationship between suspension and resulting in size fractionation. x fi C and m exists. A linearized form of Eq. (1) can be expressed as: At pH values 6, 7, and 8, the absorbance pro les of the 50, 26, and 16 nm TiO2 suspensions exposed to the silica grains became steeper. 1 1 1 fi ¼ þ : ð2Þ At pH 7, the steepening of the absorbance pro les of the suspensions x= bCN N m max max exposed to silica grains was the most pronounced (e.g., 50 nm TiO2 in Fig. 3). The increased absorbance profiles of the nanoparticle suspen- 1 1 sions exposed to silica grains compared to the unexposed standards Plotting the experimental data as x= vs. C and calculating the slope m suggest the possibility of preferential adhesion of the smaller TiO2 and intercept of the best-fit line to the data yield values for 1 bNmax and 1 , respectively, and these values were used to calculate values Nmax 0.8 for KL and Nmax. The data set included a few outlier points which occurred outside the 2σ uncertainty envelope of the linear trend of the 0.7 200 a) data. These data points were plotted in the isotherms but were ignored 0.6 150 when calculating Nmax and KL values. A number of datasets, when plotted as isotherms, did not exhibit significant departures from the linear- 0.5 ity that characterizes the low C region. When plotted using Eq. (2), 0.4 100 these datasets yielded negative y-intercepts and, hence, negative Nmax 0.3 values. These physically impossible values suggest that N was poorly absorbance 50 max 0.2 defined by the data under these conditions, and that adhesion experi- 20 ments to much higher nanoparticle concentrations would need to be 0.1 5 conducted in order to observe the non-linear isotherm behavior that 0 constrains the Nmax value. In these cases, the Kd equation was used: 350 400 450 500 550 600 650 700 750 800 wavelength (nm) x ¼ K C ð3Þ m d 0.8 0.7 b) 200 which is equivalent to the Langmuir equation at low sorbate concentra- 0.6 fi tions, to model the adhesion results. However, the rst preference, 100 where possible, was to use the Langmuir equation as it yields informa- 0.5 150 tion on Nmax. 0.4 0.3

3. Results and discussion absorbance 0.2 50 3.1. UV/vis absorbance profiles 0.1 5 20 0 3.1.1. Uncoated silica grains 350 400 450 500 550 600 650 700 750 800 The steepness of the UV/vis absorbance profiles of the nanoparticle wavelength (nm) suspensions used herein decreased with decreasing particle size (Fig. 1). At pH 4 and 5, it was observed that the absorbance profiles of Fig. 2. a) 50 nm TiO2 standards (in mg/L) at pH 4 are shown as black solid curves. The dashed curve is the absorbance profile of the 200 mg/L 50 nm TiO suspension after it the 50, 26, and 16 nm TiO suspensions exposed to the silica grains 2 2 was exposed to silica grains at pH 4. b) 16 nm TiO standards (in mg/L) at pH 4 are fi 2 were less steep than the absorbance pro les of their respective nano- shown as gray solid curves. The dashed curve is the absorbance profile of the 200 mg/L particle standards (Fig. 2a). For example, the absorbance profiles of 50 nm TiO2 suspension after it was exposed to silica grains at pH 4. 152 L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156

2.5 (see Fig. S9 in the Supplementary materials). Generally, nanoparticle adhesion to the coated and uncoated silica grains was similar from 200 2 pH 5 to pH 8, except at pH 6 where a signiﬁcantly higher concentration 150 of nanoparticles adhered to the Fe-coated grains compared to the 1.5 100 uncoated grains. Our results suggest that adhesion behavior is not controlled by elec- 1 50 trostatic forces between the TiO2 nanoparticles and the geosorbent sur-

