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


     

Published online 17 May 2007
Published in Vadose Zone J 6:363-372 (2007)
DOI: 10.2136/vzj2006.0048
© 2007 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 Similar articles in Web of Science
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 HighWire
Right arrow Citing Articles via Web of Science (3) Citing Articles via Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Seaman, J. C.
Right arrow Articles by Kaplan, D. I.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Seaman, J. C.
Right arrow Articles by Kaplan, D. I.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Seaman, J. C.
Right arrow Articles by Kaplan, D. I.
Related Collections
Right arrow Field-Scale Studies
Right arrow Colloid-Facilitated Transport
Right arrow Vadose Zone Processes and Chemical Transport

SPECIAL SECTION: SAVANNAH RIVER SITE

Spatial and Temporal Variability in Colloid Dispersion as a Function of Groundwater Injection Rate within Atlantic Coastal Plain Sediments

J. C. Seamana,*, P. M. Bertscha and D. I. Kaplanb

a Savannah River Ecology Lab., The Univ. of Georgia, Drawer E, Aiken, SC 29802
b Savannah River National Lab., Aiken, SC 29808
* Corresponding author (seaman{at}srel.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 24 March 2006.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
A subsurface injection experiment was conducted on the USDOE's Savannah River Site (SRS) to determine the influence of pump-and-treat remediation activities on the generation and transport of groundwater colloids. The impact of colloid generation on formation permeability at injection rates ranging from 19 to 132 L min–1 was monitored using a set of six sampling wells radially spaced at approximate distances of 2.0, 3.0, and 4.5 m from a central injection well. Each sampling well was further divided into three discrete sampling depths that were pumped continuously at a rate of ~0.1 L min–1 throughout the course of the injection experiment. Discrete samples were collected for turbidity and chemical analysis. Turbidity varied greatly between sampling wells and zones within a given well, ranging from <1 to 740 NTU. The two sampling wells closest to the injection well displayed the greatest response in terms of turbidity to increases in injection rate. Transient spikes in turbidity generally corresponded to incremental increases in the injection rate that were followed by a decrease in turbidity to a stable injection rate–dependent level. Mineralogical analysis of the resulting suspensions confirmed the presence of kaolinite, goethite, and to a much lesser degree, quartz and illite, with many of the particles too large (>1 µm) to be readily mobile within the formation. Turbidity measurements taken during this study indicate that colloid mobilization induced by water injection was both spatially and temporally heterogeneous. Furthermore, colloid release did not follow simple predictions based on shear force, presumably due to the complexities encountered in real heterogeneous systems. These findings have important implications to our understanding of how colloids and the co-contaminants are mobilized in the subsurface environment, as well as for the development of monitoring practices that minimize the creation of colloidal artifacts. Technical and logistical obstacles encountered in conducting such an extensive field experiment are also discussed.

Abbreviations: DO, dissolved oxygen • EDX, energy-dispersive X-ray • ICP–MS • inductively coupled plasma–mass spectrometry • ITS, injection test site • IW, injection well • SEM, scanning electron microscopy • SRS, Savannah River Site • TEM, transmission electron microscopy • XRD, X-ray diffraction.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
The mobilization and subsequent transport of soil and aquifer colloids have been an important area of research since the recognition that such processes are central to soil formation and may facilitate contaminant migration (Kretzschmar et al., 1993; McCarthy and Zachara, 1989; Ryan and Elimelech, 1996; Sprague et al., 2000). Several field and laboratory studies have focused on the physicochemical processes controlling colloid generation within both the vadose zone and saturated Atlantic Coastal Plain sediments underlying the USDOE's SRS (Bertsch and Seaman, 1999; Kaplan et al., 1993, 1994a, 1994b, 1995, 1996, 1997; Newman et al., 1993; Seaman, 2000; Seaman and Bertsch, 2000; Seaman et al., 1995a, 1997).

Until recently, turbid groundwater samples were filtered before chemical analysis to remove suspended materials generally thought to cause formation disturbance. Artifactual colloids may be introduced during well construction or development (e.g., drilling fluids, bentonite) (McCarthy and Wobber, 1993), may result from changes in chemistry or redox status due to inadequate sample preservation (Gschwend and Reynolds, 1987; Ledin et al., 1994; Ryan, 1988), or may become suspended from the immobile matrix by the shear forces associated with pumping (Backhus et al., 1993; Cusack et al., 1992; Ryan, 1988). However, growing concern that such colloidal material may act as a vector for contaminant migration has led to the adoption of pumping rates at or less than groundwater flow to minimize shear stress to the formation in an effort to avoid dispersing colloidal particles that are not inherently mobile under typical groundwater flow velocities (Backhus et al., 1993; Kearl et al., 1994; McCarthy and Degueldre, 1993; Puls and Barcelona, 1989; Ryan, 1988). Passive groundwater samplers have also been developed to eliminate the confounding impact of pumping in an effort to sample colloids that are inherently mobile in the aquifer (Magaritz et al., 1990; Ronen et al., 1992; Weisbrod et al., 1996).

