Environ. Sci. Technol. 2003, 37, 3694-3700
Light Transmission Technique for the Evaluation of Colloidal Transport and Dynamics in Porous Media
attachment and detachment from the matrix surfaces. Attachment/detachment kinetics are commonly modeled as first-order processes, using filtration theory (10). Colloidal sorption at gas-liquid-solid interfaces has recently been recognized as an important process (11-16).
N O A M W E I S B R O D , * ,† M I C H A E L R . N I E M E T , ‡,§ A N D JOHN S. SELKER‡ Department of Environmental Hydrology and Microbiology, Institute for Water Sciences and Technologies, Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede Boqer Campus, 84990, Israel, and Department of Bioengineering, Oregon State University, Corvallis, Oregon 97331
A variety of tracers have been used in colloidal transport studies, including iron and chromium hydroxide, bacteria, bacteriophages, and fluorescent polystyrene (also called latex) microspheres (5, 9, 17-23). Fluorescent latex microspheres have been intensively investigated in relation to colloidal transport (e.g., ref 10 and references therein) and are widely used due to their (i) spherical, uniform shape; (ii) commercial availability in specific sizes (typically 0.02-5 µm); (iii) variety of surface properties; (iv) relatively simple and reliable measurement techniques; and (v) variety of excitationemission wavelengths in both visible and ultraviolet spectra.
Colloidal transport in porous media has been typically studied in column experiments from which data analysis was limited to the evaluation of effluent breakthrough curves and/or destructive sampling at the end of the experiments. The internal processes occur within a “black box”, where direct observation is not possible and therefore are often poorly understood. In this paper, a nondestructive, noninvasive method is presented that allows for quantitative measurement of colloid distribution with unprecedented two-dimensional spatial and temporal resolution. This technique is well-suited to observing the effects of saturation transitions and physical heterogeneities on colloidal transport. The potential of this novel technique is explored by investigating the effect of particle size and concentration on flow dynamics under saturated and unsaturated conditions. In saturated-flow experiments, deviation from the classical advection-dispersion behavior is observed. In unsaturated systems, colloidal accumulation at the capillary fringe interface and a high deposition rate of microspheres to the unsaturated media are readily observed. The experimental system is limited to translucent porous media and fluorescent colloids and is only semiquantitative in variably saturated media; nevertheless, it holds great promise for elucidating many complex mechanisms that control or influence colloid transport in the subsurface.
Introduction and Background The transport of natural colloids in groundwater has been an area of active research over the last two decades. It has been acknowledged that colloids may enhance the transport of low-solubility contaminants, in both the vadose and the saturated zones. Rapid migration of radionuclides, metals, and organic compounds has been attributed to colloidfacilitated transport (e.g., refs 1-7). Biocolloids, such as viruses and bacteria, have also been found to be mobile under various conditions in the subsurface (e.g., refs 8 and 9). The rate and extent of colloidal movement in porous media are controlled by pore water velocity as well as * Corresponding author phone: +972-8-6596903; fax: +972-86596909; e-mail:
[email protected]. † Ben-Gurion University of the Negev. ‡ Oregon State University. § Present address: CH2M HILL, Corvallis, OR 97330. 3694
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 16, 2003
In the laboratory, colloidal transport has been studied in various porous and fractured media, mainly in column experiments. Mathematical and conceptual models were developed based on breakthrough curves (BTCs) of microspheres (12, 17, 24, 25). Wan and Wilson (13) used a glass micromodel to visualize the sorption of hydrophilic and hydrophobic fluorescent latex microspheres to air-water and glass-water interfaces. The importance of the gas-water interface was later verified in column experiments (12). Siliman (26) studied the impact of various heterogeneities in a 2-D system containing glass beads using relatively large (2-90 µm) latex spheres. Accumulation of spheres was observed mainly at the interface between different-sized beads. In both the field and the laboratory, transport properties such as the retardation factor and the dispersion and diffusion coefficients have been almost exclusively inferred from BTCs and/or destructively