VZJ sign up for etocs
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

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 Hutchison, J. M.
Right arrow Articles by Radcliffe, D. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Hutchison, J. M.
Right arrow Articles by Radcliffe, D. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Hutchison, J. M.
Right arrow Articles by Radcliffe, D. E.
Related Collections
Right arrow Laboratory Column Studies
Right arrow Dispersion
Right arrow Vadose Zone Processes and Chemical Transport
Published in Vadose Zone Journal 2:702-714 (2003)
© 2003 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

ORIGINAL RESEARCH PAPERS

Chromate Transport and Retention in Variably Saturated Soil Columns

J. M. Hutchisona, J. C. Seaman*,a, S. A. Aburimeb and D. E. Radcliffec

a Savannah River Ecology Laboratory, P.O. Drawer E, Aiken, SC 29802
b Department of Engineering, Clark Atlanta University, 223 James P. Brawley Drive, Atlanta, GA 30314
c Department of Crop and Soil Sciences, 3111 Miller Plant Sciences Building, The University of Georgia, Athens, GA 30602
* Corresponding author: (seaman{at}srel.edu).

Received 20 December 2002.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX
 REFERENCES
 
Most subsurface contamination passes through the unsaturated zone before reaching an aquifer; however, transport studies are often conducted under saturated conditions because of the difficulty in maintaining steady-state flow. Chromate migration was measured in coarse-textured, oxide-rich sediment under different water contents using vacuum and centrifuge techniques to obtain a steady-state unsaturated flow regime. Leaching solutions contained 0.5 or 1.0 mM Cr(VI) and tritium in artificial groundwater. Breakthrough curves (BTCs) were modeled using CXTFIT assuming equilibrium conditions, since evaluation of data using a "two-region" physical nonequilibrium model indicated that mobile water was >90% regardless of saturation level. Dispersivity increased nonlinearly with decreasing water content. Retardation (R) increased with decreasing water content, but water content had little effect on the distribution coefficient calculated from R, Kd-app. The average Kd-app of all Cr(VI) experiments (water content range: 0.07–0.43 cm3 cm-3) was 0.633 mL g-1, very similar to the distribution coefficient derived from batch equilibration, Kd (0.684 mL g-1). Though results in both transport systems were similar, average solute residence times in the vacuum system were 4 to 23 times longer than in the centrifuge system at comparable water contents. The centrifuge system column experiments could also be run over a greater range in volumetric water content (0.07–0.42 cm3 cm-3) than the vacuum column system (0.23–0.43 cm3 cm-3).

Abbreviations: AGW, artificial groundwater • BTC, breakthrough curve • CDE, convective–dispersive equation • SRS, Savannah River Site • UFA, Unsaturated Flow Apparatus


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX
 REFERENCES
 
CHROMIUM exists in two oxidation states in soils (III and VI) and is the second most frequently detected inorganic groundwater contaminant found at hazardous waste sites in the USA (National Research Council, 1994). The more soluble hexavalent form, Cr(VI), is relatively toxic to plants and animals, mobile in many vadose zone environments, and consequently a greater threat to groundwater supplies (Bartlett and Kimble, 1976; Griffin et al., 1977). Therefore, studies evaluating solute transport processes in the vadose zone yield important information for situations in which chemical wastes are released at or below the land surface. Despite the fact that virtually all subsurface contamination must pass through the unsaturated vadose zone before reaching an aquifer, most studies focus on processes occurring under saturated conditions due to the difficulty in maintaining steady-state flow and moisture content under unsaturated conditions. However, measurements made under saturated conditions may not mimic the natural system under which solute transport occurs.

The convective–dispersive equation (CDE) (Lapidus and Amundson, 1952) was developed in an attempt to quantify solute transport processes for one-dimensional steady-state flow in a homogeneous soil column:

[1]
where R is the retardation factor, C is the solute concentration (M L-3), D is the hydrodynamic dispersion coefficient (L2 T-1), v is the average pore-water velocity (L T-1), z is distance, and t is time. In order for this equation to be valid, it is assumed that all the fluid participates in the transport process.

Hydrodynamic dispersion (D) is expressed as (Bear, 1969)

[2]
where {lambda} and a are constants, {lambda} is referred to as dispersivity (L) and is a characteristic property of the porous media, a is taken as a value between 1 and 2 (Freeze and Cherry, 1979), and De is the effective molecular diffusion coefficient through the media. At high pore water velocities, the contribution of molecular diffusion (De) to dispersion (D) is negligible and often ignored in the calculation of hydrodynamic dispersion. Therefore, dispersivity is often calculated by dividing D by v, assuming there is a linear relationship between these two parameters (Toride et al., 2003).

Dispersivity is often considered to be an intrinsic property of the porous media, but different values have been reported under varying degrees of saturation. Table 1 is a summary of studies evaluating the effect of moisture content on transport parameters and includes the tracer, a description of the media, the column dimensions and method for controlling steady-state moisture conditions, and the saturation level over which experiments were conducted, along with a short summary of the experimental results. As seen in Table 1, dispersivity values have been shown to increase, decrease, or be unaffected by changes in average water content.