Despite recognition that flow velocity is an important factor controlling colloid mobilization, field studies have generally focused on the impact of changes in solution chemistry under natural or marginally elevated hydraulic gradients. These studies have identified several important mechanisms of colloid mobilization: (i) clay dispersion due to changes in ground-water–pore-water pH, ionic strength, or Na/Ca ratios (Bunn et al., 2002; Czigany et al., 2005; Flury et al., 2002; Laegdsmand et al., 1999; Nightingale and Bianchi, 1977; Ryan and Gschwend, 1994; Seaman et al., 1995a); (ii) the manipulation of surface charge using chemical dispersing agents (Johnson et al., 2001; Seaman and Bertsch, 2000); (iii) the manipulation of surface charge or steric properties with natural organic matter (Kretzschmar et al., 1993); (iv) the dissolution of carbonate or Fe cementing agents resulting in the release and transport of silicate clays (Gschwend et al., 1990; Ronen et al., 1992; Ryan and Gschwend, 1990; Swartz and Gschwend, 1998); and (v) the in situ precipitation of colloidal particulates resulting from induced changes in groundwater chemistry (Gschwend and Reynolds, 1987). In many instances, more than one of these mechanisms may be operable at the same time.

Relatively few field studies have evaluated the impact of pore-water velocity on colloid generation, generally focusing on the impact of rapid infiltration on colloid generation in the unsaturated zone (El-Farhan et al., 2000; Kaplan et al., 1993, 1997; Ryan et al., 1998) or evaluating the impact of pumping rates on the mobilization of colloidal artifacts in groundwater sampling studies (Backhus et al., 1993; Kearl et al., 1994; Ryan, 1988). Under ideal laminar flow conditions, the hydrodynamic shear force acting on a spherical particle attached to a flat surface can be expressed as follows:

Formula 1[1]
where V is the flow velocity at the center of the particle, {eta} is the fluid viscosity, and r is the particle radius (Ryan and Elimelech, 1996; Ryan et al., 1998; Sharma et al., 1992). Hydrodynamic shear stress is opposed by the attractive forces (i.e., van der Waals, electrostatic) between the particle and the flat immobile surface. Such adhesive forces are subject to changes in solution chemistry; thus, colloid mobilization in response to a physical perturbation increases with increasing repulsive surface charge between the attached colloid and the stationary matrix (McDowell-Boyer, 1992; Ryan and Elimelech, 1996; Sharma et al., 1992), as previously demonstrated for variably saturated soil systems on the SRS by the lysimeter studies of Kaplan et al. (1997, 1993). The critical velocity for particle release decreases with increasing particle size, even though the required hydrodynamic force increases. This has been verified in idealized experimental systems for non-Brownian particles, that is, >1 µm in diameter (Sharma et al., 1992). As one might expect, the amount of particles mobilized should generally increase with increasing flow velocity, but greater flow velocities are required to mobilize smaller particles that extend to a lesser degree into the advective stream. However, many assumptions inherent in applying such a physical approach to soil and aquifer systems are questionable. In more realistic porous matrix systems, interrelated factors such as contact area, adhesion strength, and surface charge heterogeneity are important in countering shear forces (Sharma et al., 1992). In some cases, researchers have used extremely short column systems to differentiate between mechanisms responsible for particle mobilization and those controlling subsequent transport (Ryan and Gschwend, 1994).

Rainfall infiltration experiments have generally demonstrated the importance of transient flow events in controlling colloid mobilization within the vadose zone, with the highest colloid loads observed at the onset of flow, often with lesser colloid mobilization observed in subsequent infiltration or transient flow events (El-Farhan et al., 2000; Gao et al., 2004; Kaplan et al., 1993; Laegdsmand et al., 1999; Ryan et al., 1998; Schelde et al., 2002). Colloid generation in fractured and karstic aquifer systems also appears to be controlled by storm-driven recharge events that increase flow velocities and lower pore-water ionic strength (McCarthy and Shevenell, 1998), with larger colloids being mobilized due to the lack of straining and higher flow velocities compared with granular porous media (Degueldre et al., 1989). In the long term, however, significant colloid mobilization may continue once steady flow has been achieved (Levin et al., 2006). Using an intact soil core from a site displaying significant macropore flow, Levin et al. (2006) found the diffusion-limited supply of colloids to be more important than shear forces under relatively steady flow conditions.

