| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Biological and Environmental Engineering, Riley-Robb Hall, Cornell University, Ithaca, NY 14853
b Department of Earth and Planetary Sciences, University of Tennessee, Knoxville, TN 37996
* Corresponding author (tss1{at}cornell.edu)
Received 29 January 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
Abbreviations: AW, air–water • AWS, air–water–solid • BTC, breakthrough curve • DLVO, Derjaguin–Landau–Verwey–Overbeek forces • PV, pore volume • SW, soil–water • 2D, distilled–deionized
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
In a review of colloid transport in the vadose zone, Lenhart and Saiers (2002) described the transport and distribution of colloids in the vadose zone as advection and dispersion, together with a sink–source term. The advection–dispersion is relatively well understood and could be modified to include the effects of preferential flow (Steenhuis et al., 2001). Difficulties arise from size exclusion that limits the accessibility of colloids to portions of the pore space. Determining the sink–source term is even more uncertain and currently an area of active research. As shown below, one of the major limitations in understanding and modeling this term is the difficulty of visualizing the processes in a medium where the location and extent of the AW interface is a function of many factors, such as matric potential and the previous wetting history (Lenhart and Saiers, 2002). Visualization is difficult; therefore, most studies have been limited to colloid breakthrough experiments, as well as conceptual, analytical, and/or computer models. Colloid breakthrough experiments in partly saturated media (Schäfer et al., 1998a; Jewett et al., 1999; Jin et al., 2000; Chu et al., 2001; Lenhart and Saiers, 2002) show that more hydrophobic colloids, compared with hydrophilic colloids, are retained in the porous media under otherwise similar conditions. Moisture content and interfacial energies also play an important role. While under saturated conditions all or most negatively charged hydrophilic colloids will be transported through clean sands, breakthrough diminishes with decreasing moisture contents.
Reduced transport at lower moisture contents is often attributed to retention of colloids at the AW interface, the area of which increases at lower moisture content. This interpretation arises from the pore-scale visualization studies performed by Wan and Wilson (1994a)( 1994b). Those researchers, employing etched glass micromodels, observed retention of microspheres and bacteria at the edges of air bubbles within the pores of the two-dimensional micromodel. Crist et al. (2004) questioned this interpretation on the basis of their pore-scale visualization in three-dimensional porous media. Their observations suggested that colloids were not retained at the AW interfaces, but rather near the AWS interface near the menisci of pendular rings.
In unsaturated porous media, an additional mechanism of "film straining" in thin water films was postulated by Wan and Tokunaga (1997) and Veerapaneni et al. (2000). In the conceptual model of Wan and Tokunaga (1997), when the water content is below critical moisture content, colloids are retained because the thickness of the water film connecting one pendular ring to the next falls below the diameter of the colloid (Fig. 1 in Wan and Tokunaga, 1997). However, calculations of film thicknesses for a range of matric potentials from –10 to 30 cm (Iwamatsu and Horii, 1996; Lenhart and Saiers, 2002) demonstrated that the thicknesses of the films are tens of nanometers under equilibrium conditions. Thus, such thin films can simply be interpreted as discontinuities in pendular rings between individual grains (Lenhart and Saiers, 2002; Crist et al., 2004).
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
Twelve colloid breakthrough experiments were completed (six with unmodified hydrophilic sand and six with water-repellent sand), producing six sets of replicate experiments with hydrophilic colloids, hydrophobic colloids, and no colloids. Nonfluorescent, blue-dyed polystyrene latex microspheres (Magsphere, Inc., Pasadena, CA) comparable in size to Cryptosporidium parvum oocysts were used in the experiments. The surfaces of the colloids were either negatively charged, hydrophilic 4.8-µm carboxylated microspheres or hydrophobic 5.2-µm underivatized microspheres. Hydrophilic sand consisted of translucent quartz sand (Unimin Corp., New Canaan, CT) with grain diameters equivalent to 0.85 to 1.70 mm, and was washed and rinsed 10 times in distilled water to remove loose surface impurities. The procedure of Bauters et al. (1998) was used to modify the surfaces of sand grains to make them hydrophobic and water-repellent, with a negative matric entry value. The hydrophobic water-repellent sands used in these experiments consisted of a mixture of 0.4% of the modified water-repellent grains with unmodified sand for the remainder.
