VZJ
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 1 August 2008
Published in Vadose Zone J 7:930-940 (2008)
DOI: 10.2136/vzj2007.0163
© 2008 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sharma, P.
Right arrow Articles by Flury, M.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sharma, P.
Right arrow Articles by Flury, M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Sharma, P.
Right arrow Articles by Flury, M.
Related Collections
Right arrow Colloids
Right arrow Colloid-Facilitated Transport
Right arrow Vadose Zone Processes and Chemical Transport

ORIGINAL RESEARCH

Effect of the Lower Boundary Condition and Flotation on Colloid Mobilization in Unsaturated Sandy Sediments

Prabhakar Sharmaa, Hesham M. Abdoub and Markus Flurya,*

a Dep. of Crop and Soil Sciences, Dep. of Biological Systems Engineering, Washington State Univ., Pullman, WA 99164
b Dep. of Soils and Water, Suez Canal Univ., Ismailia, Egypt
* Corresponding author (flury{at}mail.wsu.edu).

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.


Received 4 October 2007.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Theory
 Materials and Methods
 Results and Discussion
 Implications
 REFERENCES
 
In unsaturated soil columns, the boundary condition imposed at the column outlet may cause experimental artifacts. Our objective was to study in situ colloid mobilization during transient, unsaturated flow as affected by the boundary condition at the column outflow. We conducted colloid mobilization experiments by infiltrating unsaturated sandy sediment columns under different bottom boundary conditions: a seepage and a suction control. The mechanisms of colloid mobilization were investigated using force calculations (adhesive and interfacial forces), complemented with flotation experiments, where colloids in the bulk fluid and at the liquid–gas interface were measured separately. More colloids were mobilized under seepage than under suction-controlled boundary conditions. The shape of the colloid breakthrough curves also differed: for the seepage boundary, the maximum of the colloid concentration occurred at the beginning of the column outflow, but for the suction-controlled boundary, colloid concentrations in the outflow increased gradually before reaching a maximum. Colloid mobilization increased with flow rate and decreased with ionic strength for both boundary conditions; however, colloids were mobilized even at ionic strength exceeding the critical coagulation concentration (CCC). Flotation experiments showed that colloids were located both in the bulk fluid and at the liquid–gas interface at electrolyte concentrations less than the CCC, but only at the liquid–gas interface when the CCC was exceeded. Theoretical considerations confirm that interfacial forces at the liquid–gas interface exceeded adhesive forces at all ionic strengths. Both experiments and theory show that the liquid–gas interface had a dominant effect on colloid mobilization.

Abbreviations: CCC, critical coagulation concentration • DLVO, Derjaguin–Landau–Verwey–Overbeek


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Theory
 Materials and Methods
 Results and Discussion
 Implications
 REFERENCES
 
COLLOID MOBILIZATION in the vadose zone is governed by chemical as well as physical factors. Chemical factors include pH, ionic strength, and the surface properties of colloids and sediments (Kretzschmar et al., 1999). Physical factors include flow rate, water content, film straining, pore size, surface heterogeneity, and capillary forces. Increasing flow rate usually leads to increasing water content, so that these two factors usually cannot be separated. Increased water content can cause colloid mobilization by expanding water films (Wan and Tokunaga, 1997; Crist et al., 2005; Gao et al., 2006), by a greater fraction of pores contributing to flow (Levin et al., 2006), or by a reduction of the size of immobile water zones (Saiers and Lenhart, 2003; Gao et al., 2006). Kjaergaard et al. (2004) found that the effect of flow rate on colloid mobilization depended on the initial water content or water potential of the soil: increasing flow rate led to more water-dispersible clay mobilization when the soil was initially wet (high matric potential), but no effect of flow rate was observed for an initially dry soil (low matric potential). Totsche et al. (2007), on the other hand, observed increased colloid mobilization after extended periods of drying followed by heavy rainfall.

Colloids can be attracted to the air–water interface, caused by electrostatic (Wan and Tokunaga, 2002), or hydrophobic forces (Gillies et al., 2005). When particles are attached to the air–water interface, usually capillary forces are dominant (Sirivithayapakorn and Keller, 2003; Gillies et al., 2005). There is some controversy in the literature on whether colloids in a porous medium attach directly to the air–water interface or to the air–water–solid interface (Crist et al., 2004; Chen and Flury, 2005; Crist et al., 2005; Steenhuis et al., 2005; Wan and Tokunaga, 2005).

Solid particles suspended in aqueous solutions strongly interact with the air–water interface. In industrial applications, air bubbles are used to separate suspended particles in aqueous solutions, a technique known as flotation. The process is based on the attachment of small particles to the surfaces of air bubbles as they rise up inside a suspension (Scheludko et al., 1976; Ralston et al., 1999). Hydrophobic particles will attach more easily than hydrophilic particles to air bubbles and will be carried by the rising bubbles to the surface of the suspension, where they can be collected (Crawford and Ralston, 1988). The mechanisms and forces leading particle attachment to bubbles have been investigated by atomic force microscopy (Gillies et al., 2005; Johnson et al., 2006). It was found that for negatively charged particles, Derjaguin–Landau–Verwey–Overbeek (DLVO) and hydrodynamic forces hinder attachment, but hydrophobic forces favor attachment (Gillies et al., 2005). We hypothesize that these forces are also controlling the colloid interactions with the air–water interface in a porous medium.