In a series of lysimeter experiments, Kaplan et al. (1993) observed that the abundance of colloids mobilized in response to rainfall increased with increasing flow rate, while the net zeta potential, indicative of surface charge and colloidal stability, decreased with increasing flow rate, in qualitative agreement with hydrodynamic detachment theory. However, the average particle size also increased with flow rate, in opposition to the shear force theory. An increase in colloid concentration, and in some cases, average particle size, with increasing flow rate has also been observed in groundwater pumping studies (Backhus et al., 1993; Kearl et al., 1994; Puls, 1990).

During a series of rainfall simulations, Ryan et al. (1998) observed a poor correlation between particle concentration and the discharge rate into zero-tension lysimeters placed at various depths in the direct path of soil macropores (>1 mm in width). Particle size of the mobilized materials did not appear to be flow dependent, with similar size distributions observed throughout the study regardless of flow velocity. Jacobsen et al. (1997) also observed no correlation in mobile colloid levels with flow rate for intact soil cores. In a recent column study conducted under variably saturated conditions, Gao et al. (2004) observed that perturbations in flow were critical for colloid mobilization but that the impact of solution chemistry on colloid deposition depended on clay mineralogy, with kaolinite deposition being more sensitive to solution pH than illite. Such general discrepancies demonstrate the importance of factors such as surface heterogeneity and pore-water chemistry in overriding hydrodynamic shear forces in natural systems. These factors also limit our ability to predict particle deposition using idealized filtration theory (Huber et al., 2000).

The processes controlling colloid mobilization and deposition are directly relevant to groundwater remediation during the course of pump-and-treat activities. Colloid mobilization in the vadose zone above the plume, the formation of precipitates due to changes in groundwater chemistry (i.e., pH, dissolved oxygen [DO]), enhanced bacterial growth due to the introduction of nutrients, and the introduction or mobilization of indigenous colloidal materials can reduce formation permeability and eventually lead to costly well failure (Smith, 1995; Wiesner et al., 1996). Therefore, our objectives were (i) to evaluate the impact of injection rate on colloid generation as a function of depth within the formation and distance from the injection well and (ii) to evaluate the impact of colloid mobilization on water quality and formation permeability in the vicinity of the injection well. We used information gained during the initial test to plan and design subsequent long-term injection experiments to evaluate the impact of solution chemistry on colloid mobilization, as well as for several extensive groundwater tracer experiments (Seaman et al., 2007).


   Materials and Methods
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results and Discussion
 Conclusions
 REFERENCES
 
Well Installation
In the vicinity of the injection test site (ITS), the Floridan Aquifer System is divided into the Upper (water-table aquifer) and Lower Aquifer zones, separated by the Tan Clay confining layer (Wyatt and Harris, 2004). The water-table aquifer, the focus of the field injection study, consists predominantly of sands and clayey sand with interbeds of clay, sandy clay, and gravel (Strom and Kaback, 1992; WSRC, 1992). The ITS consisted of a 15-cm i.d. central injection well, designated IW (TD 19.1 m), screened over a 4.56-m interval starting at 13.5 m from the surface (Fig. 1). Well Obs-1, a previously existing well located within the study site and screened within the first confined aquifer underlying the water-table aquifer was used to monitor water depth to confirm the integrity of the confining layer. Composite drilling logs for the Upper Aquifer zone revealed that the transmissive region of the water-table aquifer consists primarily of sand, without any major confining or retarding layers (WSRC, 1992). Six additional sampling wells (7.6 cm i.d.) were constructed specifically for our study, using a technique similar to the procedure recommended by Ryan (1988) to minimize the impact of well installation on the physical and chemical properties of the aquifer adjacent to the well. The sampling wells (S1–S6) were radially spaced at approximate distances of 2.0 (S1 and S4), 3.0 (S2 and S5), and 4.5 m (S3 and S6) from the injection well. A plugged 10.8-cm hollow-stem auger was used to avoid the introduction of colloidal artifacts (i.e., drilling mud) or the creation of preferential flow paths between closely spaced wells or sampling zones within a given well (Fig. 2). After augering to the desired depth (~19 m), the well casing was inserted within the auger stem. The auger plug was displaced before withdrawing the stem from around the well casing, allowing the formation to collapse around the screened zone without the use of a filter pack.