obtained samples of the porous medium at the end of the experiment. Measurements within the system are restricted to destructive sampling at the end of the experiments. The lack of continuous internal monitoring limits the ability to explore processes that occur at textural interfaces in layered soils, along the transition zone between saturated and unsaturated porous media, and at discontinuities within the porous matrix. Light transmission techniques have been gaining popularity as nondestructive, noninvasive, laboratory measurements of water, air, and nonaqueous phase liquids (NAPLs) in both porous and fractured media using 2-D flow cells (e.g., refs 27-31). Previously, this method has mainly been used for quantification of changes in water content over time under different conditions (27-30). However, the use of similar systems for the quantification of fluorescent microsphere transport, as presented in this paper, has not yet been reported nor, to our knowledge, attempted. In this paper, a light transmission technique for the quantification of fluorescent microspheres under both saturated and unsaturated conditions is presented. This work is unique in that transmitted light, at a controlled wavelength, is used to induce a measurable fluorescence from the microspheres. The fluorescence is subsequently separated from the overall light output and corrected for spatial variations in source intensity to ultimately quantify the spatial and temporal distribution microspheres. The method is independent of grain size and applicable under a variety of hydrodynamic conditions. The main focus of this paper is to demonstrate the great potential of this method for better understanding of colloidal transport in porous media. 10.1021/es034010m CCC: $25.00
2003 American Chemical Society Published on Web 07/12/2003
FIGURE 1. Exploded and assembled views of the light transmission chamber [after Niemet and Selker (32)]. Chamber is 50 cm wide and 60 cm high (a); important spectra for the fluorescent tubes (TL835 Phillips) (b) and gel-filter sheet (Rosco, medium blue, transmission 4%) (c). Both spectra were taken from the manufacturers.
TABLE 1. Experimental Detailsa expt 1 expt 2 expt 3
left plume
center plume
right plume
Accusand used
1.0 µm (0.2%) 0.02 µm (0.04%) 0.02 µm (0.04%)
1.0 µm (0.04%) 0.2 µm (0.04%) 1.0 µm (0.04%)
1.0 µm (0.01%) 2.1 µm (0.04%)
saturated 40/50b saturated 40/50b prewetted 20/30c
a All microspheres were diluted in 0.005 M NaNO . FluoSphere (solid) concentrations are given in parentheses. b Spheres were injected under 3 no-flow conditions (i.e., after the sand was saturated but before the pumps started to create the flow). After injection, a flow rate of 3.6 mL/min was maintained. c Prewetted refers to sand that was saturated and allowed to drain for 24 h while the outflow pipe was kept 17 cm above the bottom of the chamber (see text for details).
Materials and Methods Experimental System. The 2-D light transmission flow cell (chamber) is diagrammed in Figure 1a. The chamber, light detection system, and porous media were similar to that described by Weisbrod et al. (30). The porous media consisted of translucent silica sand (Accusand, Unimin, Le Sueur, Minnesota) of different grades. The sand, commercially available in different grain sizes, was prewashed with distilled water to remove dust. The chemical, physical, and optical properties of Accusand have been well-characterized (31, 32). Fluorescent polystyrene microspheres with peak excitation/emission wavelengths of 580/605 nm in 0.02, 0.2, and
2.1 µm diameters were obtained from Molecular Probes (Fluospheres, Eugene, OR). Carboxylate modified latex (CML) microspheres were used. These have carboxyl surface functional groups, which give them a negative charge at pH >5. They also have hydrophilic surfaces, distinguishing them from other types of polystyrene microspheres. The combination of hydrophilic and negatively charged surfaces makes CML colloids stable against aggregation at much higher ionic strengths than other microspheres. All colloidal suspensions were obtained by diluting the stock solution (2% or 4% solids) with a 0.005 M NaNO3 solution. The pH of the suspensions was ∼6. Table 1 shows the type and concentrations of microspheres applied in each of the three experiments presented in this paper. VOL. 37, NO. 16, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3695
FIGURE 2. Schematic diagram of the major components of the light transmission system. Note that in the saturated system (a) the emitted light detected by the CCD camera is proportional to the concentration of microspheres, while in the unsaturated system (b) it is a combination of the microsphere concentration and the amount of trapped air in the system.