absorbance faces. At the pH values used in these experiments, the charge on the 0.5 20 silica grains is negative, and the isoelectric point of the TiO2 is between pH5andpH6(Suttiponparnit et al., 2010). If adhesion is based purely 5 0 upon electrostatics, then we would expect a decrease in TiO2 adhesion 350 400 450 500 550 600 650 700 750 800 onto the uncoated silica grains with increasing pH over the entire pH wavelength (nm) range of this study. Instead, we observed increasing adhesion from pH 3 to pH 5, with lower extents of adhesion and pH-independent be- Fig. 3. 50 nm TiO standards (in mg/L) at pH 7 are shown as black solid curves. The 2 havior at higher pH values. Furthermore, the surface charges of the dashed curve is the absorbance profile of the 200 mg/L 50 nm TiO2 suspension after it was exposed to silica grains at pH 7. uncoated and Fe-coated silica grains are quite different, with the pHzpc of the coated grains expected to be between pH 7 and pH 8 (Ams et al., 2004). Again, based upon electrostatic considerations only, we would expect to see minimal adhesion below pH 6 where both the particles within a given size fraction, the opposite of what was nanoparticles and the Fe-coated silica grains are positively charged. suggested in experiments conducted at pHb6. From pH 6 to pH 7, we might expect an increase where the charges are opposite, but above pH 7 both surfaces are negatively charged and electrostatic repulsive forces would be large. However, the adhesion be- 3.1.2. Iron-coated silica grains haviors of the two geosorbents are broadly similar. It should also be The changes in the absorbance profiles between the 26 nm TiO2 noted that the high BET surface areas measured for both types of standards and the 26 nm suspensions exposed to Fe-coated silica geosorbents suggest that the availability of surface sites should not be grains were generally similar to what was seen in the experiments a limiting factor for TiO adhesion. with uncoated silica, and we do not depict the UV/vis absorbance pro- 2 The changes in adhesion as a function of pH and the similarities in files for these samples. At pH 5, the absorbance profiles of the suspen- TiO nanoparticle adhesion onto the silica and Fe-coated silica grain sur- sions exposed to the Fe-coated silica grains exhibited a shallower 2 faces suggest that it is some property of the nanoparticles that governs slope than the standard suspensions. At pH 6, 7, and 8, there was a their adhesion. Nanoparticle agglomeration is strongly affected by sus- steepening in the absorbance profiles of the exposed suspensions pension pH, with agglomeration increasing as the pH is approached. compared to the standards, although these increases were slight at zpc We did observe changes in UV/vis absorbance as a function of pH that pH 6 and 8. are suggestive of changes in agglomeration. For these reasons, it is likely that the agglomeration state of the nanoparticles plays a dominant role 3.2. TiO2 adhesion in determining the extent of TiO2 nanoparticle adhesion onto mineral surfaces. In order to control and/or limit agglomeration, surface coat- 3.2.1. Uncoated silica grains ings are often applied to engineered nanoparticles (e.g., Mattigod et The adhered nanoparticle concentrations from each set of experi- al., 2005), and in environmental systems, natural organic matter can ad- ments were calculated by difference from the starting concentrations sorb onto nanoparticle surfaces and reduce the extent of particle ag- and the nanoparticle concentrations remaining in suspension. In gener- glomeration (e.g., Espinasse et al., 2007; Hyung et al., 2007; Domingos al, the extent of nanoparticle adhesion onto uncoated silica grains in- et al., 2009; Kim et al., 2009; Ghosh et al., 2010). Thus, the presence of creased with increasing nanoparticle concentration (Fig. 4), which is surface coatings which act to reduce nanoparticle agglomeration may typical isotherm behavior. However, the sorbent capacity for adhesion also act to reduce nanoparticle adhesion. During groundwater flow varied markedly as a function of pH, as can be seen by comparing the through porous media, agglomeration and straining are likely to be sim- – experimental data to the 100% adhesion lines shown in Fig. 4a f. As de- ilarly interrelated. scribed earlier, the pHzpc of our TiO2 ranged from 5.0 to 5.8 depending Differences in nanoparticle surface properties as a function of par- on particle size (Suttiponparnit et al., 2010). To examine pH depen- ticle size could explain the observed effects of nanoparticle size on dence more closely, we show adhesion as a function of pH for the exper- particle adhesion onto the silica grains. For example, nanoparticulate iments involving nanoparticle concentrations of 115 mg/L and 175 mg/ anatase adsorbed less Cd and Pb from solution than larger-grained fl L(Fig. 5). pH strongly in uenced the adhesion behavior of the three anatase when normalized to the nanoparticle surface area (Gao nanoparticle sizes, but it was not a straightforward relationship. At et al., 2004; Giammar et al., 2007). Similarly, Waychunas et al. and below pH 5, the concentration of adhered nanoparticle increased (2005) observed that Hg(II) adsorption onto nanoparticulate goethite with increasing pH (from 3 to 5) with almost complete removal at decreased with decreasing particle size when adsorption was normal- pH 5. From pH 5 to pH 6, all nanoparticle sizes exhibited a decrease in ized to surface area, perhaps due to increased disordering on the sur- adhesion, and adhesion was independent of pH at pH 6 and above. faces of the smaller particles. Increasing disorder with decreasing However, it should be noted that the percent of TiO2 adhered changed nanoparticle size has been observed for hematite and ferrihydrite much less between pH 6 and pH 8 than it did between pH 3 and nanoparticles (Chernyshova et al., 2007; Michel et al., 2007). Howev- pH 5. In general, the differences in adhesion between the three nano- er, Jiang et al. (2008) found that surface defect sites were more com- particle sizes at a given pH were not large. mon per unit area on larger TiO2 particles, and Zhang et al. (2009) observed conflicting effects occurring with decreasing particle size: 3.2.2. Iron-coated silica grains an increasing proportion of high energy sites led to an increase in sur- The mineralogy and/or surface charge of the geosorbent did not face free energy, but increases in internal distortion decreased the markedly affect the observed extents of adhesion (Fig. 6). As we ob- surface free energy. It is not clear at this point whether agglomeration served for the uncoated silica grains, nanoparticle adhesion onto the effects, surface reactivity effects, or other factors cause the differences Fe-coated grains increased with increasing nanoparticle concentration in the extent of adhesion observed for the different nanoparticle sizes. L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156 153