In situ colloid mobilization from unsaturated soils and sediments has been studied under both field and laboratory conditions. Colloids are usually sampled with lysimeter devices in the field and by collecting column outflow in the laboratory. In a few cases, the column outflow was suction controlled (Lenhart and Saiers, 2003; Kjaergaard et al., 2004; Laegdsmand et al., 2005; Levin et al., 2006; Zhuang et al., 2007); in most cases, however, the bottom boundary was open to the atmosphere, that is, a seepage boundary was used (Jacobsen et al., 1997; Ryan et al., 1998; Laegdsmand et al., 1999; El-Farhan et al., 2000; Schelde et al., 2002; Totsche et al., 2007). Such seepage boundaries disturb the unsaturated flow profile because the soil or sediments have to be water saturated before outflow can occur (Flury et al., 1999). Due to heterogeneity, the bottom will probably initially saturate only locally, causing water to flow horizontally until the bottom is completely saturated (Abdou and Flury, 2004). This boundary-induced increase in water content may impact colloid mobilization, as it is known that more colloids are mobilized at large water contents compared with small water contents. We hypothesize that more colloids are mobilized from a system with a seepage boundary as compared to a suction-controlled boundary and that this constitutes an experimental artifact. Further, we anticipate that the higher the flow rate, the less the impact of the bottom boundary on colloid mobilization.

Our objectives were (i) to study in situ colloid mobilization during transient, unsaturated flow as affected by the boundary condition imposed at the column outflow and (ii) to elucidate the mechanisms of colloid mobilization. Column experiments were performed, where unsaturated sediments were sprinkler irrigated. Two different boundary conditions, a seepage boundary and a suction-controlled boundary, were imposed at the column outlet. We used different infiltration rates to create different liquid–gas configurations inside the columns, and we varied the ionic strength of the infiltration solution to vary DLVO interactions. Theoretical considerations and flotation experiments were used to elucidate the colloid mobilization mechanisms.


   Theory
TOP
 ABSTRACT
 INTRODUCTION
 Theory
Materials and Methods
 Results and Discussion
 Implications
 REFERENCES
 
Colloids are mobilized in a porous medium when the detachment force exceeds the attachment force. We can calculate the net force acting on a colloid in an unsaturated porous medium as follows. We consider the forces acting on a particle exerted by gravity, buoyancy, surface tension, pressure, and adhesion for the case in which a particle in a porous medium is in contact with the liquid–gas interface. We first discuss the forces exerted by the liquid–gas interface and then discuss the forces exerted by adhesion (DLVO forces).

The maximum size of a particle that can float at a liquid–gas interface can be calculated by a force balance using gravity, buoyancy, and interfacial forces. The interfacial forces include surface tension and pressure forces, which, in the general case, have to be calculated numerically. We used the numerical results obtained by Huh and Mason (1974) to plot the maximum size of a spherical particle of 2.65 g cm–3 density that can float at a liquid–gas interface. This density is typical for aluminosilicate-type subsurface colloids. The maximum radius of a particle, as a function of contact angle, that can float at a water–air and ethanol–air interface is shown in Fig. 1 . Water and ethanol were chosen to illustrate the effect of a fluid with high and low surface tension. As the contact angle increases, so does the maximum particle size, and the relationship is approximately linear for contact angles less than 90°. The graph shows that for an air–water interface and a contact angle of 30°, particles with radii up to 650 µm can float at the interface; for a contact angle of 150°, the maximum radius is 2800 µm. For ethanol the radii are smaller because of the smaller surface tension.


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		FIG. 1. Maximum radius of a spherical particle of density of 2.65 g cm–3 that can float at a liquid–gas (air) interface as a function of contact angle for deionized (DI) water and ethanol. (Data points denoted by symbols were taken from Table 1 in Huh and Mason, 1974).

 

View this table:
[in this window]
[in a new window]

  		TABLE 1. Properties of solutions used in the experiments.

 
For particles with radii < 500 µm, the effect of gravity and buoyancy can be neglected, and the force to detach a spherical particle from the liquid–gas interface into the liquid phase is given as (Scheludko et al., 1976; Preuss and Butt, 1998; Pitois and Chateau, 2002)

Formula 1[1]
where R is the particle radius, {gamma} is surface tension of liquid, and {theta} is the advancing contact angle.

The adhesion force is given by the sum of electrostatic and van der Waals forces, which can be calculated from DLVO theory. We calculated electrostatic interaction energies for a colloid–silica sand system using equations for sphere-plate geometry. The electrostatic interaction energies are (Gregory, 1975)

Formula 2[2]
and {Gamma}i for i = 1,2 is defined as

Formula 3[3]
where {varepsilon} is the dielectric permittivity of the medium, R is the particle radius, k is the Boltzmann constant, T is the absolute temperature, z is the ion valence, e is the electron charge, {Psi}0,i is the surface potential of particles and the sediments (taken as the particle and sediment {zeta}-potentials), h is the separation distance, and {kappa} is the inverse of Debye-Hückel length, which is given as

Formula 4[4]
where nj is the number concentration of the ions in solution, and zj is the ion valence.

The van der Waals (vdW) interaction energies were calculated as (Gregory, 1981)

Formula 5[5]
where A is the effective Hamaker constant, and {lambda}0 is a characteristic length of 100 nm. The effective Hamaker constant was calculated using the individual Hamaker constants of the liquid (subscript 1) and solid (subscript 2) for homogeneous interactions (Hiemenz and Rajagopalan, 1997, p. 492):

Formula 6[6]
where A11 is the Hamaker constant of the liquid and A22 is the Hamaker constant of the solid. Finally, the DLVO forces were calculated as

Formula 7[7]
where Fad is the adhesion force, and {Delta}Gtotal is the sum of the electrostatic and van der Waals interaction energies. We assumed a separation distance of h = 0.3 nm (Elimelech et al., 1995) to calculate the values of the DLVO forces.