FIGURE 3. Schematic diagrams of the experimental setup of saturated (a) and prewetted porous (b) media. Two identical peristaltic pumps (inflow and outflow pipes) were employed to maintain a continuous and constant flow rate (a). Selection of a filter set turned out not to be a simple matter of matching the peak excitation and emission frequencies. The peak excitation and emission wavelengths for these colloids are only 25 cm apart. To our knowledge, no filters are available that can completely separate between these wavelengths without some carryover. Furthermore, the output of the fluorescent tubes (Phillips TL835; Figure 1b) was highly multi-spectral and impossible to reduce entirely to the peak excitation wavelength. Ultimately, after a lengthy trial-and-error process, a medium blue no. 83 gel-filter sheet (Rosco Laboratories, Inc., Hollywood CA; Figure 1c), was selected. Surprisingly, this sheet’s transmission minima was near the 580 nm wavelength recommended for maximum excitation of the FluoSpheres. Apparently, sufficient excitation was achieved from other off-peak wavelengths below 580 nm (33). Better results were achieved using a slightly off-peak emission filter in front of the camera as well. The peak emission frequency for the microspheres was reported to be 605 nm; however, it was found that a 620 nm filter placed in front of the camera provided higher signal-to-noise ratio than a 605 nm filter. The signal-to-noise ratio is defined here as the ratio of actual signal attributable to fluorescence to the total random noise components associated primarily with source light that “leaks” through the emission filter. This is likely to be due to the shift in the effective excitation energy below the recommended 580 nm peak excitation wavelength. A 490 nm filter was used to correct for changes in saturation 3696
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 16, 2003
during the unsaturated flow experiments to prevent interference from emitted light (31). Quantification and Image Processing. Under saturated conditions, fluorescence was quantified using the 620 nm filter in front of the camera lens (Figure 2). Since some light was measured using this filter, even in the absence of colloids (due to limitation of the gel-filter), background levels of light intensity (no colloids) were subtracted from the emission images. In addition, spatial variability due to edge effects, etc. was removed by flat-fielding (dividing) by the same background image. Fluorescence measured in this way is termed the saturated normalized fluorescence (Fsat) and was computed as follows:
Fsat )
Iem - Iem,0 Iem,0
(1)
where Iem represents the emission image and Iem,0 is the initial emission image (time zero, before the addition of colloids). In this discussion, all measured light intensities are corrected for dark and thermal noise as well as for variability in source lighting as explained by Niemet and Selker (31). Fsat was computed on a pixel-by-pixel basis over the entire image. Fluorescence was shown to be conserved, since the ∑Fsat remained constant for all images ((4%) as the plumes moved and dispersed in the chamber. It is worth mentioning that the absolute emission varies between microspheres according to their size (33).
FIGURE 4. Three images taken 1, 100, and 200 min after injection of 5 mL of 0.2, 0.04, and 0.01% of 1 µm microspheres (left to right, respectively) (expt 1; Table 1) (a). Three images taken after 1, 100, and 180 min after injection of 5 mL of 0.02, 0.2 and 2.1 µm microspheres (left to right, respectively; 0.04%) (expt 2; Table 1) (b). Both experiments were under saturated conditions with a flow rate of 3.6 mL/min downward. Under unsaturated conditions, the detected light is a function of water content as well as colloidal fluorescence (Figure 2). A similar approach was used to compute the normalized fluorescence under unsaturated conditions except that two simultaneous and independent images were required. A second image obtained through a filter in the excitation range (490 nm in this case) (Iex) corrected for the spatial distribution of excitation energy caused by the water content distribution and is independent of colloidal fluorescence. The normalized fluorescence under unsaturated conditions (F) was computed as follows:
F)
Iem Iex
determined as follows:
M)F
∑M ∑F
(3)
where ∑M is the total mass of colloids in the chamber and ∑F is the sum of the normalized fluorescence over all pixels. Concentration at a pixel (C) can subsequently be determined from
C ) Mθ
(4)
(2)
The processed images provide a pixel-by-pixel spatial distribution of the changes in apparent fluorescence. The mass of colloids present at a given pixel (M) is proportional to the normalized fluorescence and can be
where θ is the volumetric water content, determined as described by Niemet and Selker (31). Experimental Procedure. Two of the experiments were performed in saturated sand, where the impact of colloidal concentration and size on transport were investigated (Figure 3a). In both experiments, the chamber was packed with 40/ VOL. 37, NO. 16, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3697
50 sand (mean grain diameter, d50 ) 0.359 mm), saturated with distilled water from the bottom; three 5 mL colloidal suspensions were injected ∼3 cm below the surface at a rate of 2 mL/min. A constant flow rate of 3.6 mL/min was maintained. In the first experiment (expt 1), three dilutions (0.2, 0.04, and 0.01% solids) of 1 µm microspheres were used, whereas three different sizes of microspheres (0.02, 0.2, and 2.1 µm) at 0.04% solids were used in expt 2. Images were taken at 10 min intervals using a 620 nm (10 nm band-pass) filter in front of the camera until the plumes left the chamber (∼4 h). One series of images from experiments in a variably saturated porous media is included in this paper to demonstrate the potential of this system to be used in unsaturated media as well. In expt 3, 20/30 sand (mean grain diameter, d50 ) 0.713 mm) was saturated with distilled water and drained for 24 h while the outflow pipe was kept 17 cm above the bottom of the outflow manifold (Figure 3b). A detailed description of the saturation and draining procedure as well as the retention curve for the 20/30 Accusand is provided by Weisbrod et al. (30). Next, two types of microspheres, 0.02 and 1 µm at 0.04% solids, were applied to the surface of the sand at a flow rate of 1 mL/min for 5 min. To separate the fluorescence of the microspheres from the excitation energy, every image with the 620 nm filter was followed by an image with the 490 nm filter. The 490 nm center-wavelength, 10 nm band-pass filter was opaque to the light emitted by the microspheres and represented the spatial distribution of excitation energy caused by changes in water content, packing irregularities, edge effects, etc. Images were taken every 10 min for 2 h and then every 2 h for 24 h. Next, an 80 mL/min application from a 10 needle irrigation system (Figure 3b) was used to flush the chamber for 2 h. The main purpose of this stage was an attempt to mobilize the microspheres contained in the chamber. Overall, more than 15 pore volumes were used to irrigate the variably saturated sand.
Results and Discussion Saturated Media. Three images for Fsat in expts 1 and 2 (Table 1) are shown in Figure 4a,b, respectively (pseudo-colorized), for the concentration and size experiments. In both cases, the initially round plumes dispersed along the flow direction over time. In expt 1 (Figure 4a), observable amounts of microspheres from the high concentration plume were left behind near the point of injection. In expt 2 (Figure 4b), a large number of 2.1 µm spheres were left behind the migrating plume, and a tail developed behind it. Conservation of fluorescence was evaluated by the sum of Fsat for each of the three injected solutions. The ratio between the calculated total fluorescence in the system at a given time and the total fluorescence right after injection for both experiments was compared for all experiments. Conservation of fluorescence for expt 1 suggests that virtually no microspheres were left within the chamber ((4%). In expt 2, ∼15% of the 2.1 µm spheres (by fluorescence) were trapped and never left the chamber (see tail behind the 2.1 µm plume in Figure 4a). Observed transport was compared against that predicted by the advection dispersion equation (ADE). Ideal behavior was modeled using the 2-D solution to the ADE for an instantaneous point source (34; eq 2.32). The observed fluorescence pattern immediately after injection was approximated by a symmetrical 2-D normal distribution. The ideal plume was then divided into ∼1 mm elements. Thereafter, at each time step, the ADE solution was individually applied to each element and subsequently superimposed. Dispersion coefficients were determined based on the best fit to the data, and the longitudinal (flow direction) velocity was determined based on the flow rate, porosity, and chamber dimensions. 3698
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 16, 2003
FIGURE 5. Comparison between predicted transport of the microspheres according to the ADE, assuming conservative tracers (solid lines), vs the observed data (dots, expt 2; Table 1). Note the advancement of the front of the observed colloid plume relative to the ADE. Figure 5 shows the calculated versus predicted longitudinal transport of the microspheres in expt 2 at seven points in time. The figure shows that the leading edge of the observed colloid plumes advanced faster than expected relative to the ADE prediction; this effect was most pronounced for the 2.1 µm microspheres. For the same experiment, the 1st through 3rd spatial moments for the microspheres were calculated and compared to those predicted by the traditional ADE using a solution for a point slug injection into a uniform 2-D flow field (35). The 1st, 2nd, and 3rd spatial moments represent the location of the center of mass, the variance about the center of mass, and the skewness of the plume, respectively, assuming conservative Fickian behavior (e.g., ref 36). In the calculation of the moments, only pixels 2% above the background intensity were considered. Figure 6 shows the calculated versus predicted 1st through 3rd normalized spatial moments in expt 2. The predicted moments were calculated based on the solution to the ADE, and the normalized value refers to the ratio between the mean and the total values.
FIGURE 6. 1st, 2nd and 3rd normalized spatial moments of the 0.02, 0.2 and 2.1 µm microspheres (expt 2) as compared to the ADE assuming conservative tracers (solid lines).
FIGURE 7. Four images taken 3.5, 20, 45, and 1480 min after injection (expt 3, Table 1). Water saturation is represented by the black contour lines at the 0.15, 0.3, 0.5, and 0.8 levels and was taken with a 490 nm filter. Capillary fringe is just below the 0.8 saturation contour. Pseudo-colorized scale represents the level of fluorescence emission. Note that the image taken 1480 min from injection is slightly “out of focus”, probably due to a slight movement (1-2 mm) of the CCD camera. Nevertheless, this has no influence on the observed location of the fluorescent plumes. The apparent longitudinal dispersion coefficient, estimated from the slope of the 2nd spatial moment, was higher than the ADE prediction. For the 2.1 µm microspheres, the tail of colloids left behind the plume caused the method of moments to estimate an unrealistic dispersion coefficient and skewness. The 3rd moment of the observed plumes increased in magnitude as the plumes approached the bottom of the chamber, suggesting non-Fickian transport. The early arrival of mass, noted from Figure 5, and the increasing 3rd moments (Figure 6) may play a role in the rapid colloidfacilitated transport of radionuclides and other colloid-bound contaminants beyond that predicted by most models. This non-Fickian behavior in an almost ideal (homogeneous) system warrants further investigation. In the transverse direction, there appeared to be little or no dispersion. For the ADE prediction, the effect of transverse dispersion was
taken to be zero, as shown by the horizontal line in Figure 6 for the 2nd moment. Variably Saturated Media. The results of expt 3 are shown in Figure 7. Here, three pseudo-colorized images of normalized fluorescence (620 nm filter) were taken 3.5, 20, and 45 min after the application of 5 mL of 0.02 and 1 µm microspheres (0.04% solids) into prewetted sand. Superimposed on the image are saturation contours determined using transmission images (490 nm filter). Interestingly, the colloidal plume was trapped at the capillary fringe interface and never penetrated into the capillary fringe (during the 24 h of expt 3). Furthermore, while no entrapment was observed for colloids e1 µm in saturated sand, a significant number of colloids were left behind for both the 0.02 and 1 µm microspheres in the unsaturated sand. The trapped colloids, most likely at the air-water-solid interfaces, never reVOL. 37, NO. 16, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3699
mobilized even after the chamber was irrigated with a high flux of water for 2 h following the experiment (irreversible deposition). The ability to track independently variation in saturation and microsphere concentration provides a powerful tool for exploring colloidal transport in variably saturated porous media. Implications for Colloidal Transport and Future Research. The results presented in this paper are intended to demonstrate the strength and potential of the experimental system rather than to act as a detailed interpretation of the experimental results. Therefore, only a few illustrative examples are presented here. Currently, the signal-to-noise ratio is still relatively high, and the impact of saturation on total emission is not yet fully understood. These factors may be improved in future work using higher performance of excitation/emission filters and light sources, TransFluoSphere beads in which there is a large discrepancy between excitation and emission peaks (33), and more accurate calibration experiments. Additionally, one should be aware of the limitations of the presented method, including (i) the need to use translucent silica sand; (ii) only fluorescent microspheres can be used; and (iii) two-dimensional transport. In unsaturated media, a number of difficulties remain to be overcome. Specifically, a nonlinear link between the degree of saturation and the emission of fluorescent colloids prevents accurate determination of colloid concentration. Despite the qualitative nature of some of the data interpretation (especially in the variably saturated media), it was shown that transport processes within the “black box” of the matrix of interest can be explored under a fully controlled variety of hydrodynamic conditions to reveal important spatial structures. The dynamics of colloidal migration can be determined, including kinetics of transport processes. Many of the most interesting transport phenomena are likely to occur at the interface between saturated and unsaturated zones (airwater interface, capillary fringe), at the interface between different matrix properties (e.g., grain sizes), and at preferential flow pathways. These zones of interest can be directly explored with the system presented here. The importance of matrix heterogeneity on colloidal transport could also be studied in the described system using a variety of packs. Chambers could easily be packed in horizontal or slanted layers of different sand grades, representing typical conditions in many sedimentary basins. Since the saturation level at the air-water interface can be accurately evaluated using the light transmission method (37), this method may also be used to study the importance of air-water interfaces on colloidal transport.
Acknowledgments We thank Mark Rockhold from Pacific Northwest National Laboratory for fruitful discussions during the data analysis for this manuscript. We also thank Yu-Zhong Zhang from Molecular Probes (Eugene, OR) for his contribution to the selection and use of the Fluospheres. The constructive review of three anonymous reviewers is highly appreciated.
Literature Cited (1) Bates, J. K.; Bradley, J. P.; Teetsov, A.; Bradley, C. R.; Buchholtz ten Brink, M. Science 1992, 256, 649-651. (2) Kersting, A. B.; Efurd, D. W.; Finnegan, D. L.; Rokop, D. J.; Smith, D. K.; Thompson, J. L. Nature 1999, 397, 56-59. (3) McCarthy, J. F.; Zachara, J. M. Environ. Sci. Technol. 1989, 26, 496-502.
3700
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 16, 2003
(4) McDowell-Boyer, L. M.; Hunt, J. R.; Sitar, N. Water Resour. Res. 1986, 22, 1901-1921. (5) Puls, R. W.; Powell, R. M. Environ. Sci. Technol. 1992, 26, 614621. (6) Ryan, J. N.; Elimelech, M. Colloid Surf. A 1996, 107, 1-56. (7) Ryan, J. N.; Illangasekare, T. H.; Litaor, M. I.; Shannon, R. Environ. Sci. Technol. 1998, 32, 476-482. (8) Yates, M. V.; Yates, S. R. Crit. Rev. Environ. Control 1992, 17, 307-344. (9) Harvey, R. W.; George, L. H.; LeBlanc, D. R. Environ. Sci. Technol. 1989, 23, 51-56. (10) Elimelech, M.; Gregory, J.; Jia, X.; William, R. A. Particle Deposition and Aggregation, Measuring, Modeling, and Simulation; Butterworth-Heinemann: Waburn, MA, 1995. (11) Wan, J.; Wilson, J. L.; Kieft, T. L. Appl. Environ. Microbiol. 1994, 60, 509-516. (12) Wan, J.; Wilson, J. L. Water Resour. Res. 1994, 30, 857-864. (13) Wan, J.; Wilson, J. L. Water Resour. Res. 1994, 30, 11-23. (14) Williams, D. F.; Berg, J. C. J. Colloid Interface Sci. 1991, 152, 218-229. (15) Powelson, D. K.; Mills, A. L. Appl. Environ. Microbiol. 1996, 62, 2593-2597. (16) Christ, J. T.; Yuniati, Z.; Taylor, J.; Peranginangin, N.; Gao, B.; Lyon, S.; Dittrich, T. M.; Marshall, A. G.; De Alwis, D.; Baveye, P. C.; Parlange, J. Y.; McCarthy, J. F.; Steenhuis, T. S. In Proceedings of the International Workshop on Colloid and Colloid-Facilitated Transport of Contaminants in Soils and Sediments; de Jonge, L. W., Maldrup, P., Jacobson, O H., Eds.; DIAS Report; Tjele, Denmark, 2002; pp 31-39. (17) Cumbie, D. H.; McKay, L. D. J. Contam. Hydrol. 1999, 37, 139157. (18) Elimelech, M. J. Colloid Interface Sci. 1991, 146, 337-352. (19) Elimelech, M. Water Res. 1992, 26, 1-8. (20) Harvey, R. W.; Kinner, N. E.; MacDonald, D.; Metge, D. W.; Bunn, A. Water Resour. Res. 1993, 29, 2713-2721. (21) McKay, L. D.; Gillham, R. W.; Cherry, J. A. Water Resour. Res. 1993, 29, 3879-3890. (22) Weisbrod, N.; Ronen, D.; Nativ, R. Environ. Sci. Technol. 1996, 30, 3094-3099. (23) Bales, R. C.; Li, S.; Maguire, K. M.; Yahya, M. T.; Gerba, C. P.; Harvey, R. W. Ground Water 1995, 33, 653-661. (24) Ward, R. S.; Harisson, I.; Leader, R. U.; Williams, A. T. In Tracer Hydrology 97; Kranjc, A., Ed.; A. A. Balkema: Rotterdam, 1997; pp 99-103. (25) Toran, L. E.; Palumbo, A. V. J. Contam. Hydrol. 1992, 9, 289303. (26) Silliman, S. E. J. Hydrol. 1995, 167, 79-98. (27) Detwiler, R. J.; Pringle, S. E.; Glass, R. J. Water Resour. Res. 1999, 35, 2605-2617. (28) Glass, R. J.; Nicholl, M. J. Geoderma 1996, 70, 133-163. (29) Walter, M. T.; Kim, J. S.; Steenhuis, T. S.; Parlange, J.-Y.; Heilig, A.; Braddock, R. D.; Selker, J. S.; Boll, J. Water Resour. Res. 2000, 36, 841-849. (30) Weisbrod, N.; Niemet, R. M.; Selker, J. S. Adv. Water Resour. 2002, 25, 841-855. (31) Niemet, M. R.; Selker, J. S. Adv. Water Resour. 2001, 24, 651666. (32) Schroth, M. H.; Ahern, S. J.; Selker, J. S.; Istok, J. D. Soil Sci. Soc. Am. J. 1996, 60, 1331-1339. (33) Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals, 6th ed.; Molecular Probes: Eugene, OR, 1996. (34) Fetter, C. W. Contaminant Hydrogeology; Macmillan: New York, 1993. (35) De Josselin De Jong, G. EOS, Trans. Am. Geophys. Union 1958, 39. (36) Freyberg, D. L. A. Water Resour. Res. 1986, 22, 2031-2046. (37) Niemet, M. R.; Rockhold, M. L.; Weisbrod, N.; Selker, J. S. Water Resour. Res. 2002, 38, 10-11-10-12.
Received for review January 5, 2003. Revised manuscript received May 18, 2003. Accepted May 28, 2003. ES034010M
