Colloid Transport in a Heterogeneous Partially Saturated Sand

Jan 10, 2008 - Experiments were performed under three flow rates (0.1, 0.2, and 0.4 cm/min) applied by a rain simulator at the top of the column. Cons...
38 downloads 10 Views 659KB Size
Download PDF
Environ. Sci. Technol. 2008, 42, 1066–1071

Colloid Transport in a Heterogeneous Partially Saturated Sand Column MIKHAIL MISHUROV,† ALEXANDER YAKIREVICH,* AND NOAM WEISBROD Department of Environmental Hydrology and Microbiology, Zuckerberg Institute for Water Research, J. Blaustein Institutes for Desert Research, Ben Gurion University of the Negev, Sede Boker Campus, 84990, Israel

Received July 19, 2007. Revised manuscript received November 19, 2007. Accepted November 20, 2007.

Colloid transport was studied in heterogeneous sand columns under unsaturated steady-state conditions, using two sizes of acid-cleaned sand to pack the column. Heterogeneity was created by placing three continuous tubes of fine sand (3.6% of the total volume) within a column of coarse sand (mean grain diameters 0.36 and 1.2 mm, respectively). Experiments were performed under three flow rates (0.1, 0.2, and 0.4 cm/ min) applied by a rain simulator at the top of the column. Constant water-content profile in the coarse sand was achieved by applying corresponding suction at the column bottom. Three sizes of latex microspheres (1, 0.2, and 0.02 µm) and soluble tracers (LiBr), diluted in a weak base (pH 7.3, ionic strength 0.0023 M) solution, were used simultaneously. Introduction of preferential pathways reduced front-arrival time about 2-fold and increased colloid recovery which, at the 0.2 cm/min flow rate, was higher than at 0.4 and 0.1 cm/min. Maximum solution flux from coarse to fine sand (due to differences in matric pressure) at 0.2 cm/min, verified by hydrodynamic modeling, could explain this phenomenon. Results suggest that in heterogeneous soil, maximum colloid recovery does not necessarily occur at maximum water content. This has clear implications for colloid transport in natural soils, many of which are heterogeneous.

Introduction Understanding the mechanisms of colloid and colloidfacilitated transport in porous media is very important when addressing environmental problems (e.g., refs 1–3,). Many studies have explored the transport of biocolloids, mineral colloids, and latex microspheres. These experiments typically investigate the influence of solution and colloid-surface chemistry, porous medium texture, flow rate, and water content on colloid transport (4–10). Variable saturation and heterogeneity (e.g., presence of macropores, fissures, layers with different properties, etc.) are often inherent features of porous media which play a critical role in the transport processes. High colloid retention in unsaturated porous media is explained by introducing the concepts of air–water (AW) and * Corresponding author phone: +972-8-6596869; fax: +972-86596909; e-mail: [email protected]. † Current address: Department of Civil and Environmental Engineering, University College Cork, Cork, Ireland. 1066

