Influence of Ionic Strength and Cation Charge on Transport of

Yuniati Zevi, Annette Dathe, John F. McCarthy, Brian K. Richards, and Tammo S. Steenhuis. Environmental Science & Technology 2005 39 (18), 7055-7064...
0 downloads 0 Views 276KB Size
Environ. Sci. Technol. 2002, 36, 3735-3743

Influence of Ionic Strength and Cation Charge on Transport of Colloidal Particles in Fractured Shale Saprolite J O H N F . M C C A R T H Y , * ,† LARRY D. MCKAY,‡ AND DEIRDRE DIANA BRUNER‡ Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, Tennessee 37996-1610, and Department of Geological Sciences, University of Tennessee, Knoxville, Tennessee 37996-1410

The role of solution chemistry (cation charge and concentration) and particle size on colloid transport was examined in an intact monolith of fractured shale saprolite (highly weathered rock). Recovery of the microsphere tracers consistently decreased with increasing ionic strength of either mono- (Na+) or divalent- (Ca2+) dominated solutions, but a much greater concentration of Na+ in the influent solution was required to result in a similar reduction in colloid recovery as compared to Ca2+. However, composition of the solution along the flow path, and hence the degree of microsphere retention, was also strongly influenced by cation exchange and diffusive exchange between pore water in the fractures and in the finegrained, Ca- and Mg-rich matrix. The influence of “matrix diffusion” on solute transport is also evident in the 5-fold difference between the arrival of the center-ofmass of microspheres as compared to the much later arrival of a bromide tracer. Particle size affected the extent of microsphere transport, but the solution chemistry appears to be a more dominant control. While confirming the importance of ionic strength, counterion charge, and particle size on colloid migration, this study emphasizes the profound effect that pore structure and geochemical processes such as cation exchange have on solution chemistry and thus on colloid transport.

dominated soils and that loss or retention rates can widely vary (1-7). These studies suggest that colloidal contaminants, such as pathogenic microorganisms or radionuclides attached to colloidal particles (8), may also be mobile in fractured clay-rich deposits and could adversely impact underlying aquifers or nearby streams. Column transport studies in intact shale saprolite monoliths have examined the role of flow rate and particle size on colloid transport. Increased flow rate resulted in faster transport and lower retention of bacteriophage tracers (6, 9), likely due to reduced interactions with fracture walls or less diffusion into the fine pores. These results are consistent with previous findings of a strong influence of flow rate on colloid transport in macropore-dominated clay-rich soils (2, 4). Cumbie and McKay (7) examined the influence of particle size in saprolite monoliths. They observed greater recovery of 0.5- and 1.0-µm particles than for the other sizes and hypothesized that gravitational settling and straining contributed to losses of particles larger than the optimum size, whereas smaller particles experienced greater retention either because of diffusion into stagnant flow regions or because of more frequent collisions with fracture walls. These mechanisms are consistent with conceptual models developed for colloid transport in granular material (10-12) and transport in fractured rock (13, 14). This study examined the combined effect of solution chemistry and particle size on colloid transport in fractured fine-grained materials. The importance of counterion concentration and charge in controlling colloid transport has been emphasized in model systems and in small columns of repacked natural granular materials that permitted the solution chemistry to be controlled (15, 16). Our study extends this understanding to larger intact monoliths of fractured geological material and illustrates the dominant role that the structure and geochemistry of the geological material can have on colloid transport. The specific objectives of this study are to experimentally determine the relative importance of aqueous chemistry (ionic strength and charge of the dominant cation in solution) and particle diameter on colloid transport. The experiments were carried out using an intact monolith of a fractured clayrich shale saprolite, and the influences of the physical and chemical characteristics of this highly structured material are also a key part of the study.

Experimental Section Introduction Colloids can be highly mobile over significant distances in fractured soils and subsoils derived from clay-rich materials including saprolite (decomposed materials that retain the fabric of the parent bedrock), glacial till, and other sedimentary deposits. In a field tracer experiment in fractured shale saprolite at the Oak Ridge Reservation (ORR; Oak Ridge, TN), colloidal tracers (bacteriophage, bacteria, and latex microspheres) were transported in groundwater under natural flow conditions at rates of up to 200 m/d and over distances of up to 35 m (1). Other laboratory- or small-scale field tracer experiments also show that colloids can be transported rapidly in clay-rich fractured or macropore* Corresponding author telephone: (865)974-8039; fax: (865)9748086; e-mail: [email protected]. † Department of Ecology and Evolutionary Biology. ‡ Department of Geological Sciences. 10.1021/es025522o CCC: $22.00 Published on Web 07/24/2002

