Journal of Contaminant Hydrology 158 (2014) 1–13

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology journal homepage: www.elsevier.com/locate/jconhyd

Transport of stabilized engineered silver (Ag) nanoparticles through porous sandstones Christoph Neukum ⁎, Anika Braun, Rafig Azzam RWTH Aachen University, Department of Engineering Geology and Hydrogeology, Lochnerstr. 4-20, 52064 Aachen, Germany

a r t i c l e

i n f o

Article history: Received 30 September 2013 Received in revised form 13 December 2013 Accepted 16 December 2013 Available online 22 December 2013 Index terms: Engineered silver nanoparticles Sandstone Reactive transport Field-Flow Fractionation Keywords: Engineered nanoparticles Transport Column experiments

a b s t r a c t Engineered nanoparticles are increasingly applied in consumer products and concerns are rising regarding their risk as potential contaminants or carriers for colloid-facilitated contaminant transport. Engineered silver nanoparticles (AgNP) are among the most widely used nanomaterials in consumer products. However, their mobility in groundwater has been scarcely investigated. In this study, transport of stabilized AgNP through porous sandstones with variations in mineralogy, pore size distribution and permeability is investigated in laboratory experiments with well-defined boundary conditions. The AgNP samples were mainly characterized by asymmetric flow field–flow fractionation coupled to a multi-angle static laser light detector and ultraviolet–visible spectroscopy for determination of particle size and concentration. The rock samples are characterized by mercury porosimetry, flow experiments and solute tracer tests. Solute and AgNP breakthrough was quantified by applying numerical models considering one kinetic site model for particle transport. The transport of AgNP strongly depends on pore size distribution, mineralogy and the solution ionic strength. Blocking of attachment sites results in less reactive transport with increasing application of AgNP mass. AgNPs were retained due to physicochemical filtration and probably due to straining. The results demonstrate the restricted applicability of AgNP transport parameters determined from simplified experimental model systems to realistic environmental matrices. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The growing production and application of engineered nanoparticles (ENP) are raising concerns regarding their risk as potential contaminants or carriers for colloid-facilitated contaminant transport. Only little is known about the release and fate of nanomaterials in the natural environment (Klaine et al., 2008; Wiesner et al., 2006). Silver nanoparticles (AgNPs) are used for their antibacterial properties in everyday products such as clothing, food packaging or washing machines, as well as in medical products and other product groups. According to the Woodrow Wilson database (WWICS, 2012, nanotechproject.org), silver is the most commonly used nanomaterial. In 2011, 313 out of 1317 listed consumer products contained AgNPs. ⁎ Corresponding author. E-mail address: [email protected] (C. Neukum). 0169-7722/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jconhyd.2013.12.002

Recent studies show the relevance of AgNP transformations in different environments. Lowry et al. (2012) investigated AgNP transformation and fate in a simulated large scale freshwater wetland and concluded that erosion and runoff are potential pathways for AgNP to enter waterways. Although sulfidation of AgNP into Ag2S and Ag-sulfhydryl compounds occurs, high body burdens of Ag were found in mosquito fish and chironomids, proving the bioavailability of Ag+ from NP even after partial sulfidation. Levard et al. (2012) emphasize the importance of AgNP reactions with sulfur in natural systems. Kaegi et al. (2011) observed AgNP in the effluent of a pilot wastewater treatment plant only during an initial pulse spike. Their measurements indicate that most of the Ag in the sludge and the effluent was present as Ag2S. However, studies from Reinsch et al. (2012) suggest that the initial properties of AgNP can affect sulfidation products, which affect microbial growth inhibition. Due to their high reactivity, AgNP will transform in the environment

2

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

by oxidation, aggregation, sulfurization and chlorination. In the presence of dissolved organic matter, dissolved Ag+ can also be reduced to AgNP. Although progress has been made in the monitoring of AgNP, it is still hard to track them in the environment and processes of fate and transport are not fully understood (Yu et al., 2013). According to general particle-transport theory, particles are transported by advection and hydrodynamic dispersion and are eventually removed from the water phase by pore straining or filtration due to attachment to a collector (Penell et al., 2008; Tufenkji and Elimelech, 2004; Yao et al., 1971). Classical clean-bead filtration theory considers the attachment of particles onto collectors by interception, diffusion and sedimentation. Advanced filtration approaches consider both hydrodynamic and van der Waals forces for the calculation of particle filtration in porous media (Tufenkji and Elimelech, 2004). These approaches express particle attachment by the attachment efficiency or deposition rate coefficient. Other approaches consider particle blocking of attachment sites additionally (e.g. Bradford et al., 2003; Simunek and van Genuchten, 2008; Tosco and Sethi, 2009). The blocking of favorable attachment sites causes time dependent attachment due to differences in the attachment behavior of colloids on clean solid surfaces and on media already containing attached colloids (Bradford and Bettahar, 2006). Straining, another mechanism of colloid retention, is the trapping of colloids or colloid clusters in pore throats which are too narrow for particle passage. Experimental identification of deposition processes requires knowledge of effluent concentrations with time and spatial distribution of attached colloids (Tufenkji et al., 2003). Numerous factors are reported to contribute to deviations between filtration theory predictions and experimental observations. These factors include straining (Li et al., 2004) and aggregation (Bradford et al., 2006), the presence of stabilizing agents causing short range repulsive forces (Wang et al., 2012), surface charge heterogeneities of particles and collectors (Tufenkji and Elimelech, 2005) and hydrodynamic drag (Li and Johnson, 2005). Local heterogeneities of surface charge may result in local differences in surface forces, causing differences not only in magnitude but also in direction and resulting in attachment of particles under unfavorable chemical conditions (Lin et al., 2011; Taboada-Serrano et al., 2005). Particle retention may decrease over time due to blocking of favorable attachment sites, resulting in increased particle

transport with time. The rate of filling favorable attachment sites depends on the particle concentration of the pore fluid (Liang et al., 2013). A number of recent studies examined the transport of engineered nanoparticles in artificial porous media (glass beads, quartz sand) (e.g., Fang et al., 2009; Guzman et al., 2006; Jaisi and Elimelech, 2009; Jaisi et al., 2008; Lecoanet et al., 2004; Leocanet and Wiesner, 2004; Li et al., 2008; Solovitch et al., 2010; Wang et al., 2008). Others investigated the aggregation behavior of AgNPs under various environmental conditions (Lin et al., 2011, 2012; Song et al., 2011; Thio et al., 2012). El Badawy et al. (2010), for instance, found that AgNPs tend to aggregate at high ionic strengths and acidic pH values and that the presence of a divalent ion increases this effect. Other studies with similar findings include Liu et al. (2011) and Huynh and Chen (2011). Liang et al. (2013) showed the sensitivity of stabilized AgNP to physicochemical factors in quartz sand, where the distribution of retention profiles depends on solution ionic strength, grain size, particle velocity and input concentration. AgNP were interacting largely irreversibly in a primary minimum associated with microscopic heterogeneity, while only a small portion of the retained AgNP was retained in a secondary minimum, depending on solution ionic strength. Only a few studies have focused on NP transport in natural porous media, including Fang et al. (2009), who investigated transport of TiO2 NP in natural soils, or Sagee et al. (2012) and Tian et al. (2010), who conducted column experiments in natural soils to study the impact of grain size and humic acids on transport of AgNPs. The latter reported that AgNPs might have high mobility in natural soils depending on grain size, Darcy velocity and the presence of humic acid. The presence of surfactants or humic acids supports the mobility of AgNP (Lin et al., 2011; Thio et al., 2012; Tian et al., 2010), but higher ionic strength and divalent ions support aggregation and retention. Tellam et al. (2011) found major degree of blocking of SiO2 NPs in redbed sandstones. However, no previous study has addressed the transport of AgNP in consolidated porous rocks, despite their importance for water supply in many regions of the world. Westerhoff and Nowack (2012) emphasized the need for quantitative values concerning engineered nanoparticle transport in porous media in order to facilitate comparison across studies. In the present study, transport of engineered AgNP through three different sandstone matrices (Table 1) with differences in mineralogy and pore size distribution is investigated in

Table 1 Composition and characteristics of the sandstones according to Fitzner and Kownatzki (1991).

Quartz (%) Feldspar (%) Fragments (%) Clay minerals (%) Iron oxide/hydroxide (%) Carbonate (%) Micas (coarse-grained) (%) Chloride (coarse-grained) (%) Heavy minerals (%) Opaque substances (%) Density (g/cm3) Average grain size (mm) Porosity (%)

Herzogenrather sandstone (4)

Obernkirchner sandstone (14)

Solling sandstone (10)

95

82.4 1.6 3.8 10.4 (kaolinite) 0.2 0.8

53 14.1 5.4 6.4 (chlorite) 14.2 3.2 2.3

0.5 0.3 2.18 0.1 ~18

0.5 0.9 2.34 0.12 ~12

5

Accessory 2.1 ~0.2 ~23–26

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

laboratory experiments. Although particle transport through sandstone aquifers may often mainly occur through fractures and fissures, the interdependency between the rock matrix and the fracture system as well as the transport of particles through the rock matrix are important aspects of particle transport behavior in consolidated rocks. Transport experiments were conducted on saturated sandstone drilling cores, applying multiple AgNP injections to investigate long-term AgNP transport behavior under conditions with favorable attachment sites being gradually filled. In addition, different ionic strengths and mono- as well as divalent cations were applied to examine the effect of physicochemical factors on long-term AgNP transport. Numerical models for colloid transport are applied to AgNP breakthrough curves (BTCs) to identify relevant transport processes and to quantify the related parameters. AgNP size distribution in the effluent of the transport experiments is analyzed by Flow Field-Flow Fractionation. 2. Materials and methods 2.1. Engineered silver nanoparticles (AgNPs) The AgNP suspension (10.16% w/w, AgPURE-W10, ras materials GmbH, Regensburg, Germany) was diluted with filtered water (100 nm cut-off) to input concentrations of approximately 0.8 and 0.5 mg mL−1 (Table 2) due to analytical reasons. AgPURE corresponds to the official OECD standard material NM-300 Silver which is currently used for nanomaterial research. The physical properties of the material have been characterized by several independent European laboratories. Measurements using transmission electron microscopy and scanning electron microscopy show that 99% of the AgNPs are smaller than 20 nm. Measurements using dynamic light scattering have yielded hydrodynamic diameters of approximately 50 nm (Klein et al., 2011). The silver particles are stabilized by polyoxyethylene glycerol triolate and polyoxyethylene (20) sorbitan mono-laurat, 4% w/w each and the unbounded surfactant is around 5% in the stock suspension. The surfactants form steric repulsion barriers between AgNP and stabilize the suspension (Liang et al., 2013). The AgNP suspensions for each experiment were freshly prepared by dilution of the stock

3

suspension (120 mg mL−1) into selected electrolyte solutions (Table 2) and then sonicated for 5 min in a sonication bath. Repeated measurements of the hydrodynamic diameter of AgNP samples at different solution ionic strengths demonstrate the high stability of the AgNP suspension (Liang et al., 2013). The stability of NM-300 Silver diluted in water was demonstrated for up to 12 months (Klein et al., 2011). Steric interaction from the surfactants contributes significantly to the total energy barriers, considering the extended DLVO theory between AgNPs and between AgNP and quartz sand. Consequently, unfavorable conditions can be predicted for AgNP retention and aggregation (Liang et al., 2013). Unfavorable conditions might also be expected in comparable systems where quartz is the dominant mineral phase. The calculations presented by Liang et al. (2013) represent mean properties of the AgNP and quartz sand and do not account for physical or chemical heterogeneity on AgNP retention. Lin et al. (2012) reported that coated AgNP (poly(vinylpyrrolidone) and gum Arabic) can also attach to silica collector surfaces. Asymmetric Flow Field-Flow Fractionation (AF4) analysis was performed for determination of AgNP concentration and particle size on an AF2000 MT model coupled to an UV/Vis detector and a multi-angle laser light scattering detector (MALLS) (Postnova Analytics GmbH, Landsberg/Lech, Germany). The asymmetrical channel of the AF4 was equipped with a 10 kDa membrane made of regenerated cellulose and a 350 μm spacer providing a channel volume of 1147.125 mm3. The carrier liquid methanol was diluted with filtrated de-ionized water to a concentration of 10%. The samples were injected with an auto sampler using a 100 μL injection loop. Latex size standards were used for calibration of the residence time–particle size–function, which was verified with the MALLS detector. The size of AgNPs in the effluent was determined by using the linear residence time– particle size–function. AgNP concentration was measured by UV/ Vis detection at an adsorption wavelength of 400 nm, which is the main adsorption of AgPURE-W10 with a detection limit of 2.95 · 10−4 mg mL−1 determined by measurements of AgNP dilution at five different concentrations. Together with each sample batch containing up to 20 samples from one transport experiment, an AgNP standard was analyzed for correction of the concentration measurement due to time dependent variability of

Table 2 Experimental conditions and hydraulic parameters of the transport experiments. From measurements of electric conductivity of the deionized H2O a corresponding concentration of 0.01 mM of monovalent ions was calculated. Sandstone/effective porosity [−]

Sample no.

Experiment No.

Electrolyte

Concentration [mM]

AgNP input concentration [mg mL−1]

Darcy flux [10−6 m s−1]

Pore water velocity [10−5 m s−1]

Influent AgNP radius [nm]

4/0.189

1

4/0.188

2

10/0.088 14/0.122

1 1

14/0.118

2

1 2 3 4 5 1 2 3 4 5 6 1 1 2 1

H2O H2O H2O NaNO3 Ca(NO3)2 H2O H2O Ca(NO3)2 Ca(NO3)2 Ca(NO3)2 Ca(NO3)2 H2O H2O H2O, NaNO3 NaNO3

0.01 0.01 0.01 10 1 0.01 0.01 1 10 10 10 0.01 0.01 0, 1 1

0.800 0.545 0.536 0.474 0.477 0.486 0.482 0.479 0.494 0.547 0.566 0.441 0.441 0.500 0.556

5.14 5.28 5.12 5.51 5.22 5.04 5.16 24.9 24.7 4.89 24.9 8.43 · 10−3 0.143 0.147 0.105

2.72 2.79 2.71 2.92 2.76 2.68 2.79 13.3 13.1 2.60 13.3 9.54 · 10−2 0.117 0.121 8.94 · 10−2

27.4 22.6 25.6 25.5 26.8 27.5 27.1 24.6 23.9 27 24.4 n.d. n.d. 21.4 21.5

4

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

UV/Vis detection. The size dependent retention time of the AgNP within the channel depends on one hand on the operation period of the membrane (total number of AgNP analyses) and on the other hand on the overall ionic strength of the sample matrix. The standard deviation of the size measurement was determined using the above-mentioned AgNP standard and was quantified to 2 nm. Characterization of the AgNP suspension at 10 mM NaNO3, 1 mM Ca(NO3)2 and deionized water (ionic strength of 0.01 mM) regarding electrophoretic mobility and hydrodynamic particle size at varying pH values was carried out with a Malvern Zetasizer Nano ZS and a Malvern MPT-2 for titration with 2 M HNO3 and 0.1 M NaOH. Zeta potentials were calculated from measured electrophoretic mobility using the Smoluchowski approximation. 2.2. Sandstone samples and transport experiments Samples with diameters of 5 cm and heights of 3 cm of three different sandstones were used to characterize the rock matrix regarding hydraulic conductivity and solute as well as AgNP transport. The sample dimensions had to be selected in spite of the unknown transport behavior of AgNP. However, strong attachment of AgNP on the rock samples was partly expected. Consequently, transport length was kept small in order to achieve AgNP breakthrough in the effluent for quantitative interpretation. The Triassic Solling (Solling Formation, Buntsandstein, rock no. 10) and the Lower Cretaceous (Berrias) Obernkirchner (rock no. 14) sandstones are important regional and local aquifers, respectively, in Germany and are both frequently used for water supply. The Herzogenrather sandstone (Pliocene, rock no. 4) consists mainly of quartz and represents a welldefined model system. Some key characteristics of these three rock types are listed in Table 1. The main differences between the three rock types refer to the mineralogy. Except for the differences in the quartz content, the Obernkirchner and Solling sandstone contain around 6 and 10% clay minerals respectively. The red colored Solling sandstone contains 14% iron-oxides/hydroxides in addition. Two series of triplicate measurements of the surface charge characteristics of the sandstone samples were performed using a Malvern Zetasizer after crushing and milling the rock samples to a powder for different solution ionic strengths (deionized water, 1 and 10 mM NaNO3, 1 and 10 mM Ca(NO3)2). All cores were drilled perpendicular to the stratification but distinct sedimentary structures are not visible on the sample scale. AgNP mobility was tested with two different samples for each of the Herzogenrather and Obernkirchner sandstone and one sample of the Solling sandstone. The pore size distribution of the sandstones was characterized by mercury porosimetry (Micrometrics, Autopore IV, Series 9500). The hydraulic conductivity of each sample was measured using the constant head method. Transport experiments were implemented to characterize solute and AgNP transport in synthentic triaxial pressure cells equipped with a polyethylene filter material (Poroplast Type 162, Durst Filtertechnik, Germany) on the top and bottom of the rock samples in order to establish a uniform fluid flow over the entire cross-section of the samples. The rock samples were separated from the surrounding water pressure by a latex rubber sleeve of 5 cm diameter. A high-pressure multichannel dispenser (ICP,

ISMATEC, Germany) was used to inject solutions at a steady velocity through the rock samples in an upflow mode. Before starting the transport experiments, the rock samples were conditioned to the desired electrolyte solution. A nonreactive tracer experiment was then conducted by injecting between 3 and 6 pore volumes of sodium chloride solution depending on the rock type and applied flow velocity. Effluent samples were collected continuously in plastic bottles. Sodium chloride concentrations were determined by measuring the electrical conductivity of the effluent with a calibration using five standard solutions. The measured BTCs were inversely fitted using a one-dimensional advection–dispersion transport equation (see Section 2.3) using Hydrus-1D. The resulting transport parameters were considered at the inverse modeling of the AgNP BTCs. Prior to the first injection of AgNP suspension, the rock samples were flushed by deionized water until steady state solute concentrations corresponding to an electric conductivity of the fluid to less than 100 μS cm−1 were achieved. Subsequently, the pore fluid of the rock samples was conditioned to the individual electrolyte concentration applied in the AgNP transport experiment (Table 2). Between 3 and 5 pore volumes of AgNP were injected into the samples, depending on the rock type and the particular boundary conditions, followed by flushing with several pore volumes of particle-free electrolyte solution. The electrolyte concentration and flow rate were kept constant during the individual experiments. Between individual experiments, the pore fluids of the rock samples were conditioned to the specific solute concentrations applied in the following experiment (Table 2). All experiments were conducted at pH values between 6 and 7.6. AgNP concentrations and particle size of both the effluent and the influent were determined by AF4 using the method described in Section 2.1. Influent samples were taken shortly after AgNP application to the rock sample to control the monodisperse state of the suspension during the experiments. Unfortunately, no retention profiles of AgNP in the rock samples could be measured due to the lack of appropriate non-destructive methods. The application of excavation methods in combination with chemical digestion usually used with soil samples (e.g. Liang et al., 2013) at the end of the experiments is not applicable, since sawing of the sample in defined increments would be required and the associated loss of AgNP is not quantifiable. In addition to the tracer and AgNP transport experiments, preliminary transport studies on empty pressure cells (only equipped with the Poroplast filter material) were conducted using sodium chloride tracer and AgNP to determine losses of AgNP due to the experimental setup on the one hand and to determine the transport properties of the experimental setup by the one-dimensional form of the advection–dispersion transport equation using Hydrus-1D on the other hand. The fitted transport parameters for the experimental setup were considered during fitting tracer and AgNP BTCs of the transport experiments in Hydrus-1D. 2.3. Mathematical modeling Parameters of the solute transport, effective porosity and dispersion coefficient were obtained using inverse mathematical modeling of the BTCs, considering the applied boundary conditions of the experiments (water and solute flux).

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

For inverse modeling of solute and AgNP transport through the porous sandstones, the software package HYDRUS-1D (Simunek and van Genuchten, 2008) was used. Inverse fitting of the experimental solute and AgNP BTC was performed using a nonlinear least square optimization routine (MarquardtLevenberg), which is included in the HYDRUS-1D software package. The solute BTC for each sandstone sample was fitted using the advective dispersive equation (ADE). Porosity and dispersivity were obtained by inverse optimization. The inflow boundary conditions were defined using the measured flow rate for each sampling interval and the input concentration of solute tracer and AgNP suspension (concentration flux) respectively. The outflow boundaries were defined with a constant pressure head of 0 m and a concentration flux respectively. The Crank–Nicholson-Scheme was used for time weighting and Galerkin finite element scheme for space weighting. To ensure stable numerical solutions time and space discretization were determined considering the grid Peclet number and the Courant number to be below one for all simulation runs. The transport of AgNP was simulated using the ADE and the frequently used one kinetic site retention,   ∂ðθCÞ ∂S ∂ ∂C ∂ðqCÞ þρ ¼ θD − ð1Þ ∂t ∂t ∂x ∂x ∂x where θ [−] is the volumetric water content, C [ML−3] is the particle concentration, ρ is the bulk density of the porous medium [ML−3], t is the time [T], x is the spatial coordinate [L], S is the solid phase particle concentration [MM−1], D is the hydrodynamic dispersion coefficient [L2T−1] and q is the flow rate [LT−1]. Mass transfer between solid and aqueous phase is given as: ρ

∂S ¼ θka ψC−kd ρS ∂t

ð2Þ

where ka is the first-order deposition coefficient [T−1], kd is the first-order detachment coefficient [T−1] and ψ is a dimensionless colloid retention function [−]. Parameter ψ gives the reduction in the deposition coefficient due to filling of attachment sites using a Langmuirian approach:   S ψ ¼ 1− ð3Þ Smax where Smax is the maximum solid phase particle concentration [MM−1]. This blocking term implies the decrease of retention with time. The first-order deposition coefficient and AgNP dispersivity were simultaneously determined by inverse optimization. First-order detachment and blocking of attachment sites were considered for BTCs which cannot be represented by first-order deposition only. The lower fitting limit for ka and kd was set to 1.0 ∙ 10−8 s−1 and the upper limit was set to 1 for both parameters. Clean-bed conditions were assumed as initial conditions of all models, including also the repetitive experiments with the Herzogenrather samples. One reason for this assumption is the unknown distribution of retained AgNP, with the consequences for its consideration in the transport model and remobilization of AgNP in a quartz–water–AgNP system might occur only on a low level Liang et al. (2013). However, the modification of the surface properties of the collector due to attached AgNP is an important aspect for the long-term

5

behavior of AgNP transport. Since the long-term transport behavior of AgNP through the sandstones is of interest in this study, the effect of attached AgNP on mobile AgNP in the pore fluid and the consequences for particle filtration which is expressed in variations of the model parameters ka and Smax are of major importance. 3. Results and discussion 3.1. Characterization of sandstone samples and pressure cell The pore size distributions of the different sandstones span three orders of magnitude (Fig. 1). The Solling has the smallest pore sizes, with a median radius of 0.3 μm. The median pore radius of the Obernkirchner sandstone is 3.8 μm and 21.5 μm for the Herzogenrather sandstone. Consequently, the ratio of the mean AgNP radius to the median pore radii of around 25 nm (DLS and AF4) is 0.08, 0.006 and 0.001 respectively. As indicated in Bradford et al. (2002), straining is traditionally considered when particle diameter is greater than 5% of the median grain size diameter. The pore radius distribution of the Solling and Obernkirchner sandstones indicates portions of pores smaller than the hydrodynamic radius of AgNP are approximately 20 and 7% respectively. For these portions straining is anticipated. Steric hindrance may result in even larger portions of pore space with straining effect. The measured zeta potential of the sandstone samples is dependent on the ionic strength of the solution. It varies between −55 ± 0.76 (deionized water) and −22 ± 1.57 mV (1 mM calcium nitrate) for the Herzogenrather sandstone and is −49 ± 1.01 mV for the Solling in deionized water. For the Obernkirchner sandstone with deionized water it is −41 ± 2.79 mV. The pH values of the measurements are in the range of those measured during the transport experiments. The preliminary transport experiments on the empty pressure cells yielded a recovery of more than 99% for solute transport. The first AgNP transport experiment with deionized water achieved a particle recovery at the effluent of 95%. The two following transport experiments with AgNP

Fig. 1. Cumulative pore size distributions of the sandstones determined by mercury porosimetry.

6

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

suspension in 1 mM and 10 mM sodium nitrate solution recovered more than 98%, indicating no significant losses of AgNP due to the experimental setup. The dispersivity of the AgNP through the empty pressure cell was determined by inverse numerical modeling to 4 mm. The correlation coefficients (R2) between the measured and modeled BTCs are higher than 0.99 for solute as well as for the AgNP experiments. 3.2. AgNP characterization The particles in the prepared AgNP suspensions are spherical in shape and have a narrow size distribution with a mean hydrodynamic particle radius of 25.0 ± 2.1 nm, as determined by the retention time in the AF4 channel. The AgNP hydrodynamic radius measured by dynamic light scattering with ionic strength of 10 mM NaNO3 and 1 mM Ca(NO3)2 and with a particle concentration of 0.06 mg mL−1 is 26.0 ± 5.7 nm. The size measurements taken from the influent suspension of approximately 0.5 mg mL−1 AgNP in 10 mM Ca(NO3)2 (Herzogenrather sandstone 2, experiment nos. 4, 5 and 6) at the end of the experiments prove the high stability of AgNP suspension, even at high divalent ionic concentration and demonstrate the effective steric stabilization. The zeta potential of the AgNP suspension is negative between pH = 3 and pH = 10 (Fig. 2). Under the experimental conditions (pH = 7–7.6), it is − 9 mV for deionized water, − 7 mV for 10 mM sodium nitrate and − 5 mV for 1 mM calcium nitrate, which is close to the point of zero charge. However, steric interactions contribute significantly to the total energy barriers of the AgNP (Liang et al., 2013). Consequently, unfavorable conditions were predicted for AgNP retention and aggregation comparable to other studies using the same AgNP product (e.g. Liang et al., 2013). Note that the influence of physical and chemical heterogeneity on retention is not taken into account by this interpretation. 3.3. Hydraulic characterization of the rock samples The hydraulic conductivity determined by flow experiments and the transport parameters determined by inverse

Fig. 2. Measured zeta potentials of AgNP suspensions (0.06 mg mL−1) at varying pH values and electrolyte solutions with different ionic strengths: deionized water (H2O) 10 mM sodium nitrate and 1 mM calcium nitrate.

modeling of solute transport experiments are summarized in Table 3. The hydraulic conductivity varies between 1.53 · 10−6 and 2.2 · 10−10 m s−1 and correlates with the pore size distributions of the sandstones. The determined effective porosities of the rock matrices are in the range of previously published values (Fitzner and Kownatzki, 1991). They were set for the mathematical modeling of the AgNP transport experiments. The determined coefficients of dispersion were taken as an initial estimate for the inverse modeling of the measured AgNP BTCs. Measurement of the hydraulic conductivity for rock samples 10 and 14-1, which showed very limited AgNP transport (Section 3.4), revealed no reduction of permeability due to particle attachment or straining.

3.4. Transport of AgNP through Herzogenrather sandstone The measured BTCs of AgNP in the effluent are plotted as normalized concentrations as a function of pore volumes flushed through the sandstone samples. AgNP sizes are plotted as normalized particle sizes as a function of flushed pore volumes, since the initial AgNP sizes of the various experiments are slightly different. The results of the inverse models and the mass balance information are presented in Table 4. The breakthrough of AgNP in the first sample of the Herzogenrather sandstone with deionized water at a flow rate of 0.6 mL min−1 and C0 of 0.8 mg mL−1 is delayed compared to the conservative solute, while relative concentrations are in the same order of magnitude (Fig. 3). The tailing of the AgNP BTC is very similar to that of sodium chloride. Despite the retention of around 2.5 pore volumes, the AgNP BTC is similar to the solute reference after injection of 4 pore volumes, which indicates conservative transport after the blocking of available attachment sites. AgNP recovery is 58% with an effluent of pH 7. AgNP breakthrough can be adequately modeled by an attachment model considering a maximum AgNP attachment on the sandstone surface. The related inverse model yields a value of ka = 4.01 · 10−3 s−1 and a value for Smax = 2.08 · 10−4, which is in accordance with the value of 3 · 10−4 calculated by mass balancing. Similar effects of time dependent retentions on particle BTCs with diminished retention with time are reported in literature (e.g. Bradford et al., 2009; Liang et al., 2013; Tellam et al., 2011). In order to investigate the long-term transport behavior of AgNP with filled favorable attachment sites and to test physicochemical factors on AgNP transport, additional AgNP transport experiments with the previously used rock sample were executed. The following two experiments (nos. 2 and 3) with deionized water show almost conservative transport of AgNP, with recoveries of 96 and 99% respectively, which is confirmed by inverse transport modeling which revealed negligible deposition rates. Subsequent application of AgNP with 10 mM NaNO3 and 1 mM Ca(NO3)2 electrolyte concentration results in reactive transport with slightly lower relative AgNP concentrations and prolonged breakthrough at 1 mM NaNO3, compared to the conservative transport of AgNP with deionized water. The results of the inverse transport models of experiments 4 and 5 of the first sample of the Herzogenrather sandstone suggest remobilization of AgNP during the experiments due to the change of the hydrochemical conditions, although the standard errors for the fitted parameters ka and kd

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

7

Table 3 Hydraulic and transport parameters of the sandstone samples determined from solute tracer experiments and inverse numerical modeling. Standard errors of measurements and parameter determination respectively are plotted in italics. Rock-sample

Hydraulic conductivity [m s−1]

4-1

1.53 1.89 7.67 5.1 2.21 5.85 2.80 6.98 5.73 7.44

4-2 10 14-1 14-2

× × × × × × × × × ×

10−6 10−7 10−6 10−8 10−10 10−11 10−9 10−10 10−10 10−11

Hydraulic conductivity after AgNP application [m s−1]

Dispersivity [m]

Effective porosity [%]

R2

n.d.

1.7 × 10−3

0.991

n.d.

6.9 × 10−4

18.7 0.6 18.8 0.5 8.8 1.46 12.2 0.08 11.8 0.7

−10

2.01 × 10 3.87 × 10−11 4.23 × 10−9 1.12 × 10−9 n.d.

6.4 × 10

−3

7.6 × 10

−3

9.8 × 10−3

0.992 0.973 0.992 0.979

hydrochemical conditions. The inverse models of experiments 3 and 4 on the second Herzogenrather sample have relatively low correlation coefficients (R2) and their representativeness is limited. Standard errors of the model parameters of experiment 5 are high, which indicates relatively low confidence in the remobilization of AgNP, while model results without consideration of remobilization show even poorer results. The results correspond with observations by Liang et al. (2013), where high influent AgNP concentrations result in rapidly filled deposition sites. On the other hand, low retention in the initial experiment on the second sample indicates differences in available deposition sites: these differences may be due to variations in the mineralogy of the rock samples and associated differences in surface properties and/or heterogeneities of surface charge and variations in pore size distribution. This demonstrates the differences of AgNP studies between well-defined porous

of the inverse model for experiment 5 are relatively high, indicating limited confidence in remobilization. The two initial transport experiments with AgNP suspended in deionized water through the second Herzogenrather sandstone sample proved conservative transport, with AgNP recoveries of 92 and 99% respectively (Fig. 4). The following experiments on the same rock sample with 1 mM Ca(NO3)2, 10 mM Ca(NO3)2 and flow rates of 0.6 and 3 mL min−1 again show reactive transport with decreased recovery rates (43, 80 and 91%) and decreasing deposition coefficients (7.23 × 10−3, 7.88 × 10−4 and 2.35 × 10−4 s−1) with increasing AgNP mass application. The change from deionized water to 1 mM Ca(NO3)2 in experiment 3 is associated with a blocking behavior and a maximum AgNP concentration Smax = 1.39 ∙ 10−4. After the fourth experiment, changes in fluid velocity cause only minor changes to AgNP attachment, indicating a maximum number of deposition sites (Smax) that are dependent on the

Table 4 Fitted model parameters and mass balance information. Standard errors of the fitting parameter are plotted in italics. Model results are for first order deposition, or first order deposition and detachment, or first order deposition and blocking (“−” marks the unconsidered processes). Recovery

Dispersivity αl

[%]

[m]

Rock— sample

Experiment

Solute chemistry

4—1

1

Deionized water

58

2

Deionized water

96

3

Deionized water

99

4

10 mM NaNO3

5

1 mM Ca(NO3)2

4—2

102 96

1

Deionized water

92

2

Deionized water

99

3

1 mM Ca(NO3)2

43

4

10 mM Ca(NO3)2

80

5

10 mM Ca(NO3)2

91

6

10 mM Ca(NO3)2

95

[s−1] × × × × × × × × ×

10−3 10−4 10−3 10−4 10−3 10−4 10−3 10−4 10−3

3.05 7.60 2.39 1.04 3.00

× × × × ×

−3

10 10−4 10−3 10−3 10−3

9.49 1.52 2.07 6.88 5.82 1.04

× × × × × ×

10−4 10−3 10−3 10−4 10−3 10−3

1.80 7.40 1.55 4.94 1.37 2.70 1.33 2.40 1.40

Deposition rate ka

4.01 7.50 1.00 2.50 1.00 1.50 9.90 1.20 4.06 2.20 1.00 2.81 1.00 3.93 7.23 1.29 7.88 1.31 2.35 1.34 1.00 4.42

× × × × × × × × × × × × × × × × × × × × × ×

10−3 10−4 10−8 10−17 10−8 10−17 10−5 10−5 10−5 10−5 10−8 10−17 10−8 10−17 10−3 10−3 10−4 10−4 10−4 10−4 10−4 10−5

Detachment rate kd

Smax

R2

[s−1]

–

–

–

0.992

–

2.04 × 10−4 5.94 × 10−6 –

–

–

0.997

2.70 × 10−3 5.60 × 10−5 7.22 × 10−6 3.58 × 10−5 –

–

0.993

–

0.981

–

0.990

–

–

0.982

–

1.39 × 10−4 1.64 × 10−5 –

0.915

–

0.994

–

0.996

– −3

1.45 × 10 1.20 × 10−3 –

0.992

0.980

8

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

Fig. 3. a) Transport of sodium chloride and AgNPs through the Herzogenrather sandstone matrix, BTCs in deionized water in the first experiment with AgNP transport model. b) Normalized BTCs of repetition experiments 2 and 3 with related models and related particle sizes (open symbols). c) Normalized BTCs of experiments 4 and 5 with related models and relative particle sizes (open symbols). Solute breakthrough is given for comparison. Note that solute breakthrough of experiment 1 differs from those of experiments 2–5.

media where the properties of collector surfaces can be reproduced to a certain level, and real rock samples where naturally occurring differences of mineralogical composition and their often randomly spatial distribution are ubiquitous. The fitted BTC and the recovered AgNP masses show significant changes with increased ionic strength. After changing the ionic strength from deionized water to 10 mM NaNO3 with the first sample and to 1 mM Ca(NO3)2 with the second sample respectively, the deposition rate coefficients increased and for the latter case Smax was increased too. This marks a significant decline of AgNP transport. The following experiments on both rock samples again show decreasing deposition rate coefficients and increasing recovery rates. This observation is consistent with the compression of the electrostatic double layer at higher ionic strength, as reported in previous studies (Liang et al., 2013; Lin et al., 2011; Liu et al., 2011). Liang et al. (2013) reported that the secondary

minimum plays a minor role in AgNP–collector interaction and most of the particles are attached in a primary minimum. The effect of microscopic heterogeneity on ka and Smax increases with ionic strength because of a smaller zone of electrostatic influence with compression of the double layer (Torkzaban et al., 2008), and Liang et al. (2013) concluded from this observation that increased values of ka and Smax are expected to largely reflect the effects of microscopic heterogeneity. The change of the deposition rate coefficient and Smax in the second sample (experiments 3–6) indicates a minor role of flow velocity on particle deposition in comparison to the change of ionic strength, which shows a significant impact on particle deposition, as discussed above. The particle sizes measured in the effluent of the Herzogenrather sandstone samples vary within the range of measurement accuracy for almost all experiments. An exception is experiment 4 of the second sample, where the ionic

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

9

Fig. 4. Solute and AgNP BTC of a): experiments 1 & 2 with deionized water and flow rate = 0.6 mL min−1. b) Experiments 3 & 4 with 1 and 10 mM Ca(NO3)2 respectively and flow rate = 3 mL min−1. c) Experiments 5 & 6 with 10 mM Ca(NO3)2 and flow rate = 0.6 and 3 mL min−1 respectively through the second sample of the Herzogenrather sandstone. Solute chemistry and fluid flow rate are indicated. Open symbols are related to the measured relative AgNP size. The arrows mark the changes from AgNP dispersions to AgNP-free influent.

strength changed from 1 mM to 10 mM Ca(NO3)2, resulting in larger particle sizes or aggregates at the end of the breakthrough, up to relative particle sizes of 1.24. 3.5. Transport of AgNP through Obernkirchner and Solling sandstones For the Obernkirchner and Solling sandstones, no AgNP breakthrough was observed in initial experiments with injection of 11 and 30 pore volumes AgNP suspension in deionized water respectively. Visual examination of the samples after the first AgNP application indicates small transport length. As

indicated by the results from mercury injection porosimetry, straining of AgNP in small pore throats and subsequent attachment on the rock surface is probably one of the relevant transport processes. Filtration and straining of AgNPs in these sandstone matrices is effective. The importance of primary minimum interaction is demonstrated by the irreversible attachment of AgNP on the inflow side of the rock sample, which was visible after removal of the sample from the pressure cell. Neither treatment of the samples in a sonication bath nor conditioning with deionized water was able to remove a detectable amount of AgNP from the rock surface. Considering the attachment of AgNP on the sample inflow, retention profiles

10

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

are probably hyperexponentially shaped, which might be due to straining and/or filtration when the flux adjacent to the solid surface is the dominant mass transfer mechanism. Among other factors, decreased grain size and associated smaller pore sizes promote particle attachment (Bradford et al., 2011). In addition, the strong filtration and probably straining results in partial retention of AgNP in the lower filter material. This effect does not occur during the experiments with the Herzogenrather sandstone, which indicates the conservative behavior of the filter material, as also shown by the preliminary experiments with the empty pressure cells. An active interaction of the AgNPs with the filter plate material is therefore precluded and the effect is interpreted as emerging from pore space filling of the sandstone samples and subsequent partial clogging of the filter pore space by AgNP. Strong physicochemical filtration of AgNP on the surface of the Solling sandstone, with its relatively high amount of iron oxide, is in accordance with observations from previous studies (e.g. Li and Logan, 2004; Vadillo-Rodríguez and Logan, 2006), where positively charged metal oxides caused increased attachment of negatively charged particles. The hydraulic conductivity of both samples does not change significantly after AgNP application, which might be caused by the relatively low injected mass of 1.87 mg (Solling) and 4.48 mg (Obernkirchner). Repetition of the experiment with the first sample of the Obernkirchner sandstone with AgNP in deionized water for 30 pore volumes and subsequent rinsing of the sample with AgNP-free influent with 1 mM NaNO3 for another 33 pore volumes at varying fluid flow rates, and deionized water for another 10 pore volumes at constant fluid flow velocity, resulted in a distinct AgNP breakthrough in the effluent of the rock sample (Fig. 5a). No AgNP breakthrough could be detected during the injection of the first 30 pore volumes of AgNP application. Shortly after changing the solute ionic conditions to AgNP-free influent with 1 mM NaNO3, AgNP could be detected in the effluent with a sharp rise in concentration that is followed by a nearly continuous decrease of AgNP, indicating mobilization. The change of the volumetric flow rate from

0.013 mL min−1 to 0.7 mL min−1 results in a decrease of AgNP concentration that is reversed by changing the flow rate back to 0.013 mL min−1. The overall recovery of AgNP during the experiment is 3.7%. No mathematical model could be fitted adequately to the experimental data. This observation indicates that a certain portion of AgNP might be mobilized due to changes in the physicochemical conditions and/or fluid velocity that control mass transfer between the bulk phase and the solid phase. These findings were tested on a second sample of the Obernkirchner sandstone, applying AgNP for 25.5 pore volumes at a constant solute concentration of 1 mM NaNO3. The resulting AgNP breakthrough starts after 15 pore volumes and reaches a plateau of C/C0 ~ 0.1 at about 19 pore volumes. The overall recovery is 8.5%. Thus, the experimental data of the second sample indicates also increased mobility at moderate electrolyte concentrations compared to conditions with low ionic strength. Due to the variability of sandstone composition and associated behavior concerning surface charge heterogeneity for instance and related effects on particle transport more experimental results are needed to verify this interpretation. Application of inverse mathematical modeling yielded no representative transport parameters. The first model (Model 1 in Fig. 5b) takes into account a first-order deposition coefficient and a maximum solid phase concentration. The correlation coefficient R2 is only 0.39. Introduction of a first-order detachment coefficient (Model 2 in Fig. 5b) improves the fitting result slightly (R2 = 0.50), although the model does not represent the measured AgNP transport and the tailing of the BTC in particular. The increased mobility of AgNP in the Obernkirchner sandstone at moderate ionic strength might be explained by the compression of the double layer and associated decrease of the long range energy barrier, which reduces the effective pore space available for AgNP transport. Assuming a plate– sphere interaction model (Elimelech et al., 1995) to calculate the DLVO energy profile between the Obernkirchner sandstone and AgNP (Hamaker constant = 1.3 × 10−20 J) in the experimental conditions considering the measured collector

Fig. 5. a) AgNP BTC and eluted particle sizes for the first Obernkirchner sample with indication of the fluid flow rate, injection function and application of deionized water (shaded areas) and 1 mM NaNO3 (blank area). b) Solute and AgNP breakthrough and the eluted relative particle sizes for the second Obernkirchner sample at 1 mM NaNO3 with indication of the flow rate. Model 1 considers a first-order deposition coefficient and a maximum solid phase concentration. Model 2 considers a first-order detachment coefficient in addition.

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

surface charge = − 45 mV, the measured surface charge of AgNP = − 7 mV (both at pH = 6.5), and a measured AgNP radius of 21 nm with monovalent deionized water (molarity of 0.01 mM) and 1 mM NaNO3, the range of repulsive energy from the collector surface is 1500 nm and 69 nm respectively. This demonstrates the effect of double layer compression on particle transport through sandstones with high proportions of small and intermediate pore sizes. Visible AgNP retention on the inflow of the second sample after AgNP application is not observable, in contrast to the first sample. This fact indicates, as well as the strong influence of solute ionic strength on AgNP transport, the relative importance of physicochemical filtration on AgNP transport through the Obernkirchner sandstone. The eluted particle sizes with the first Obernkirchner sample show variations within the scope of measurement precision. The eluted particles in the experiment with the second sample show decreasing particle sizes with the eluted fluid volume. At the end of the experiment, particle sizes increase significantly. 4. Conclusion This study discusses the mobility of AgNPs in the three types of sandstone and shows that transport depends on the ionic strength of the pore fluid, the size of the pores, the mineralogy and the specific total mass of the applied AgNP. The highest mobility occurs in sandstones with uniform and relatively large pore spaces, whereas mobility is very limited for sandstones with small and intermediate pore sizes. The experimental results indicate enhanced transport of AgNP through sandstones with intermediate pore sizes by higher monovalent ionic strength, probably due to the compression of the electrostatic double layer but more experimental results are needed to test this hypothesis. The BTCs observed at the outflow of Herzogenrather sandstone show retention of AgNPs and subsequent equilibrium transport after blocking of deposition sites at the beginning of the experiment. This behavior differs from transport of other nanoparticles, e.g. C60 in silicate glass beads (Espinasse et al., 2007) or single-walled carbon nanotubes in quartz sand (Jaisi et al., 2008), where deposition of NP is increased by NPs already attached to collector surfaces (ripening). Physicochemical filtration dominates the transport of AgNP through the sandstone matrix. The deposition of AgNP onto the rock surface is predominantly irreversible, which indicates attachment in the primary energy minimum. Subsequent clogging of the pore space with associated limited AgNP transport might be a secondary important effect. However, a reduction of the hydraulic conductivity of the sandstone samples could not be observed. Blocking of attachment sites has been reported frequently in literature for transport of colloids, microorganisms and nanoparticles in porous media (e.g., Bradford and Bettahar, 2006; Gargiulo et al., 2007; Kasel et al., 2013; Wang et al., 2011, 2012). The blocking of attachment sites changes the transport behavior of AgNP towards less reactive transport (higher mobility of AgNP). This behavior differs considerably from AgNP transport through quartz sand or soil mixtures, where blocking of sorption sites is not considered (Sagee et al., 2012; Tian et al., 2010). Consequently, the applicability of

11

AgNP transport parameters from experiments with model systems to realistic environmental matrices with physical and chemical heterogeneity might be problematical. Local surface charge heterogeneity has been reported for silica surfaces (Elimelech et al., 1995). Spatial distribution of AgNPs along the flow path remains unknown, since appropriate measurement techniques for visualization or concentration determination are not available. The importance of taking into account the retention profiles in transport process interpretation is emphasized by e.g. Kasel et al. (2013). Although fitting of the BTCs gained reasonable results, the unknown attached particle distribution and the associated missing information in the model results regarding particle retention restrict the quality of the applied models. The development of techniques for determination of particle retention on nanometer scale in rock samples is of particular importance for the interpretation of AgNP transport experiments in environmental studies, since the combination of modeling the BTC and modeling the spatial retention of NP for the same experimental dataset gives important information for reliable interpretation of transport processes associated with ENP. The experimental results partly indicate the high mobility of AgNP through porous sandstones, and the spread of AgNP in aquifers may be the consequence. The results of the experiments in consolidated material with intermediate and small pore sizes indicate that restricted transport under the applied boundary conditions and associated effects of interdependency between fissures and matrix in AgNP transport might be important for spreading. Hence, more research is needed concerning AgNP mobility in fissured media, since water flow and contaminant transport in consolidated rocks occur mainly in those compartments which are characterized by high hydraulic conductivities and interaction between fissures and matrix may also play an important role in contaminant transport. Acknowledgments This work was funded by the German Ministry of Education and Research (BMBF) within the WING/NanoNature program under contract 03X0077A. The authors are solely responsible for the content of this publication. We thank three anonymous referees for their comments on the manuscript and John H. Tellam for his suggestions to improve our paper. References Bradford, S.A., Bettahar, M., 2006. Concentration dependent transport of colloids in saturated porous media. J. Contam. Hydrol. 82 (1–2), 99–117. http://dx.doi.org/10.1016/j.jconhyd.2005.09.006. Bradford, S.A., Yates, S.R., Bettahar, M., Simunek, J., 2002. Physical factors affecting the transport and fate of colloids in saturated porous media. Water Resour. Res. 38 (12), 63-1–63-12. http://dx.doi.org/10.1029/ 2002WR001340. Bradford, S.A., Simunek, J., Bettahar, M., van Genuchten, M.T., Yates, S.R., 2003. Modeling colloid attachment, straining, and exclusion in saturated porous media. Water Resour. Res. 37 (10), 2242–2250. http://dx.doi.org/ 10.1021/es025899u. Bradford, S.A., Simunek, J., Walker, S.L., 2006. Transport and straining of E. coli O157:H7 in saturated porous media. Water Resour. Res. 42 (12), W12S12. http://dx.doi.org/10.1029/2005WR004805. Bradford, S.A., Kim, H.N., Haznedaroglu, B.Z., Torkzaban, S., Walker, S.L., 2009. Coupled factors influencing concentration-dependent colloid

12

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13

transport and retention in saturated porous media. Environ. Sci. Technol. 43 (18), 6996–7002. http://dx.doi.org/10.1021/es900840d. Bradford, S.A., Torkzaban, S., Simunek, J., 2011. Modeling colloid transport and retention in saturated porous media under unfavorable attachment conditions. Water Resour. Res. 47 (10), W10503. http://dx.doi.org/ 10.1029/2011WR010812. El Badawy, A.M., Luxton, T.P., Silva, R.G., Scheckel, K.G., Suidan, M.T., Tolaymat, T.M., 2010. Impact of environmental conditions (pH, ionic strength, and electrolyte type) on the surface charge and aggregation of silver nanoparticles suspensions. Environ. Sci. Technol. 44, 1260–1266. http://dx.doi.org/10.1021/es902240k. Elimelech, M., Gregory, J., Jia, X., William, R.A., 1995. Particle Deposition and Aggregation: Measurement, Modelling and Simulation. ButterworthHeinemann, London. Espinasse, B., Hotze, E.M., Wiesner, M.R., 2007. Transport and retention of colloidal aggregation of C60 in porous media: effects of organic macromolecules, ionic composition, and preparation method. Environ. Sci. Technol. 41, 7396–7402. http://dx.doi.org/10.1021/es0708767. Fang, J., Shan, X., Wen, B., Lin, J., Owens, G., 2009. Stability of titania nanoparticles in soil suspensions and transport in saturated homogeneous soil columns. Environ. Pollut. 157, 1101–1109. http://dx.doi.org/ 10.1016/j.envpol.2008.11.006. Fitzner, B., Kownatzki, R., 1991. Porosity characteristics and weathering behaviour of sedimentary ashlar. Bauphysik 13 (4), 111–119. Gargiulo, G., Bradford, S., Simunek, A.J., Ustohal, P., Vereecken, H., Klumpp, E., 2007. Bacteria transport and deposition under unsaturated conditions: the role of matrix grain size and the bacteria surface protein. J. Contam. Hydrol. 92 (3–4), 255–273. http://dx.doi.org/10.1016/j.jconhyd.2007.01.009. Guzman, K.A.D., Finnegan, M.P., Banfield, J.F., 2006. Influence of surface potential on aggregation and transport of titania nanoparticles. Environ. Sci. Technol. 40, 7688–7693. http://dx.doi.org/10.1021/es060847g. Huynh, K.A., Chen, K.L., 2011. Aggregation kinetics of citrate and polyvinylpyrrolidone coated silver nanoparticles in monovalent and divalent electrolyte solutions. Environ. Sci. Technol. 45, 5564–5571. http:// dx.doi.org/10.1021/es200157h. Jaisi, D.P., Elimelech, M., 2009. Single walled carbon nanotubes exhibit limited transport in soil columns. Environ. Sci. Technol. 43, 9161–9166. http://dx.doi.org/10.1021/es901927y. Jaisi, D.P., Saleh, N.B., Blake, R.E., Elimelech, M., 2008. Transport of singlewalled carbon nanotubes in porous media: filtration mechanisms and reversibility. Environ. Sci. Technol. 42, 8317–8323. http://dx.doi.org/ 10.1021/es801641v. Kaegi, R., Voegelin, A., Sinnet, B., Zulegg, S., Hagendorfer, H., Burkhardt, M., Siegrist, H., 2011. Behavior of metallic silver nanoparticles in a pilot wastewater treatment plant. Environ. Sci. Technol. 45, 3902–3908. http://dx.doi.org/10.1021/es1041892. Kasel, D., Bradford, S.A., Simunek, J., Heggen, M., Vereecken, H., Klumpp, E., 2013. Transport and retention of multi-walled carbon nanotubes in saturated porous media: effects of input concentration and grain size. Water Res. 47, 933–944. http://dx.doi.org/10.1016/j.watres.2012.11.019. Klaine, S.J., Alvarez, P.J.J., Batley, G.E., Fernandes, T., Handy, R.D., Lyon, D.Y., Mahendra, S., McLaughlin, M.J., Lead, J.R., 2008. Nanomaterials in the environment: behavior, fate, bioavailability, and effects. Environ. Toxicol. Chem. 27, 1825–1851. http://dx.doi.org/10.1897/08-090.1. Klein, C.L., Comero, S., Stahlmecke, B., Romazanov, J., Kuhlbusch, T.A.J., Van Doren, E., De Temmerman, P.-J., Mast, J., Wick, P., Krug, H., Locoro, G., HundRinke, K., Kördel, W., Friedrichs, S., Maier, G., Werner, J., Linsinger, Th., Gawlik, B.M., 2011. NM-300 silver—characterisation, stability, homogeneity. NM-Series of Representative Manufactured Nanomaterials, European Commission Joint Research Centre, Institute for Health and Consumer Protection. ISBN: 978-92-79-19068-1. Lecoanet, H.F., Bottero, J.Y., Wiesner, M.R., 2004. Laboratory assessment of the mobility of nanomaterials in porous media. Environ. Sci. Technol. 38, 5164–5169. http://dx.doi.org/10.1021/es0352303. Leocanet, H.F., Wiesner, M.R., 2004. Velocity effects on fullerene and oxide nanoparticles deposition in porous media. Environ. Sci. Technol. 38, 4377–4382. http://dx.doi.org/10.1021/es035354f. Levard, C., Hotze, E.M., Lowry, G.V., Brown Jr., G.E., 2012. Environmental transformations of silver nanoparticles: impact on stability and toxicity. Environ. Sci. Technol. 46 (13), 6900–6914. http://dx.doi.org/10.1021/ es2037405. Li, X., Johnson, W.P., 2005. Nonmonotonic variations in deposition rate coefficients of microspheres in porous media under unfavorable deposition conditions. Environ. Sci. Technol. 39 (6), 1658–1665. http:// dx.doi.org/10.1021/es048963b. Li, B., Logan, B.E., 2004. Bacterial adhesion to glass and metal-oxide surfaces. Colloids Surf. B: Biointerfaces 36 (2), 81–90. http://dx.doi.org/10.1016/ j.colsurfb.2004.05.006. Li, X., Scheibe, T.D., Johnson, W.P., 2004. Apparent decreases in colloid deposition rate coefficients with distance of transport under unfavorable