Figure 1
View larger version (15K):
[in this window]
[in a new window]

  		FIG. 1. Relative location of the injection well (IW) and various sampling wells (S1–S6) at the injection test site (ITS).

 

Figure 2
View larger version (26K):
[in this window]
[in a new window]

  		FIG. 2. Well construction diagram for the injection well (IW), sampling wells (S1–S6 and Obs-1–Obs-3), and modified multidepth Waterloo sampling system.

 
Each well was screened (0.025-cm slot size) over a 6.1-m interval that spanned the screened interval of the IW, directly above the terminal well plug. A layer of fine sand was tremied above the water table, followed by the required bentonite seal at the surface. Following installation, the wells were minimally developed using bailers to remove solids suspended during the auguring process without significantly increasing the permeability of the formation adjacent to the wells.

Installation and Modification of the Multilevel Groundwater Sampling System
The screened interval for each monitoring well was further divided into three discrete sampling zones or depths using a modified Waterloo multilevel groundwater sampler (Solinst Canada, Ltd., Georgetown, ON). The screened sampling intervals were approximately 0.15 m in length and centered at 14.0, 16.6, and 18.6 m below the land surface (Fig. 2), with the deepest zone designated as Zone 1. The Waterloo system consists of a watertight 5.1-cm (2-inch) o.d. inner casing extending the full depth of the well, which encloses the double-valve pumps, pressure transducers, sampling lines, and drive-gas lines. Sections of the well casing were equipped with expandable packers to isolate sampling zones within the screened interval of the outer well casing, pto rovide greater data resolution with respect to groundwater chemistry as a function of depth, and to reduce the volume within the outer well casing to minimize the required purge time.

During the installation and testing of the Waterloo system, we encountered several unforeseen obstacles, the most significant being the failure of the double-valve pumps, which readily clogged when pumping turbid groundwater solutions. When efforts to unclog the pumps by pressure surging at the surface failed, we concluded that the pumps were inadequate for the long-term field experiments. The double-valve pumps were also susceptible to surging and often expelled drive gas through the sampling line following clogging of the recharge valve, making it difficult to monitor groundwater-quality parameters (e.g., turbidity, pH, specific conductance, DO) using inline meters at the surface. A single-valve pumping system also clogged under field testing when pumping turbid samples, and gear-driven positive displacement pumps tended to overheat and clog during long-term pumping when isolated within the confines of the Waterloo well casing. The failure of the pumps demonstrated the potential need to remove and repair the Waterloo system; however, the standard packers were designed to permanently expand when contacting water, negating the ability to remove and upgrade or repair the system once it was installed. Fortunately, the initial prototype was removed from the well before full packer expansion.

In response to such obstacles, Solinst Canada developed inflatable packers that could be expanded from the surface using compressed air and then released as necessary to remove the system for repairs. QED Environmental Systems, Inc. (Ann Arbor, MI), a supplier of 2.5-cm (1-inch) o.d. pneumatic bladder pumps that fit within the inner casing of the Waterloo system, was contacted, and initial field testing using a modified prototype confirmed the ability of the bladder pumps to sample turbid wells without clogging. QED modified the inlet, drive-gas, and sample fittings to accommodate inlet port and drive and sampling lines of the existing Waterloo system. The pneumatic bladder pumps use gas pressure to upwardly displace water held within the inner sleeve and then release the pressure to allow the bladder to refill from the bottom with fresh groundwater. Obviously, the pulsing inherent to gas-driven pumps, either double valve or bladder pumps, can induce additional shear stresses that may suspend nonmobile colloidal material. However, the limited space within the well casing, problems associated with cooling isolated electrical pumps, and the potential for mechanical failure and fouling during long-term continuous pumping (7–10 d) precluded the use of electric centrifugal pumps.

Injection Methods
The experiment consisted of a step injection test similar to those conducted to assess the ability of the formation to accept treated wastewater. The timetable for the experiment is summarized in Table 1. Water from a nonimpacted sampling well screened within the same aquifer, yet sufficient distance from the test site so as not to hydraulically impact the injection study, was used as the injectate without additional chemical manipulation. Before injection or pumping, baseline water depths were recorded for the IW, each of the six monitoring wells (S1–S6), Well Obs-1 screened within the underlying confined aquifer, and two additional monitoring wells in the vicinity of the test site. After recording water depths, we initiated pumping within all zones of the six monitoring wells to optimize sampling rates and establish background turbidity levels. During this initial period, pumping rates were frequently monitored and adjusted as necessary. Establishing the same pumping rate for all three zones within a given well was impossible using one flow control unit due to the differences in depth (hydraulic head) between the three zones. However, pressure and recharge rates for the control units were optimized to maintain a minimal flow rate (target pumping rate ~100 mL min –1) within the deepest sampling depth (Zone 1). Six flow-through water-quality meters (YSI, Inc., Yellow Springs, OH) were placed inline for all three zones within Wells S1 and S4 to monitor the pH, electrical conductivity, and DO content of the sampled groundwater throughout the duration of the experiment.


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

  		TABLE 1. Chronological summary of the injection experiment.

 
To minimize the impact force of water free falling down the bore hole, water was injected through a small-diameter pipe suspended within the IW and slotted within the screened interval. The well bore was open to the atmosphere to eliminate pressure buildup and enable us to evaluate mounding within the IW. Injection flow rate was controlled at the surface using a valve manifold with a calibrated flow meter.

During injection, turbidity was monitored and recorded at regular intervals for all zones within a given monitoring well. Before each increase in injection rate, discrete groundwater samples (~250 mL) were collected for each sampling zone and then immediately transported to the lab and stored at 4°C. A fraction of each sample was immediately filtered (0.2-µm pore-size polycarbonate) and acidified (1% HNO3) for preservation before metal analysis. The remainder was stored in an unacidified state for anion analysis by ion chromatography according to USEPA Method 300.0 (USEPA, 1993). Aliquots of the unfiltered samples were digested using two different methods: USEPA Method 3015 (USEPA, 1994a) (nitric acid–based) and the same method with the addition of hydrofluoric acid. Care was taken to ensure that turbid samples were well suspended when subsampled to maintain the correct colloid load within the remaining sample and the analyzed fractions. The filtered or digested samples were then analyzed for metals by inductively coupled plasma–mass spectrometry (ICP–MS; Elan 6000, PerkinElmer Corp., Norwalk, CT), following the quality assurance–quality control protocols outlined in USEPA Method 6020 (USEPA, 1994b). Statistical analysis was performed on the results using the JMP V3.1 statistical software package developed by the SAS Institute, Inc. (Cary, NC). Digested metal concentrations were correlated with sample turbidity levels and concentration levels for other metals as indicators of colloidal-phase suspension components.

Discrete Particle Analysis
The high polydispersivity and relatively unstable nature of the aggregates in suspension precluded sizing by photon correlation spectroscopy (Schurtenberger and Newman, 1993; Seaman et al., 2003). Therefore, select groundwater suspensions were deposited by filtration through 0.2-µm pore-size polycarbonate filters for subsequent analysis by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). After a given suspension was deposited on the filter to yield a well-dispersed coating, an additional 5- to 10-mL aliquot of dionized water was filtered through the sample to remove any readily soluble salts that may have accumulated with the particles. A 5-mm by 10-mm portion of the filter was secured with carbon tape to an SEM stub and coated with Au-Pd or carbon before imaging and X-ray microanalysis, respectively. Filters were observed with a Hitachi S 800 SEM equipped with a field-emission-gun electron source and a Tracor 5500 EDS (Tracor Northern, Inc., Middleton, WI) system for compositional analysis. Carbon-coated filters were systematically scanned, and energy-dispersive X-ray (EDX) analysis was performed randomly on approximately 20 to 25 particles or small aggregates of particles per filter. Selected micrograph images were taken of representative particles on the carbon-coated filters, but the resolution was inadequate to accurately assess primary particle morphology or the degree of aggregation. After EDX analysis, sections of the same filters were Au-Pd coated for better visual resolution, and numerous photomicrographs were collected of representative regions of the colloid-coated filters. Synthetic and reference mineral analogs were also analyzed for comparison with the natural colloids.

Concentration for X-Ray Diffraction and Thermal Analysis
Turbid samples were composited to provide sufficient material for X-ray diffraction (XRD) and thermal analysis. Composite samples were vacuum-filtered onto 0.2-µm pore-size polycarbonate filters and mounted for XRD analysis using the Drever method (Drever, 1973). The mineralogy of the suspensions was qualitatively determined by standard XRD techniques (Brindley and Brown, 1980) using a Scintag X2 Advanced Diffraction System (ThermoARL, formerly Scintag, Inc., Cupertino, CA).