 2002 American Chemical Society

Geologic Setting for Saprolite Monolith. The colloid transport experiments were conducted in an undisturbed saprolite monolith excavated from a depth of 1.2 m at Solid Waste Storage Area 7 (SWSA-7) on the ORR. The material is a highly weathered and fractured shale saprolite formed in situ by extensive leaching of the parent material, which consists of shale and siltstone from the Dismal Gap formation of the middle to upper Cambrian Conasauga Group (17). The saprolite retains the bedding, fractures, and structure of the parent bedrock, which was intensely deformed by regional tectonic activity and can also contain root holes and other fractures or macropores related to weathering, desiccation, or freezing. Where exposed, the formation has weathered to a saprolite varying in thickness from 20 PV of 5 mM NaCl through the column. Prior to each microsphere injection experiment, the column was equilibrated by flushing with approximately 10 PV of the new influent solution of the desired ionic strength and composition. The 1000-mL pulse of tracer solution containing the four sizes of microspheres suspended in the same influent solution was then injected into the column, followed by an extended flush of the same solution except containing no microspheres, all at a flow rate of 2 mL/min. The concentration history of colloids recovered in the effluent was determined, and the percent of colloids recovered during each of the tracer experiments was calculated from the area under the breakthrough curve (BTC). During and at the end of each tracer experiment, effluent samples were analyzed for major ion concentrations (Na+, K+, Ca2+, and Mg2+ were the dominant cations). The first set of four microsphere tracer experiments was carried out using a monovalent cation influent solution, NaCl. The cation concentration of the influent solutions was increased in successive experiments in the order of 5, 7, 15, and 30 mM NaCl (experiments 1-4, respectively). A total of 20 PV of 0.1 mM CaCl2 was then eluted through the column prior to initiating experiments 5-8, which used divalent cation influent solution, Ca2+, at 0.1, 0.3, 0.4, and 1.0 mM CaCl2, respectively. In experiment 9, the influent solution was abruptly changed from 1.0 to 0.1 mM CaCl2 to determine if decreasing the ionic strength could cause release of microspheres retained during the previous eight experiments. Experiment 10, a repeated injection of microspheres at 0.1 mM CaCl2, followed this. Experiment 10 was identical to experiment 5 and was intended to determine the reproducibility of the data and potential ripening or blocking effects (29) that might alter colloid attachment because of the deposition of colloids in the previous nine experiments.

Results Characteristics of Column and Microspheres. The hydraulic conductivity of the saprolite column prior to any tracer injections was 2.2 × 10-6 m/s, which is near the low end of the range of values (2.0 × 10-7-2.4 × 10-4 m/s) measured previously in undisturbed monoliths from the same site (6, 22-24). Measurements were repeated after every tracer experiment, and over the 10-month period of the experiments the hydraulic conductivity values declined slowly from an initial value of 2.2 × 10-6 m/s to a final value of 7.4 × 10-7 m/s. This small decline suggests that clogging of the fractures or tubing during the experiments was relatively minor. The mean porosity of the 15 samples collected from the excavation was 0.49 with a variance of 0.05, which is similar

TABLE 2. Results of Colloid Tracer Experiments effluent (mM) influent exp

divalent (Ca + Mg)

colloid diameter (µm)