400 400 50 nm 50 nm 350 a) pH 3 350 b) pH 4 26 nm 26 nm 300 16 nm 300 16 nm 250 250 200 200 150 150 adhered / g sorbent adhered / g sorbent

2 100 2 100 50 50

mg/L TiO 0 0 mg/L TiO 0 50 100 150 200 0 50 100 150 200

initial mg/L TiO2 initial mg/L TiO2

400 400 50 nm 50 nm 350 c) pH 5 350 d) pH 6 26 nm 26 nm 300 300 16 nm 16 nm 250 250 200 200 150 150 adhered / g sorbent adhered / g sorbent

2 2 100 100 50 50

0 mg/L TiO 0 mg/L TiO 0 50 100 150 200 0 50 100 150 200

initial mg/L TiO2 initial mg/L TiO2

400 400 50 nm 350 e) pH 7 350 50 nm f) pH 8 26 nm 26 nm 300 300 16 nm 16 nm 250 250 200 200

150 adhered / g sorbent 150 adhered / g sorbent

2 2 100 100 50 50 mg/L TiO mg/L TiO 0 0 0 50 100 150 200 0 50 100 150 200

initial mg/L TiO2 initial mg/L TiO2

Fig. 4. TiO2 nanoparticle adhesion to uncoated silica grains at pH 3–8(a–f). Open squares represent the 50 nm TiO2, black diamonds the 26 nm TiO2, and gray triangles the 16 nm

TiO2. The black line represents 100% adhesion. In cases where adhesion was close to the detection limit, the data are not presented.

3.2.3. TiO2 adhesion isotherms reflect changes in agglomeration behavior. For example, the Nmax values The Langmuir adsorption isotherm equation provided a good fitto for the data shown in Fig. 7a for the 26 and 16 nm TiO2 experiments are most of the experimental measurements of TiO2 nanoparticle adhesion. 6.2 and 2.8 mg/g, respectively. The available surface area for adhesion is An example of the model fits for the pH 5 data is shown in Fig. 7 in both the same in these two sets of experiments, and TEM imaging showed 1 1 the isotherm format (Fig. 7a for all particle sizes) and the x= vs. format that the three sizes of TiO2 nanoparticles are spherical. Thus, the larger m C (Fig. 7b for the 26 nm particle experiments only). The calculated iso- Nmax for the 26 nm particle experiments may reflect greater agglomer- therm parameters for all of the experiments are compiled in Table 1. ation, and hence greater nanoparticle packing on the surface, than oc-

For cases with Nmax values reported in Table 1, there was a good linear curs for the 16 nm system. Alternatively, the 16 nm particles may 1 1 2 correlation between x= and (with R values ranging from 0.73 to 0.99), agglomerate to a greater extent than the 26 nm particles, and increased m C suggesting that the Langmuir model provided a good ﬁt to the experi- agglomeration could prevent close packing of the agglomerates on the mental data. For the cases with Nmax values that were not deﬁned by mineral surface, leading to decreased adhesion. Previous research has x our data, there was a good linear correlation between m and C,and shown that larger TiO2 nanoparticles formed smaller agglomerates hence the Kd approach provided a better representation of the adhesion than did smaller TiO2 nanoparticles (Lecoanet et al., 2004; Pettibone behavior in these cases (e.g., the 50 nm particle experiments depicted et al., 2008; French et al., 2009), and this was also the case for some in Fig. 7a), with R2 values ranging from 0.70 to 0.99. other metal oxide nanoparticles (He et al., 2008; Darlington et al.,

A calculated Nmax value represents the maximum extent of adhesion 2009). The second explanation is consistent with these observations, possible under a particular set of conditions. Because Nmax is thought to as single spheres should pack more closely on a surface than large ag- represent monolayer coverage of the sorbent surface, and because we glomerates of spheres. However, conﬁrmation of this mechanism re- are comparing results from experiments with the same concentration quires more direct observation of the morphology of the adhered ofsorbentavailableforadhesion,thechangesinNmax values likely particles. In general, the Nmax values that were obtained parallel the 154 L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156

100 5 50 nm a) 115 mg/L a) pH 5 50 nm 80 26 nm 4 26 nm 16 nm 16 nm 60 3

40 2 adhered / g sorbent 2 adhered (normalized) 2 20 1 mg TiO % TiO 0 0 23456789 0 20 40 60 80 100 mg/L TiO in suspension pH 2

100 1.2 50 nm b) 175 mg/L b) pH 5, 26 nm 1 80 26 nm 16 nm 0.8 60 y = 5.71x + 0.16 0.6 2 R = 0.98 40 0.4 1/(x/m) (g/mg) adhered (normalized) 2 20 0.2

% TiO 0 0 0 0.05 0.1 0.15 0.2 23456789 1/C (L/mg) pH

Fig. 7. a) Adhesion isotherms from experiments involving uncoated silica grains at Fig. 5. The extent of nanoparticle adhesion onto uncoated silica grains as a function of pH, pH 5 for all three nanoparticle sizes. The experimental data are shown as squares shown for all three TiO2 nanoparticle sizes at two initial concentrations: a) 115 mg/L and (50 nm), diamonds (26 nm), and triangles (16 nm). The calculated Langmuir iso- b) 175 mg/L. Experiments were conducted with 0.5 g of silica grains in 10 mL of nanopar- therms are shown as the black (26 nm) and gray (16 nm) lines, and the Kd ﬁt is the ticle suspension. dashed line (50 nm). Experiments were conducted with 0.5 g of silica grains in 10 mL of nanoparticle suspension. b) The Langmuir linearization of the 26 nm experimental data at pH 5 with the calculated linear regression equation.

complex nanoparticle adhesion behaviors that we observed as a function of pH and particle size and that are shown in Figs. 5 and 6.Under 100 some conditions, there was a clear trend with pH and/or particle size, silica 115 mg/L a) and under other conditions there was none. The Nmax values for the ex- 80 Fe-silica periments involving uncoated silica grains were similar to correspond-

ing Nmax values for the experiments involving Fe-coated silica grains, 60 conducted at the same pH and particle size conditions.

KL or Kd values are deﬁned by the slope of the data at low nanopar- 40 ticle concentrations and represent the distribution of nanoparticles between aqueous suspension and the mineral surface under these con-

adhered (normalized) − − 2 20 ditions. The K values ranged from 6.6×10 3 to 3.4×10 1 L/g, and while there was no consistent trend with pH or particle size, some gen-

% TiO 0 eralizations can be made. As can be seen in Fig. 8,theK values for the 26 23456789 and 16 nm TiO2 were virtually identical to each other across the pH pH range studied, but the 50 nm particles exhibited markedly different behavior from the two smaller nanoparticle sizes at pH 3 and 5. From pH 6 100 to pH 8, the K values for all three nanoparticle sizes were similar, and silica b) 175 mg/L there was little change with pH. The values for Fe-coated silica grains 80 Fe-silica were also similar to those calculated for the uncoated silica grains in this pH range. 60 K values represent the distribution of nanoparticles between aqueous and solid phase, normalized per gram of sorbent and per 40 liter of suspension. Changes in nanoparticle behavior reflected by fi adhered (normalized) changes in K value can indicate signi cant differences in how 2 20 nanoparticles might be transported through the environment. For the experimental sorbent concentrations and suspension volumes % TiO 0 examined here, K values were calculated for systems in which 10%, 23456789 50%, or 90% of the particles would be adhered to the sorbent. These K pH − − values at 10, 50, and 90% adhesion are 2.2×10 3 L/g, 2.1×10 2 L/g, and 1.9×10−1 L/g, respectively (each is labeled with a “_% adhered” Fig. 6. The extent of 26 nm TiO adhesion onto uncoated and Fe-coated silica grains as a 2 fl function of pH at two initial concentrations: a) 115 mg/L and b) 175 mg/L. Experiments line in Fig. 8). The calculated K values re ect the wide range of were conducted with 0.5 g of silica grains in 10 mL of nanoparticle suspension. partitioning behavior that we observed under the experimental L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156 155

Table 1 transport experiments. We used batch experiments to isolate some of Isotherm fit parameters. the parameters that influence nanoparticle adhesion. Our results indi- 2 pH KL (L/g) Kd (L/g) Nmax (mg/g) R cate that pH exerts a strong, but complex, effect on the adhesion behavior of TiO2 nanoparticles onto uncoated silica grains. Below the expected 50 nm TiO2 on silica grains − – pH 3 1.7×10 1 0.75 0.99 pHzpc of 5.0 5.8, TiO2 nanoparticle adhesion increased with increasing pH 4 7.7×10 −2 3.8 0.76 pH; at and above pH 6, adhesion was pH independent with increasing − pH 5 3.4×10 1 NR 0.99 pH. In general, the differences in adhesion between the three nanopar- pH 6 8.9×10 −3 4.7 0.97 −2 ticle sizes at a given pH were not large. The UV/visible light absorbance pH 7 4.0×10 2.2 0.96 fi pH 8 1.2×10 −2 NR 0.89 pro les from our experiments also suggest the possibility that preferential adhesion of the larger TiO2 particles or agglomerates within a given

26 nm TiO2 on silica grains size fraction occurred onto the silica grains below pH 6 and that prefer- pH 3 6.6×10 −3 3.3 0.99 ential adhesion of the smaller TiO2 particles or agglomerates occurred at pH 4 4.8×10 −2 NR 0.70 − and above pH 6. The uncoated and Fe-coated silica grain surfaces pH 5 1.8×10 1 6.2 0.98 − fi fi pH 6 2.9×10 2 NR 0.98 exhibited similar af nities to adhere the nanoparticles, despite signi - pH 7 2.4×10 −2 NR 0.80 cant differences in surface charge properties of the sorbents. This simi- − pH 8 2.1×10 2 NR 0.89 larity in adhesion suggests that factors related to the nanoparticles themselves, rather than to the characteristics of the geosorbents, govern 26 nm TiO2 on Fe(III) oxide-coated silica grains pH 5 3.4×10 −1 4.0 0.98 the adhesion of TiO2 nanoparticles under the conditions of these exper- pH 6 8.2×10 −2 NR 0.99 iments. Because of the predominance of agglomeration and the changes pH 7 5.3×10 −2 9.5 0.99 in agglomeration that are known to occur with changes in pH, it seems −2 pH 8 2.1×10 NR 0.87 likely that agglomeration plays a significant role in controlling adhesion behavior. 16 nm TiO2 on silica grains pH 3 6.9×10 −3 0.65 0.73 pH 4 9.6×10 −2 1.2 0.94 pH 5 1.9×10 −1 2.8 0.95 Acknowledgments pH 6 2.7×10 −2 NR 0.96 pH 7 2.4×10 −2 NR 0.88 Funding for this work was provided by an NSF—Environmental Mo- −2 pH 8 1.9×10 1.3 0.95 lecular Science Institute grant to the University of Notre Dame, as well 2 Note: KL, Nmax, and R were calculated from the linearized Langmuir equation. In cases as by a grant from the U.S. Department of Defense to Washington Uni- where a negative Nmax was obtained, the value is not reported (NR) and values for the versity in Saint Louis (AFOSR 20 MURI Grant, FA9550-04-1-0430). 2 linear distribution coefficient (Kd) and its fit to the data (R ) are given instead.

Appendix A. Supplementary data conditions. The K values for the pH 6–8 experiments indicated nearly equal partitioning of the nanoparticles between the suspension and the Supplementary data to this article can be found online at http:// sorbent surfaces. Some of the experiments yielded K values between dx.doi.org/10.1016/j.chemgeo.2012.09.043. the 10% and 50% adhered lines, while most of the others were between the 50% and 90% adhered lines. All experimental K values were below −1 References the 95% adhered value of 4.0×10 L/g. Although Langmuir and Kd modeling parameters are difﬁcult to extrapolate to realistic conditions Adams, L.K., Lyon, D.Y., Alvarez, P.J.J., 2006. Comparative eco-toxicity of nanoscale TiO2, (e.g., Koretsky, 2000),thevaluescalculatedheresuggestthatawide SiO2, and ZnO water suspensions. Water Research 40, 3527–3532. range of partitioning behaviors is possible in the environment. Ams, D.A., Fein, J.B., Dong, H., Maurice, P.A., 2004. Experimental measurements of the adsorption of Bacillus subtilis and Pseudomonas mendocina onto Fe-oxyhydroxide- coated and uncoated quartz grains. Geomicrobiology Journal 21, 511–519. 4. Conclusions Brunet, L., Lyon, D.Y., Hotze, E.M., Alvarez, P.J.J., Wiesner, M.R., 2009. Comparative

photoactivity and antibacterial properties of C60 fullerenes and titanium dioxide – Because nanoparticle agglomeration and adhesion processes occur nanoparticles. Environmental Science and Technology 43 (12), 4355 4360. fi Chernyshova, I.V., Hochella Jr., M.F., Madden, A.S., 2007. Size-dependent structural simultaneously, it can be dif cult to distinguish between the effects of transformations of hematite nanoparticles. 1. Phase transition. Physical Chemistry adhesion and agglomerate straining in the results of nanoparticle Chemical Physics 9, 1735–1750. Choy, C.C., Wazne, M., Meng, X., 2008. Application of an empirical transport model to simulate retention of nanocrystalline titanium dioxide in sand columns. 0.4 Chemosphere 71, 1794–1801. 50 nm Darlington, T.K., Neigh, A.M., Spencer, M.T., Nguyen, O.T., Oldenburg, S.J., 2009. Nano- 26 nm particle characteristics affecting environmental fate and transport through soil. En- 0.3 – Fe-26 nm vironmental Toxicology and Chemistry 28 (6), 1191 1199. Domingos, R.F., Tufenkji, N., Wilkinson, K.J., 2009. Aggregation of titanium dioxide 16 nm nanoparticles: role of a fulvic acid. Environmental Science and Technology 43 – 0.2 (5), 1282 1286. 90% adhered Dunphy Guzman, K.A., Finnegan, M.P., Banfield, J.F., 2006. Influence of surface potential K (L/g) on aggregation and transport of titania nanoparticles. Environmental Science and Technology 40 (24), 7688–7693. 0.1 Espinasse, B., Hotze, E.M., Wiesner, M.R., 2007. Transport and retention of colloidal ag-

gregates of C60 in porous media: effects of organic macromolecules, ionic composi- 50% adhered tion, and preparation method. Environmental Science and Technology 41 (21), 0 10% adhered 7396–7402. 23456789 Fang, J., Shan, X.-Q., Wen, B., Lin, J.-M., Owens, G., 2009. Stability of titania nanoparticles pH in soil suspensions and transport in saturated homogeneous soil columns. Environ- mental Pollution 157, 1101–1109.

Fatisson, J., Domingos, R.F., Wilkinson, K.J., Tufenkji, N., 2009. Deposition of TiO2 Fig. 8. K (either KL or Kd;seeTable 1) values as a function of pH from experimental mea- nanoparticles onto silica measured using a quartz crystal microbalance with dissi- surements involving all three nanoparticle sizes and the uncoated silica grains, and from pation monitoring. Langmuir 25 (11), 6062–6069. experiments involving the 26 nm nanoparticles and the Fe-coated silica grains. The French, R.A., Jacobson, A.R., Kim, B., Isley, S.L., Penn, R.L., Baveye, P.C., 2009. Inﬂuence of dashed lines represent K values for various nanoparticle distributions between suspension ionic strength, pH, and cation valence on aggregation kinetics of titanium dioxide and adhesion. nanoparticles. Environmental Science and Technology 43 (5), 1354–1359. 156 L.A. Seders Dietrich et al. / Chemical Geology 332–333 (2012) 148–156

Gao, Y., Wahi, R., Kan, A.T., Falkner, J.C., Colvin, V.L., Tomson, M.B., 2004. Adsorption of Michel, F.M., Ehm, L., Antao, S.M., Lee, P.L., Chupas, P.J., Liu, G., Strongin, D.R., Schoonen, cadmium on anatase nanoparticles—effect of crystal size and pH. Langmuir 20, M.A.A., Phillips, B.L., Parise, J.B., 2007. The structure of ferrihydrite, a nanocrystalline 9585–9593. mineral. Science 316, 1726–1729.

Ghosh, S., Mashayekhi, H., Bhowmik, P., Xing, B.S., 2010. Colloidal stability of Al2O3 Nel, A., Xia, T., Mädler, L., Li, N., 2006. Toxic potential of materials at the nanolevel. Sci- nanoparticles as affected by coating of structurally different humic acids. Langmuir ence 311, 622–627. 26, 873–879. Ottofuelling, S., Von Der Kammer, F., Hofmann, T., 2011. Commercial titanium dioxide Giammar, D.E., Maus, C.J., Xie, L., 2007. Effects of particle size and crystalline phase on nanoparticles in both natural and synthetic water: comprehensive multidimensional lead adsorption to titanium dioxide nanoparticles. Environmental Engineering Sci- testing and prediction of aggregation behavior. Environmental Science and Technol- ence 24 (1), 85–95. ogy 45, 10045–10052. He, Y.T., Wan, J., Tokunaga, T., 2008. Kinetic stability of hematite nanoparticles: the ef- Oyama, T., Yanagisawa, I., Takeuchi, M., Koike, T., Serpone, N., Hidaka, H., 2009. Remedi-

fect of particle sizes. Journal of Nanoparticle Research 10, 321–332. ation of simulated aquatic sites contaminated with recalcitrant substrates by TiO2/ Hyung, H., Fortner, J.D., Hughes, J.B., Kim, J.H., 2007. Natural organic matter stabilizes ozonation under natural sunlight. Applied Catalysis B: Environmental 91, 242–246. carbon nanotubes in the aqueous phase. Environmental Science and Technology Pena, M.E., Korﬁatis, G.P., Patel, M., Lippincott, L., Meng, X., 2005. Adsorption of As(V) 41, 179–184. and As(III) by nanocrystalline titanium dioxide. Water Research 39, 2327–2337. Iler, R.K., 1979. The Chemistry of Silica. Wiley, New York. Petosa, A.R., Jaisi, D.P., Quevedo, I.R., Elimelech, M., Tufenkji, N., 2010. Aggregation and Jiang, J., Chen, D.-R., Biswas, P., 2007. Synthesis of nanoparticles in a ﬂame aerosol re- deposition of engineered nanomaterials in aquatic environments: role of physico- actor with independent and strict control of their size, crystal phase and morphol- chemical interactions. Environmental Science and Technology 44, 6532–6549. ogy. Nanotechnology 18, 1–8. Pettibone, J.M., Cwiertny, D.M., Scherer, M., Grassian, V.H., 2008. Adsorption of organic

Jiang, J., Oberdörster, G., Elder, A., Gelein, R., Mercer, P., Biswas, P., 2008. Does nanopar- acids on TiO2 nanoparticles: effects of pH, nanoparticle size, and nanoparticle ag- ticle activity depend upon size and crystal phase? Nanotoxicology 2 (1–4), 33–42. gregation. Langmuir 24, 6659–6667. Jiang, J., Oberdörster, G., Biswas, P., 2009. Characterization of size, surface charge, and Ridley, M.K., Hackley, V.A., Machesky, M.L., 2006. Characterization and surface-reactivity agglomeration state of nanoparticle dispersions for toxicological studies. Journal of nanocrystalline anatase in aqueous solutions. Langmuir 22, 10972–10982. of Nanoparticle Research 11, 77–89. Sahu, M., Suttiponparnit, K., Suvachittanont, S., Charinpanitkul, T., Biswas, P., 2011.

Joo, S.H., Al-Abed, S.R., Luxton, T., 2009. Inﬂuence of carboxymethyl cellulose for the Characterization of doped TiO2 nanoparticle dispersions. Chemical Engineering transport of titanium dioxide nanoparticles in clean silica and mineral-coated Science 66, 3482–3490. sands. Environmental Science and Technology 43 (13), 4954–4959. Simon-Deckers, A., Gouget, B., Mayne-L'Hermite, M., Herlin-Boime, N., Reynaud, C., Kaegi, R., Ulrich, A., Sinnet, B., Vonbank, R., Wichser, A., Zuleeg, S., Simmler, H., Brunner, S., Carrière, M., 2008. In vitro investigation of oxide nanoparticle and carbon nanotube

Vonmont, H., Burkhardt, M., Boller, M., 2008. Synthetic TiO2 nanoparticle emission toxicity and intracellular accumulation in A549 human pneumocytes. Toxicology from exterior facades into the aquatic environment. Environmental Pollution 156, 253, 137–146. 233–239. Solovitch, N., Labille, J., Rose, J., Chaurand, P., Borschneck, D., Wiesner, M.R., Bottero, J.-Y.,

Keller, A.A., Wang, H., Zhou, D., Lenihan, H.S., Cherr, G., Cardinale, B.J., Miller, R., Ji, Z., 2010. Concurrent aggregation and deposition of TiO2 nanoparticles in a sandy porous 2010. Stability and aggregation of metal oxide nanoparticles in natural aqueous media. Environmental Science and Technology 44, 4897–4902. matrices. Environmental Science and Technology 44, 1962–1967. Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry. John Wiley and Sons, Inc., New York. Kim, J., Shan, W.Q., Davies, S.H.R., Baumann, M.J., Masten, S.J., Tarabara, V.V., 2009. Inter- Suttiponparnit, K., Jiang, J., Sahu, M., Suvachittanont, S., Charinpanikul, T., Biswas, P., 2010.

actions of aqueous NOM with nanoscale TiO2: implications for ceramic membrane Role of surface area, primary particle size, and crystal phase on titanium dioxide filtration–ozonation hybrid process. Environmental Science and Technology 43 nanoparticle dispersion properties. Nanoscale Research Letters http://dx.doi.org/ (14), 5488–5494. 10.1007/s11671-010-9772-1. Koretsky, C., 2000. The significance of surface complexation reactions in hydrologic Theron, J., Walker, J.A., Cloete, T.E., 2008. Nanotechnology and water treatment: appli- systems: a geochemist's perspective. Journal of Hydrology 230, 127–171. cations and emerging opportunities. Critical Reviews in Microbiology 34, 43–69. Lecoanet, H.F., Bottero, J.-Y., Wiesner, M.R., 2004. Laboratory assessment of the mobil- Tkachenko, N.H., Yaremko, Z.M., Bellmann, C., Soltys, M.M., 2006. The influence of ionic ity of nanomaterials in porous media. Environmental Science and Technology 38 and nonionic surfactants on aggregative stability and electrical surface properties (19), 5164–5169. of aqueous suspensions of titanium dioxide. Journal of Colloid and Interface Sci- Limbach, L.K., Wick, P., Manser, P., Grass, R.N., Bruinink, A., Stark, W.J., 2007. Exposure ence 299, 686–695.

of engineered nanoparticles to human lung epithelial cells: influence of chemical Wang, J.J., Sanderson, B.J.S., Wang, H., 2007. Cyto- and genotoxicity of ultrafine TiO2 composition and catalytic activity on oxidative stress. Environmental Science and particles in cultured human lymphoblastoid cells. Mutation Research 628, 99–106. Technology 41 (11), 4158–4163. Waychunas, G.A., Kim, C.S., Banfield, J.F., 2005. Nanoparticle iron oxide minerals in soils Long, T.C., Saleh, N., Tilton, R.D., Lowry, G.V., Veronesi, B., 2006. Titanium dioxide (P25) and sediments: unique properties and contaminant scavenging mechanisms. Jour- produces reactive oxygen species in immortalized brain microglia (BV2): implica- nal of Nanoparticle Research 7, 409–433. tions for nanoparticle neurotoxicity. Environmental Science and Technology 40 Wiesner, M.R., Lowry, G.V., Alvarez, P., Dionysiou, D., Biswas, P., 2006. Assessing the (14), 4346–4352. risks of manufactured nanomaterials. Environmental Science and Technology 40 Mattigod, S.V., Fryxell, G.E., Alford, K., Gilmore, T., Parker, K., Serne, J., Engelhard, M., (14), 4336–4345.

2005. Functionalized TiO2 nanoparticles for use in situ anion immobilization. Envi- Zhang, H., Chen, B., Banﬁeld, J.F., 2009. The size dependence of the surface free energy ronmental Science and Technology 39 (18), 7306–7310. of titania nanocrystals. Physical Chemistry Chemical Physics 11, 2553–2558.