View this table:
[in this window]
[in a new window]
  		Table 1. Review of studies evaluating changes in transport parameters with water content.{dagger}

 
De Smedt and Wierenga (1979)(1984) found dispersion to be linearly related to average pore water velocity (D = 1.2 + 0.021v) for saturated and unsaturated columns using a two-region immobile water model. Corey et al. (1963) also found dispersivity to be constant with water content. However, these studies were conducted on columns of glass beads (De Smedt and Wierenga, 1979, 1984) and intact sandstone (Corey et al., 1963) and therefore may not reflect the conditions present in soil, as glass beads represent a very uniform media and sandstone has few interconnecting pores.

Studies on packed sand columns have found dispersivity to be higher under unsaturated conditions (Maraqa et al., 1997; Jin et al., 2000), even when using a two-region immobile water model (De Smedt et al., 1986; Padilla et al., 1999). In a plot of D vs. v for saturated and unsaturated columns, Maraqa et al. (1997) found that the slope of the regression line for unsaturated columns was 1.9 to 2.8 times greater than the slope for the saturated soil, indicating greater dispersivity under unsaturated conditions. Though experiments were not conducted under a uniform water content distribution, dispersivity in sand was seven times higher under unsaturated conditions than saturated conditions, even after accounting for immobile water (De Smedt et al., 1986). Padilla et al. (1999) improved their linear relationship of D with v by dividing v by the mobile water fraction of the soil.

In contrast, many studies have shown dispersivity to decrease with desaturation (James and Rubin, 1986; Seyfried and Rao, 1987; Wierenga and van Genuchten, 1989; Jardine et al., 1993). For intact soil columns, this can be explained by the draining of macropores on desaturation, decreasing preferential flow and dispersivity (Seyfried and Rao, 1987; Jardine et al., 1993). Other studies have found a decrease in dispersivity with desaturation in packed sand columns (James and Rubin, 1986; Wierenga and van Genuchten, 1989). This packed, sandy media is similar to that used in studies by De Smedt et al. (1986) and Padilla et al. (1999) where an increase in dispersivity with decreasing water content was observed. One explanation for this difference may be that James and Rubin (1986) used different column lengths for saturated and unsaturated experiments and measured concentrations profiles in the column as opposed to BTCs. However, James and Rubin (1986) explained that their findings were the result of a reduction in the possible range in pore sizes available for transport as the larger pores drained during desaturation, though this explanation may be more applicable to structured and intact columns. Wierenga and van Genuchten (1989) found a good linear relationship between v and D, which may indicate that changes in dispersivity with water content are mostly the result of velocity differences.

Many factors contribute to these changes in the hydrodynamic properties of soil with desaturation, including the existence of physical nonequilibrium (regions of immobile water), a wider variation in pore water velocities when the media is desaturated, or an increase in air-filled pore space that increases the tortuosity of the solute flow path with desaturation (Yule and Gardner, 1978; De Smedt and Wierenga, 1984; Fesch et al., 1998).

The degree of saturation may also influence the availability of surface sites for reaction with the transported solutes, influencing the apparent retardation. The retardation factor (R) is often estimated from the distribution coefficient, Kd, derived from batch equilibration studies through the equation:

[3]
where Vw is the velocity of the carrier fluid, Vsp is the velocity of the solute species, {rho}b is the dry bulk density, and {theta} is the volumetric water content. If a solute is nonreactive (Kd = 0), then R = 1 for that species and the solute will travel with the water at the same transport velocity (Hillel, 1980).

The distribution coefficient (Kd) is a specific type of batch equilibration study and is defined as

[4]
where x is the mass of the sorbed solute, m is the mass of the sorbent, and C is the solute concentration. The distribution coefficient can also be derived from dynamic column test retardation coefficients (R) using Eq. [3]. This distribution coefficient is different than that derived from batch equilibration and will be termed the apparent distribution coefficient, Kd-app.

The retardation factor has been shown to increase with decreasing water content (Fesch et al., 1998; Maraqa et al., 1999), as expected from Eq. [3], due to an increase in the solid/liquid ratio with desaturation. However, retardation coefficients observed in column studies often differ from values determined by batch equilibration (Porro et al., 2000; Gamerdinger et al., 2001a). This could be due to a change in Kd with water content, differences in soil/solution ratio in each system (Maraqa et al., 1999), no removal of reaction products in batch studies, limitations in availability of sorption sites in batch vs. column experiments, rate-limited geochemical reactions that become significant only under flowing conditions (Gamerdinger et al., 2001a), and physical nonequilibrium present under flowing vs. static conditions (Wagenet and Chen, 1998).

This study will further the research of Celorie et al. (1989), who compared sorption of phenol on kaolinite in batch and centrifuge systems; Gamerdinger and Kaplan (2001) who compared U sorption in batch, centrifuge, and saturated systems; and Porro et al. (2000), who compared Sr sorption in batch, saturated, and vacuum systems. The current study evaluates chromate transport parameters using batch, saturated, vacuum and centrifuge methods for a range of moisture contents. There is limited data comparing unsaturated transport in the vacuum and centrifuge systems (Lindenmeier et al., 1995) and few studies focusing on unsaturated transport for Cr(VI).

The objectives of this research are to (i) evaluate the impact of water content on the hydrodynamic dispersion of a conservative tracer (tritium) under unsaturated conditions; (ii) evaluate the impact of water content on the sorption properties of Cr(VI) under various water contents; (iii) compare results from batch, saturated, vacuum and centrifuge systems; and (iv) implement modeling approaches to evaluate the impact of water content on various solute transport processes.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX
 REFERENCES
 
Soil
A coarse-textured, vadose zone sediment from the Tobacco Road Formation, displaying characteristics typical of the Atlantic Coastal Plain, was collected on the Department of Energy's Savannah River Site (SRS) near Aiken, SC (Table 2). Samples were air dried and sieved before batch and column experiments. Percentage moisture of the air-dried soil was found to be <1%. The material is coarse in texture (92.65% sand) and has a citrate-dithionite-bicarbonate extractable Fe content of 0.37 g 100 g-1 soil. Kaolinite was found to be the dominant clay mineral, but the clay fraction also contained quartz, goethite, and trace amounts of mica, gibbsite, and hematite.


View this table:
[in this window]
[in a new window]
  		Table 2. Physical and chemical characteristics of Tobacco Road sand used in column and batch studies.

 
Solutions
All solutions for sorption isotherms and column studies were prepared in artificial groundwater (AGW). The AGW, formulated based on routine groundwater monitoring data from the SRS (Strom and Kaback, 1992), contained the following (mg L-1): 1.00 Ca2+, 0.37 Mg2+, 0.21 K+, 1.40 Na+, and 0.73 SO2-4. Tritium was used as the conservative tracer for all column studies and chromate [Cr(VI)] solutions were prepared from reagent-grade K2Cr2O7.

Batch Sorption Isotherm
Batch sorption isotherms were completed before column experiments to determine the range over which Cr(VI) sorption was linear. Chromate concentrations for column experiments were then selected in this linear range.

Five-gram samples of soil were weighed into centrifuge tubes with three treatment replicates. Thirty milliliters of solution containing 0.0, 0.5, 1.0, 1.5, or 2.0 mM of Cr(VI) was added to the appropriate tubes. As a control, equivalent treatment levels were replicated in soil-free tubes to account for Cr(VI) losses in the absence of the sorbing matrix. The tubes were placed on a reciprocating shaker for about 18 h. Previous studies on similar sediment materials indicated that this was sufficient time to achieve equilibrium with respect to Cr(VI) sorption (Seaman et al., 1999).

After equilibration, the tubes were centrifuged for 15 min at 26 000 g (15 000 rpm) on a Sorvall 5B Plus centrifuge (Kendro Laboratory Products, Newtown, CT). The supernatant was then passed through a 0.22-µm pore size nylon membrane filter before Cr(VI) analysis using the diphenylcarbazide method (Clesceri et al., 1989). The method detection limit for Cr(VI) in this study was 0.00012 mM. Samples were analyzed using a Cary 500 Scan UV-Vis NIR Spectrophotometer (Varian, Palo Alto, CA) at a wavelength of 540 nm. Chromate concentrations were compared with the soil-free equivalents to determine sorption based on their difference.

Vacuum-Based Transport Experiments
One method used in this study to maintain unsaturated steady-state flow is the Wierenga vacuum column system (Fig. 1A). Disadvantages to using this technique include the difficulty in maintaining a stable moisture content below 30% of saturation for soil materials, the long experimental time required for a single breakthrough experiment, and the potential for mechanical failure due to laboratory power outages during the extended testing period (Lindenmeier et al., 1995).



View larger version (47K):
[in this window]
[in a new window]
  		Fig. 1. (A) Diagram of the Wierenga column setup for unsaturated flow [adapted from (Jin et al., 2000)] and (B) diagram of the Unsaturated Flow Apparatus components (Khaleel et al., 1995).

 
The vacuum-based column experiments were conducted following the procedure of van Genuchten and Wierenga (1986) using equipment obtained from Soil Measurement Systems, Tucson, AZ. The system consists of a peristaltic pump, packed soil column, vacuum chamber, vacuum pump and regulator, and a fraction collector. The columns are constructed of Plexiglas (30.5 cm long, 2.5-cm i.d.) with bolted flanges at both ends and two tensiometers installed approximately 10 cm from each end. For unsaturated experiments, the lower end of the column had a 1.2-µm pore size membrane filter with a bubbling pressure of greater than 0.06 MPa while the upper end of the column had a 30-µm pore size nylon membrane filter with a bubbling pressure of 0.03 MPa. For saturated experiments, 30-µm pore size nylon membrane filters were used at both ends of the column.

Columns were packed with air-dried soil by adding the soil in approximately 1-cm increments and tapping lightly. Care was taken to avoid obvious layering of the material or segregation of the soil by particle size. Column characteristics and experimental conditions are given in Table 3. The average bulk density, which ranged from 1.43 to 1.52 g cm-3, was calculated as the weight of solid added to the column per unit of column volume.


View this table:
[in this window]
[in a new window]
  		Table 3. Characteristics and experimental conditions for columns run on the vacuum and UFA systems.{dagger}

 
Columns were saturated slowly in an upward fashion with AGW at flow rates <=3.9 cm3 h-1 to minimize the amount of entrapped air in the column. Water content at saturation was determined by weighing the column before and after saturation. Once saturated, the inlet solution line was switched to the top of the column while the bottom of the column was secured to the vacuum chamber. Artificial groundwater flow was established at the top of the column with vacuum applied to the bottom of the column through the vacuum chamber. Since consistent relationships between water content, flow rates, and vacuum pressures were not observed (Table 3), flow rates and vacuum pressures were adjusted until a desired water content was obtained.

Column tensiometer data were erratic and could not be used to indicate uniform water potential. This may have been due to poor contact between the tensiometer and the soil, although care was taken to avoid this occurrence. It was also a matter of concern in using tensiometers for these relatively short experiments, since tensiometers may need days or weeks to reach equilibrium (Sisson et al., 2002). Instead, average water content was determined by repeatedly weighing the column until an equilibrium water content was reached. Columns were also weighed at the end of the experiment to ensure the persistence of a stable average water content and then segmented to verify the vertical water content distribution. Average water contents for each vacuum experiment are listed in Table 3.

Once a stable moisture content was achieved, the inlet solution was switched to the tracer solution containing tritium and Cr(VI). The two concentrations used for Cr(VI) were 0.5 and 1.0 mM. Whenever possible, the inlet solution was displaced through the column until the tracer concentration in the effluent was equivalent to that in the inlet solution. At this point, the inlet solution was switched back to AGW and leached through the column until effluent tracer levels eventually fell below the detection limit.

Fresh sediment materials were used for all eight unsaturated columns run on the vacuum column apparatus.

Centrifugation-Based Transport Experiments
Centrifuge techniques were also employed to simulate unsaturated flow through geologic materials and have the advantage of being able to achieve stable, low water contents in a relatively short time, even for fine-grained materials (Gamerdinger and Kaplan, 2000). The centrifuge apparatus, initially described by Nimmo et al. (1987)( 1992), was modified by researchers at Pacific Northwest National Laboratory (Conca and Wright, 1992) and is referred to as an Unsaturated Flow Apparatus (UFA, model L8-UFA, Beckman Coulter, Inc., Fullerton, CA; Fig. 1B). Possible problems and disadvantages associated with centrifuge column methods are maintaining stable water content along the length of the column, physical compaction problems due to the centrifugal forces, and disruption of solute transport processes during necessary stoppage for collection of effluent samples (Gamerdinger and Kaplan, 2000).

The UFA, described in detail elsewhere (Conca and Wright, 1990; Khaleel et al., 1995; Gamerdinger and Kaplan, 2000), consists of two infusion pumps (AVI 200A, 3M, St. Paul, MN) and a modified Beckman J6-MI centrifuge (UFA Ventures, Richland, WA), including a rotating seal, rotor, and column assembly. The column assembly consists of the sample holder and the effluent collection cup. The packed soil columns, contained in the sample holder, are 5 cm in length with an internal diameter of 3.3 cm. The effluent cup is placed at the end of the sample holder and serves to collect the column effluent during centrifugation. The rotating seal is at the center of the rotor and is the conduit through which fluid is delivered by the infusion pumps to the columns by two separate flow paths. This enables two columns to be run simultaneously with different inlet solutions.

Columns were packed with air-dried soil, weighed, and allowed to saturate with AGW before being placed in the centrifuge. Column characteristics and experimental conditions are given in Table 3. The average bulk density, which ranged from 1.47 to 1.58 g cm-3, was calculated as the weight of solid added to the column per unit of column volume. Columns were flushed with AGW at experimental flow rates and centrifuge speeds for about 2 h before the start of the experiment. Experiments were initiated when steady-state water contents were reached, based on the gravimetric analysis of the packed columns.

Inlet solutions were then switched to those containing tritium and Cr(VI) (0.5 or 1.0 mM). Flow was not initiated until the centrifuge speed reached 75% of its maximum value. This ensured that steady-state flow conditions were maintained (UFA Ventures, Inc., 1996). Whenever possible, the inlet solution was displaced through the column until the effluent solute concentration equaled that of the inlet solution. The column was then flushed with AGW by switching solutions again until effluent tracer levels eventually fell below the detection limit.

Since the effluent collection cups can only hold about 20 mL, centrifugation must be interrupted for sample collection at specified intervals. According to Gamerdinger and Kaplan (2000), stopping centrifugation and flow for sampling purposes does not adversely affect breakthrough for nonsorbing solutes. They also reported water content as stable throughout the soil column except for the final segment at the outflow, in contact with the water-absorbent filter paper. Therefore, since this study used similar sandy materials, UFA columns were not segmented for water content distribution.

Fresh sediment materials were used for each Cr(VI) experiment. Twenty-two columns were run in the UFA [12 columns with 0.5 mM Cr(VI) and 10 with 1.0 mM Cr(VI)], four of which were replicates.

Saturated Columns
Saturated experiments were conducted for both vacuum and UFA systems, using columns of the same dimensions as those used for the unsaturated experiments. These columns were saturated from bottom to top with AGW to remove as much air from the column as possible. For the UFA columns, this method achieved adequate saturation levels. However, vacuum columns could not be saturated more than 75% using this technique. Therefore, the saturated vacuum columns were purged with CO2 before saturation at low flow rates with AGW that had been boiled to remove excess dissolved air. This technique improved saturation from 75 to about 100%. The columns were then positioned horizontally and the AGW replaced with the appropriate tracer solution. For the saturated experiments, the effluent pH and EC were monitored using flow-through electrodes.

Analysis of Samples
Tritium analysis was performed by liquid scintillation counting using a Packard 2550 TR/AB Liquid Scintillation Analyzer (Agilent, Palo Alto, CA). Two milliliters of sample were added to 10 mL of scintillation cocktail (Packard Ultima Gold, Agilent, Palo Alto, CA) and counted for 20 min. Chromate concentrations were determined using the diphenylcarbazide method, described previously.

Modeling
Tritium and Cr(VI) BTCs were modeled using CXTFIT, a program in the Stanmod software package (Version 2.0), developed by researchers at the USDA Salinity Lab, Riverside, CA. This program is an extension and update of earlier versions published over the years by van Genuchten (1981), van Genuchten and Cleary (1979), Parker and van Genuchten (1984), and Toride et al. (1995). Tritium curves did not exhibit tailing, an indication of the presence of immobile water. However, both the deterministic equilibrium and two-region physical nonequilibrium CDE models were used in describing the breakthrough data.

For the equilibrium model, tritium curves were used to fit values of D, which were then used to find appropriate R values for Cr(VI). When using the nonequilibrium model, values of D, ß (mobile water fraction), and {omega} (mass transfer coefficient) were fit to tritium data, and these values were then used to determine R for Cr(VI) data.

When modeling solute BTCs, parameters were slightly sensitive to input values. When fitting D values to the tritium BTCs, values of v were calculated from experimental conditions using the water flux (Jw) and volumetric water content (v = Jw{theta}-1). Initial estimates of D were then obtained by multiplying v by 2. Output values of D were not very sensitive to input values as long as appropriate values of v were used and initial estimates of D could be varied as much as a factor of 10 or more without changing the model results. When fitting R using tritium-based v and D values, it was found that input estimates of R values were less sensitive than D and could be varied by a factor of 4 to 500 before changes in the model results occurred.

Residual Cr(VI) Extraction
Soil from both the saturated and unsaturated column systems was removed on completion of each experiment and allowed to air dry. The soil was then thoroughly mixed and subsampled for extraction of residual Cr(VI). Five-gram samples of soil were added to centrifuge tubes with 15 mL of 10 mM KH2PO4 extracting solution, adjusted to pH 10 (Bartlett and James, 1996). The samples were placed on a reciprocating shaker for 24 h and then centrifuged at approximately 26 000 g (15 000 rpm) for 15 min. The supernatant was filtered with a 0.22-µm pore size nylon filter before analysis of Cr(VI) using the diphenylcarbazide method.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
SUMMARY AND CONCLUSIONS
 APPENDIX
 REFERENCES
 
Batch Sorption Isotherm
Chromate sorption was linear up to about 71 µg mL-1 Cr(VI) equilibrium concentration, where it began to level off (Fig. 2). The linear range of the isotherm encompasses the solute concentration range used in the column studies [0.5 mM (26 µg mL-1) and 1.0 mM (52 µg mL-1) Cr(VI)]. Taking data over the linear range and forcing a regression line through zero produced a Kd for this isotherm of 0.684 mL g-1 (95% confidence interval: ± 0.060; p < 0.001; r2 = 0.9558).



View larger version (13K):
[in this window]
[in a new window]
  		Fig. 2. Cr(VI) sorption isotherm on sandy soil at a soil/solution ratio of 1:6 by mass equilibrated for 24 h, where Se is sorbed Cr(VI) and Ce is aqueous Cr(VI) at equilibration.

 
Transport Experiments
Information pertaining to column experiments [flow rate, bulk density, average volumetric water content, pore volume, pulse duration, retention time of solute in column, and percentage recovery of tritium and Cr(VI)] is included in Table 3. The experiments are named to include the column method (V = vacuum, U = UFA), Cr(VI) concentration (mM), and percent moisture saturation. For example, V-0.5-100 is a column run with 0.5 mM Cr(VI) on the vacuum-based column system at a saturation level of 100%. In discussing transport experiments, "vacuum" and "UFA" will refer to the vacuum-based and centrifuge-based column experiments, respectively.

General results of the vacuum and UFA experiments will be discussed first, followed by the results of the analysis of the tritium data for all experiments (vacuum and UFA) as it pertains to the physical transport properties of the media. Lastly, the results of the analysis of Cr(VI) data for all experiments (vacuum and UFA) and both concentrations (0.5 and 1.0 mM) will be discussed in light of the chemical transport properties of the media.

Vacuum-Based Transport Experiments
Column saturation for all vacuum experiments ranged from 51 to 100%. Average solute residence time ranged from 0.89 to 5 h, depending on flow rate. The experiment duration ranged from 17 to 145 h.

A graph of the water content distribution in a vacuum column on completion of the experiment, done similarly to the method of Maraqa et al. (1997), is shown as Fig. 3. The data presented in this graph is for V-0.5-56 and is typical of vacuum column experiments in this study. The water content was calculated on a gravimetric basis since precise measurement of sample volumes was difficult when removing and segmenting the column materials. The vertical water content distribution varied little when compared with average water content values as determined from the difference in weight of the column in its dry and moist state (shown in Fig. 3 as a solid line). The largest difference in water content between any two column segments was 0.04 g g-1. Although some variation in water content along the length of the column existed during the experiment, average water content measurements were consistent, based on weighing the entire column before and after the experiments were conducted. Furthermore, despite somewhat nonuniform water content distributions, BTCs of solutes in unsaturated systems have been adequately predicted using average water content measurements (De Smedt and Wierenga, 1978; De Smedt et al., 1986).



View larger version (17K):
[in this window]
[in a new window]
  		Fig. 3. Gravimetric water content with column length for V-0.5-56 under -0.05 MPa of vacuum pressure. The solid line represents the assumed gravimetric water content based on the difference in weight of the entire column when dry and moist.

 
Centrifuge-Based Transport Experiments (UFA)
Inlet flow rates in the UFA columns varied little, as different water contents were obtained mainly by adjusting the centrifugation speed (Table 3). Two sets of columns, run simultaneously within the UFA, were treated as replicates and produced very similar BTCs (not shown). The column saturation ranged from 16 to 94%. Average solute residence time ranged from 0.06 to 0.40 h while experiment duration ranged from 3 to 12 h.

Physical Transport Parameters—Tritium Results
Tritium was used as a conservative tracer to evaluate the physical properties of the media. Selected tritium BTCs for vacuum and UFA column experiments at different saturation levels are shown as Fig. 4. The conventional CDE equation adequately described the tritium breakthrough curves, indicating that immobile water did not play a significant role in solute transport (CDE fit shown as solid line for solid circles and dotted line for open circles in Fig. 4). Some variation in tritium data was noticed when the effluent concentration was close to the influent concentration (C/C0 close to 1.0). However, the cause of this variation could not be determined. Estimates of experimental error for column data have been based on the mass balance of conservative tracers (Gamerdinger and Kaplan, 2000). Percentage recoveries averaged 100 ± 3% with a range of 94 to 107% (Table 3), indicating an experimental error of about 13%.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 4. Selected tritium breakthrough curves in (A) vacuum and (B) UFA columns. Solid lines correspond to fit with the convective–dispersive equation (CDE) for solid circles and dotted lines correspond to CDE fit for open circles.

 
Tritium BTCs were described with the CDE using average pore water velocity measurements calculated from experimental conditions to obtain values of dispersion (Table 4). Dispersion was plotted with average pore water velocity, yielding a significant (p < 0.001), but weakly correlated result (r2 = 0.4940; Fig. 5). A regression line forced through zero (De Smedt et al., 1986; Wierenga and van Genuchten, 1989; Maraqa et al., 1997) yielded a slope of 5.308, corresponding to the mean dispersivity in centimeters. This is comparable to reported values for unsaturated sand of 0.93 cm (Wierenga and van Genuchten, 1989), 1.354 cm (Maraqa et al., 1997), and 7.3 cm (De Smedt et al., 1986).


View this table:
[in this window]
[in a new window]
  		Table 4. Experimental and fitted parameters for tritium and Cr(VI) transport in vacuum and UFA column systems.{dagger}

 


View larger version (11K):
[in this window]
[in a new window]
  		Fig. 5. Hydrodynamic dispersion (D) with average pore water velocity (v) for all saturated and unsaturated vacuum and UFA experiments. The slope of the regression line gives the average dispersivity (5.308 cm).

 
Therefore, these data do not strongly support the assertion that the dispersion coefficient is linearly related to the average pore water velocity by a factor known as dispersivity. The relationship between D and v is stronger (p < 0.001; r2 = 0.5303) for the UFA data alone than the vacuum data (p = 0.068; r2 = 0.0618), though this may be due to the fact that the vacuum columns were run for a much smaller range in water content (0.23–0.43 cm3 cm-3) than the UFA columns (0.07–0.41 cm3 cm-3).

A probable explanation for the weakly linear plot of D vs. v is that the dispersivity is changing with water content. The dispersivity ({lambda}) was calculated from the dispersion coefficient ({lambda} = Dv-1) and plotted with water content to determine if there was a change in the transport pathways of the media with desaturation (Toride et al., 2003; Fig. 6A). The data showed a better relationship when dispersivity was expressed in the log scale (overall r2 = 0.3996 vs. r2 = 0.5897 log plot).



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 6. (A) Log of dispersivity and (B) Peclet number with volumetric water content for UFA and vacuum column tritium data. Regression lines encompass all data points.

 
When the data were separated by column method, the vacuum data had a weaker negative relationship of log of dispersivity with water content (p = 0.064; r2 = 0.3657) than the UFA data (p < 0.001; r2 = 0.5277). The regression line in Fig. 6A encompasses all data points from the vacuum and UFA experiments (p < 0.001; r2 = 0.5897). This plot indicates that the log of dispersivity increases as the media is desaturated. This is consistent with results from other research on packed soil columns that found dispersivity to increase with decreasing water content (De Smedt et al., 1986; Maraqa et al., 1997; Fesch et al., 1998; Padilla et al., 1999). De Smedt et al. (1986) attributed this increase in dispersivity with decreasing water content to the presence of immobile water, while Fesch et al. (1998) explained that the increase in dispersivity could be the result of the increase in the tortuosity of the solute flow path as the media is desaturated.

The variation in dispersion (D) as a result of changing water content can also be examined using the Peclet number (P) to account for differences in column length (L) and average pore water velocity (v):

[5]

The data were plotted as Peclet number vs. water content, but showed a better relationship when the Peclet number was expressed on the log scale (overall r2 = 0.4365 vs. r2 = 0.7127 log plot). The data are shown in Fig. 6B, separated by column method. Overall, the data indicated that the Peclet number decreases with desaturation (p < 0.001; r2 = 0.7127), corresponding to an increase in hydrodynamic dispersion as the media becomes desaturated since these two parameters are inversely related (Eq. [5]). The vacuum data alone had a weaker relationship (p = 0.064; r2 = 0.3657), than the UFA data (p < 0.001; r2 = 0.7034).

To account for the possibility of immobile water in these systems, the two-region physical nonequilibrium model was fit to all tritium data. In addition to dispersion (D), ß and {omega} parameters were included. Parameters obtained using both models were similar. All vacuum and UFA experiments had mobile water contents >90%. It was concluded that the nonequilibrium model was not needed to describe these data since immobile water did not play a significant role in the transport of solutes in this system. This would follow the criteria of Gamerdinger and Kaplan (2000), who only used parameters determined with the nonequilibrium model if the mobile water fraction, ß, was <90%.

Chemical Transport Parameters—Cr(VI) Results
A main objective of this research was to determine whether Cr(VI) sorption parameters changed with degree of water saturation. For the saturated Cr(VI) experiments, the pH of the column effluent changed from about 4.40 to 4.18 during breakthrough of the Cr(VI) tracer solution. The pH leveled off at 4.55 when the tracer solution was changed back to AGW. Hexavalent chromium sorption has been shown to increase with decreasing pH for iron oxyhydroxides and aluminum hydroxide minerals (Griffin et al., 1977; Davis and Leckie, 1980; Leckie et al., 1980; Rai et al., 1986) and soils (James and Bartlett, 1983; Rai et al., 1988; Selim and Amacher, 1988). Example Cr(VI) BTCs for vacuum and UFA columns at various levels of saturation are included in Fig. 7. Data were well described using the conventional CDE for Cr(VI) curves in most cases, although the CDE did not adequately describe the long tailing of the Cr(VI) desorption curve (CDE fit shown as solid line for solid circles and dotted line for open circles in Fig. 7). This was not improved by using a "two-region" immobile water model in fitting the BTCs. Sorption parameters are given in Table 4.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 7. Selected Cr(VI) breakthrough curves in vacuum and UFA columns. Solid lines correspond to CDE fit for solid circles and dotted lines correspond to convective–dispersive equation (CDE) fit for open circles.

 
Figure 8A shows the retardation coefficients as a function of water content for all saturated and unsaturated vacuum and centrifuge experiments and both concentrations of Cr(VI) solution. The distribution coefficient, Kd, determined from the batch sorption isotherm (0.684 mL g-1), was used to calculate the expected retardation coefficient for each experiment, using the relationship described by Eq. [3] and the average bulk density of all column experiments, 1.52 g cm-3. This relationship is shown as a solid line in Fig. 8A and labeled "Batch fit."



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 8. (A) Retardation factor and (B) apparent distribution coefficient with volumetric water content for UFA and vacuum column systems at 0.5 and 1.0 mM Cr(VI) concentration levels. The relationship R = 1 + [({rho}bKd){theta}-1] is graphed as a solid line in (A) using the batch isotherm Kd and the average column bulk density. The solid line in (B) represents the batch sorption isotherm Kd, and the dotted line represents the average of all column experiment Kd-app.

 
The R data from the vacuum column experiments leached with 1.0 mM Cr(VI) had no significant linear relationship with water content (p = 0.414; r2 = 0.2300), while the 0.5 mM level had a better relationship (p = 0.036; r2 = 0.8153). The UFA experiments showed a significant increase in R with decreasing water content for both 0.5 and 1.0 mM concentration levels (p < 0.001, r2 = 0.6889; p = 0.024, r2 = 0.5784, respectively). The general increasing trend of R with decreasing water content in these two systems agrees with the relationship described by Eq. [3] and represented as a solid line in Fig. 8A.

Retardation data from the vacuum and UFA experiments were not significantly different. Also, the 0.5 mM Cr(VI) experiments had generally higher retardation factors as opposed to the 1.0 mM Cr(VI) experiments, possibly indicating nonlinear sorption (Seaman et al., 1999), although there was no significant difference between the treatment levels and the isotherm did not indicate nonlinear behavior for this concentration range (Fig. 2).

Since data for both systems were similar and the UFA solute residence times were much shorter than the vacuum experiments at comparable water contents (Table 3), it would appear that residence time had little effect on transport parameters over this range in water content (0.07–0.43 cm3 cm-3) and solute residence time (0.06–5 h). This is also evidenced by the fact that stopping the centrifuge to collect samples had no apparent effect on Cr(VI) BTCs. This is probably due to the physically homogeneous media of packed sand in which the experiments took place and the fact that Cr(VI) sorption was not kinetically limited. Batch sorption studies done with similar soils found that most of the Cr(VI) sorbs during the first few hours of reaction (Amacher et al., 1988; Zhou and Chen, 2000). However, solutes that are kinetically limited (Gamerdinger et al., 2001a, 2001b) and display concentration perturbations when flow is interrupted (Murali and Aylmore, 1980; Reedy et al., 1996; Brusseau et al., 1997; Mayes et al., 2000) may not display similar sorption properties in vacuum and UFA columns (as seen here with chromate) and may not be appropriate for the UFA technique due to the necessity of flow interruption for collection of samples and short residence times.

To examine whether sorption was changing with water content, the Kd-app of each column experiment was calculated from the R value, using bulk density and water content measurements from each experiment (Eq. [3]; Tables 3 and 4). The results are shown in Fig. 8B. There is not a significant relationship between the Kd-app and water content for either vacuum or UFA data and either level of Cr(VI) concentration. Also, there was no significant difference between Kd-app values determined with the UFA and vacuum systems. Generally, the 0.5 mM data points are higher than those of the 1.0 mM data, again indicating a possible deviation from the linear sorption, although the treatment levels were not significantly different.

The average of all Cr(VI) column experiment Kd-app values (0.5 and 1.0 mM) was 0.633 mL g-1, represented in Fig. 8B as a dotted line, similar to the batch sorption isotherm Kd value of 0.684 mL g-1 (solid line). In fact, the 95% confidence interval for the batch sorption Kd (0.624– 0.744 mL g-1) includes the average of the column Kd-app (0.633 mL g-1). Therefore, it appears that for this media, under these conditions, with this solute, average sorption parameters in dynamic column studies were similar to those obtained with batch sorption tests. These results are different than those of Gamerdinger et al. (2001a), who found a change in sorption at very low water contents that was unexplained by velocity or kinetic effects.

Effluent recoveries of Cr(VI) in the columns ranged from 63.7 to 110.4% (Table 3), with low recoveries resulting from insufficient leaching with AGW to remove the sorbed fraction. Average effluent recovery was 81.6%. Therefore, soil from each column experiment was extracted with 10 mM KH2PO4. Average recovery after the extraction was 93.1%. There was no relationship between the experimental recovery and any transport parameters.


   SUMMARY AND CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
APPENDIX
 REFERENCES
 
The objective of these solute transport experiments was to evaluate the effect of water content on the dispersion and sorption properties of tritium and Cr(VI) in a sandy sediment characteristic of the Atlantic Coastal Plain. Two methods were employed to create a steady-state unsaturated flow regime within the packed soil columns: a vacuum-based Wierenga column system and a centrifuge-based UFA.

It was found that there was no significant difference in solute transport parameters derived with these two systems. However, the duration of the longest UFA experiment was shorter than that of the shortest vacuum experiment. Each UFA experiment could be finished in 1 d, though the system had to be stopped for sampling purposes at specified intervals. This represents an advantage of the UFA system over the vacuum system, as the experiment duration for the vacuum-based system was 4 to 23 times longer than those run with the centrifuge-based system at comparable water contents. Another difference between the UFA and vacuum system was the range of water content that could be evaluated. The UFA had a much greater achievable range in water content in this study (0.07–0.42 cm3 cm-3) and could get as low as 16% of saturation, while the range in the vacuum system was much smaller (0.23–0.43 cm3 cm-3), with a low of 51% saturation.

Dispersion and dispersivity increased with decreasing water content, though these relationships were not strongly linear in nature. Retardation, as expected, increased with decreasing water content; however, calculation of the apparent distribution coefficient, Kd-app, from the retardation factor indicated little trend with water content. The average of all Cr(VI) column experiment Kd-app values, 0.633 mL g-1, was similar to the batch sorption isotherm Kd value of 0.684 mL g-1, indicating that batch sorption experiments were similar to average sorption parameters in dynamic column studies.


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX
REFERENCES
 


   ACKNOWLEDGMENTS
 
This research was supported by the Environmental Remediation Sciences Division of the Office of Biological and Environmental Research, U.S. Department of Energy through the Financial Assistance Award no. DE-FC09-96SR18546 to the University of Georgia Research Foundation. The authors would like to acknowledge the thoughtful comments of Dr. Marianne Guerin and Julian Singer on an earlier version of the manuscript, the assistance of Dr. Machelle Wilson in conducting statistical analysis, and the laboratory assistance of J. Logan, J. McIntosh, A. Kelsey-Wall, and Dr. M. Uddin.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX
 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 Hutchison, J. M.
Right arrow Articles by Radcliffe, D. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Hutchison, J. M.
Right arrow Articles by Radcliffe, D. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Hutchison, J. M.
Right arrow Articles by Radcliffe, D. E.
Related Collections
Right arrow Laboratory Column Studies
Right arrow Dispersion
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