peak C/C0

% recovery

1

5 mM Na1+

4.107

0.36

0.1 0.5 1.0 2.1

0.021 0.106 0.077 0.045

1.34 11.7 7.19 4.51

2

7 mM Na1+

6.21

0.38

0.1 0.5 1.0 2.1

0.009 0.124 0.101 0.055

0.790 13.8 8.82 5.05

3

15 mM Na1+

13.14

0.69

0.1 0.5 1.0 2.1

0.004 0.026 0.017 0.005

0.258 2.88 1.38 0.673

4

30 mM Na1+

25.13

1.22

0.1 0.5 1.0 2.1

0.001 0.005 0.003 0.002

0.079 0.633 0.369 0.238

5

0.1 mM Ca2+

0.15

0.06

0.1 0.5 1.0 2.1

0.133 0.365 0.273 0.18

6

0.3 mM Ca2+

0.12

0.25

0.1 0.5 1.0 2.1

0.004 0.039 0.013 0.027

0.458 3.42 1.25 1.50

7

0.4 mM Ca2+

0.05

0.34

0.1 0.5 1.0 2.1

0.002 0.01 0.003 0.003

0.207 0.875 0.342 0.315

8

1.0 mM Ca2+

0.04

1.1

0.1 0.5 1.0 2.1

4.7E-05 1.9E-04 5.8E-05 1. 8E-04

0.005 0.023 0.008 0.016

9

nonequilibrium

0.1a 0.5a 1.0a 2.1a

1.1E-07 2.4E-07 1.5E-07 1.6E-07

5.0E-04 1.3E-03 8.0E-04 9.0E-04

0.1 0.5 1.0 2.1

0.054 0.346 0.184 0.158

10

a

colloid

monovalent (Na + K)

0.1 mM Ca2+

0.01

0.09

12.6 29.2 22.8 14.7

5.45 36.3 16.0 12.5

Refers to plateau C/C0 and total percent recovery of previously deposited colloids following the change in ionic strength for 1.0-0.1 mM Ca2+.

to those measured in other columns from the same site by Cumbie (24) and Cropper (22). The calculated pore volume for the column was 3900 mL. The dominant macropores in the sample were fractures, with relatively few root holes, as determined from both hand examination of specimens and microscopic examination of epoxy-impregnated thin sections. The aperture, or width of opening between the fracture walls, was estimated based on a mathematical solution for the resistance of flow between two parallel surfaces (30, 31). This parameter, with a value of 30 µm, is typically referred to as the “hydraulic aperture” and was calculated based on the measured hydraulic conductivity at the start of the experiments (2.2 × 10-6 m/s) and the average fracture spacing measured on the face of the excavation (0.025 m). The fracture porosity, which is calculated based on the hydraulic aperture and the fracture spacing, was very low (approximately 0.003), indicating that the total porosity is dominated by pores in the fine-grained matrix. These results are consistent with previous measurements of hydraulic aperture and fracture porosity at the site (6, 22-24) as well as measurements of macropore drainage (22).

The pH of the influent and effluent solutions remained fairly constant, ranging from 4.4 to 5.0. The electrophoretic mobility (EM) of the three largest sizes of microspheres was determined at the approximately ambient pH of 5.0 to evaluate differences in the surface potential between the colloidal tracers (Table 1). The 0.1-µm colloids were too small for reliable measurements of EM. The colloids had the expected negative charge, with the EM values being more negative with increasing microsphere diameter. Ionic Composition of the Influent and Effluent Solutions. The concentrations of cations in the effluent (Table 2) were constant throughout each microsphere injection but differed from the influent concentrations. In general, NaCl influent solutions became depleted in Na+ and enriched in Ca2+, Mg2+, and K+, likely because of cation exchange with the saprolite and diffusion from immobile pore water in the matrix. For the ease of presentation, discussion of the experiments will refer to influent solutions unless otherwise specified. Bromide Tracer Experiments. The two bromide tracer experiments, conducted 3 months apart, yielded similar BTC (Figure 1). Using temporal moments analysis, the mean arrival VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3737

FIGURE 1. Breakthrough curves for the first and second bromide experiments (lines) and microspheres (symbols) for the second bromide experiment. The vertical lines indicate the arrival times of the center-of-mass of the two bromide experiments as well as the colloid tracers. times of the center-of-mass of the two bromide tracer experiments were calculated to both be approximately 1 PV as compared to the mean arrival time of the center-of-mass of the microspheres at approximately 0.2 PV. There was also extensive tailing of the descending limb of the bromide BTC. Experiments 1-8: Microsphere BTC and Percent Recovery. BTC for the first eight microsphere tracer injections are shown in Figure 2. For each injection there was rapid initial arrival of all sizes of microspheres occurring within the first 100 mL of the tracer injection. This initial arrival corresponds to approximately 0.025 PV, and the center-ofmass of the colloids arrived between 0.15 and 0.25 PV. The rapid transport demonstrates that flow occurred mainly through fractures or macropores. Concentrations increased until the end of the pulse injection and then declined during the subsequent flushing portion of each experiment. Very low concentrations of microspheres from the previous experiments were often observed in effluent prior to beginning a new experiment. This persistent low concentration of mobile colloids generally amounted to less than 10-4 of the injection concentration and was subtracted as a background correction for each experiment. For all of the tracer experiments, the 0.5-µm microspheres were preferentially transported in terms of having both the highest maximum C/C0 (effluent/influent concentration) and the greatest total percent recovery of colloids (area under the BTC). The preferential recoveries of 0.5-µm microspheres were followed by the other colloid sizes, usually in the order 1.0-, 2.1-, and 0.1-µm diameter (Table 2; Figure 3). For both the Na+ and Ca2+ injections, the percent recovery decreased as the ionic strength increased (Figure 3), with total recovery ranging from 0.005 to 29.2%. However, for the Na+ influent solution, recovery of similar levels of microspheres occurred at cation concentrations more than 1 order of magnitude higher than for Ca2+. 3738

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002

Experiment 9: Step Decrease in Ionic Strength. Experiment 9 involved an abrupt decrease in the ionic strength of the influent solution (from 1.0 to 0.1 mM CaCl2) without the injection of any colloidal tracers for a 5-week (80 PV) equilibration period. The release of each size of microsphere over time is presented in Figure 4a as C/Cr, which is the ratio of C (the number of microspheres in the effluent at a given time) and Cr (the total number of microspheres retained in the column) (calculated as the difference between the number recovered in the effluents from experiments 1-8 and the total number injected). Concentrations of microspheres in the effluent increased slowly after the drop in ionic strength and continued to increase for about 5-10 PV until a nearly steady-state concentration was achieved that was about 3-5-fold higher than prior to the ionic strength adjustment (Figure 4a). However, the total number of microspheres released from the column was equivalent to only 0.0035% of the total number of microspheres calculated by mass balance to have been retained after experiments 1-8. Although the ionic composition of the influent solution changed abruptly, the effluent concentration (Figure 4b) exhibited extensive tailing, requiring more than 5 PV before the effluent concentration reached a new equilibrium level, just below that of the influent concentration. Turbidity was monitored in the effluent from experiment 9 to determine if the abrupt decline in ionic strength caused release of natural colloids along with the microspheres. No turbidity was visible in the samples, and turbidity readings (32) remained low and relatively constant at 0.006-0.26 NTU over the period of 15 min to 41 d after the step decrease in ionic strength. This confirms that the release of autochthonous colloids was minimal. Experiment 10: Repeat Experiment 5. Experiment 10 was a second microsphere injection at 0.1 mM CaCl2 to verify

FIGURE 2. Breakthrough curves of microspheres for (a) experiments 1-4 (Na1+) and (b) experiments 5-8 (Ca2+). The results of experiment 8 are same as those in Figure 1 but are shown here to facilitate comparison with the results of the other experiments. VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3739

FIGURE 3. Recovery of different sized microspheres is plotted as a function of the concentration of Na1+ (dotted lines) or Ca2+ (solid lines) in the influent tracer solution. The open symbols represent the percent recovery of microspheres in Experiment 10.

FIGURE 5. Recovery of microspheres is plotted as a function of the concentration of divalent cations (sum of Ca2+ and Mg2+) in the column effluent. The open symbols represent the percent recovery of microspheres in experiment 10. ment of microspheres. The cation concentration, peak concentration, percent recovery of microspheres, and trends among the different size colloids in experiments 5 and 10 (Table 2 and Figure 5) are very similar and within the expected range of variation of these types of tracer experiments. This suggests that the short pulses of microspheres had a relatively small effect on attachment of colloids and that multiple injection experiments in a single saprolite monolith was an acceptable technique.

Discussion FIGURE 4. Results of experiment 9 are shown. (a) Breakthrough curve for the release of microspheres relative to the total number of microspheres retained in the column during experiments 1-8. (b) Concentration of Ca2+ in the influent solution and concentration of monovalent and divalent cations in the effluent during experiment 9. the results of experiment 5 and to determine if the particles deposited in experiments 5-8 altered the extent of attach3740

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002

Comparison of Colloid and Solute Transport Mechanisms. The microspheres, which are small relative to the fracture aperture size (approximately 30 µm) but large relative to the size of the matrix pores (much less than 1 µm), were expected to be transported mainly by advection through the fractures. This was confirmed by the distribution of microspheres in the epoxy-impregnated thin sections of the saprolite, which were prepared from the dismantled column following the experiments. Examination of the saprolite thin sections under

an epifluorescent microscope showed that virtually all of the retained microspheres occurred along fractures or root holes. This is consistent with a previous study of microsphere transport in shale saprolite from the same site by Cumbie and McKay (7), in which microsphere distribution in a column used for tracer experiments was examined using UV light and microscale sampling methods. The bromide tracer and the cations in the influent solution are transported by a combination of advection along the fractures combined with diffusion-controlled exchange between the fracture pore water and the relatively immobile water in the pores of the fine-grained matrix. In the case of the cations, exchange between cations in solution and those attached to mineral surfaces is also an important transport mechanism. “Matrix diffusion” is expected to be most dominant in fractured, fine-grained materials, such as saprolite, where the matrix porosity is much greater than the fracture porosity (1, 33, 34). The role of matrix diffusion in the saprolite column is confirmed by the 5-fold difference in arrival times of the centers-of-mass of pulses of the bromide tracer and the microspheres (Figure 1), which is almost certainly due to diffusion of the bromide but not the microspheres into the matrix. After the pulse injection, when water containing no bromide is pumped through the column, the bromide slowly diffuses out of the matrix causing the long “tails” that are typical of solute tracers in sedimentary rock saprolite (1, 33, 34). These tails can persist for months or even years after a change in the chemistry of water entering the saprolite and are one of the principal causes of lingering groundwater contamination around waste pits in saprolite at the ORR in Tennessee (18). Effect of Ionic Strength and Composition on Colloid Mobility. For both the Na+ and Ca2+ experiments, as the ionic strength increased the peak C/C0 and percent recovery decreased (Table 2, Figure 3). This was the expected result arising from compression of the electric double layer and decreased repulsive interactions at higher counterion concentrations. Likewise, a much greater concentration of Na+ in the influent solution was required to result in a similar reduction in colloid recovery as compared to Ca2+. This effect has also been described for particle aggregation by the classic Schultze-Hardy rule (35). The results of this study thus appear to be consistent with similar studies in less heterogeneous materials, which have demonstrated similar dependence of cation charge and/or concentration on colloid deposition (15, 16, 36, 37). However, closer inspection of the results reveals the profound effect that pore structure and soil chemistry have on colloid transport. Note, for example, that the recovery of the three largest particles sizes was very similar at 5 and 7 mM Na+ but considerably lower than at the lowest Ca2+ ionic strength. The composition of the influent monovalent cation solution was altered along the flow path through the Ca-rich saprolite by processes of cation exchange and diffusioncontrolled exchange between pore water in the fractures and in the fine pore structure of the saprolite matrix (Table 2). Decreased colloid transport with increasing concentrations of monovalent solutions is attributed primarily to exchange and release of divalent cations at concentrations sufficiently high to permit the divalent cations to dominate the surface charge and interaction energies and, thus, control the extent of colloid retention. Theoretical considerations suggest that the surface potential will be determined solely by the divalent cations once their concentration exceeds about 3% of the monovalent concentration (38). The dominance of the divalent cations in colloid transport through the saprolite column is illustrated by the similarity in the slopes of the colloid stability curve (log of the percent recovery vs ionic strength) plotted with respect to the divalent cation concentration (Ca2+ + Mg2+; Figure 5). The slopes of the stability

curves for the different sized colloids are very similar, and the average slope for all four of the Ca2+ curves is identical to the averaged slope of the Na+ curves (-2.47 ( 0.17 and -2.45 ( 0.19, respectively). However, the colloid stability curves for experiments 1-4 (Na+ influent solutions) are transposed upward (i.e., toward greater colloid recoveries) as compared to the Ca2+ experiments. For example, the divalent cation concentrations in the effluent of experiments 1 (with Na+ influent solution) and 7 (with Ca2+ influent solution) are very similar (ca 0.35 mM), as are those in the effluents of experiments 4 and 8 (ca. 1.15 mM). Recoveries of the different size colloids for those two pairs of experiments average approximately 20-fold greater for the experiments where the microspheres were introduced in the Na+ solutions (experiments 1 and 4) as compared to the Ca solution experiments 7 and 8. We attribute the greater recoveries in the Na experiments to the creation of a gradient of Na- to Ca-dominated chemistry along the flow path, resulting in a shorter effective travel distance for colloids within a Ca-dominated solution chemistry. Given the much greater colloid deposition rate coefficients expected for the Ca-dominated portion of the flow path, most of the colloid retention is expected to occur in the downgradient segment of the column where Ca2+ concentrations exceed a few percent of the total cation composition. Thus, the transport and the retention of the injected microspheres were not controlled solely by the chemistry of the influent solutions. Instead, the extent of modification of the chemical conditions within the intact formation appears to have dominated these processes. Even prolonged inputs of Na+ solutions through this 21cm saprolite monolith did not lead to the dominance of Na+ over Ca2+. The cation exchange capacity (CEC) of this material is 140 mequiv kg-1 or a total of 1800 mmol of charge in the monolith calculated from the dimensions and bulk density of the column. However, only a small fraction of the total CEC is readily accessible to the mobile water in the hydraulically conductive fractures. Over the course of experiments 1-4, approximately 3000 mmol of Na+ was flushed through the column, yet the divalent cation concentration in the effluent solution did not drop below approximately 5% of the monovalent cation concentration. We attribute the failure to achieve a monocationic system to diffusive exchange between mobile water in the fractures and immobile water in the Ca-rich matrix solution. The diffuse tailing observed during the change from 1.0 to 0.1 mM Ca2+ (Figure 4b) demonstrates the slow exchange kinetics that accompanies attempts to alter the solution chemistry within the saprolite. Effect of Particle Size on Colloid Transport in Saprolite. All of the microsphere tracer injections show that microspheres of 0.5 µm diameter are preferentially transported in this material, despite varying solution chemistry (this study and ref 7). Previous studies (7) have suggested that the losses of the larger than optimum sizes were due to gravitational settling or straining, whereas the smaller particles were retained due to higher diffusion coefficients and more frequent collisions with the fracture walls (10). These findings do not appear to be related to the different surface charge of the different-size microspheres because the EM of the 0.5-µm particles is less negative than that of the larger size particles (Table 1). Thus, a reduced level of electrostatic repulsion and greater interactions with surfaces rather than diminished ones are predicted for the 0.5-µm particles, both expected to cause enhanced retention instead of the observed preferential transport. Colloid diameter has less effect on colloid transport than the solution chemistry. Ten-fold changes in ionic strength resulted in a 3 order of magnitude change in colloid transport through the column, while varying the colloid diameter by VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3741

a factor of 20 had a much smaller impact on colloid recovery. The ratio (rsize) of the percent recovery of the optimum 0.5µm particles as compared to that of the two larger microspheres is only a factor of 2.5, regardless of the chemistry. There is a greater difference between the 0.5- and 0.1-µm particles, with a greater disparity for the Na+ experiments (rsize ) 11) than for the Ca2+ experiments (rsize ) 5). This likely reflects the much higher diffusion coefficient of the smallest colloids as compared to the other colloids, which leads to a greater number of collisions of the 0.1-µm colloids with fracture walls. The particle size for optimal transport is likely to be different for dissimilar fractured formations. For example, Vilks and Bachinski (39) found size to have no effect on the mobility of colloids in fractured granite, whereas Becker et al. (40) found that very small colloids (0.19-µm microspheres) were recovered to a greater extent than 1-µm particles. In field tracer experiments with different size colloids, McCarthy (41) reported preferential transport of 0.5-µm colloids in the shallow saturated zones of the Dismal Gap saprolite, but larger colloids exhibited greater transport in the deeper zones. The deeper horizon represented a transition zone between the bedrock and overlying saprolite and is characterized by fewer, but larger, fractures as compared to the more weathered shallow zones from which the monolith used in the current study was derived. Thus, even within a single saprolite unit having a typical solution chemistry, the characteristics of the fracture network at a specific depth may have a critical control on the optimal size of colloids for transport. Microsphere Mobilization in Response to Reduced Ionic Strength. The total number of particles released from the column by the chemical perturbation experiment was very small relative to the total number of microspheres attached to the column and available to be released. Previous experimental studies in granular media have also shown that electrostatic attachment of particles is often largely irreversible, especially in systems dominated by divalent cations (42, 43). Although both the injected microspheres and the natural colloids appeared to be strongly attached to the Cadominated formation even after a sharp ionic strength adjustment, there was a small but finite background level of mobile microspheres observed in the column effluent throughout the experimental series. There appeared to be a continuous process of colloid release that persisted even after 20 or more PV of influent solution had been flushed through the column. The microsphere population observed prior to the beginning of experiments 2-10 (i.e., after 10 or more PV passed through the column since the previous injection) remained relatively constant throughout the experimental series and was not systematically related to either the solution conditions or the total number of colloids retained within the column. This resident level of mobile colloids was very small as compared to the injection concentration (C/C0 < 10-4), but the observations suggest that there is a small rate of colloid release even after extended periods. This is consistent with the video imaging and determination of a constant flux of colloids detaching from surfaces by AbdelFattah et al. (44). Grolimund et al. (45) demonstrated that colloid release was nonexponential and could be more appropriately described in terms of a distribution of release rates, which seems appropriate for portraying the behavior of colloids in heterogeneous natural materials such as the saprolite used here. Environmental Implications. While confirming the previously documented effects of ionic strength, counterion charge, and particle size on colloid migration, this study emphasizes the profound effect that the pore structure and geochemistry of the formation can impose on solution chemistry and thus on colloid transport. Various strategies 3742

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002

to manipulate colloid retention or mobilization have been discussed with the goal of either immobilizing colloid-bound contaminants, releasing them for purge-and-treat remediation or enhancing delivery of microorganisms to promote bioremediation (46). These strategies often propose to alter colloid behavior by imposing changes in the groundwater chemistry. Our results illustrate the difficulty of implementing manipulations of major cation chemistry in subsurface media such as saprolite, even in a laboratory setting in which the hydrological parameters of the experimental setup can be well controlled. Seasonal and storm-related changes in groundwater chemistry can also alter the ionic strength of the groundwater and thus potentially result in mobilization and enhanced migration of colloids. However, the largely irreversible attachment of previously deposited colloids observed here suggests that this effect may not be important in Cadominated formations. This is consistent with field observations that storm events resulting in changes in ionic strength had no consistent effect on turbidity in a Ca-dominated karstic formation at Oak Ridge, TN (47). Fundamental studies on colloid interactions are critical to understanding and predicting colloid mobility in the subsurface. However, field experiments and laboratory studies using complex geological materials are equally important to understanding how key processes will be expressed in natural subsurface environments.

Acknowledgments We are indebted to Dr. S. Driese at the University of Tennessee (UT) for carrying out the microscopic investigation of saprolite pore structure and microsphere distribution. We also thank C. R. Knight for technical assistance. The U.S. Department of Energy (DOE) Environmental Management Science Program (Project 55036) provided the main support for this research, with additional support provided by the UT Center for Environmental Biotechnology and the UT Waste Management Research and Education Institute.

Literature Cited (1) McKay, L. D.; Sanford, W. E.; Strong, J. M. Ground Water 2000, 38 (1), 139. (2) Smith, M. S.; Thomas, G. W.; White, E.; Ritonga, D. J. Environ. Qual. 1985, 14 (1), 87. (3) McKay, L. D.; Gillham, R. W.; Cherry, J. A. Water Resour. Res. 1993, 29 (12), 3879. (4) Kretzschmar, R.; Robarge, W. P.; Amoozegar, A. Water Resour. Res. 1995, 31 (3), 435. (5) Hinsby, K.; McKay, L. D.; Jorgensen, P.; Lenczewski, M.; Gerba, C. P. Ground Water 1996, 34 (6), 1065. (6) Harton, A. D. M.S. Thesis, University of Tennessee at Knoxville, 1996. (7) Cumbie, D. H.; McKay, L. D. J. Contam. Hydrol. 1999, 37, 139. (8) McCarthy, J. F.; Zachara, J. M. Environ. Sci. Technol. 1989, 23 (5), 496. (9) McKay, L. D.; Harton, A. D.; Wilson, G. V. J. Environ. Qual. 2002, 31 (4), 1095. (10) Yao, K.-M.; Habibian, M. T.; O’Melia, C. R. Environ. Sci. Technol. 1971, 5 (11), 1105. (11) Elimelech, M.; O’Melia, C. R. Environ. Sci. Technol. 1990, 24 (10), 1528. (12) Jacobsen, O. H.; Moldrup, P.; Larsen, C.; Konnerup, L.; Petersen, L. W. J. Hydrol. 1997, 196, 185. (13) Reimus, P. W. Ph.D. Thesis, University of New Mexico, 1995. (14) Grindrod, P. J. Contam. Hydrol. 1993, 13, 167. (15) Kretzschmar, R.; Sticher, H. Environ. Sci. Technol. 1997, 31 (12), 3497. (16) Grolimund, D.; Elimelech, M.; Borkovec, M.; Barmettler, K.; Kretzschmar, R.; Sticher, H. Environ. Sci. Technol. 1998, 32 (22), 3562. (17) Hatcher, R. D., Jr.; Leminski, P. J.; Dreier, R. B.; Ketelle, R. H.; Lee, R. R.; Leitzke, D. A.; McMaster, W. M.; Foreman, J. L.; Lee, S. Y. Status Report on the Geology of the Oak Ridge Reservation; Oak Ridge National Laboratory Publication ORNL/TM-12074; ORNL: Oak Ridge, TN, 1992.

(18) Solomon, D. K.; Moore, G. K.; Toran, L. E.; Dreier, R. B.; McMaster, R. B. Status Report: A Hydrologic Framework for the Oak Ridge Reservation; Oak Ridge National Laboratory Publication ORNL/ TM-12026; ORNL: Oak Ridge, TN, 1992. (19) Jardine, P. M.; Jacobs, G. K.; Wilson, G. V. Soil Sci. Soc. Am. J. 1993, 57 (4), 945. (20) Driese, S. G.; McKay, L. D.; Penfield, C. P. J. Sediment. Res. 2001, 71 (5), 843. (21) Wilson, G. V.; Jardine, P. M.; O’Dell, J. D.; Collineau, M. J. Contam. Hydrol. 1992, 145, 83. (22) Cropper, S. C. M.S. Thesis, University of Tennessee at Knoxville, 1998. (23) Pitner, A. M.S. Thesis, University of Tennessee at Knoxville, 2000. (24) Cumbie, D. H. M.S. Thesis, University of Tennessee at Knoxville, 1997. (25) Jardine, P. M.; Wilson, G. V.; Luxmoore, R. J. Soil Sci. Soc. Am. J. 1988, 52 (5), 1252. (26) Blake, G. R.; Hartge, K. H. In Methods of Soil Analysis: Part 1-Physical and Mineralogical Methods, 2nd ed.; Klute, A., Ed.; Soil Science Society of America: Madison, WI, 1986; pp 371373. (27) Harvey, R. W.; George, L. H.; Smith, R. L.; LeBlanc, D. R. Environ. Sci. Technol. 1989, 23, 51. (28) Kretzschmar, R.; Barmettler, K.; Golimund, D.; Yan, Y.; Borkovec, M.; Sticher, H. Water Resour. Res. 1997, 33 (5), 1129. (29) Ryan, J. N.; Elimelech, M. Colloids Surf. A 1996, 107, 1. (30) Snow, D. T. Water Resour. Res. 1969, 5, 1273. (31) McKay, L. D.; Cherry, J. A.; Gillham, R. W. Water Resour. Res. 1993, 29 (4), 1149. (32) Bruner, D. D. Thesis, University of Tennessee at Knoxville, 1997. (33) Reedy, O. C.; Jardine, P. M.; Wilson, G. V.; Selim, H. M. Soil Sci. Soc. Am. J. 1996, 60, 1376.

(34) McKay, L. D.; Stafford, P. L.; Toran, L. E. Ground Water 1997, 35 (6), 997. (35) Hunter, R. J. Foundations of Colloid Science; Clarendon Press: Oxford, 1986; Vol. 1, p 673. (36) Vinten, A. J. A.; Nye, P. H. J. Soil Sci. 1985, 36, 531. (37) Amirbahman, A.; Olsen, T. M. Environ. Sci. Technol. 1993, 27 (13), 2807. (38) Israelachvilli, J. N. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, 1991; p 449. (39) Vilks, P.; Bachinski, D. B. J. Contam. Hydrol. 1996, 21 (1-4), 269. (40) Becker, M. W.; Reimus, P. W.; Vilks, P. Ground Water 1999, 37 (3), 387. (41) McCarthy, J. F. Abstracts of Papers, 217th National Meeting of the American Chemical Society, Anaheim, CA; American Chemical Society: Washington, DC, 1999. (42) Kuo, R. J.; Matijevic, E. J. Chem. Soc., Faraday Trans. 1979, 75, 2014. (43) Hahn, M. W. Ph.D. Thesis, Johns Hopkins University, 1995. (44) Abdel-Fattah, A.; El-Genk, M. S.; Reimus, P. W. J. Colloid Interface Sci. 2002, 246 (2), 241. (45) Golimund, D.; Barmettler, K.; Borkovec, M. Water Resour. Res. 2001, 37 (3), 571. (46) McCarthy, J. F.; Wobber, F. J. Manipulation of Groundwater Colloids for Environmental Restoration; Lewis Publishers: Chelsea, MI, 1993; p 320. (47) Shevenell, L.; McCarthy, J. F. J. Hydrol. 2002, 255 (1/4), 50-68.

Received for review January 10, 2002. Revised manuscript received May 30, 2002. Accepted June 19, 2002. ES025522O

VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3743