The infiltration chamber was constructed from 0.5-cm-thick, clear acrylic sheets. Interior dimensions of the chamber were 26.0 cm high, 4.8 cm wide, and 0.5 cm deep. The front plate interfered with the image analysis and was mounted with bolts and wing nuts for disassembly after the sand was added. The infiltration chamber was supported on a mounting assembly at a 45° incline from horizontal and perpendicular to the focus of the camera. A 45° inclination was chosen to maximize gravitational effects while preventing erosion of the packed sand layers during infiltration and drainage. To allow visualization of different locations, the front plate was removed and the viewing area was adjusted across the camera by sliding the chamber along rails on the mounting assembly. Through a sampling port at the bottom of the chamber, effluent samples for the colloid BTCs were collected simultaneously with the visualizations.
The infiltration chamber was prepared by filling it with sand; then one pore volume (PV, 
26 mL) of water (0.1 mM CaCl2, pH 5.6) was delivered through the sampling port at an inlet flow rate of 2 mL min–1. The pore volume was determined in initial experiments at a flow rate of 2 mL min–1 by measuring the volume required to obtain 50% of the initial Cl– concentration in the leachate. With the sand completely wetted, the chamber was placed on the inclined mounting assembly and left to drain undisturbed for 30 min. The front plate was removed, and using a peristaltic pump, a suspension of either hydrophilic or hydrophobic microspheres at a concentration of approximately 3 x 105 particles mL–1 (in a solution of 0.1 mM CaCl2 and pH 5.6) was applied as a point source on the top layer of sand. One pore volume of colloidal suspension was delivered at a flow rate of 2 mL min–1. Two pore volumes of colloid-free solution of the same ionic strength and pH were applied at the same application rate immediately following the input of the colloidal suspension. Effluent from the sampling port was collected every minute during the 3-PV injection sequence. The samples were analyzed by measuring absorbance at 380 nm using a spectrophotometer (Bausch and Lomb, Inc., Rochester, NY). No correction for background levels was required because absorbance in effluent samples from the control experiments was negligible. At the conclusion of the experiments with the unmodified hydrophilic sand, the vertical distribution of retained colloids was determined by sectioning the sand in the chamber at 1-cm intervals. Each layer was oven-dried at 105°C, and the retained colloids resuspended by mixing with 7 mL of distilled–deionized (2D) water for 30 min in a slow-speed agitator. The released colloids were quantified using the spectrophotometer. The sand was oven-dried at 105°C a second time and reexamined with the electro-optical equipment. The efficiency of colloid recovery in the effluent and mean arrival time of the colloids were evaluated using moments analysis of the observed BTCs for each replicate.
Interfacial Potential Energies
The interactions of colloids approaching each other or the AW or SW interfaces were evaluated as the sum of Derjaguin–Landau–Verwey–Overbeek (DLVO) forces, including van der Waals and double layer potential energies. Additionally, hydrophobic forces were considered, although these interactions were important only for interactions of colloids with the AW interface and of hydrophobic colloids with each other. The total potential energy, 
tot, to these interactions was evaluated as a function of the separation distance, x:
 | [1] |
vdW, was estimated using Eq. [2]. Interactions of a colloid with another colloid or a grain surface were formulated as an unretarded sphere–sphere interaction, assuming pair-wise additivity of the interatomic potentials (Eq. [2a]; Hamaker, 1937). Colloid interactions with a macroscopically flat surface (the AW interface) were approximated using Eq. [2b] (Norde and Lyklema, 1989).
 | [2a] |
 | [2b] |
 | [3] |
For the interaction of the colloids with each other, the double layer potential, edl, was calculated for sphere–sphere interaction for the constant potential case (Hogg et al., 1966):
 | [4] |
is the dielectric constant of water (dimensionless), 0 is the permittivity of free space, 
0c is the surface potential of the colloid, and 
is the reciprocal double layer thickness calculated from the valence and ionic strength of the electrolyte solution. The surface potentials, 0, were calculated based on the 
-potential. For small potentials, the potential decays exponentially in the diffuse double layer, and the surface potential is related to the -potential by
 | [5] |
. However, the calculated interaction energy profile did not change substantially over a several-fold range of values of z. The -potential of the colloids in the electrolyte solution used in the experiments was measured using a Zetasizer (Malvern Instruments, Southbough, MA) and found to be –18.6 and –23.4 mV for the hydrophilic and hydrophobic colloids, respectively. The -potential of the quartz sand was taken to be –60 mV (Elimelech, 1985; Elimelech et al., 2000).