During infiltration of water into a porous medium, liquid–gas interfaces are moving and exerting forces on particles. By comparing detachment forces exerted by the liquid–gas interface with DLVO forces, we can assess the likelihood that particles are scoured from sediment surfaces during infiltration.


   Materials and Methods
TOP
 ABSTRACT
 INTRODUCTION
 Theory
 Materials and Methods
Results and Discussion
 Implications
 REFERENCES
 
Sediments
We used unconsolidated sediments from the Hanford site (south-central Washington State) for our experiments. The sediments were obtained from 17-m depth below ground surface at the Environmental Restoration Disposal Facility, which is located 8 miles from the Columbia River between the 200 East and 200 West areas of the Hanford Site. The sediments were air dried and stored under ambient laboratory conditions until use. The bulk sediments consisted mainly of quartz, feldspar, mica, magnetite, pyroxene, hornblende, kaolinite, illite, and smectite (Mashal et al., 2004). The fine fraction (diameter < 2 µm) was mainly quartz, kaolinite, illite, and smectite (Czigany et al., 2005). The median particle diameter of the bulk sediments was 797 µm, and the sand (50–2,000 µm), silt (2–50 µm), and clay (<2 µm) fractions were 92, 6, and 2% by weight, respectively. The sediments had a pH of 8.9 (1:2.5 soil/H2O weight ratio), a cation exchange capacity of 25.4 ± 0.4 mmolc kg–1, a base saturation of 87.2%, and total organic carbon content of 146 ± 85 mg kg–1.

Four undisturbed sediment samples (5.37 cm i.d. and 3 cm length) were obtained from the same location as the disturbed samples using a hammer-driven auger. The cores were sealed with tape and stored in a refrigerator at 4°C.

Experimental Setup
The disturbed sediments were packed into a brass column of 5.37 cm i.d. and 3 cm length. Before packing the sediments into the columns, the sediments were moisturized to a water content of 0.04 m3 m–3 to mimic the in situ water content of the sediments at the Hanford Site. In addition, the moisturizing helped to bind the fine material to the coarse fraction of the sediments. The bulk density of the packed sediments was 1.46 ± 0.03 g cm–3, and the porosity was 0.45 ± 0.03 cm3 cm–3. We determined the soil moisture characteristic of the packed sediments by using the hanging water column method (Dane and Hopmans, 2002). The saturated hydraulic conductivity was determined by the constant-head method (Reynolds et al., 2002) and found to be 2.2 ± 0.12 cm min–1.

For the colloid mobilization experiments, the inflow solution was supplied from the top using a sprinkler made of 12 hypodermic needles (22 gauge) and a peristaltic pump (Ismatec IP4, Glattburg, Switzerland). A filter paper (8 µm pore opening) was placed on the top of the column to prevent splashing and to enhance uniformity of the sprinkling application. The bottom of the column consisted of two layers of a nylon membrane (45 µm pore opening, Gilson Company, Lewis Center, OH). The entire column setup was placed on an electronic load-cell connected to a data logger (CR-7X, Campbell Scientific, Inc., Logan, UT) to monitor the overall gravimetric water content during the experiment. The outflow from the column was collected with a fraction collector.

We used two different setups for the bottom boundary: (i) seepage and (ii) suction control (Fig. 2a, b ). For the seepage boundary, we mounted the two nylon membranes at the bottom to the brass column using rubber bands. The membranes were rigid enough to support the sediments inside the column. A glass funnel was mounted below the column to collect outflow, which was then routed via Tygon tubing to a fraction collector. For the suction-controlled boundary, we used the bottom piece of a Tempe cell (Soil Measurement Systems, Tucson, AZ). The two nylon membranes, inserted into the bottom piece of the Tempe cell, and a hanging water column were used as suction-control device. Suction was varied with a hanging water column according to the sprinkling rate.


Figure 2
View larger version (25K):
[in this window]
[in a new window]

  		FIG. 2. Experimental setup for the mobilization experiments: (a) seepage boundary, (b) suction boundary, and (c) flotation experiments. Vertical arrows indicate direction of flow.

 
Mobilization Experiments
Packed Sediments
The mobilization experiments were conducted by infiltrating aqueous solutions under different flow rates and different ionic strengths into the packed sediment columns. To assess the effect of the flow rates on colloid mobilization, we infiltrated deionized water (initial pH 7 and electric conductivity 1 µS cm–1) at flow rates of 0.018, 0.035, 0.071, 0.142, and 0.284 cm min–1. For each flow rate, new sediment was packed into the column. An inert tracer experiment, using 0.4 mM NaNO3 in the infiltration solution, was used to check the column performance. For this tracer experiment, the column effluent was filtered through a 0.2 µm syringe filter, and nitrate in the filtrate was quantified photometrically (HP 8452A Diode Array Spectrophotometer, Hewlett-Packard, Palo Alto, CA) at a wavelength of 220 nm.

We chose one flow rate (0.071 cm min–1, the median flow rate) to study the effect of the ionic strength on colloid mobilization. The ionic strength of the inflow was adjusted with 1, 10, 100, 500, and 1000 mM CaCl2 solutions, corresponding to ionic strengths of 3, 30, 300, 1500, and 3000 mM, respectively (Table 1 ). We measured electrical conductivity in the inflow and outflow to monitor changes in electrolyte concentrations during the experiments.