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008

air–water-solid (AWS) interfaces (11–15). Retention mechanisms considered include sorption at the AW interface (12), film straining (13), and a combination of thin-film straining and immobile water storage at low moisture content (14). Recent visualization experiments by Christ et al. (11) and Zevi et al. (15) have highlighted the importance of the AWS interface in colloid sorption. The solid phase affects the AWS interface geometry in the region between grains where water menisci diminish to thin water films on the grain surfaces, an area termed the air–water-meniscus-solid interface (AWmS). The sorption of hydrophilic and hydrophobic particles is more pronounced in these areas where the thickness of the water film covering the medium grains is in the range of the particles’ diameter and the curvature of the film is changing from convex to concave (15). A recent review on colloid transport in porous media (16) emphasizes the role of soil structure in the transport of colloid-sized particles. Nevertheless, all of the aforementioned studies carried out in unsaturated porous media have focused solely on homogeneous conditions. Preferential flow is described as a phenomenon in which flow occurs along some limited and often volumetrically insignificant pathways inside a porous medium bypassing large parts of the porous matrix. The types of preferential flow (macropore-driven, gravity-driven, heterogeneitydriven, and oscillatory) are characterized by specific conditions and spatial scales (17). The presence of macropores causes preferential flow at the pore to pedon scale (18), whereas fingering flow induced by profile heterogeneities or water repellency, and funneling flow caused by textural differences of sloping soil layers both occur at larger scales (19). Porous medium heterogeneity and preferential flow result in high spatial variability of the water content in the soil profile and faster transport of solutes and colloids relative to uniform flow conditions. Inorganic silica particles were used in research performed by Saiers et al. (20), who employed both homo and heterogeneous packing in saturated experiments. These authors found that over 60% of the total particulate mass could be transmitted through a preferential flow path created by installing a slender tube of coarse sand (comprising 11% of the total volume) within a bulk of fine sand. In laboratory experiments, Ullum (21) revealed that a network-like system of macropores is essential for fast transport of polystyrene latex microspheres under saturated conditions. Under unsaturated conditions, colloid mobility was strongly diminished by the presence of an air phase. In structured columns, conservative soluble-tracer and colloid breakthrough curves (BTCs) were similar, meaning that the same processes governed the transport of both tracers and colloids. A series of experiments on colloid transport in saturated heterogeneous porous medium (22, 23) highlighted the importance of the textural interface between different types of sand, and the size-dependent nature of the straining and attachment phenomena. It was found that with a decrease in particle size, straining affects transport to a lesser extent, whereas attachment processes become more important. While previous studies have focused on homogeneous unsaturated (3, 10–13, 15) or heterogeneous saturated porous media (4, 20–23), the main objectives of this research were to address (1) the effects of preferential flow pathways and (2) the impact of water flux and water content on colloid transport in unsaturated heterogeneous porous media. We conducted column transport experiments with colloids in homogeneous and heterogeneous sand columns under unsaturated steady-state conditions. The results of our research suggest that in the presence of preferential flow, 10.1021/es071780f CCC: $40.75

 2008 American Chemical Society

Published on Web 01/10/2008

maximum colloid recovery will occur at an intermediate water flux and water content in which the lateral exchange of solution between the fine and coarse grains is optimal.

Materials and Methods Porous Media. Two types of natural sand were used as porous media: sand with a mean grain diameter of 1.2 mm (grade 14/18) and sand with a mean grain diameter of 0.36 mm (grade 40/50). These are referred to hereafter as “coarse” and “fine” sand, respectively. The mineralogical composition of the sands was close to pure quartz with traces of ferrous minerals. The cleanliness of the grain surface is one of the most important factors controlling the reproducibility of sand-packed column experiments (24). To decrease external effects on the experiment, an elaborate sand-washing procedure (12), including boiling in concentrated acid and sonication to remove organic matter, was used. Hydraulic characteristics of the given sands were obtained in steady-flow column experiments under unsaturated and saturated conditions. Simultaneous readings of water content and pressure head taken by time-domain reflectometry (TDR) and tensiometers, respectively, enabled a determination of the experimental water-retention curves for fine and coarse sands. Experimental retention curves obtained in column experiments were fitted using the van Genuchten (25) model: Se ) (1 + (R|h|)n)-m

(1)

where Se ) (θ - θr)/(θs - θr) is effective water saturation, θ is volumetric water content, r and s indexes indicate residual and saturated water contents, respectively, h is the matric pressure head, n and R are parameters, and m ) 1 - 1/n. The parameters in (1) for fine sand were as follows: R ) 0.044 cm-1, n ) 3.454, θs ) 0.37, and θr ) 0.023, while those for coarse sand were R ) 0.345 cm-1, n ) 2.803, θs ) 0.37 and θr ) 0.045. Averaged values of saturated hydraulic conductivity obtained in Darcy experiments were 1.3 ( 0.02 and 8.4 ( 2.4 cm/min for fine and coarse sand, respectively. Tracers. To investigate the transport of colloids relative to that of solutes, both conservative (Li and Br) tracers and colloids (latex microspheres) were used. Concentrations in the tracer solution corresponded to 23 ppm for Br and 2 ppm for Li. Chemical analyses of ion concentration were carried out by standard methods: phenol red colorimetric method (26) with spectrophotometer (U-2000, Hitachi, Inc., Tokyo, Japan) measurement at 590 nm for Br and atomic absorption at 670.8 nm on a fast sequential atomic absorption spectrometer (AA280FS, Varian, Inc., Palo Alto, CA) for Li. Colloids. FluoSpheres were selected as the colloid tracers due to their negatively charged surfaces at neutral pH and their maximum stability (Molecular Probe, Eugene, OR). They are almost ideally spherical colloids dyed with fluorescent dye, which are commercially available in a variety of sizes and excitation/emission ranges. FluoSpheres of three size groups, with nominal diameters of 1, 0.2, and 0.02 µm, were used. Each size group has a different excitation/emission range, which enabled us to measure concentrations in a mixed solution simultaneously, without interference. The surface charges of the carboxylate-modified microspheres used in the experiments ranged between 0.1 and 2.0 meq/g with a particle density of 1.055 g cm-3 (data provided by the manufacturer). The ζ-potentials of the different-sized FluoSpheres were measured using a ZetaMaster (Malvern Instruments LTD, Malvern, UK) and were found to be -32.6, -21.2, and -19.2 mV for 1, 0.2, and 0.02 µm, respectively, at the experimental pH (∼7–8). Note that the same combination of FluoSpheres has been previously used in other published works (24, 27). The size distribution of colloids in groundwater has been found to be exponential (27) in many cases, with small