For the interaction of colloids with the sand grains or with the AW interface, the double layer potential for a sphere and flat surface was approximated (Norde and Lyklema, 1989):
 | [6] |
0s is the surface potential of the flat surface; the -potential of the AW interface was taken to be –60 mV (Schäfer et al., 1998b).
In addition to the DLVO interactions, hydrophobic forces act between particles and AW interfaces. Asymmetrical hydrophobic interactions between two surfaces can be calculated based on the respective water contact angles (Yoon et al., 1997; Schäfer et al., 1998b). The hydrophobic interaction energy between small particles and a flat surface thus can be described by (Schäfer et al., 1998b)
 | [7] |
The force constant, K123, for asymmetric interactions between macroscopic bodies 1 and 2 in medium 3 can be described as (Yoon et al., 1997; Schäfer et al., 1998b)
 | [8] |
1 and 2 are the water contact angles for the AW interface (180°; Schäfer et al., 1998b) and water–colloid interface (10° and 100° for the hydrophilic and hydrophobic latex microsphere, respectively; Wan and Wilson, 1994a, 1994b; Butt et al., 2002). The terms a and b are system-specific constants. For a system of silica surfaces with different contact angles, a was –7 and b was –18 (Schäfer et al., 1998b). We adjusted a and b until the hyd approached zero at large separation distances, which yielded a = –6 and b = –22. |
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
|
As expected from the results of Bauters et al. (2000), the presence of a few water-repellent grains affected the water flow pattern. Infiltration in the unmodified hydrophilic sand produced one fingered flow path for colloid and water movement, measuring approximately 2 cm wide in the upper packed sand layers and increasing to the width of the chamber below the 11- to 13-cm depth (Fig. 3) . In contrast, for infiltration in the weakly water-repellent sand, flow across the whole width of the chamber was established within 2 to 4 cm below the point of application. The type of colloid did not affect the infiltration pattern, although, as will be discussed, differences in the flow pattern may have had an effect on the extent of colloid retention.
|
|
70% retained in the columns), compared with the hydrophilic sand (25.6 ± 0.5% recovery in the effluent) (Fig. 4). It is unlikely that the differences in the flow pattern between the two types of sand played a role in the extent of the retention of hydrophobic colloids because the hydrophilic colloids should have been affected in a similar way. Thus, the difference may be related to greater attraction between the hydrophobic colloids and the strongly water-repellent grains despite that these grains constituted only 0.4% of the porous media. We did not attempt to estimate the potential energy of interaction of the colloids with the water-repellent grains. The depth distribution of the colloids retained in the porous media but capable of being detached by extraction with 2D water was determined for the unmodified hydrophilic sand (Fig. 5) . No measurements were made for the water-repellent sand. For the unmodified hydrophilic sand the relative greatest amount of colloids retained was around the 14-cm depth, where the capillary fringe begins. Besides this similarity, the trends of retention with depth varied with colloid type. The retention of hydrophobic colloids did not show a clear trend with depth in the first 10 cm, where the moisture contents are the smallest, and then decreased farther down (open and solid circles in Fig. 5). The concentration hydrophilic colloids retained on the sand increased first and then became approximately constant below the 15-cm depth (open and solid squares in Fig. 5). Although the trend was correct, the absolute concentrations are underestimated for the hydrophilic colloids because when the amount of colloids in the effluent water and that retained in the soil were summed, the mass balance could only account for 73 and 75% of the total amount of colloids for the replicate columns. The mass balance for the hydrophobic colloids was 95 and 116%. That is, after drying and resuspension in 2D water, all the hydrophobic colloids could be released, but only one-half of the hydrophilic colloids retained on the unmodified sand grains could be removed by the same procedure.
|
Selected video images of the experiment where hydrophilic colloids were added to the unmodified hydrophilic sands are provided in the supplemental material. These three files are large and can best be downloaded with a fast internet connection. Two videos were taken in the viewing area shown in Fig. 6 , which was located 6 cm below the top of the column. In these videos (Videos 1 and 2), we observe four grains (Fig. 6) with four pendular rings partly visible. The large space between the grains is devoid of water and filled with air. Colloids moved through three of the pendular rings (labeled A, B, and C), with water flowing down from the top of the figure. These pendular rings are apparently connected at a level below the visual capabilities of the experimental setup. Several processes affecting colloid transport are evident in the still images and videos. These retention and mobilization processes are discussed below.
|
|
|
|
In addition, Video 1 shows that colloids are unlikely to be retained at the AW interface as proposed originally by Wan and Wilson (1994a). The water in the AW interface, except close to the edges near the solid interface, is in motion, and any colloid at the AW interface will move with the water in the main flow direction. The water in the pendular ring labeled D is stationary and, at the same time, is unconnected during infiltration and drainage. Therefore, colloids were not present at any time, and the pendular ring cannot retain any colloids.
Finally, in Video 2 another mechanism of colloid retention can be observed that was not visible in the still images. In the images of Video 2, which is at the location depicted in Fig. 6 and appears later in the flow experiment, aggregated particles are seen accumulated in the lowest part of the pendular ring labeled A. This is likely due to gravitational settling of the aggregates near the meniscus even though it is apparent that the water is flowing upward at this time. It is interesting to see that sometimes the water movement stops and even moves backward. This has a large effect on the stability of the accumulated colloids. It is not clear if a blockage upstream or the pump caused the change in flow conditions. In this video, filtering of colloids can also be seen in pendular rings labeled B and C in Fig. 6. It is a dynamic process where colloids sometimes move slowly and sometimes quickly, almost similar to rocks as bed load in a fast flowing river.
Retention of Hydrophobic Colloids
For the hydrophobic colloids (Fig. 8b), retention at the AWS interface was present but minor compared with the hydrophilic colloids. The major deposition mechanisms are due to the strong attachment force that exists between hydrophobic colloids, resulting in the formation of larger colloid aggregates that can be more easily filtered by the relatively narrow pore spacing close to where the grains are touching. Also, the colloidal aggregates can attach to those colloids already present at the grain surface. The dynamic nature of this process was not captured on video, but it is well illustrated in still images (Fig. 8b). Even though the injection solution contained colloids of a uniform size, it is apparent from visual observations during the experimental run that the mobile colloids occur in a range of sizes, suggesting that the colloids have aggregated during transport.
The images of colloid retention are in agreement with the BTCs. The presence of fewer hydrophobic than hydrophilic colloids in the drainage water is consistent with rapid aggregation of the hydrophobic colloids and retention of the aggregates by straining at pore throats as postulated by Bradford et al. (2002)(2003). Although some aggregate formation was observed for the hydrophilic colloids, the images suggest that aggregation was far less extensive than for the hydrophobic colloids, which is consistent with the existence of a relatively small repulsive energy barrier for hydrophilic colloid–colloid interactions.
Absence of Film Straining
The resolution of the visualization equipment made it difficult to prove the absence or presence of film straining, because water films thinner than approximately 1 µm could not be observed. Despite that, we did not find any microspheres on the grains itself away from the AWS interface. Thus, the relatively large 5-µm microspheres were not strained by the films where water moves from one pendular ring to the next via film flow. Further, we would expect that, as water covering the grain thins to films during drainage and films approach the diameter of the colloids, capillary pressure would push the colloids toward the bulk solution, as shown in Sur and Pak (2001) for suspended films.
Mobilization of Colloids Deposited at the Air–Water–Solid Interface
During a trial run, the water flow was increased, and the accompanying video (Video 3) demonstrates colloid behavior consistent with the proposed mechanism of Saiers and Lenhart (2003a) that immobilization of the hydrophilic colloids at the AWS interface was easily reversible. In this video, coagulated colloids are deposited at the AWS interface, but as soon as the water surface expands due to an increase in flow rate, the deposited colloids are swept away. Moreover, the video images show an air bubble trapped in the pore space. While a few colloids are trapped within the ring, the majority of the colloids pass by the bubble.
Mechanisms of Colloid Retention and Transport
On the basis of the results of the Darcy-scale BTCs and the video and still images at the pore scale, we can identify the mechanisms of colloid transport in our partially saturated porous media.
The observed retention of a greater amount of hydrophobic than hydrophilic colloids is consistent with the calculated balance of attractive and repulsive electrostatic forces. Repulsive energy barriers limit attachment between the colloids and the hydrophilic surfaces of the unmodified sand grains, as well as with the AW interface (Fig. 2). The absence of an energy barrier between hydrophobic colloids and attractive interactions at shorter separation distances favor aggregation of the hydrophobic colloids. Rapid aggregation would be expected in the absence of energy barriers, resulting in formation of the extended structures seen in Fig. 8b. Kim and Berg (2000) also observed that as the aggregate grew, new particles were more likely to contact and immediately attach to the periphery of existing aggregates. This is presumably the most significant mechanism of hydrophobic colloid retention.