We determined the concentration of the colloids in the effluent by turbidity measurements at a wavelength of 300 nm (HP8452A Diode Array Spectrophotometer). Before measurement, the vials containing the colloid suspensions were shaken by hand for about 10 s to disperse colloids. A calibration equation was developed by dispersing the Hanford sediments and separating the colloidal fraction < 2 µm in diameter by gravity settling. Standards were then prepared by dilution of the original suspension and gravimetric measurement of colloid concentrations. Two calibration equations were developed, one for colloid concentrations between 5 and 40 mg L–1 and one for 40 and 400 mg L–1. Linear calibration equations were used in the two ranges, and the R2s of the regressions were 0.97 and 1.0, respectively. Samples were diluted when concentrations exceeded the calibration range. The analytical detection limit of the concentration measurements was determined using the standard deviation from several blank measurements (Skoog et al., 1996). The detection limit was less than 0.42 mg L–1 and highest for the 1000 mM CaCl2 solution. The electrophoretic mobility of the colloids in the eluent solution was measured by dynamic light scattering (Zetasizer 3000HAS, Malvern Instruments, Ltd., Malvern, UK).

All experiments were done at ambient laboratory conditions (~22°C). Each experiment was repeated three times to check reproducibility of experiments. We also tested the uniformity of the water flow by using a dye tracer to visualize the infiltration patterns. Visualization tests were performed for each flow rate, using a Brilliant Blue FCF dye solution (Flury and Wai, 2003); the results showed no evidence of preferential flow within the columns.

The concentration data are presented as averages and standard deviations of three replicates. Breakthrough curves are plotted as a function of pore volume, that is, cumulative water outflow normalized by the column pore volume. The pore volume for each experiment was estimated by the volumetric water content at steady state. For the suction-controlled boundary, there was a dead volume between the sediment column and the fraction collector, which we subtracted from the pore volumes when plotting the breakthrough curves. The outflow tubing also caused additional dispersion (Taylor dispersion), and we estimated this dispersion by conducting a tracer test with NO3– in the outflow tubing alone. The volume of outflow affected by Taylor dispersion was determined from the tracer breakthrough curve; we indicate this volume when plotting the colloid breakthrough curves with a shaded pattern. The volume affected by Taylor dispersion was negligible on the scale of interest for our experiments.

Undisturbed Sediments
The four undisturbed columns were used for selected tests of the effect of flow rate on colloid mobilization. The experimental procedure was similar to the packed sediment experiments for both seepage and suction boundary conditions. Since we only had four undisturbed samples, we chose the highest and the lowest flow rates with deionized water as the infiltration solution.

Flotation Experiments
To examine the effect of the liquid–gas interface on colloid mobilization, we conducted a series of flotation experiments, where the column was saturated from the bottom. The top of the column was open, and a second, empty brass cylinder was mounted on top of the sediment column, to allow the liquid to rise above the sediment surface (Fig. 2c). We used the same solutions for these experiments as described above: 0, 1, 10, 100, 500, and 1000 mM CaCl2. In addition, we used ethanol (reagent histological alcohol, ethanol 90% v/v, methanol 5% v/v, isopropanol 5% v/v; Fisher Scientific, Waltham, MA). Ethanol was used because of its low surface tension (we measured a value of {sigma} = 22.42 ± 0.02 mN m–1 at 20°C for the ethanol used), and we expected that less colloids would be mobilized because the interfacial forces are smaller at the ethanol–air interface than at the water–air interface. Ethanol experiments were conducted under ambient laboratory conditions and also in a closed chamber that allowed us to saturate the air with ethanol vapor to minimize ethanol evaporation.

The liquids were pumped into the columns from the bottom at a flow rate of 0.071 cm min–1, and pumping continued until the liquid completely filled the top cylinder, just so that no overflow occurred. Due to surface tension, the liquid–gas interface protruded slightly above the brim of the cylinder without overflowing. We used a stainless-steel gliding rod attached to a collector tube to remove the top layer of the liquid by slicing the gliding rod once over the brim of the column. With this procedure, we collected about 1 mL of solution. We then inserted a hypodermic needle (18 gauge) into the bulk fluid of the top cylinder and extracted about 1 mL of solution. The fluid was not stirred prior the sampling, and the needle was inserted to a depth of 1 cm below the free water surface. The collected suspensions were then filtered through a 45-µm nylon membrane, and the colloid concentrations in the filtrate were quantified by turbidity measurements as described above.

We also tested the effect of the flow velocity on colloid mobilization but used only one ionic strength for these tests (100 mM CaCl2). The flow rates tested were 0.018, 0.071, and 0.284 cm min–1. Colloids were sampled and quantified as described above.

The differences of colloid concentrations between and among different treatments were analyzed by a paired t test at a significance level of p = 0.05 (SAS Institute, 1990). The surface tensions of the fluids were measured for the inflow and outflow solutions using the Wilhelmy plate method at 20°C (K100 Tensiometer, Krüss GmbH, Hamburg, Germany), and differences were analyzed with a paired t test (p = 0.05).

Parameters for Interfacial Force Calculations
We used the Washburn method to determine the colloid–liquid contact angle for the colloids. Sieved sediments (<100 µm) were packed into a Krüss powder sample holder, and the liquid penetration weight gain was measured with a tensiometer (K100, Krüss GmbH, Hamburg, Germany). Contact angles were calculated using the Washburn equation (Chen and Flury, 2005).