   Results and Discussion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
Conclusions
 REFERENCES
 
Two months after well construction and before the installation of the multilevel sampling system, baseline water-quality data were recorded in the field, and groundwater samples were collected for comparison with the injectate chemistry (Table 2). The high pH noted in Well Obs-1 was attributed to the errant placement of grouting material close to the screen interval. Uncharacteristically elevated pH values were never observed for groundwater samples collected from the six monitoring wells installed specifically for the current study, suggesting that the well installation artifact evident in Obs-1 had not impacted the injection zone. With the exception of Well Obs-1, the concentrations of major dissolved constituents were consistent with reported values for existing wells in the vicinity (Strom and Kaback, 1992). Suspended material (<150 µm fraction) collected from the monitoring well casings during background sampling consisted largely of quartz, kaolinite, goethite, and illite as identified by XRD, again consistent with previous mineralogical characterization of the vadose zone and aquifer sediments typical of the study site (Seaman et al., 1995b, 1997; Strom and Kaback, 1992).


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

  		TABLE 2. Baseline water-quality data collected approximately 2 months after installation of the sampling wells, S1–S6.

 
Our study was designed to evaluate the impact of elevated flow velocities on colloid mobilization and groundwater quality within an injection well system under ambient pore solution chemistries. However, previous column experiments using coarse-textured sediments from the SRS that are typical of the vadose zone and water-table aquifer at the ITS have demonstrated that colloid mobilization is extremely sensitive to pore solution composition, with colloid release occurring under two drastically different chemical regimes (Bertsch and Seaman, 1999; Seaman et al., 1995a, 1997). The introduction of dilute solutions containing polyvalent cations, Ca2+ and Mg2+, resulted in a downward shift in solution pH and the preferential release of positively charged iron oxides from the highly weathered Coastal Plain sediments, the bulk clay fraction of which consists largely of kaolinite, goethite, and illite. The presence of polyvalent counter anions such as SO42– was found to inhibit both colloid release and the downward shift in pH that was attributed to Al exchange reactions and subsequent hydrolysis. Colloids mobilized from the low pH mechanism failed to cause any discernable clogging and appeared to be generally stable with prolonged storage. Significant colloid mobilization that resulted in a reduction in column hydraulic conductivity was also observed for alkaline solutions dominated by Na+. Colloidal suspensions generated under the high pH conditions tended to more accurately reflect all components of the bulk clay fraction, in contrast to the selective mobilization of iron oxides observed under low pH conditions. As might be expected for a system displaying amphoteric surface charge, minimal colloid generation occurred when the pH (~5.0–5.5) and solution composition were similar to the low ionic strength groundwater typical of the water-table aquifer on the SRS. Additionally, the duration of our study was insufficient to alter the groundwater chemistry in all but the closest monitoring wells, as confirmed by subsequent groundwater tracer studies (Seaman et al., 2007). Therefore, colloid mobilization here likely resulted from a combination of formation disturbance during well installation and the altered hydraulic gradient associated with injection and groundwater sampling.

During the injection study, no detectable change in groundwater depth over the course of injection was observed for Well Obs-1, despite the close proximity to the IW, confirming the integrity of the confining layer in restricting injection to the water-table aquifer. An immediate increase in mounding was observed within the IW and the sampling (S1–S6) and observation wells (Obs-2 and Obs-3) screened within the water-table aquifer for each increase in injection rate. Although it was not the intent of the study, the short duration of the injection test precluded an evaluation of changes in aquifer hydraulic conductivity over the course of injection.

Groundwater flow velocities as a function of distance from the central IW were estimated using the following approach. Given the length of the screened interval and diameter of the injection well, the average linear velocity, u (cm s–1), was estimated by

Formula 2[2]
where Q is the injection rate (cm3 s–1), n is the well screen or formation porosity, r is the radius from the center of the injection well, and h is the length of the screened interval or the depth of the transmissive zone. For the injection well (r = 7.62 cm), with a 4.57-m screened interval, the average linear velocity for an injection rate of 19 L min–1 would be 2.2 cm min–1 at the well casing. When calculating flow velocities within the formation, the transmissive depth was assumed to extend from the base of the injection well screen to the water-table depth at a given sampling well, assuming an average formation porosity of 0.392 (Looney et al., 1987) (Table 3). Based on the natural hydraulic gradient in the vicinity of the test site and a hydraulic conductivity of 7.6 m d–1, the natural flow rate at the test site was estimated at 15.7 ± 6.0 cm d–1. Even at the lowest injection rate, the resulting flow velocities within the study site were at least twice that of the natural groundwater flow.


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

  		TABLE 3. Estimated groundwater flow velocity due to injection at each sampling well.

 
During the course of injection, turbidity varied greatly between wells and within different zones of a given sampling well, with data for S3 and S6, the two wells spaced the greatest distance from the IW, given as examples (Fig. 3). In general, higher turbidity values were observed with depth for a given sampling well, but no clear trends in maximum turbidity levels or cumulative colloid mobilization were observed with respect to distance from the IW, which covaried with groundwater velocity.