The retention of hydrophilic colloids in the very thin film of water at the edge of the menisci is probably the result of hydrodynamic processes. Saiers and Lenhart (2003b) also reasoned that silica colloids were trapped in the narrow wedges near the three-phase contact of pore-corner menisci and at the termini of discontinuous corner water, whereas colloid retention did not occur at water films. The centrifugal motion within the pendular rings observed in the videos would tend to force colloids toward the AW interface, and then any deviation from the primary direction of flow could move the particle toward the AWS interface, as shown in Video 1. Under laminar flow, the particle velocity approaches zero near the grain surface and increases at distances away from the surface. Thus, colloids propelled toward the edge of the meniscus would encounter the very slow moving water near the AWS interface and become immobilized. Alternately, or in addition, the retention in the thinnest portion of the SW interface at the edge of the meniscus may represent a form of film straining, with colloids becoming immobilized in films as the film thickness approaches that of the colloid diameter. Assuming a contact angle for water and quartz sand of 30° (Freitas and Sharma, 1999), the meniscus thickness will equal that of the colloid diameter (5 µm) at a distance of only 5 to 7 µm from the edge of the AWS interface. However, when the immobilized colloids are aggregated, we find that for an aggregate 10-fold larger than the primary particle, the theoretical distance from the AWS interface (50 µm) is in better agreement with the observed distances (e.g., Fig. 8A).
Lenhart and Saiers (2002) found that colloid transport in unsaturated porous media depended principally on the degree of pendular ring discontinuity, pore water velocity, and the retention capacity of the AW interface. Since we worked with relatively large colloid sizes (5 µm), our interpretation of the transport processes in the vadose zone is not inconsistent with this description, but requires the following modifications. Retention occurs at the AWS interface, and depends on the retention capacity of this triple-point interface. Second, particle motion within the curved pendular rings may be important in attachment at the AWS interface, as shown in the video images. Furthermore, the mechanisms for colloid retention at the AWS interface are not clear, but may be related to factors such as pore water motion with the pendular rings, low laminar flow velocities near the grain surface, and/or retention of colloids or colloidal aggregates in the thin water films near the AWS interface. The latter is affected by changes in flow regime (Video 3). More studies are needed to examine the importance of these processes.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
Video 1. Retention of hydrophilic colloids at AWS interface.
Video 2. Mechanisms of colloid retention in unsaturated porous media.
Video 3. Instability of colloid retention at AWS interface with changing flow rate.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. Sharma, H. M. Abdou, and M. Flury Effect of the Lower Boundary Condition and Flotation on Colloid Mobilization in Unsaturated Sandy Sediments Vadose Zone J., August 1, 2008; 7(3): 930 - 940. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. Bradford and S. Torkzaban Colloid Transport and Retention in Unsaturated Porous Media: A Review of Interface-, Collector-, and Pore-Scale Processes and Models Vadose Zone J., May 27, 2008; 7(2): 667 - 681. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Flury and H. Qiu Modeling Colloid-Facilitated Contaminant Transport in the Vadose Zone Vadose Zone J., May 27, 2008; 7(2): 682 - 697. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Gargiulo, S. A. Bradford, J. Simunek, P. Ustohal, H. Vereecken, and E. Klumpp Bacteria Transport and Deposition under Unsaturated Flow Conditions: The Role of Water Content and Bacteria Surface Hydrophobicity Vadose Zone J., May 1, 2008; 7(2): 406 - 419. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Torkzaban, S. M. Hassanizadeh, J. F. Schijven, H. A. M. de Bruin, and A. M. de Roda Husman Virus Transport in Saturated and Unsaturated Sand Columns Vadose Zone J., July 26, 2006; 5(3): 877 - 885. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Wan and T. K. Tokunaga Comments on "Pore-Scale Visualization of Colloid Transport and Retention in Partly Saturated Porous Media" Vadose Zone J., October 10, 2005; 4(4): 954 - 956. [Full Text] [PDF]
|
|
|
|
|
|
T. S. Steenhuis, J. F. McCarthy, J. T. Crist, Y. Zevi, P. C. Baveye, J. A. Throop, R. L. Fehrman, A. Dathe, and B. K. Richards Reply to "Comments on 'Pore-Scale Visualization of Colloid Transport and Retention in Partly Saturated Porous Media'" Vadose Zone J., October 10, 2005; 4(4): 957 - 958. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||