We calculated the {zeta}-potentials from the electrophoretic mobilities using the von Smoluchowski equation (Hiemenz and Rajagopalan, 1997). The effective Hamaker constant for water and Hanford sediment–colloid was chosen as 0.83 x 10–20 (Israelachvili, 1992). The effective Hamaker constant for ethanol and Hanford sediment–colloid was calculated with Eq. [6]. The Hamaker constants for Hanford sediments and colloids were assumed to be equal and were approximated as that of silica sand, taken as 4.14 x 10–20 J (Hiemenz and Rajagopalan, 1997). The Hamaker constant of ethanol was taken as 4.2 x 10–20 J (Israelachvili, 1992). Hence, the calculated effective Hamaker constant for ethanol and Hanford sediment–colloid system was 0.22 x 10–23 J.

Water Flow Modeling
We numerically simulated the infiltration of water into initially moist Hanford sediment columns using the HYDRUS-1D code (Simunek et al., 2005). The parameters for the soil moisture characteristic and the unsaturated hydraulic conductivity functions were obtained by fitting the Mualem–van Genuchten model to experimental water characteristic and saturated hydraulic conductivity data using the RETC code (van Genuchten et al., 1991). The upper boundary condition for the simulations was set as constant flux equal to the inflow rates used in the colloid mobilization experiments (Table 2 ), and the initial condition was set to a water potential of –100 cm H2O.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Initial and boundary conditions used for water flow modeling.

 
Two different cases of the lower boundary were simulated. In the first case, the lower boundary condition was set to be a seepage face, and the boundary had to reach saturation before outflow occurred. In the second case, we used a suction-controlled lower boundary condition. Different suctions, varying from 0 to –30 cm H2O, were applied at the lower boundary, and simulations were run until water flow was steady state. The suctions that provided a uniform water content distribution within the sediments were then used as the boundary conditions for the colloid mobilization experiments (Table 2).


   Results and Discussion
TOP
 ABSTRACT
 INTRODUCTION
 Theory
 Materials and Methods
 Results and Discussion
Implications
 REFERENCES
 
Hydraulic Conditions of Mobilization Experiments
Table 2 summarizes the measured and modeled hydraulic conditions of the columns at the different flow rates and bottom boundary conditions. For the seepage boundary condition, both the simulations and the experimental measurements indicate that the sediments were close to saturation (measured effective saturation ranged from 93 to 97%) (Table 2). For the suction-controlled boundary, the water contents decreased with increasing suction, ranging from 46 to 75% effective saturation (Table 2). The water content distributions inside the columns simulated with HYDRUS-1D are shown in Fig. 3 . For the seepage boundary, the water content increased from about 0.418 to 0.422 cm3 cm–3, representing a small, but distinct increase of water content toward the outflow boundary. Overall, the water content distribution along the column was fairly uniform.


Figure 3
View larger version (15K):
[in this window]
[in a new window]

  		FIG. 3. Water content distribution along the depth of the columns for different flow rates simulated with HYDRUS-1D: (a) seepage boundary and (b) suction-controlled boundary. (Note the different scales of the x axes).

 
Effect of Flow Rate and Boundary Condition on Colloid Mobilization
For both the seepage boundary and the suction boundary, more colloids were mobilized as the flow rate was increased (Fig. 4 and 5 ). As the flow rates increased from 0.018 to 0.284 cm min–1, the maximum colloid concentrations in the outflow increased by a factor of five to six for both the seepage and the suction-controlled boundary conditions (Fig. 5a). Similarly, the larger the flow rate, the larger the total mass of colloids mobilized (Fig. 5b). This positive effect of flow rate and water content on in situ colloid mobilization during infiltration under unsaturated flow is consistent with previous results (Saiers and Lenhart, 2003; Gao et al., 2004; Gao et al., 2006). Also under drainage conditions, where water is displaced by air, increasing drainage rates have been found to lead to increased colloid mobilization (Saiers et al., 2003; Zhuang et al., 2007). As in unsaturated porous media, water content and flow rate cannot be adjusted independently, it is difficult to elucidate the relative contribution of the two variables, and the literature does not allow to differentiating whether the flow rate or the water content is the dominant factor for colloid mobilization.


Figure 4
View larger version (30K):
[in this window]
[in a new window]

  		FIG. 4. Colloid breakthrough curves from in situ colloid mobilization at different water flow rates for (a) seepage and (b) suction-controlled bottom boundary. Infiltration solution was deionized water. Tracer breakthrough curves for step input (NO3–) for (c) seepage and (d) suction-controlled bottom boundary. Error bars represent standard deviations of three replicates; for colloids breakthrough curves, only every fifth error bar is shown. Shaded area represents the mixing zone affected by Taylor dispersion in the outflow tubing of the suction-controlled system.

 

Figure 5
View larger version (16K):
[in this window]
[in a new window]

  		FIG. 5. (a) Maximum colloid concentrations and (b) total colloid masses in column outflow as a function of water flow rates for seepage and suction-controlled bottom boundary conditions. Infiltration solution was deionized water. Error bars represent standard deviations of three replicates.