deposition conditions: a general phenomenon. Environ. Sci. Technol. 38 (21), 5616–5625. http://dx.doi.org/10.1021/es049154v. Li, Y., Wang, Y., Pennell, K.D., Abriola, L.M., 2008. Investigation of the transport and deposition of fullerene (C60) nanoparticles in quartz sands under varying flow conditions. Environ. Sci. Technol. 42, 7174–7180. http://dx.doi.org/10.1021/es801305y. Liang, Y., Bradford, S.A., Simunek, J., Vereecken, H., Klumpp, E., 2013. Sensitivity of the transport and retention of stabilized silver nanoparticles to physicochemical factors. Water Res. 42 (7), 2572–2582. http:// dx.doi.org/10.1016/j.watres.2013.02.025. Lin, S., Cheng, Y., Bobcombe, Y., Jones, L.K., Liu, J., Wiesner, M.R., 2011. Deposition of silver nanoparticles in geochemically heterogeneous porous media: predicting affinity from surface composition analysis. Environ. Sci. Technol. 45 (12), 5209–5215. http://dx.doi.org/10.1021/es2002327. Lin, S., Cheng, Y., Liu, J., Wiesner, M.R., 2012. Polymeric coatings on silver nanoparticles hinder autoaggregation but enhance attachment to uncoated surfaces. Langmuir 28 (9), 4178–4186. http://dx.doi.org/ 10.1021/la202884f. Liu, W., Zhou, Q., Liu, J., Fu, J., Liu, S., Jiang, G., 2011. Environmental and biological influences on the stability of silver nanoparticles. Chin. Sci. Bull. 56 (19), 2009–2015. http://dx.doi.org/10.1007/s11434-010-4332-8. Lowry, G.V., Espinasse, B.P., Badireddy, A.R., Richardson, C.J., Reinsch, B.C., Bryant, L.D., Bone, A.J., Deonarine, A., Chae, S., Therezien, M., Colman, B.P., Hsu-Kim, H., Bernhardt, E.S., Matson, C.W., Wiesner, M.R., 2012. Long-term transformation and fate of manufactured Ag nanoparticles in a simulated large scale freshwater emergent wetland. Environ. Sci. Technol. 46, 7027–7036. http://dx.doi.org/10.1021/es204608d. Penell, K.D., Costanza, J., Wang, Y., 2008. Transport and retention of nanomaterials in porous media. In: Grassian, V.H. (Ed.), Nanoscience and Nanotechnology: Environmental and Health Impacts. John Wiley & Sons, Hoboken, NJ. http://dx.doi.org/10.1002/9780470396612.ch5 (469 pp.). Reinsch, B.C., Levard, C., Li, Z., Ma, R., Wise, A., Gregory, K.B., Brown Jr., G.E., Lowry, G.V., 2012. Sulfidation of silver nanoparticles decreases Escherichia coli growth inhibition. Environ. Sci. Technol. 46 (13), 6992–7000. http:// dx.doi.org/10.1021/es203732x. Sagee, O., Dror, I., Berkowitz, B., 2012. Transport of silver nanoparticles (AgNPs) in soil. Chemosphere 88 (5), 670–675. http://dx.doi.org/ 10.1016/j.chemosphere.2012.03.055. Simunek, J., van Genuchten, M.Th, 2008. Modeling nonequilibrium flow and transport processes using HYDRUS. Vadose Zone J. 7 (2), 782–797. http://dx.doi.org/10.2136/wzj2007.0074. Solovitch, N., Labille, J., Rose, J., Chaurand, P., Borschneck, D., Wiesner, M.R., Bottero, J., 2010. Concurrent aggregation and deposition of TiO2 nanoparticles in a sandy porous media. Environ. Sci. Technol. 44, 4897–4902. http://dx.doi.org/10.1021/es1000819. Song, J.E., Phenrat, T., Marinakos, S., Xiao, Y., Liu, J., Wiesner, M.R., Lowry, G.V., 2011. Hydrophobic interactions increase attachment of gum arabicand PVP-coated Ag nanoparticles to hydrophobic surfaces. Environ. Sci. Technol. 45 (14), 5988–5995. http://dx.doi.org/10.1021/es200547c. Taboada-Serrano, P., Vithayaveroj, V., Yiacoumi, S., Tsouris, C., 2005. Surface charge heterogeneities measured by atomic force microscopy. Environ. Sci. Technol. 39 (17), 6352–6360. http://dx.doi.org/10.1021/es050100a. Tellam, J.H., Greswell, R.B., Riley, M.S., Rahman, S.H., 2011. Manufactured nanoparticle movement in the groundwaters of a redbed sandstone: laboratory experiments and field observations. In: Schirmer, M., Hoehn, E., Vogt, T. (Eds.), GQ10: Groundwater Quality Management in a Rapidly Changing World. IAHS Publ, 342. ISBN: 978-1-907161-16-2, pp. 50–54. Thio, B.J.R., Montes, M.O., Mahmoud, M.A., Lee, D.W., Zhou, D., Keller, A.A., 2012. Mobility of capped silver nanoparticles under environmentally relevant conditions. Environ. Sci. Technol. 46 (13), 6985–6991. http:// dx.doi.org/10.1021/es203596w. Tian, Y., Gao, B., Silvera-Bastista, C., Ziegler, K.J., 2010. Transport of engineered nanoparticles in saturated porous media. J. Nanoparticle Res. 12, 2371–2380. http://dx.doi.org/10.1007/s11051-010-9912-7. Torkzaban, S., Tazehkand, S.S., Walker, S.L., Bradford, S.A., 2008. Transport and fate of bacteria in porous media: coupled effects of chemical conditions and pore space geometry. Water Resour. Res. 44 (4), W04403. http://dx.doi.org/10.1029/2007WR006541. Tosco, T., Sethi, R., 2009. MNM1D: a numerical code for colloid transport in porous media: implementation and validation. Am. J. Environ. Sci. 5 (4), 516–524. http://dx.doi.org/10.3844/ajessp.2009.517.525. Tufenkji, N., Elimelech, M., 2004. Correlation equation for predicting singlecollector efficiency in physicochemical filtration in saturated porous media. Environ. Sci. Technol. 38, 529–536. http://dx.doi.org/10.1021/es034049r. Tufenkji, N., Elimelech, M., 2005. Breakdown of colloid filtration theory: role of the secondary energy minimum and surface charge heterogeneities. Langmuir 21, 841–852. Tufenkji, N., Redman, J.A., Elimelech, M., 2003. Interpreting deposition patterns of microbial particles in laboratory scale column experiments. Environ. Sci. Technol. 37 (3), 616–623. http://dx.doi.org/10.1021/es025871i.

C. Neukum et al. / Journal of Contaminant Hydrology 158 (2014) 1–13 Vadillo-Rodríguez, V., Logan, B.E., 2006. Localized attraction correlates with bacterial adhesion to glass and metal oxide substrata. Environ. Sci. Technol. 40 (9), 2983–2988. http://dx.doi.org/10.1021/es052365v. Wang, Y., Li, Y., Fortner, J.D., Hughes, J.B., Abriola, L.M., Pennell, K.D., 2008. Transport and retention of nanoscale C60 aggregates in water-saturated porous media. Environ. Sci. Technol. 42 (10), 3588–3594. http:// dx.doi.org/10.1021/es800128m. Wang, D., Paradelo, M., Bradford, S.A., Peijnenburg, W.J.G.M., Chu, L., Zhou, D., 2011. Facilitated transport of Cu with hydroxyapatite nanoparticles in saturated sand: effects of solution ionic strength and composition. Water Res. 45 (18), 5905–5915. http://dx.doi.org/10.1016/j.watres.2011.08.041. Wang, Y., Kim, J.H., Baek, J.B., Miller, G.W., Pennell, K.D., 2012. Transport behaviour of functionalized multi-wall carbon nanotubes in watersaturated quartz sand as a function of tube length. Water Res. 46 (14), 4521–4531. http://dx.doi.org/10.1016/j.watres.2012.05.036.

13

Westerhoff, P., Nowack, B., 2012. Searching for global descriptors of engineered nanomaterial fate and transport in the environment. Acc. Chem. Res. http://dx.doi.org/10.1021/ar300030n. Wiesner, M.R., Lowry, G.V., Dionysiou, D., Biswas, P., 2006. Assessing the risk of manufactured nanomaterials. Environ. Sci. Technol. 40, 4336–4345. http://dx.doi.org/10.1021/es062726m. WWICS (Woodrow Wilson International Center for Scholars), 2012. Project on Emerging Nanotechnology. http://www.nanotechproject.org/ (assessed May 2012). Yao, K.M., Habibian, M.T., O'Melia, C.R., 1971. Water and waste water filtration. Concepts and applications. Environ. Sci. Technol. 5, 1105–1112. http:// dx.doi.org/10.1021/es60058a005. Yu, S., Yin, Y., Liu, J., 2013. Silver nanoparticles in the environment. Environ. Sci. Process. Impact 15, 78–92. http://dx.doi.org/10.1039/ c2em30595j.