Figure 3
View larger version (24K):
[in this window]
[in a new window]

  		FIG. 3. Sample turbidity at various depths within (A) Well S3 and (B) Well S6 during groundwater injection ranging from 19 to 132 L min–1 (Zone 1, deepest; Zone 3, shallowest; the location of wells is presented in Fig. 1).

 
The shear stress, G (s–1), induced by injection can be estimated using the approach outlined by Ryan (1988):

Formula 3[3]
where {varepsilon} is the energy dissipation rate (cm–2 s–1), and vf is the kinematic viscosity of fluid (cm2 s–1). For flow through porous media, the energy dissipation rate can be equated to the energy of head loss as expressed:

Formula 4[4]
where g is the acceleration due to gravity (980 cm s–2), u is the average linear groundwater velocity (cm s–1), and dh/dl is the hydraulic gradient (Table 4). The shear forces encountered at the closest sampling wells, even at the maximum injection rate, 132 L min–1, were relatively weak compared with the values estimated by Ryan (1988) when sampling groundwater at various conventional pumping rates. Apparently, the high mobile colloid loads we observed resulted from a combination of the shear forces associated with subsurface injection and groundwater sampling and formation disturbance caused by monitoring well installation, as generally less colloid mobilization was observed in subsequent injection experiments. Turbidity values appeared to decrease with continued pumping for sampling zones that displayed high initial colloid loadings, as illustrated for Zone 1 in Fig. 3B, possibly due to the depletion of readily mobilized fines in the vicinity of the sampling port. Depletion of fines near the sampling wells may bias subsequent results, including the mobilization of colloids with each increase in injection rate.


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

  		TABLE 4. Shear rate estimates as a function of injection rate compared with estimated values for various groundwater pumping velocities reported by Ryan (1988).

 
High variability in mobile colloid loads as a function of depth within a well has also been observed for passive groundwater sampling systems (Ronen et al., 1992; Weisbrod et al., 1996). In our study, however, it is unclear if this variability reflects differences in formation texture (e.g., clay content) with depth or particle settling within the radius of influence for each sampling well, considering that a dominant portion of the collected suspensions were not inherently stable and flocculated with even minor storage. Generally lower pumping rates were maintained for the deeper sampling zones within each well, further confounding the interpretation of the turbidity results. Maximum turbidity values observed for each of the sampling wells as a function of injection rate are provided in Table 5. Again, maximum turbidity values did not change in a consistent manner with distance from the IW or with depth. In fact, a great deal of variability can be seen in these maximum values. This may in part be due to our selection of a 15-cm effective screened zone, compared with the more conventional wells with a screened zone of 500 to 1000 cm, since smaller screened zones may have result in less sample averaging.


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

  		TABLE 5. Maximum turbidity (NTU) levels and estimated colloid concentration (mg L–1) observed for each monitoring well during the injection experiment.

 
Despite the variability, transient spikes in turbidity generally corresponded to incremental increases in the injection rate for Wells S1 and S4, located approximately 1.8 and 2.0 m, respectively, from the IW, which were followed by a decrease in turbidity to a stable injection rate–dependent level (Fig. 4A). To evaluate the impact of turbidity on metal loadings, we collected 72 groundwater samples with turbidity values ranging from <1.0 to 740 NTU at the elevated flow rates. As observed during preliminary sampling, the dispersed material consisted largely of kaolinite and goethite, with varying amounts of illite and quartz. However, the wide size distribution and unstable nature of the turbid samples precluded size characterization by dynamic light scattering or the evaluation of zeta potential by laser dopler velocimetry (Schurtenberger and Newman, 1993; Seaman et al., 2003). In subsequent electron microscopy analysis, numerous particles with diameters >1 µm that would not be inherently mobile within the aquifer formation were observed.


Figure 4
View larger version (22K):
[in this window]
[in a new window]

  		FIG. 4. (A) Sample turbidity (NTU) at three depths within Well S1, and (B) groundwater mounding in injection well, IW, as a function of injection rate.

 
Correlation Coefficients for Groundwater Samples
Correlation coefficients for filtered (0.22-µm pore size) and digested groundwater samples are reported in Table 6. Poor correlations with turbidity, the highest being 0.269 for Tl, were observed for each analyte in the filtered samples, including Na, suggesting that colloid mobilization was not the result of dispersion due to Na levels. Of the 18 metals considered, 9 were highly correlated (r ≥ 0.8) with sample turbidity for nitric acid–digested samples, with 6 having correlation coefficients (r) greater than 0.9 (r2 = 0.81). Not surprisingly, the elements Ba, Be, Cs, Fe, K, Pb, Rb, Sr, and U generally consist of trace and relatively insoluble metals. In contrast, common soluble cations such as Na, Ca, and Mg displayed poor correlations with sample turbidity for filtered samples and samples digested using both procedures. The high correlations with turbidity observed for K and Cs most likely reflect the influence of illitic type phyllosilicate clay minerals, which have been reported within the vadose zone and aquifer sediments (Seaman et al., 1995b, 1997; Strom and Kaback, 1992). However, the poor correlation observed for Al with turbidity for nitric digests illustrates the ineffectiveness of nitric acid in solubilizing phyllosilicate clay minerals. The high correlation observed between Al and turbidity for triple acid digestions of the same samples using hydrofluoric acid confirmed that Al was largely associated with silicate clays that were not completely dissolved by USEPA Method 3015 (USEPA, 1994a).


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

  		TABLE 6. Correlation coefficients for various handling and digestion protocols for turbid groundwater samples.

 

   Conclusions
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
REFERENCES
 
Minimizing experimental artifacts and interpreting results can be quite difficult in the case of field-scale experiments. Colloid mobilization during subsurface injection and groundwater sampling is heavily influenced by local hydrogeologic conditions that cannot be explicitly controlled, varied, or even quantified at the field scale. For example, our study was designed to evaluate the impact of elevated flow velocities on colloid mobilization and groundwater quality within an injection well system under ambient pore solution chemistries, which requires the installation of monitoring wells and the collection of groundwater samples. Six monitoring wells were installed in a manner designed to minimize both the introduction (bentonite drilling mud) and the removal of colloidal fines during well construction. In preliminary tests, only bladder pumps performed reliably under the extended pumping duration required for our study. Despite such precautions, colloid concentrations varied greatly between wells and within different sampling zones of the same well. Shear force estimates resulting from the elevated hydraulic gradient due to injection were quite moderate compared with the forces induced by conventional groundwater sampling procedures.

In general, colloid levels were greater with depth within the aquifer, but it remains unclear if this was a consequence of differences in formation texture or an artifact associated with well installation. Turbidity results for Wells S1 and S4 indicate that subsurface injection alone may, in part, increase colloid mobilization at relatively close distances to the IW, but the field-scale mobility of the resulting suspension is likely to be quite limited. Temporal and spatial variability of colloids did not change in a consistent trend with distance from the IW or with depth. Quite possibly, a large degree of variability may be attributed in part to the small effective screened zone (15 cm) associated with the Waterloo multilevel sampling system. Most conventional monitoring wells have screened zones of 1000 cm, which can result in spatial mixing or averaging of turbidity results. Another important result from our study is that at this finer scale, colloid release did not follow simple predictions based on shear force. Even so, our study illustrates the difficulty in discriminating between colloid mobilization resulting from subsurface injection and that caused by a combination of formation disturbance during well installation, elevated shear forces associated groundwater sampling, and the surging inherent to bladder pumps. Together, these findings have important implications to our understanding of how colloids and the co-contaminants move in the subsurface environment and for the development of monitoring practices that minimize the creation of colloidal artifacts.


   ACKNOWLEDGMENTS
 
This research was funded by Financial Assistance Award Number DE-FC09-96SR18546 from the USDOE to the University of Georgia–Research Foundation. The authors would like to acknowledge the assistance of R. Masion, J. McIntosh, J. Noonkester, T. Rea, W. Simmons, and especially Dr. R. Strom in planning and conducting this study. The authors would like to thank A. Kelsey-Wall and J. Derrick for their assistance in manuscript preparation.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
J. C. Seaman, B. B. Looney, and M. K. Harris
Research in Support of Remediation Activities at the Savannah River Site
Vadose Zone J., May 17, 2007; 6(2): 316 - 326.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
J. C. Seaman, P. M. Bertsch, M. Wilson, J. Singer, F. Majs, and S. A. Aburime
Tracer Migration in a Radially Divergent Flow Field: Longitudinal Dispersivity and Anionic Tracer Retardation
Vadose Zone J., May 17, 2007; 6(2): 373 - 386.
[Abstract] [Full Text] [PDF]

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 Similar articles in Web of Science
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 HighWire
Right arrow Citing Articles via Web of Science (3) Citing Articles via Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Seaman, J. C.
Right arrow Articles by Kaplan, D. I.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Seaman, J. C.
Right arrow Articles by Kaplan, D. I.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Seaman, J. C.
Right arrow Articles by Kaplan, D. I.
Related Collections
Right arrow Field-Scale Studies
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