 
The bottom boundary condition had a pronounced effect on the shape of the colloid breakthrough curves (Fig. 4a, b). For the seepage boundary, the maximum colloid concentrations occurred right with the first outflow sample, and colloid concentration gradually decreased afterward. For the suction-controlled boundary, colloid concentrations in the outflow gradually increased, until reaching a peak, and then decreased continuously. The Taylor dispersion in the outflow tube, as indicated with the shaded area in Fig. 4, had a negligible effect on the breakthrough curve. A second important feature is that the maximum colloid concentrations, as well as the total amounts of colloids mobilized, were considerably greater under the seepage boundary than under suction-controlled boundary conditions (Fig. 5a, b).

The tracer breakthrough curves (Fig. 4c, d) show that for the seepage boundary, the tracer arrived with its maximum concentration at the beginning of the outflow, whereas for the suction-controlled boundary, the tracer concentrations in the outflow increased gradually, mirroring the pattern of the colloid breakthrough. No trend as a function of flow rate was observed for the tracer.

The results from undisturbed sediments show a similar type (shape) of breakthrough curves as those from the packed sediments for both seepage and suction boundary conditions (Fig. 6 ). However, the magnitude of the colloid concentrations was not the same between the undisturbed and the packed sediments; larger concentrations were observed for the packed sediments.


Figure 6
View larger version (12K):
[in this window]
[in a new window]

  		FIG. 6. Colloid breakthrough curves from in situ colloid mobilization at two different water flow rates using undisturbed sediments for (a) seepage and (b) suction-controlled bottom boundary. Infiltration solution was deionized water. Shaded area represents the mixing zone affected by Taylor dispersion in the outflow tubing of the suction-controlled system.

 
Inspecting the literature on in situ colloid mobilization under unsaturated flow shows that under seepage boundary conditions, the colloid breakthrough curves often have a similar shape as ours (Jacobsen et al., 1998; Schelde et al., 2002; de Jonge et al., 2004). In a few cases where suction-controlled boundaries were used, the type of colloid breakthrough curves as shown in Fig. 4b were observed (Saiers and Lenhart, 2003; Gao et al., 2004). Some inconsistencies, however, exist in the literature. Zhuang et al. (2007) reported breakthrough curves, from a suction-controlled boundary, that are similar in shape to our breakthrough curves for a seepage boundary. No consistent shapes of colloid breakthrough curves for different ionic strengths infiltration solutions were reported by Laegdsmand et al. (2005), who also used a suction-controlled boundary.

The different outflow behavior between seepage and suction-controlled boundary is caused by different water contents reached at a given flow rate for the two boundary conditions: a higher effective water saturation was reached for the seepage boundary than for the suction boundary (Table 2). On the contrary, the pore water velocities for the seepage boundary were less than those for the suction boundary (Table 2). If flow rate would control colloid mobilization, we would expect more colloid mobilization under suction boundary conditions. However, because we observed more colloids mobilization under seepage than under suction boundary, we infer that the water content had a dominant effect on colloid mobilization.

Effect of Ionic Strength and Boundary Condition on Colloid Mobilization
The electrical conductivity measurements showed no significant change between inflow and outflow for solutions with ionic strengths larger than 10 mM CaCl2 (Table 1). However, for the 0 and 1 mM CaCl2 solutions, the electrical conductivity of the outflow was elevated for the first few outflow samples and then decreased steadily (Table 1). The electrophoretic mobilities of the colloids in the effluent showed no significant differences among samples within the same breakthrough curve and no significant differences between seepage and suction boundary conditions when comparing respective ionic strengths. However, significant differences were observed among the different ionic strengths: average values were –1.94, –1.45, –1.34, and –0.61 µm s–1/(V cm–1) for 0, 1, 10, and 100 mM CaCl2, respectively. No measurements could be made for 500 and 1000 mM CaCl2 because of coagulation.

Increasing the ionic strength of the infiltration solution caused a decrease in the concentrations and total mass of colloids mobilized under both seepage and suction boundary conditions (Fig. 7 and 8 ). The decrease in colloid mobilization with increasing ionic strength is well known from other studies as well. Most of these observations came from saturated porous media (Ryan and Gschwend, 1994; Nocito-Gobel and Tobiason, 1996; McCarthy et al., 2002; Lenhart and Saiers, 2003), but similar observations were also reported from unsaturated porous media (Laegdsmand et al., 2005; Zhuatng et al., 2007). As the CaCl2 concentration increased from 0 to 10 mM, we observed a drastic reduction of the amounts of colloids mobilized, both in terms of maximum concentrations and total mass (Fig. 8a, b). Nonetheless, even at the highest ionic strength used, 1000 mM CaCl2, we still observed colloid mobilization. The maximum concentrations were still in the range of 25 to 45 mg L–1 (Fig. 7a, b insets). Under high ionic strength, colloid breakthrough curves under unsaturated flow did not show a pronounced tailing, as was observed for low ionic strengths. We suspect that under high ionic strength, the mobile colloids were located mainly at the liquid–gas interface and that they were eluted from the porous medium as the liquid–gas interface was displaced during an infiltration event. Most of the colloids were therefore eluted when the liquid–gas interface reached the bottom of the porous medium.


Figure 7
View larger version (16K):
[in this window]
[in a new window]

  		FIG. 7. Colloid breakthrough curves from in situ colloid mobilization at different CaCl2 concentrations for (a) seepage and (b) suction-controlled bottom boundary. Flow rate was Jw = 0.071 cm min–1. Error bars represent standard deviations of three replicates. Shaded area represents the mixing zone affected by Taylor dispersion in the outflow tubing of the suction-controlled system.

 

Figure 8
View larger version (16K):
[in this window]
[in a new window]

  		FIG. 8. Maximum colloid concentrations (a) and total colloid masses (b) in column outflow as function of CaCl2 concentrations for seepage and suction-controlled bottom boundary conditions. Flow rate was Jw = 0.071 cm min–1. Error bars represent standard deviations of three replicates.