particles being more abundant than larger ones. We used a mixture in the quantitative proportion of 1:4:8 (1:0.2:0.02 µm) to simulate colloid content in underground systems. This ratio corresponds to actual concentrations of 2.26 × 106, 1.13 × 109, and 2.26 × 1012 particles/ml for 1-, 0.2-, and 0.02 µm microspheres, respectively. Microsphere concentrations were measured by fluorescence spectrophotometer (Cary Eclipse, Varian, Inc.); measurements required only 0.3 mL of sample and were carried out within one hour of the end of the experiment. Solution Chemistry. Artificial rainwater (ARW) was used as the blank solution in all experiments. Its chemical composition was chosen based on average Israeli rainwater data (28): concentrations (ppm) of ions were Ca2+ 12.2, Cl12.7, SO42- 13.8, Na+ 13.1, Mg2+ 3.5, HCO3- 35, NO3- 15.3, ionic strength 0.0023 M, total dissolved solids 105.6 mg/L, pH 7.3 ( 0.1. The same background solution was used in a previous colloid-transport study carried out in a fractured system (29). Column Transport Experiments. The cylindrical column was made of Plexiglas, 35 cm in length, and 11.5 cm in diameter (Figure 1). TDR probes and tensiometers were helically installed at levels of 7.5, 15, and 22.5 cm from the bottom. A data logger (CR10X, Campbell Scientific, Inc. Logan, UT) and TDR unit (TDR100, Campbell Scientific, Inc.) were used along with a multiplexer (SDMX50, Campbell Scientific, Inc.). TDR and pressure sensors (26PC Series, Honeywell Int., Inc., Morristown, NJ) used with tensiometers were calibrated prior to the experiment. Tensiometers (column tensiometers, 0.65 cm outer diameter, Soil Measurements Systems, Tucson, AZ) and porous membrane (pore size 25–50 µm, thickness 10 mm, diameter 120 mm, C filter disk, Ace Glass, Inc., Vineland, NJ) were suitably saturated prior to the experiments according to the manufacturers’ instructions. The column was filled with media up to 30 cm in height by wet-packing technique. The sand was slowly poured into water standing in the bottom of the column; water level was maintained at about 0.5 cm above the surface of the sand during the entire procedure. To develop preferential flow pathways, tubes of fine sand were created inside the bulk of coarse sand. Under saturated conditions, the hydraulic conductivity value for coarse sand is higher than that for fine sand. Under unsaturated conditions, a decrease in pressure head (and subsequently in degree of saturation) causes a decrease in conductivity, but this decrease occurs at different rates for different kinds of sand; at some point, the conductivity of the fine sand becomes higher than that of the coarse sand and their “roles” switch. To obtain the range of preferential flow conditions to be used in the experiments, prior simulations of water flow with the Hydrus 2D model (33) were performed. Thin-walled plastic tubes were inserted into the column in the configuration shown in Figure 1(see cross-section A-A) and bulk medium was added. Then, the fine sand was poured into these tubes as they were slowly withdrawn from the column. The inner diameter of each tube was 1.25 cm, resulting in a volumetric proportion of fine sand of 3.6%. This setup is similar in principle to the one developed by Saiers et al. (20). The simple configuration of three narrow fine sand tubes provides a relatively high surface area between the two sand grades. Additionally, it allows for repetitions if repacking is needed. Sand porosity was determined to be 0.37 ( 0.01 based on bulk density at column packing. Tensiometers were subsequently installed during packing to avoid loss of water from them. The edge of the filter was sealed with nonreactive silicon grease to restrict water flow to the membrane. The rain simulator was comprised of 22 hypodermic needles (outer diameter 1.1 mm) positioned at approximately 5 cm above the sand surface. To prevent damage to the medium surface from the drops, a thin VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1067

