Transport of Fullerene Nanoparticles (nC60) in ... - ACS Publications

Jun 6, 2012 - ... (nC60) in Saturated Sand and Sandy Soil: Controlling Factors and Modeling ...... Peide Sun , Altantuya Shijirbaatar , Jing Fang , Ga...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/est

Transport of Fullerene Nanoparticles (nC60) in Saturated Sand and Sandy Soil: Controlling Factors and Modeling Lunliang Zhang,† Lei Hou,† Lilin Wang,† Amy T. Kan,‡ Wei Chen,*,† and Mason B. Tomson‡ †

College of Environmental Science and Engineering/Ministry of Education Key Laboratory of Pollution Processes and Environmental Criteria/Tianjin Key Laboratory of Environmental Remediation and Pollution Control, Nankai University, Wei Jin Road 94, Tianjin 300071, China ‡ Department of Civil and Environmental Engineering, Rice University, 6100 Main Street, Houston, Texas 77005, United States S Supporting Information *

ABSTRACT: Understanding subsurface transport of fullerene nanoparticles (nC60) is of critical importance for the benign use and risk management of C60. We examined the effects of several important environmental factors on nC60 transport in saturated porous media. Decreasing flow velocity from approximately 10 to 1 m/d had little effect on nC60 transport in Ottawa sand (mainly pure quartz), but significantly inhibited the transport in Lula soil (a sandy, low-organic-matter soil). The difference was attributable to the smaller grain size, more irregular and rougher shape, and greater heterogeneity of Lula soil. Increasing ionic strength and switching background solution from NaCl to CaCl2 enhanced the deposition of nC60 in both sand and soil columns, but the effects were more significant for soil. This was likely because the clay minerals (and possibly soil organic matter) in soil responded to changes of ionic strength and species differently than quartz. Anions in the mobile phase had little effect on nC60 transport, and fulvic acid in the mobile phase (5.0 mg/L) had a small effect in the presence of 0.5 mM Ca2+. A two-site transport model that takes into account both the blocking-affected attachment process and straining effects can effectively model the breakthrough of nC60.



soils,10,11 and the transport behaviors are even more complex. Cheng et al.11 observed that while in general higher flow velocity leads to greater breakthrough of nC60 from a sandy soil column, the breakthrough of nC60 dropped to zero after 57 pore volumes (PV) under a pore-water velocity of 0.86 m/d, and this phenomenon was attributed to filter ripening or particle straining. Recently, Wang et al.10 studied nC60 transport in two soils, and found that at a pore-water velocity of 8 m/d, breakthrough of nC60 was observed in neither soil even after a continuous pumping up to 65 PV. The complex transport properties of nC60 in soils likely stem from the heterogeneous nature of soil grains, both physically (e.g., size, shape, roughness) and chemically (e.g., surface charge). For example, both wedging (retention of particles at two bounding surfaces) and bridging (when multiple particles collide and are retained in a pore constriction)12referred to collectively as “straining” hereafter and argued as an important colloid deposition mechanism under unfavorable attachment conditions13−17 are likely more significant for natural soils. Furthermore, we

INTRODUCTION With the increasing production and use of buckminsterfullerene (C 60) and its derivatives, the potential environmental implications of these engineered carbon nanomaterials have received much attention.1−3 In particular, ecological and human-health risks of stable colloidal suspensions of C60, that is, fullerene nanoparticles (nC60), have been studied extensively, because nC60 is likely the most important form of C60 in aqueous environments.1 The environmental impact of nC60 is largely dependent on its transport and mobility.1,4,5 Thus, understanding the transport of nC60 is of critical importance for the benign use and risk management of C60. Thus far, only a few studies have been conducted to understand the transport of nC60 in saturated porous media, mainly using homogeneous materials such as glass beads6,7 and pure sand.5,8,9 However, even among these homogeneous porous materials, the transport properties can vary significantly. For example, Lecoanet and Wiesner7 found that varying Darcy velocity from 120 m/d to 34.6 m/d had little influence on the migration and deposition of nC60 in a glass-bead column. Nonetheless, Li et al.9 reported that changing pore-water velocity had a significant effect on nC60 transport in sand columns, especially for the columns packed with finer sands. To date, only two reports are available on nC60 transport in natural © 2012 American Chemical Society

Received: Revised: Accepted: Published: 7230

March 29, 2012 May 10, 2012 June 6, 2012 June 6, 2012 dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Table 1. Experimental Protocols of Column Tests column properties

length(cm)

bulk density (g/cm3)

porosity (−)

influent properties nC60 conc. (mg/L)

pH

figure

6.0

6.8

Figure 1a

6.0 6.0

7.2 7.3

Figure 1b Figure 1c

6.0

7.0

6.0

7.5

DI water →1 mM NaCl → 2 mM NaCl → 5 mM NaCl → 10 mM NaCl DI water→ 1.5 mM NaCl → 0.5 mM CaCl2 DI water→ 1.5 mM NaCl → 0.5 mM CaCl2 1.5 mM NaCl → 0.5 mM Na2SO4

6.0

7.1

6.0

6.7

6.0

7.0

5.8

6.5

Figure 2a SI Figure S5a Figure 2b SI Figure S5b Figure 2c SI Figure S5c Figure 4a SI Figure S7a Figure 4b SI Figure S7b SI Figure S6a

1.5 mM NaCl → 0.5 mM Na2SO4 1.5 mM NaCl + 5.0 mg/L fulvic acid (FA) → 1.5 mM NaCl → 0.5 mM CaCl2 → 0.5 mM CaCl2 + 5.0 mg/L FA

5.9 6.0

7.1 7.1

SI Figure S6b Figure 5

pore-water velocity (m/d)

column no.

porous medium

1

Ottawa sand Lula soil Lula soil

6.4

1.63

0.38

pulse

8.8 → 0.7

6.1 6.1

1.45 1.40

0.44 0.46

pulse continuous

6.3

1.67

0.36

pulse

5

Ottawa sand Lula soil

6.0

1.40

0.46

pulse

10 → 1.0 1 mM NaCl 1 mM NaCl 10 → 5.0 → 3.0 → 2.0 → 1.0 9.1 Deionized (DI) water →1 mM NaCl → 6 mM NaCl 8.6 DI water→ 1 mM NaCl → 6 mM NaCl

6

Lula oil

6.1

1.41

0.46

continuous

10

7

Ottawa sand Lula soil

6.8

1.63

0.38

continuous

10

6.1

1.45

0.45

continuous

10

Ottawa sand Lula soil Lula soil

6.7

1.65

0.37

continuous

10

6.2 6.0

1.45 1.44

0.45 0.45

continuous continuous

10 10

2 3

4

8 9 10 11

injection mode

background solution 1 mM NaCl



MATERIALS AND METHODS Materials. Sublimed fullerene powder (C60, >99.5%) was purchased from SES Research (Houston, TX). Tritiated water (3H2O) was purchased from Amersham Co. (Arlington Heights, IL). Ottawa sand, with an average grain size of 250 μm, was obtained from EM Science (Gibbstown, NJ). Lula soil, from a ranch near Lula, OK, was provided by Dr. John Wilson of Robert S. Kerr Environmental Research Lab. The average grain size of Lula soil is approximately 120 μm. The fractional organic carbon of the soil is 0.36%.4 The soil contains 45% sand, 36% silt, and 19% clay. The Brunauer−Emmett−Teller surface area of the soil is 8.38 m2/g, and the cation exchange capacity is 10.9 meq/100 g (A&L Plains Laboratories Inc., Lubbock, TX). Preparation of nC60 Stock Suspension. Stock suspension of nC60 was prepared using a previously developed solventexchange method.19 The detailed procedures are given in Supporting Information (SI). The concentration of nC60 in the stock suspension was determined with an oxidation−toluene extraction procedure by Fortner et al.20 Concentration of C60 in the toluene phase was determined by measuring UV absorbance at 344 nm of the extract based on a pre-established calibration curve of C60 in toluene. The total carbon content of the stock suspension was also checked, using a high sensitivity total organic carbon analyzer (Shimadzu Scientific Instruments, Columbia, MD). Particle size (175 ± 50 nm) and ζ potential (−45 ± 10 mV) of the nC60 stock suspension were measured by dynamic light scattering and electrophoretic mobility analyses, respectively, using a Zetasizer Nano (Marlvern Instruments Ltd., Worcestershire, UK). Column Transport Experiments. The apparatus of the column experiments is depicted in SI Figure S1. Ottawa sand or Lula soil was dry-packed into Omnifit borosilicate glass columns (10 × 0.66 cm, Bio-Chem Valve Inc., Boonton, NJ) with 10 μm stainless-steel screens (Valco Instruments Inc., Houston, TX) on both ends. Each column contained

hypothesize that nC60 transport in soils may respond to the changes of hydrodynamic forces and solution chemistry very differently than transport in more homogeneous porous media. In several earlier reports, the clean-bed filtration theory (CFT)18 was applied to interpret the retention of nC60 by glass beads.5,7,9 According to the CFT, the deposition process consists of two simple steps: transport of particles to the vicinity of collectors, and attachment of particles to the collector surfaces. However, the CFT cannot simulate the observed asymmetry in breakthrough curves (BTCs) of nC60 even in glass bead columns.5 Thus, a modified CFT model was used to simulate the BTCs of nC605,9by taking into account the blocking effect on nanoparticle attachment (i.e., the surface exclusion phenomenon during particle deposition due to the repulsion of earlier deposited particles on grain surfaces), this model can provide reasonable fits to most of the BTCs of nC60 in glass beads and pure sand.5,9 To accurately simulate the transport of nC60 in heterogeneous porous media (such as soils and aquifer materials), the straining effects should probably be incorporated in transport modeling; however, this has not been tested. The objectives of this study were (1) to verify the hypothesis that changes of flow conditions and solution chemistry affect nC60 transport in natural soils differently than in homogeneous porous media; and (2) to demonstrate that two-site transport models that incorporate both the blocking-affected attachment process and the straining effects can better model nC60 transport in heterogeneous porous media. The effects of flow velocity, ionic strength, ionic species, and dissolved organic matter on the transport of nC60 were examined, with special attention paid to the differences between a representative homogeneous porous medium (Ottawa sand) and a representative heterogeneous porous medium (Lula soil). The traditional CFT, the modified CFT model by incorporating the blocking effect, and a two-site transport model were used to fit the BTCs under various flow and water chemistry regimes. 7231

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Figure 1. Effect of flow velocity on transport of nC60. (a) and (b): BTCs in Ottawa sand (Column 1) and in Lula soil (Column 2), under pulse injection mode using two different pore-water velocities; (c): BTC in Lula soil (Column 3) under continuous injection mode, using five different pore-water velocities (vertical dotted lines indicate where pore-water velocities were changed). Solid lines (), dash dotted lines (−••−), and dashed lines (− −) were plotted by fitting the BTCs with the two-site transport model, the clean-bed filtration theory (CFT) model, and the modified CFT model, respectively. The thick gray lines show the injected concentration of nC60.

the influent and effluent were determined to understand the potential aggregation of nC60 particles at higher ionic strength or in the presence of Ca2+. No significant changes in nC60 particle size were observed (SI Figure S2). Both the pulse-injection mode and the continuous-injection mode were used in the column tests. For the pulse-injection mode, an nC60-containing influent was first pumped into the column for a predetermined number of PV; then an nC60-free influent with the same electrolyte was pumped into the column until no nC60 could be detected in the effluent; next, without changing the column, the same nC60-containing influent at a different flow velocity, or another nC60-containing influent (in a different background solution from the first one) was pumped through the column (e.g., see Figure 1a). For the continuousinjection mode, the flushing step with the nC60-free solution was not included between the two pumping steps that involve different nC60-containing influents, or the same nC60-containing influent pumped at different flow velocities (e.g., see Figure 1c).

approximately 3 g of soil or 4 g of sand (dry-weight) with an average length of 6.0−6.8 cm. The porosity and dead volume were determined with the 3H2O-tracer test. The packed columns were flushed with deionized (DI) water at a flow rate of 2 mL/h for several days, and then saturated with 180 mL of 0.5 mM NaCl (this was not done for the experiments using DI water as the background solution). The detailed experimental protocols of the column tests are summarized in Table 1. The influents were prepared by diluting the stock nC60 suspension with a certain electrolyte (Table 1), followed by a 3 h stirring. In a typical column experiment the influent was pumped into the column from one of the two 100 mL glass syringes (SGE Analytical Science, Victoria, Australia). The effluent from the column was collected in 2 mL glass vials at an interval of approximately 1 sample per PV. The concentrations of nC60 in the effluents were determined with a DR/4000 UV−visible spectrophotometer (HACH Company, Loveland, CO) at a wavelength of 344 nm. For selected samples (Columns 6 to 8, Table 1), the particle sizes of nC60 in 7232

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Mass Balance. At the end of each column experiment, the porous medium together with the two steel screens were transferred to a 20 mL glass beaker and air-dried at room temperature for 7 days. Then, 10 mL of toluene was added to the beaker, and the mixture was sonicated for 4 h. Afterward, the mixture was transferred to a centrifugation tube, and was centrifuged at 2000 rpm for 1 h. Then, the toluene supernatant was withdrawn to measure the concentration of C60. The mass balance of the column experiments was within the range of 91 to 99%. Mathematical Modeling. The BTCs of the conservative tracer (3H2O) were fitted with the one-dimensional steady-state advection−dispersion equation (SI Figure S3). The hydrodynamic dispersion coefficient of each column was obtained by fitting the experimental BTCs of 3H2O using the CXTFIT code.21 A two-site transport model developed by Bradford et al.22 was used to fit the transport data of nC60. This model has been applied to model bacteria transport under unsaturated conditions (assuming that the blocking effect was negligible).23,24 The model divides the deposition sites of nC60 into an attachment site and a straining site: ρ ∂S1 ρ ∂S2 ∂ 2C ∂C ∂C + + =D 2 −v ∂t θ ∂t θ ∂t ∂x ∂x

nanoparticle spatial distribution. A value of 0.432 was assigned for β.22 The BTCs of nC60 were fitted with the HYDRUS-1D software,25 using three different models: (1) two-site transport model; (2) modified CFT model; and (3) CFT model. For the two-site model, Katt, Smax, and Kstr were the fitting parameters; for the modified CFT model, Katt and Smax were the fitting parameters; and for the CFT model, Katt was the only fitting parameter. Different sections of a BTC, corresponding to different flow velocities or solution chemistry, were fitted separately, each using a fixed set of parameters (see SI Tables S1 and S2). The detailed equations of the modified CFT model and the CFT model are given in the SI.



RESULTS AND DISCUSSION Effect of Flow Velocity. The effects of flow velocity on nC60 transport in Ottawa sand and Lula soil are shown in Figure 1. Varying flow velocity had little effect on nC60 transport through Ottawa sand, in that the breakthrough curves of nC60 under the two different velocities (8.7 and 0.7 m/d) are nearly identical (Figure 1a)under both velocities, the concentration of nC60 in the effluent increased quickly and the breakthrough reached over 90% after 8 PV. However, flow velocity had a significant effect on nC60 transport through Lula soil (Figure 1b): while the breakthrough reached over 87% under the high velocity (10 m/d), it reached only 43% under the low velocity (1.0 m/d), indicating that more nC60 particles were retained in the soil column under the low flow velocity; this velocity effect is consistent with literature reports on colloid transport.9,26,27 The distinctly different effects of flow velocity between nC60 transport in soil and transport in sand can probably be attributed to the smaller gain size, more irregular and rougher shape, and greater heterogeneity of Lula soil compared with those of Ottawa sand. Li et al.9 observed enhanced nC60 retention under low pore-water velocities (∼1 m/d) than higher velocities (∼8 m/d), and found that this velocity effect was more significant for finer sands. Johnson et al.15 proposed that retention in flow stagnation zones is an important mechanism of colloid retention under unfavorable attachment conditions, and the volumes of flow stagnation zones decrease with increasing flow velocity. It is likely that the flow-velocitydependent effect of flow stagnation zones is more profound for irregularly shaped materials such as Lula soil. Furthermore, Johnson and Tong26 argued that the decrease in deposition efficiencies with increasing flow velocity is consistent with the mechanism of deposition in heterogeneous domains in the presence of an energy barrier to deposition. Accordingly, the effect of flow velocity should be greater for more heterogeneous materials. To further understand the effect of flow velocity on nC60 transport in Lula soil, an additional test was conducted by continuously pumping nC60 and changing flow velocities stepwise. As shown in Figure 1c, nC60 breakthrough responded reasonably with the change of flow velocitybreakthrough of nC60 reached 81% at 30 PV under the velocity of 10 m/d, and then decreased gradually to 46% when the velocity was reduced stepwise to 1.0 m/d (interestingly, the 46% breakthrough is similar to the 43% breakthrough observed in the pulse-injection experiment (Figure 1b), when the velocity was changed directly from 10 m/d to 1.0 m/d). The BTCs of nC60 in both the pulse-injection and continuous-injection experiments were fitted with the CFT

(1)

where C (mg/L) is nC60 concentration in the aqueous phase at time t (h) and a distance x (cm); ρ (g/cm3) is the dry bulk density of the packed column; θ (−) is porosity of the packed column; S1 (mg/kg) and S2 (mg/kg) are nC60 concentrations in the attachment site and the straining site, respectively; D (m2/ d) is hydrodynamic dispersion coefficient; and v (m/d) is the pore-water velocity. For the attachment site (site 1), nC60 attachment can be described as follows:5,9 ρ ∂S1 = K attψ1C θ ∂t

(2)

where Katt (h−1) is the attachment rate; and ψ1 (−) is the blocking factor. Assuming the release of attached nC60 is negligible,5 the detachment term is not included in eq 2. Furthermore, ψ1 can be expressed as follows: ψ1 =

Smax − S1 Smax

(3)

where Smax (mg/kg) is the maximum retention capacity of nC60 on the attachment site. For the straining site (site 2), nC60 deposition can be described as follows: ρ ∂S2 = K strψ2C θ ∂t

(4)

−1

where Kstr (h ) is the straining rate; and ψ2 (−) is the straining factor. Straining is a function of distance, and can be expressed with a depth-dependent power law function as follows:22 ⎛ d + z ⎞−β ψ2 = ⎜ c ⎟ ⎝ dc ⎠

(5)

where dc (cm) is mean grain diameter of the packed material; z (cm) is the down gradient distance from the porous medium inlet; and β (−) is a fitting parameter that controls the shape of 7233

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Figure 2. Effect of ionic strength on transport of nC60. (a) and (b): BTCs in Ottawa sand (Column 4) and Lula soil (Column 5) under pulse injection mode using three different ionic strength values; (c) BTC in Lula soil (Column 6) under continuous injection mode, using five different ionic strength values (vertical dotted lines indicate where ionic strengths were changed). Solid lines () and dashed lines (− −) were plotted by fitting the BTCs with the two-site transport model and the modified clean-bed filtration theory (CFT) model, respectively. The thick gray lines show the injected concentration of nC60.

The better fits by the two-site transport model indicates the importance of the straining effects on the transport of nC60 through porous media,24 especially for more heterogeneous materials. Compared with the relatively homogeneous Ottawa sand, Lula soil is characterized with smaller grain size, wider particle size distribution, and more irregular grain shape and rougher grain surfaces. These factors could result in more complex flow pathways, narrower pore throat, and dead-end pores in the columns packed with Lula soil. Accordingly, the likelihood of nC60 retention by straining in Lula soil is much greater than that in Ottawa sand. In SI Figure S4 the fitted values of Katt, Smax, and Kstr are plotted against the pore-water velocities. Both Katt and Kstr show near linear correlations with the velocity, whereas Smax increases only slightly with velocity. This seems to indicate that the attachment rate and straining rate can be affected greatly by flow conditions, whereas Smax is controlled mainly by water chemistry (see discussion below).

model, modified CFT model, and two-site transport model, and the results are shown in Figure 1, and SI Tables S1 and S2. The CFT model failed to fit the BTCs, even for the case of Ottawa sand. This is consistent with the findings of other researchers.5,9,10 The modified CFT model (which incorporates the blocking effect) provided better fits, but is still not sufficient to simulate the BTCs in Lula soil (Figure 1b and c). For example, according to the modified CFT model the breakthrough of nC60 through Lula soil should reach 100% under the high flow velocity, whereas the actual breakthrough only reached 87% (Figure 1b). More significant deviation can be seen for the breakthrough under the low flow velocity, in that the modified CFT model failed to capture the plateau data (Figure 1b). Figure 1 and SI Table S1 show that the two-site transport model, which takes into account the straining effects in addition to particle attachment, can much more accurately model the BTCs of nC60, both in Ottawa sand and in Lula soil. 7234

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Figure 3. Correlations between fitted two-site transport model parameters (based on breakthrough data of Columns 3, 4, and 5) and ionic strength. CNaCl is concentration of NaCl in background solution.

Effect of Ionic Strength. The effects of ionic strength on the transport of nC60 are shown in Figure 2 and SI Table S1. Ionic strength had significant effects nC60 transport in Ottawa sand and Lula soilincreasing ionic strength caused more significant deposition of nC60, resulting in retarded nC60 breakthrough. For example, Figure 2a shows that when nC60 in DI water was pumped into an Ottawa sand column, the breakthrough of nC60 reached essentially 100% within a few PV; when the influent contained 1 mM NaCl, breakthrough reached 90% after 7 PV; when the influent contained 6 mM NaCl, breakthrough only reached 88% after 22 PV. Thus, not only a greater percentage of nC60 was retained per unit PV when ionic strength increased, but the BTC also became more asymmetric. Similar phenomenon has been observed for nC60 transport through columns of glass beads6 and quartz sand.8 The BTCs of nC60 from Lula soil show a similar response to the increase of ionic strength, both in the pulse-injection experiment (Figure 2b) and the continuous-injection experiment (Figure 2c) (except that at a given ionic strength, a greater amount of nC60 was retained by the soil column and the BTC from the soil column was more asymmetric). Increasing ionic strength compresses double layer thickness and reduces double layer repulsion between nanoparticles and grain surfaces.28 Additionally, retention of colloids in flow stagnation zones can be more significant at higher ionic strength.15 The fitting results with the modified CFT model and the two-site transport model are compared in Figure 2 (the CFT model cannot provide reasonable fits to most of the BTCs; see SI Figure S5). Again, the two-site model provides much better fits to the BTCs than the modified CFT model (SI Tables S1 and S2), except when the influent was nC60 in DI water (for which both models work well). A general trend is that for both the column packed with Ottawa sand and the columns packed with Lula soil, the fitted values of Katt, Smax, and Kstr increase with the increase of ionic strength (Figure 3). This indicates that increasing ionic strength in general results in greater attachment rate, more attachment sites, and greater straining effects. An interesting observation (Figure 2c) was that a

sudden dip of breakthrough occurred when the ionic strength of the influent was increased, indicating that the abrupt increase of ionic strength resulted in a drastic increase in the amount of favorable deposition sites on the soil surfaces, and accordingly, a significant short-time increase of nC60 deposition. Note that such sudden dip was not observed when the flow velocity was changed suddenly (Figure 1c). This further indicates that the amount of available deposition sites is controlled mainly by water chemistry and is less affected by hydrodynamic conditions. Figure 3 clearly shows that increasing ionic strength caused more significant increase of Smax value for the columns packed with Lula soil than for the column packed with Ottawa sand, indicating that when ionic strength increases, more attachment sites are created on the surfaces of soil than on the surfaces of sand. This can probably be linked to the more complex compositions of Lula soil than Ottawa sand. Ottawa sand is mainly composed of quartz, while Lula soil also consists of significant amount of clay, silt, and soil organic matter. Compared with the simple Si−O tetrahedral structure of quartz, soil clay fraction is commonly composed of aluminosilicate minerals (e.g., kaolinite and montmorillonite), oxides (e.g., goethite and gibbsite), amorphous materials (e.g., imogolite and allophane), and sulfur and carbonate minerals.29 Different components of soil may acquire surface charges via different mechanisms than quartz. For example, the surface charge of SiO2 is developed by the adsorption of proton or hydroxyl groups to SiOH, yielding discrete charged surface sites of SiOH2+ or SiO−;25 however, clay minerals acquire surface charges mainly through isomorphous lattice substitutions, broken bonds at crystal faces and ends, and leaching out of interlayer cations,30 and therefore, can possess permanent surface charge. Accordingly, soils may respond to the increase of cation concentrations differently than pure sand. Recently, Fortner et al.31 found that the special surface interactions between clay minerals and nC60 could play an important role in the transport of nC60. 7235

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Figure 4. Effect of cationic species on transport of nC60. (a) and (b): BTCs in Ottawa sand (Column 7) and Lula soil (Column 8) under continuous injection mode, using two different cationic species (vertical dotted lines indicate where cationic species were changed). Solid lines () and dashed lines (− −) were plotted by fitting the BTCs with the two-site transport model and the modified clean-bed filtration theory (CFT) model, respectively. The thick gray lines show the injected concentration of nC60.

of nC60 more significantly than Na+; this effect might facilitate straining (via enhanced bridging). Second, Ca2+ can neutralize the surface charge of both nC60 and sand/soil grains more effectively than Na+, resulting in more significant reduction of electrostatic repulsion.33 Third, Ca2+ may serve as a bridging agent by complexing with both nC60 and the functional groups of the grain surfaces,34 causing enhanced deposition. The greater effect of Ca2+ on nC60 transport in Lula soil than in Ottawa sand is probably due to the presence of clay minerals and organic matter in Lula soil. First, sand and soil may respond differently to the changes of cation species, via the mechanisms similar to those controlling the different responses between sand and soil to the changes of ionic strength. Second, compared with pure quartz, clay minerals and organic matter of soil may have more complexation sites for Ca2+. Thus, the bridging effect of Ca2+ can be more profound for soil than for pure sand. Figure 4 shows that even though the two-site transport model fits the BTCs better than the modified CFT model (the CFT model cannot fit the BTCs; see SI Figure S7), it cannot accurately simulate the observed BTC in Lula soil column when the background solution was first switched from Na+ to Ca2+ (i.e., the section of the BTC from 82 to 92 PV, Figure 4b). This is because once Ca2+ was introduced into the column, it started to exchange the adsorbed Na+ on the grain surfaces, thereby creating more available deposition sites for nC60. In particular, this cation exchange process is expected to be slower for Lula soil than for Ottawa sand, because of the higher cation exchange capacity of the soil. Because a constant Smax was assumed in the two-site transport model from 82 PV and thereafter (whereas theoretically the Smax value should be

Effect of Ionic Species. The changes of nC60 breakthrough profiles in response to the change of cation species are shown in Figure 4 (the continuous-injection approach was used in this set of experiments), and the fitted model parameters are summarized in SI Tables S1 and S2. For both the Ottawa sand column and the Lula soil column, Ca2+ had a much stronger effect on nC60 transport than Na+ at the same ionic strength. As shown in Figure 4, when the background solution was switched from DI water to 1.5 mM NaCl, a sudden dip in BTC occurred and then the BTC climbed gradually back to almost the same maximum value (97% for sand and 90% for soil). Additionally, switching DI water to NaCl only made the BTCs slightly more asymmetric. However, when the background solution was switched from 1.5 mM NaCl to 0.5 mM CaCl2 (same ionic strength), the shapes of the BTCs changed notably, in that the rise, following the sudden dip, was much slower and the maximum breakthrough plateau values were lower than those when using DI water or NaCl as the background solution. This stronger effect of Ca2+ on nC60 breakthrough was even more significant for the transport through Lula soil (Figure 4b) when the influent was switched to CaCl2, the decrease of breakthrough lasted for 17 PV, and the breakthrough recovered only to 69%, as compared with 90% when the influent was NaCl. In contrast to the strong effects of divalent cations on the transport of nC60, divalent anionstested by switching from 1.5 mM NaCl to 0.5 mM Na2SO4had negligible effects on nC60 transport in both Ottawa sand and Lula soil (see SI Figure S6). The greater effect of Ca2+ than Na+ on nC60 transport is likely due to the following mechanisms. First, by binding to the surface of nC60,32 Ca2+ can reduce the electrophoretic mobility 7236

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

Figure 5. Effect of fulvic acid (FA) on transport of nC60 in Lula soil (Column 11) under continuous injection model. Vertical dotted lines indicate where background electrolyte (with or without FA) were changed. Solid lines () and dashed lines (− −) were plotted by fitting the BTCs with the two-site transport model and the modified clean-bed filtration theory (CFT) model, respectively. The thick gray lines show the injected concentration of nC60.



increased gradually to this constant Smax value for the section of the BTC from 82 to 92 PV), the model cannot capture the fine detail of the profile of BTC during this slow cation exchange phase. Effect of Fulvic Acid. The effects of fulvic acid (selected as a representative dissolved organic matter) in the influent on the transport of nC60 were examined briefly in this study, and the results are shown in Figure 5. The continuous-injection approach was used and only the transport in Lula soil was tested. No significant changes in breakthrough were observed when switching the background solution from 1.5 mM NaCl with 5 mg/L fulvic acid to 1.5 mM NaCl only. However, breakthrough of nC60 increased slightly when switching the background solution from 0.5 mM CaCl2 to 0.5 mM CaCl2 with 5 mg/L fulvic acid. The observation is consistent with the findings of Chen and Elimelech,34 and is likely due to a combination of two factors: adsorption of fulvic acid to nC60 and soil grains enhances steric repulsion (which inhibits nC60 deposition), but Ca2+ may serve as a bridging agent between fulvic acid adsorbed to nC60 and functional groups of grain surfaces (which enhances deposition).34 Once again, the twosite transport model appears to be far superior to the modified CFT model in simulating the BTCs (Figure 5). Findings of this study indicate that nC60 has a high potential for migration through soil and aquifer materials. The greater effects of hydrodynamic conditions and solution chemistry on nC60 transport in Lula soil than in Ottawa sand are in line with the proposed key mechanisms controlling colloid deposition under unfavorable attachment conditions, including surface charge heterogeneity and roughness, straining, and the presence of flow stagnation zones.15,28 One limitation of this study is that due to the highly heterogeneous nature of soils, the relative contribution of different mechanisms to the deposition of nC60 cannot be distinguished, and further studies that yield direct microscopic evidence and detailed retention profiles (e.g., in sequentially extracted soils) are much needed. Finally, even though this study clearly indicates the advantage of the two-site transport model, methods that can be used to estimate the values of Katt, Smax, and Kstr in response to different environmental conditions are needed to enhance the applicability of such models.

ASSOCIATED CONTENT

S Supporting Information *

Tables S1 and S2 showing fitted parameters of two-site model, CFT model, and modified CFT model from BTCs, Figure S1 showing apparatus of column tests, Figure S2 showing particle sizes of nC60 in selected influents and effluents, Figure S3 showing BTCs of tracer, Figure S4 showing correlations between fitted parameters of two-site model and flow velocities, Figure S5 showing fits of BTCs (under varied ionic strength) with CFT model, Figure S6 showing effects of anionic species on nC60 transport, and Figure S7 showing fits of BTCs (under varied cationic species) with CFT model. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Phone/fax: 86-22-66229516; e-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This project was supported by National Natural Science Foundation of China (Grants 21177063 and 20977050), Tianjin Municipal Science and Technology Commission (10SYSYJC27200), China−U.S. Center for Environmental Remediation and Sustainable Development, Brine Chemistry Consortium, and Advanced Energy Consortium. We thank Professor Cheng Gu, Nanjing University, Professor Brain Teppen, Michigan State University, and two anonymous reviewers for their comments.



REFERENCES

(1) Wiesner, M. R.; Hotze, E. M.; Brant, J. A.; Espinasse, B. Nanomaterials as possible contaminants: The fullerene example. Water Sci. Technol. 2008, 57, 305−310. (2) Wiesner, M. R.; Lowry, G. V.; Alvarez, P.; Dionysiou, D.; Biswas, P. Assessing the risks of manufactured nanomaterials. Environ. Sci. Technol. 2006, 40, 4336−4345. (3) Colvin, V. L. The potential environmental impact of engineered nanomaterials. Nat. Biotechnol. 2003, 21, 1166−1170. (4) Zhang, L.; Wang, L.; Zhang, P.; Kan, A. T.; Chen, W.; Tomson, M. B. Facilitated transport of 2,2′,5,5′-polychlorinated biphenyl and

7237

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238

Environmental Science & Technology

Article

unsaturated flow conditions: The role of water content and bacteria surface hydrophobicity. Vadose Zone J. 2008, 7, 406−419. (25) Simunek, J.; Van Genuchten, M. T.; Sejna, M. The HYDRUS-1D Software Package for Simulating the One-Dimentional Movement of Water, Heat, And Multiple Solutes in Variably-Saturated Media: Version 3.0; Department of Environmental Science, University of California Riverside: Riverside, CA, 2005. (26) Johnson, W. P.; Tong, M. Observed and simulated fluid drag effects on colloid deposition in the presence of an energy barrier in an impinging jet system. Environ. Sci. Technol. 2006, 40, 5015−5021. (27) Shen, C.; Huang, Y.; Li, B.; Jin, Y. Predicting attachment efficiency of colloid deposition under unfavorable attachment conditions. Water Resour. Res. 2010, 46, W11526. (28) Ryan, J. N.; Elimelech, M. Colloid mobilization and transport in groundwater. Colloids Surf., A 1996, 107, 1−56. (29) Sparks, D. L., Environmental Soil Chemistry, 2nd ed.; Academic Press: London, UK, 2002. (30) Sharma, M. M.; Yen, T. F. Interfacial electrochemistry of oxide surfaces in oil-bearing sands and sandstones. J. Colloid Interface Sci. 1984, 98, 39−54. (31) Fortner, J. D.; Solenthaler, C.; Hughes, J. B.; Puzrin, A. M.; Plötze, M. Interactions of clay minerals and a layered double hydroxide with water stable, nano scale fullerene aggregates (nC60). Appl. Clay Sci. 2012, 55, 36−43. (32) Chen, K. L.; Elimelech, M. Aggregation and deposition kinetics of fullerene (C60) nanoparticles. Langmuir 2006, 22, 10994−11001. (33) Kuznar, Z. A.; Elimelech, M. Adhesion kinetics of viable Cryptosporidium parvum oocysts to quartz surfaces. Environ. Sci. Technol. 2004, 38, 6839−6845. (34) Chen, K. L.; Elimelech, M. Interaction of Fullerene (C60) nanoparticles with humic acid and alginate coated silica surfaces: Measurements, mechanisms, and environmental implications. Environ. Sci. Technol. 2008, 42, 7607−7614.

phenanthrene by fullerene nanoparticles through sandy soil columns. Environ. Sci. Technol. 2011, 45, 1341−1348. (5) Wang, Y.; Li, Y.; Fortner, J. D.; Hughes, J. B.; Abriola, L. M.; Pennell, K. D. Transport and retention of nanoscale C60 aggregates in water-saturated porous media. Environ. Sci. Technol. 2008, 42, 3588− 3594. (6) Espinasse, B.; Hotze, E. M.; Wiesner, M. R. Transport and retention of colloidal aggregates of C60 in porous media: Effects of organic macromolecules, ionic composition, and preparation method. Environ. Sci. Technol. 2007, 41, 7396−7402. (7) Lecoanet, H. F.; Wiesner, M. R. Velocity effects on fullerene and oxide nanoparticle deposition in porous media. Environ. Sci. Technol. 2004, 38, 4377−4382. (8) Wang, Y. G.; Li, Y. S.; Pennell, K. D. Influence of electrolyte species and concentration on the aggregation and transport of fullerene nanoparticles in quartz sands. Environ. Toxicol. Chem. 2008, 27, 1860−1867. (9) Li, Y.; Wang, Y.; Pennell, K. D.; Abriola, L. M. Investigation of the transport and deposition of fullerene (C60) nanoparticles in quartz sands under varying flow conditions. Environ. Sci. Technol. 2008, 42, 7174−7180. (10) Wang, Y.; Li, Y.; Kim, H.; Walker, S. L.; Abriola, L. M.; Pennell, K. D. Transport and retention of fullerene nanoparticles in natural soils. J. Environ. Qual. 2010, 39, 1925−1933. (11) Cheng, X. K.; Kan, A. T.; Tomson, M. B. Study of C60 transport in porous media and the effect of sorbed C60 on naphthalene transport. J. Mater. Res. 2005, 20, 3244−3254. (12) Bradford, S. A.; Torkzaban, S. Colloid transport and retention in unsaturated porous media: A review of interface, collector, and porescale processes and models. Vadose Zone J. 2008, 7, 667−681. (13) Bradford, S. A.; Simunek, J.; Bettahar, M.; van Genuchten, M. T.; Yates, S. R. Significance of straining in colloid deposition: Evidence and implications. Water Resour. Res. 2006, 42. (14) Jaisi, D. P.; Saleh, N. B.; Blake, R. E.; Elimelech, M. Transport of single-walled carbon nanotubes in porous media: Filtration mechanisms and reversibility. Environ. Sci. Technol. 2008, 42, 8317−8323. (15) Johnson, W. P.; Li, X.; Yal, G. Colloid retention in porous media: Mechanistic confirmation of wedging and retention in zones of flow stagnation. Environ. Sci. Technol. 2007, 41, 1279−1287. (16) Li, X.; Lin, C.-L.; Miller, J. D.; Johnson, W. P. Role of grain-tograin contacts on profiles of retained colloids in porous media in the presence of an energy barrier to deposition. Environ. Sci. Technol. 2006, 40, 3769−3774. (17) Tong, M.; Johnson, W. P. Excess colloid retention in porous media as a function of colloid size, fluid velocity, and grain angularity. Environ. Sci. Technol. 2006, 40, 7725−7731. (18) Yao, K. M.; Habibian, M. M.; Omelia, C. R. Water and waste water filtration: Concepts and applications. Environ. Sci. Technol. 1971, 5, 1105−1112. (19) Andrievsky, G. V.; Kosevich, M. V.; Vovk, O. M.; Shelkovsky, V. S.; Vashchenko, L. A. On the production of an aqueous colloidal solution of Fullerenes. J. Chem. Soc., Chem. Commun. 1995, 12, 1281− 1282. (20) Fortner, J. D.; Lyon, D. Y.; Sayes, C. M.; Boyd, A. M.; Falkner, J. C.; Hotze, E. M.; Alemany, L. B.; Tao, Y. J.; Guo, W.; Ausman, K. D.; Colvin, V. L.; Hughes, J. B. C60 in water: Nanocrystal formation and microbial response. Environ. Sci. Technol. 2005, 39, 4307−4316. (21) N. Toride, F. J. L.; Van Genuchten, M. T. The CXTFIT Code for Estimating Transport Parameters from Laboratory or Field Tracer Experiments; U.S. Salinity Laboratory: Riverside, CA, 1999. (22) Bradford, S. A.; Simunek, J.; Bettahar, M.; Van Genuchten, M. T.; Yates, S. R. Modeling colloid attachment, straining, and exclusion in saturated porous media. Environ. Sci. Technol. 2003, 37, 2242−2250. (23) Gargiulo, G.; Bradford, S. A.; Simunek, J.; Ustohal, P.; Vereechen, H.; Klumpp, E. Transport and deposition of metabolically active and stationary phase Deinococcus radiodurans in unsaturated porous media. Environ. Sci. Technol. 2007, 41, 1265−1271. (24) Gargiulo, G.; Bradford, S. A.; Simunek, J.; Ustohal, P.; Vereecken, H.; Klumpp, E. Bacteria transport and deposition under 7238

dx.doi.org/10.1021/es301234m | Environ. Sci. Technol. 2012, 46, 7230−7238