 
As is shown in the next section, no colloids were present in the bulk fluid at electrolyte concentrations larger than 10 mM CaCl2, corroborating that at high ionic strength, the eluted colloids were located at the liquid–gas interface. It is expected that as the ionic concentration of the infiltration solution exceeds the CCC, colloids in the bulk pore water will coagulate and sediment out (Grolimund et al., 1996; Grolimund et al., 2001; Czigany et al., 2005).

Effect of the Liquid–Gas Interface on Colloid Mobilization
The results of the flotation experiments indicate that colloids were captured at the liquid–gas interface. For all solutions used, a considerable amount of colloids was found at the liquid–gas interface (Fig. 9 ). The colloid concentrations at the liquid–gas interface were significantly larger than in the bulk fluid for all cases. Although the absolute values of the colloid concentrations at the liquid–gas interface are somewhat subject to the measurement method used, our data indicate that a substantial fraction of the colloids mobilized was located at the liquid–gas interface. We would expect that the larger the surface tension, the greater the amount of colloids captured at the liquid–gas interface, because the capillary and interfacial forces increase with surface tension. There were differences in colloid concentrations at the liquid–gas interface among the different liquids (see significance matrix in Fig. 9); however, the differences did not show a consistent trend. Unexpectedly, the solution with the lowest surface tension (ethanol) had the significantly highest concentrations of colloids at the liquid–gas interface.


Figure 9
View larger version (34K):
[in this window]
[in a new window]

  		FIG. 9. Colloid concentrations at the liquid–gas interface and in the bulk liquid measured for the flotation experiments with flow rate of 0.071 cm min–1. Error bars represent standard deviations of three replicates. Insert matrix shows statistical significance of colloid concentrations at the liquid–gas interface among different solutions (NS: not significantly different; S: significantly different; numbers represent CaCl2 concentrations in mM).

 
The importance of the liquid–gas interface on colloid mobilization has been described in previous studies, where colloids were mobilized through transient flow events (El-Farhan et al., 2000; Saiers et al., 2003; Zhuang et al., 2007). However, in these studies, the mechanisms by which the liquid–gas interface mobilizes colloids have not been identified, nor have mobilized colloids directly been quantified at the liquid–gas interface. Using pore-scale visualization techniques, Gao et al. (2006) found that colloid mobilization occurred from immobile water zones when thin films expanded during increase of flow rate and water content. Particle removal caused by a moving liquid–gas interface was reported by Noordmans et al. (1997) and Gomez-Suarez et al. (1999a). These authors used gas bubbles to detach adhered particles from glass slides and quantified particle removal by microscopy. Their experimental and theoretical results clearly demonstrate that particles can be detached from a solid surface by a moving liquid–gas interface.

The analysis of the bulk liquid showed that deionized water had the largest colloid concentration in the bulk liquid (Fig. 9). Already at 1 mM CaCl2, the colloid concentration dropped significantly, and for ≥10 mM, the measured colloid concentrations were below the analytical detection limit. The drop in colloid concentrations agrees with the CCC for coarse Hanford sediment colloids, which were reported to range from 0.7 to 1.4 mM CaCl2 (Czigany et al., 2005). The colloid concentrations in the bulk liquid dropped with increasing ionic strength, but the interfacial concentrations did not drop. This shows that the colloids at the liquid–gas interface were protected against sedimentation.

Saiers et al. (2003) reported an inverse relationship between colloid scouring by the air–water interface and ionic strength: for ionic strengths ranging from 1 to 50 mM NaCl, the scouring probability decreased with increasing ionic strength. They hypothesized that at the higher ionic strength, colloids formed aggregates at the silica surfaces and were then less susceptible for mobilization by the air–water interface. Our data also showed decreasing colloid mobilization with increasing ionic strength as long as the CCC was not exceeded; however, no consistent trends were evident at concentrations larger than the CCC.

For ethanol, the colloid concentrations in the bulk fluid were smaller than for 0 and 1 mM CaCl2 solutions (Fig. 9). No significant differences were observed between colloid concentrations at the ethanol–air interface for ambient conditions and evaporation control. Generally, charge-stabilized colloidal suspensions are less stable in alcohol than in water because alcohol reduces the electrical repulsion between particles (Permien and Lagaly, 1994). For instance, the CCC of Na montmorillonite decreased with increasing alcohol contents of the suspensions; montmorillonite-alcohol complexes formed at high alcohol contents, and colloidal suspensions became unstable (Permien and Lagaly, 1994). In our ethanol experiment, colloids were therefore probably not in stable suspension, and only a small amount of colloids is expected to be in the bulk fluid.

The results of the velocity experiments did not show a significant effect of flow velocity on the colloid concentrations at the air–water interface for the range of velocities tested. The colloid concentrations were 312 ± 17, 307 ± 37, and 310 ± 24 mg L–1 for flow rates of 0.018, 0.071, and 0.284 cm min–1, respectively.