FIGURE 1. General setup design for unsaturated experiments. synthetic mesh was placed over it. A peristaltic pump (Masterflex, Cole-Palmer Instrument Co., Vernon Hills, IL) was used to control the applied flow, and an internally constructed vacuum pump was used to maintain a constant pressure head at the bottom boundary. By regulating pump suction, it was possible to change the pressure head at the bottom of the column. The outflow pump was calibrated before the experiment to remove all effluent; outflow collection was employed by positioning the fraction collector (Spectra/Chrom CF-1, Spectrum Chromatography, Inc., Houston, TX) lower than the vacuum chamber (Figure 1). Samples were collected at equal intervals, up to a volume of 9 mL each. After the experiment, samples were transferred to disposable tubes, which were stored in a dark cold room until analysis. The tensiometers and TDR probes were placed inside the bulk medium (coarse sand); no measurements were taken inside the fine sand due to size limitations. Uniform water content was maintained in the coarse sand throughout each experiment by applying a particular flow rate at the top boundary and the corresponding suction at the bottom boundary. Three column experiments were carried out under steadystate conditions with flow rates of 0.1, 0.2, and 0.4 cm/min and a pressure head at the column bottom of -8, -6, and -5.5 cm, respectively. These conditions allowed close to uniform volumetric water-content profiles in the coarse sand: 0.11, 0.12, and 0.14 for flow rates of 0.1, 0.2, and 0.4 cm/min, respectively. Conducting experiments under steady-state conditions allowed us to define the notion of column water volume (Vcθ): Vcθ )

∫ θdv

(2)

Vc

where Vc is column volume. Column water volume will be used hereafter as a scale to define the dimensionless time variable. This definition is similar to the “pore volume” notion for saturated conditions. Initially, ARW solution was used to reach steady conditions. This was followed by injection of 1 column water volume of solution and an additional 2-3 volumes of ARW. The BTCs obtained in these experiments were compared with those of a tracer test obtained for the homogeneous coarse sand column. The flow rate and water content in the latter experiment were 0.2 cm/min and 0.11, 1068

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008

respectively. Each experiment was carried out in two replicates. Agreement between repetitions was good and, therefore, the BTCs from one replicate are presented. Overall, seven experiments were carried out and analyzed for this paper.

Results and Discussion Homogeneous Coarse Sand Column. The BTCs from unsaturated homogeneous experiments with coarse sand show that first arrival (C/C0 > 0.01), BTC fronts of soluble tracers and large/medium colloids (C/C0 ) 0.5), and maximum concentration of most tracers and colloids, occurred at about 0.5, 1.0, and 1.5 column water volumes, respectively (Figure 2a). The BTCs exhibited small concentration fluctuations at some time intervals, presumably due to experimental errors associated with measuring concentrations and small variations in water flux. Overall recoveries of tracers were 93% for Br, 85% for Li, and 68, 68, and 23% for 1, 0.2, and 0.02 µm particles, respectively (Table 1). The higher retention of the smallest particles as compared to the conservative tracers and larger colloids can be explained by AW-interface sorption. In particular, the 0.02 µm particles can be more affected by the presence of an AW interface than the larger particles because their small size provides them with a greater area for sorption (30). Another plausible explanation is the higher Brownian diffusion of smaller particles (2.36 × 10-7, 2.36 × 10-8, and 4.72 × 10-9 cm2/s for the 0.02, 0.2, and 1.0 µm FluoSpheres, respectively), which results in large differences between the collision probabilities of small versus large colloids. Increasing collision of the 0.02 µm FluoSpheres will increase the deposition of small colloids within the media (31). The 0.2 µm colloids sometimes exhibited slightly higher recovery than the 1.0 µm colloids (Table 1), confirming the widely reported suggestion of the existence of “most mobile” colloid sizes (29, 30), which is close to 0.2 µm in many porous and fractured systems. Heterogeneous Sand Column. Tracers and colloids arrived earlier, by up to 0.25 column water volume, in the heterogeneous sand column, which also exhibited an asymmetric BTC shape (Figure 2b), in contrast to the symmetrical shape in the homogeneous sand (Figure 2a). Under unsaturated conditions, the fine sand in the heterogeneous column may act as a preferential path, potentially explaining the early tracer and colloid arrival. Additionally, there was

FIGURE 2. Breakthrough curves (BTCs) from (a) the homogeneous coarse sand column and (b) the heterogeneous sand column at a water flux of 0.2 cm/min under unsaturated conditions. Br, black hollow circles; Li, black filled circles; 1 µm particles, orange circles; 0.2 µm particles, blue squares; 0.02 µm particles, red triangles. See text for explanation of “column water volume”.

TABLE 1. Overall Recoveries (%) in Column Experiments under Unsaturated Conditions after Three Column Water Volumes (Values in Brackets Are Given to Characterize Range of Recoveries in Duplicated Experiments) flow rate, cm/min a

0.2 0.1 0.2 0.4 a

colloids Br

Li

1 µm

0.2 µm

0.02 µm

93 105 ((5) 99 ((3) 102 ((5)

85 100 ((5) 86 ((3) 89 ((3)

68 60 ((5) 98 ((5) 78 ((1)

68 65 ((5) 99 ((3) 78 ((2)

23 29 ((3) 64 ((2) 37 ((2)

Homogeneous packing.

appreciably higher recovery of colloids in the heterogeneous column at the 0.2 cm/min flow rate, whereas the recovery of the soluble tracers was similar (around 100% for Br- and 86–100% for Li+, Table 1). The higher recovery of colloids in the heterogeneous column implies lateral exchange of solution between fine and coarse sands. Under unsaturated conditions in the heterogeneous column, a lateral water flux and corresponding advective flux of solutes and colloids occurred because the matric pressure in the fine sand was lower than that in the coarse sand. The average water content in the coarse sand was 0.12, whereas the fine sand tubes were expected to be close to full saturation, as estimated using the average pressure head in the column and the retention curve of fine sand. This was verified by hydrodynamic modeling with the Hydrus 2D code (see subsection “Water Flow Simulations with the Hydrus 2D Model” and Figure S1(b), Supporting Information (SI)). The calculated average water content in fine sand was 0.367 (this was not measured directly). At this water content, the area of the AWmS interface in the fine sand tubes is likely to be relatively small compared to that in the coarse sand because fine sand is close to full saturation. Therefore, this can decrease colloid retention at the AWmS interface in the fine sand tubes and increase colloid recovery in the heterogeneous column relative to the homogeneous one. As recovery of conservative tracers was around 100% in both homogeneous and heterogeneous sand experiments, the effect of lateral flux on these tracers could be concluded by early breakthrough but not by changes in mass balance. Effect of Flow Rate on Colloid Transport. To assess the influence of flow rate on colloid transport in a heterogeneous sand column, BTCs of each type of colloid were compared at the different flow rates of 0.1, 0.2, and 0.4 cm/min (Figure 3). Note that here, in each figure (a, b, c), we compare the behavior of one size of colloid at the three flow rates and in the homogeneous packing (0.2 cm/min). Colloid arrival time was twice as early and their recovery was higher relative to

FIGURE 3. Breakthrough curves (BTCs) of colloids at different flow rates: 0.1 cm/min, triangles; 0.2 cm/min, circles; 0.4 cm/ min, squares; and microsphere sizes (a) 1 µm; (b) 0.2 µm; (c) 0.02 µm. Hollow symbols, homogeneous packing; filled symbols, heterogeneous packing. VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1069

the homogeneous column test (for the same flow rates). It is important to remember that the graphs are presented in “column water volume” rather than absolute time units; therefore, the concentration front at a flow rate of 0.2 cm/ min appears to arrive faster than at 0.4 cm/min. If a time scale in absolute units had been used, the concentration arrival times would have been consistent with the water flow rates (data not shown). Recoveries of all colloid sizes were significantly higher in the 0.2 cm/min experiments, which was not expected based on theoretical considerations and previous studies in homogeneous unsaturated porous media. Studies of homogeneous media under steady-state conditions (e.g., refs 31 and 32) have demonstrated that a reduction in water content (and correspondingly, flow rate) leads to a decrease in colloid recovery. Apparently, the presence of preferential flow pathways is responsible for the existence of an optimal water content, below and above which colloid recovery decreases and arrival time is higher. The size-dependent retention of particles noted in the homogeneous column experiments (Table 1) could also be seen in the heterogeneous column experiments. Most of the mobile colloid size behavior was repeated in the heterogeneous column, with 1 and 0.2 µm particles often having similar recoveries. Water Flow Simulations with the Hydrus 2D Model. To explain the enhanced mass recovery in the unsaturated heterogeneous column with a water flux of 0.2 cm/min versus 0.1 and 0.4 cm/min, we carried out simulations with the Hydrus 2D code (33). An axis-symmetrical problem of water flow in part of a column composed of one preferential flow tube (fine sand) embedded in coarse sand was considered. The portion of fine sand out of the total volume was 3.6% as in the experimental setup. Top and bottom boundary conditions, namely water flux of 0.1, 0.2, and 0.4 cm/min and pressure head of -8, -6, and -5.5 cm, respectively, were chosen to simulate the experiments. A mesh discretization of 0.1 cm in the horizontal dimension and 0.2 cm in the vertical dimension was used. To obtain good fit between the observed and simulated water content and matric pressure head, we had to adjust the hydraulic conductivity of the coarse sand by changing the value of its saturated hydraulic conductivity from the experimentally based one. The unsaturated hydraulic conductivity calculated from the van Genuchten-Mualem relation (25), which is imbedded in the Hydrus 2D code, did not always fit experimental data for the whole range of water contents. Since our experiments were conducted for steady flow and a constant water-content profile, adjusting the saturated hydraulic conductivity allowed us to validate the model for each specific experiment. Simulations proved that in the upper 4–6 cm, the matric pressure head in fine sand is smaller than that in coarse sand, and water flow is from coarse to fine sand (See SI Figure S1). This leads to an increase of the matric pressure in fine sand in the rest of the column, the difference in matric pressure between fine and coarse sand is decreased, the water exchange between fine and coarse sands is negligible, and the flow field is essentially vertical downward. We note that most changes in the flow velocity field occur in the upper 2 cm of the column (See SI Figure S2). Simulated flow velocities quantify the portion of preferential flow as the percentage of volumetric water flux flowing out of the fine sand tubes relative to the volumetric water flux through the entire crosssectional area of the column. The volumetric fluxes were caculated by integrating the Darcy velocities at the column outlet over corresponding areas. The percentages of preferential flow were 23.9, 15.0, and 8.9 for fluxes of 0.1, 0.2, and 0.4 cm/min, respectively. These results in particular, and the notion of “preferential” flow in general, suggest water redistribution inside the heterogeneous column. Data produced by the Hydrus 2D were used to calculate water transfer 1070

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008

FIGURE 4. Cumulative water-transfer rate from coarse to fine sand along the column length. Dashed line, water flux 0.1 cm/ min; dotted line, 0.4 cm/min; solid line, 0.2 cm/min. between domains (Figure 4). The horizontal components of velocity vectors along the boundary nodes (between domains) were taken into account and the integral cumulative watertransfer rate was calculated as follows: L-z

Qx ) 2πRf

∫ v dz, x

(3)

0

where Qx is total water-transfer rate from coarse to fine sand, Rf is the radius of a fine tube, L is column length, vx is the horizontal component of the velocity vectors at r ) Rf, and dz is grid resolution. At a flow rate of 0.2 cm/min, cumulative lateral water flux was 0.503 cm3/min higher than those at 0.1 and 0.4 cm/min (0.405 and 0.454 cm3/min, respectively). This could explain the observed higher overall recovery at a flow rate of 0.2 cm/min relative to the other cases. Sensitivity analysis was performed by changing the retention curve parameters a and n (1) by 5% of their initial values in various permutations. In all cases we obtained the highest lateral water transfer rate for a water flux of 0.2 cm/min. The higher colloid recovery obtained at a flow rate of 0.2 cm/min compared to 0.4 cm/ min was unforeseen. Such a phenomenon could be explained by the presence of intermediate water-content conditions, optimal for lateral mass transfer between the coarse and fine sand, and thus enhancing colloid transport. Current experimental and theoretical work assume that colloid recovery will increase with increasing level of saturation in porous media. This study suggests that for the common field situation in which preferential flow pathways exist, this assumption is incorrect and there is an optimal level of saturation for transport. Obviously, this observation must be incorporated in predictive models.

Acknowledgments This work was funded by the Israeli Science Foundation (ISF), grant no. 131-03. The comments provided by three anonymous reviewers were very helpful in improving this manuscript.

Supporting Information Available Hydrus 2D-simulated distributions of matric pressure head, water content and Darcy velocity in fine and coarse sands of the heterogeneous column are shown in Figure S1. Hydrus 2D-simulated flow velocity vectors in the upper 2 cm of the heterogeneous column for different water flow rates are shown in Figure S2. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Corapcioglu, M. Y.; Jiang, S. Colloid-facilitated groundwater contaminant transport. Water Resour. Res. 1993, 29 (7), 2215– 2226. (2) Gschwend, P. M.; Reynolds, M. D. Monodisperse ferrous phosphate colloids in an anoxic groundwater plume. J. Contam. Hydrol. 1987, 1 (3), 309–327. (3) Ryan, J. N.; Illangasekare, T. H.; Litaor, M. I.; Shannon, R. Particle and plutonium mobilization in macropore soils during rainfall simulations. Environ. Sci. Technol. 1998, 32, 476–482. (4) Silliman, S. E. Particle transport through two dimensional saturated porous media: influence of physical structure of the medium. J. Hydrol. 1995, 167, 79–98. (5) Harvey, R. W.; Kinner, N. E.; MacDonald, D.; Metge, D. W.; Bunn, A. Role of physical heterogeneity in the interpretation of small scale laboratory and field observations of bacteria, microbial sized microspheres, and bromide transport through aquifer sediments. Water Resour. Res. 1993, 29, 2713–2721. (6) Gschwend, P. M.; Backhus, D. A.; MacFarlane, J. K.; Page, A. L. Mobilization of colloids in groundwater due to infiltration of water at a coal ash disposal site. J. Contam. Hydrol. 1990, 6 (4), 307–320. (7) Kretzschmar, R.; Barmettler, K.; Grolimund, D.; Yan, Y.-d.; Borkovec, M.; Sticher, H. Experimental determination of colloid deposition rates and collision efficiencies in natural porous media. Water Resour. Res. 1997, 33 (5), 1129–1137. (8) Sirivithayakorn, S.; Keller, A. Transport of colloids in saturated porous media: A pore-scale observation of the size exclusion effect and colloid acceleration. Water Resour. Res. 2003, 39 (4), 1109. (9) Grolimund, D.; Borkovec, M. Release of colloidal particles in natural porous media by monovalent and divalent cations. J. Contam. Hydrol. 2006, 87 (3–4), 155–175. (10) Gao, B.; Saiers, J. E.; Ryan, J. N. Deposition and mobilization of clay colloids in unsaturated porous media. Water Resour. Res. 2004, 40 (8), W08602. (11) Christ, J. T.; Zevi, Y.; Taylor, J.; Peranginangin, N.; Gao, B.; Lyon, S.; Dittrich, T. M.; Marshall, A. G.; De Alwis, D.; Baveye, P. C.; Parlange, J.-Y.; McCarthy, J. F.; Steenhuis, T. S. Colloid and Colloid-Facilitated Transport of Contaminants in Soils and Sediments. In Visualization of Colloidal Transport in Porous Media,Tjele, Denmark, September 19–20, 2002; de Jonge, L. W., Moldrup, P., Jacobsen, O. H., Eds.; Tjele, Denmark, 2002; pp 31–39. (12) Wan, J.; Wilson, J. L. Colloid transport in unsaturated porous media. Water Resour. Res. 1994, 30, 857–864. (13) Wan, J.; Tokunaga, T. K. Film straining of colloids in unsaturated porous media: conceptual model and experimental testing. Environ. Sci. Technol. 1997, 31 (8), 2413–2420. (14) Gao, B.; Saiers, J. E.; Ryan, J. N. Pore-scale mechanisms of colloid deposition and mobilization during steady and transient flow through unsaturated granular media. Water Resour. Res. 2006, 42, W01410. (15) Zevi, Y.; Dathe, A.; McCarthy, J. F.; Richards, B. K.; Steenhuis, T. S. Distribution of colloid particles onto interfaces in partially saturated sand. Environ. Sci. Technol. 2005, 39 (18), 7055–7064. (16) DeNovio, N. M.; Saiers, J. E.; Ryan, J. N. Colloid movement in unsaturated porous media: Recent advances and future directions. Vadose Zone J. 2004, 3, 338–351. (17) Nieber, J. L. Water movement and chemical transport in the environment, Honolulu, HI, 2001. In the Relation of Preferential

(18) (19)

(20) (21) (22) (23) (24) (25) (26)

(27)

(28)

(29) (30)

(31) (32)

(33)

Flow to Water Quality, And Its Theoretical and Experimental Quantification; American Society of Agricultural Engineers: Honolulu, HI, 2001; pp 1–10. Flury, M.; Fluehler, H.; Jury, W. A.; Leuenberger, J. Susceptibility of soils to preferential flow of water: A field study. Water Resour. Res. 1994, 30, 1945–1954. Hendrickx, J. M. H.; Flury, M. Uniform and preferential flow mechanisms in the vadose zone. In Conceptual Models of Flow and Transport in the Fractured Vadose Zone; National Academy Press: Washington, DC, 2001. Saiers, J. E.; Hornberger, G. M.; Harvey, R. W. Colloidal silica transport through structured heterogeneous porous media. J. Hydrol. 1994, 163, 217–288. Ullum, M. Effects of water content and soil structure on colloid transport in porous media. Ph. D. dissertation,Technical University of Denmark, Copenhagen, 2001. Bradford, S. A.; Simunek, J.; Bettahar, M.; Tadassa, Y. F.; van Genuchten, M. T.; Yates, S. R. Straining of colloids at textural interfaces. Water Resour. Res. 2005, 41, W10404. Bradford, S. A.; Bettahar, M.; Simunek, J.; van Genuchten, M. T. Straining and attachment of colloids in physically heterogeneous porous media. Vadose Zone J. 2004, 3 (2), 384–394. Shani, C.; Weisbrod, N.; Yakirevich, A. Colloid transport through saturated sand columns: influence of physical and chemical surface properties on deposition. Colloids Surf., A 2007. van Genuchten, M. T. A closed-form equation for predicting the hydraulic conductivity of unsaturated soil. Soil Sci. Soc. Am. J. 1980, 44, 892–898. Eaton, A. D.; Clesceri, L. S.; Greenberg, A. E., Eds. Standard Methods for the Examination of Water and Wastewater, 19 ed.; American Public Health Association Publications: Washington, DC, 1995. Ledin, A.; Karlsson, S.; Duker, A.; Allard, B. Measurements in situ of concentration and size distribution of colloidal matter in deep groundwaters by photon correlation spectroscopy. Water Res. 1994, 28 (7), 1539–1545. Livshitz, Y. The influence of natural and artificial factors on the chemical composition of the groundwater in the northwestern Negev and the southern portion of the Shfela. Ph. D. dissertation,Ben-Gurion University, Beer-Sheva, 1999. Zvikelsky, O.; Weisbrod, N. Impact of particle size on colloid transport in discrete fractures. Water Resour. Res. 2006, 42, W12S08. Zhuang, J.; Qi, J.; Jin, Y. Retention and transport of amphiphilic colloids under unsaturated flow conditions: Effect of particle size and surface property. Environ. Sci. Technol. 2005, 39 (20), 7853–7859. Lenhart, J. J.; Saiers, J. E. Transport of silica colloids through unsaturated porous media: experimental results and model comparisons. Environ. Sci. Technol. 2002, 36, 769–777. Keller, A. A.; Sirivithayakorn, S. Transport of colloids in unsaturated porous media: Explaining large-scale behavior based on pore-scale mechanisms. Water Resour. Res. 2004, 40, W12403. Simunek, J.; Sejna, M.; van Genuchten, M. Th. The HYDRUS-2D Software Package for Simulating Two-Dimensional Movement of Water, Heat, and Multiple Solutes in Variably Saturated Media, Version 2.0, IGWMC-TPS-53; International Ground Water Modeling Center, Colorado School of Mines: Golden, CO, 1999.

ES071780F

VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1071