Interfacial Force Considerations
We calculated the forces exerted on a particle attached to the solid phase during an infiltration event. Before infiltration, the particles attached to the solid phase are partially exposed to the air phase. Because of the prewetting of the sediments, a thin liquid film formed strong attractive capillary forces pinning the particles to the solid surfaces. As we infiltrated fluid under saturated conditions (infiltration from the bottom), these liquid films expanded and ultimately caused a repulsive force on the particles away from the solid phase as the liquid–gas interface moved along with the displacement of the gas phase. Particles detached from the solid surface when adhesion (DLVO) forces were less than detachment (capillary) forces. Our calculations showed that the detachment forces exceeded the adhesion forces for all particle radii (Fig. 10 ). The larger the particle radius, the larger was the force difference between detachment and adhesion. We calculated the forces for all the different fluids used in our experiments, using the surface tensions and densities shown in Table 1. The measured contact angles for the colloids with air–water and air–ethanol were 26° and 10°, respectively. The measured fluid surface tensions did not significantly differ between inflow and outflow solutions. The forces among the electrolyte solutions were similar; however, both the capillary and the DLVO forces were much smaller for ethanol (Fig. 10).


Figure 10
View larger version (22K):
[in this window]
[in a new window]

  		FIG. 10. Comparison of detachment (capillary) and adhesion (DLVO) forces for our experimental system. Surface tension forces were calculated with Eq. [1], DLVO forces with Eq. [7].

 
These calculations indicate that the forces exerted by the liquid–gas interface can indeed detach colloids from the sediment surface. As the liquid–gas interface moves through the porous medium, colloids that are attached to the liquid–gas interface are displaced along with the liquid–gas interface. As colloids are captured at the liquid–gas interface, they are floating on the liquid surface, as in a typical mineral flotation process (Schulze, 1977; Leja, 1982; Ralston et al., 1999). The maximum size of spherical particles that can float on a liquid–gas interface (air–water and air–ethanol) based on a force balance is shown in Fig. 1, and the specific radii for our experimental system are listed in Table 1. These data show that the floatable particle size is on the order of hundreds of micrometers, consistent with what is known from mineral flotation (Scheludko et al., 1976; Pitois and Chateau, 2002).

These force calculations support our experimental observations of high particle concentrations at the liquid–gas interface (Fig. 9). The different ionic strengths of the aqueous solutions did not result in significant changes of the DLVO and capillary forces (Fig. 10), which corroborates our experimental finding of a nonconsistent trend of particle concentrations at the liquid–gas interface as a function of ionic strength. The force calculations also show that particles up to a few hundred micrometers can float on the liquid–gas interface (Table 1). In our experiment, we filtered the samples through a 45-µm membrane, eliminating larger particles. Therefore, the measured particle concentrations at the liquid–gas interface are only approximate. Nonetheless, our data show the pronounced effect of flotation.

That particles attach to liquid–gas interfaces is known from flotation and bubbling experiments (Scheludko et al., 1976; Ralston et al., 1999; Wan and Tokunaga, 1998; Wan and Tokunaga, 2002) and from micromodel studies (Sirivithayapakorn and Keller, 2003; Lazouskaya et al., 2006). An important difference in our experiments is that our colloids are forced to contact the liquid–gas interface, whereas in previous bubbling and visualization experiments, colloids were in the liquid phase and had to penetrate the liquid–gas interface first before strong capillary forces became active. Our experimental scenario is more similar to the air-bubble experiments described by Gomez-Suarez et al. (1999a) and Gomez-Suarez et al. (1999b), where the liquid–gas interface was physically forced to move over attached particles. These latter experiments corroborate that colloids can be scoured from sediment surfaces by moving liquid–gas interfaces.


   Implications
TOP
 ABSTRACT
 INTRODUCTION
 Theory
 Materials and Methods
 Results and Discussion
 Implications
REFERENCES
 
Our experiments showed that colloid mobilization in the sandy sediments was strongly affected by the type of boundary condition imposed at the column outflow. Not only did the amounts of colloids mobilized differ (the colloid mass mobilized under the suction-controlled boundary was 20 to 60% less than the one under seepage), but also the shapes of the colloid breakthrough curves differed between the boundary conditions. This difference in colloid mobilization was caused by saturation of the sediments at the seepage boundary, which is an experimental artifact in unsaturated flow experiments. The smaller the water flow rates (or the smaller the water content), the more pronounced the artifacts will be. This has important implications for sampling colloids in the vadose zone: any sampling device that causes local saturation, such as a zero-tension lysimeter, will probably cause sampling artifacts, as demonstrated here for a sandy sediment.

Our results further show the importance of interfacial forces associated with the liquid–gas interface on colloid mobilization. Even at ionic strengths larger than the CCC, colloid mobilization occurred. At low ionic strength, the mobilized colloids were located both in the bulk fluid and at the liquid–gas interface, whereas at high ionic strength (larger than the CCC), colloids were associated mainly with the liquid–gas interface and no colloids were found in the bulk fluid. Moving liquid–gas interfaces are effective in mobilizing colloids, and any changes in liquid–gas interface configurations, as caused by sampling devices or outflow boundaries, can cause experimental artifacts.

The results from this study point to the relevance of moving air–water interfaces for colloid mobilization and transport in the vadose zone. Such moving air–water interfaces are common in soils and near-surface sediments, where rainfall, snowmelt, or irrigation cause infiltration and drainage. Current theory for colloid transport in unsaturated porous media does not consider the effect of moving air–water interfaces.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Theory
 Materials and Methods
 Results and Discussion
 Implications
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sharma, P.
Right arrow Articles by Flury, M.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sharma, P.
Right arrow Articles by Flury, M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Sharma, P.
Right arrow Articles by Flury, M.
Related Collections
Right arrow Colloids
Right arrow Colloid-Facilitated Transport
Right arrow Vadose Zone Processes and Chemical Transport

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome