Introduction to special section on Colloid Transport in Subsurface Environments

Authors


Abstract

[1] The Water Resources Research special section on Colloid Transport in Subsurface Environments presents new knowledge that is critical to solving problems related to groundwater pollution by microbial pathogens and hazardous chemicals. This introduction to the special section surveys fourteen manuscripts that advance current understanding of the transport of biocolloids (e.g., bacteria, viruses, and protozoa), mineral colloids, and colloid-associated contaminants in the vadose zone and in groundwater. These papers present new techniques for elucidating mechanisms that govern colloid mobility, propose mathematical models appropriate for quantifying colloid and colloid-associated contaminant transport, and report pore-scale and column-scale observations requisite for evaluating these models. Together, the papers of this special section illuminate the complexity of the colloid transport problem and describe progress toward understanding this complexity.

[2] Groundwater is the sole source of drinking water for more than one quarter of the world's population and comprises more than half of the drinking water consumed in the United States and Europe [Clark and King, 2004]. Agriculture, industry, mining, and urbanization have not only stressed the world's supply of groundwater, but have diminished its quality, leading to debilitating disease and illness [Gleick, 2004]. The agents of pollution are in some cases dissolved in groundwater, but in other cases consist of colloid-sized (0.001 to 10 μm) pathogens, such as viruses, bacteria, and protozoa, or toxic chemicals that are bound to colloid-sized mineral particles, such as clays and oxide precipitates. The papers that appear in this special section focus on advancing understanding of how these colloid-sized microbes and colloid-associated contaminants move in subsurface environments.

[3] Microbes may enter groundwater via the vadose zone or with seepage from surface water bodies (lakes and rivers). The delivery of microbes through the vadose zone and across the water table depends, in part, on infiltration rates, dispersive mixing, and processes that affect microbial deposition (immobilization) and release from stationary interfaces. One contribution to this special section is aimed at advancing knowledge of microbial deposition and release within the vadose zone. Torkzaban et al. [2006] report that attachment to air-water interfaces, in addition to solid-water interfaces, controls the deposition of the bacteriophages MS2 and ΦX174 (used as surrogates for pathogenic viruses) in columns of partially saturated quartz sand. According to the authors, the sensitivity of bacteriophage deposition to changes in pH indicates that electrostatic interactions, rather than hydrophobic interactions, regulate virus attachment to both air-water and solid-water interfaces. Bacteriophages retained within the unsaturated sand packs were mobilized upon column imbibition, presumably in response to the elimination of bacteriophage-scavenging air-water interfaces. These observations suggest that pore water flow transients and resulting changes in air-water configuration are important drivers of microbial mobilization in the vadose zone.

[4] The contributions of Tufenkji [2006], Cortis et al. [2006], and Bradford et al. [2006a] similarly address the microbial transport problem but are distinguished from the work of Torkzaban et al. [2006] by their focus on water-saturated porous media. The basic framework for describing the transport and deposition of colloid-sized microbes in saturated porous media was developed in the 1970s by engineers interested primarily in improving the performance of packed bed filters used in water treatment [Yao et al., 1971; Rajagopalan and Tien, 1976]. At the heart of this colloid filtration theory is a method for calculating deposition rates from measurements of the physical and chemical properties of the colloid-water-sediment system.

[5] Tufenkji [2006] examines the appropriateness of colloid filtration theory for describing the deposition kinetics of bacteria. She reports that theoretical estimates of deposition rates substantially underestimate those measured for Eschericia coli in column experiments with quartz sand. Comparison of retained and eluted bacterial concentrations with corresponding model predictions reveal that bacterial deposition is characterized by a broad distribution in rates, rather than by one or two discrete rates. Tufenkji [2006] hypothesizes that this broad distribution in rates results from charge heterogeneity of the bacterial cell surface imparted by membrane-bound proteins or lipopolysaccharide-associated functional groups. The findings from this work imply that advancement of predictive approaches for bacterial transport in groundwater will rely, at least in part, on improvement in methods used to characterize bacterial cell surfaces and on derivation of relationships that link cell surface properties (in addition to zeta potential) to deposition rates.

[6] Cortis et al. [2006] also investigate the influences of heterogeneity on microbial mobility within water-saturated aquifer materials. A key contribution of this study is the presentation of a continuous time random walk model that accounts for complexities stemming from microscopic-scale heterogeneity in mineral grain surface composition and nonuniformity in biocolloid surface charge. This stochastic model treats biocolloid attachment and detachment phenomenologically by defining these processes in terms of a sticking rate and sticking time probability density function. Evaluation of this model against experimental data on the transport of Cryptosporidium parvum illuminates the importance of physicochemical heterogeneity in governing the slow, extended elution of C. parvum that persisted long after the cessation of the injection. This theoretical framework has applicability outside the range of experimental conditions tested by Cortis et al. [2006], as late time tailing of breakthrough curves has been observed in published studies that use viruses, bacteria, and mineral particles as the colloids [Harvey et al., 1995; Schijven et al., 1999; Zhang et al., 2001].

[7] Bradford et al. [2006a] contend that biocolloid transport is influenced by straining, a mechanism not considered by traditional colloid filtration theory. These investigators present results from pore-scale visualization experiments showing immobilization of bacteria (E. coli) by straining, which occurred when colloids were trapped within small pores and within pore space constrictions associated with grain-grain contacts. Results from complementary experiments with sand columns demonstrate that this straining is sensitive to variation in specific discharge and sand grain texture. Bradford et al. [2006a] observe that strained E. coli accumulate in aggregates, and hypothesize that, once aggregates grow to sufficient size, they are disrupted by shear forces, leading to the release and downgradient transport of the biocolloids. A mathematical model developed by the authors to include this phenomenon describes measurements made in the column experiments with good success.

[8] Auset and Keller [2006] expand the theme that straining is an important contributor to colloid retention within water-saturated geologic materials. They used etched silicon micromodels with rough and smooth grain surfaces to distinguish, in an unambiguous way, the contributions of straining and attachment (as described by traditional colloid filtration theory) to the retention of latex microspheres. The authors demonstrate that classical colloid filtration theory is suitable for quantifying colloid retention when the ratio of pore throat diameter (T) to colloid diameter (C) exceeds 2.5; however, straining contributes significantly to colloid immobilization for T/C < 2.5 and measured retention rates deviate from those calculated on the basis of colloid filtration theory. Auset and Keller [2006] also show that greater concentrations of colloids adhere to rough than smooth grains. They suggest rough surfaces perturb fluid streamlines in a way that increases the frequency of colloid-collector collisions and that surface irregularities may lower repulsive colloid-collector interaction forces, thereby increasing the probability of attachment.

[9] Xu et al. [2006] seek to advance quantitative descriptions of straining by identifying relationships between straining rates and measurable properties of the colloids and porous media. The authors report that straining kinetics of latex microspheres can be described with a mathematical model that predicts a nonlinear decline in straining rates with increasing concentrations of strained colloids. Best fit values of the model coefficients that quantify clean bed straining rates and the capacity of the porous medium to retain colloids vary linearly with the ratio of colloid diameter (dp) to grain size (dg). Although variation in the model parameter estimates can be described in terms of a single predictor (i.e., dp/dg), Xu et al. [2006] point out that factors not tested in their study, such as nonuniformity in colloid shape, colloid size, and grain size, will further complicate descriptions of straining kinetics in natural groundwater environments.

[10] Bradford et al. [2006b] review and synthesize various lines of evidence that underpin the assumption that straining is a significant and, in some cases, a leading mechanism of mineral colloid and biocolloid immobilization within aquifer sediments. The authors identify deficiencies in our understanding of straining and suggest that experiments designed to test the effects of pore water chemistry, hydrodynamics, physical heterogeneity, and volumetric moisture content are critical to improving our understanding of the phenomenon.

[11] As described above for straining [Auset and Keller, 2006], the application of visualization techniques has improved our understanding of colloid transport [Wan and Wilson, 1994; Keller and Sirivithayapakorn, 2004; Gao et al., 2006]. Visualization techniques, unlike column experiments, permit positive identification of the mechanisms that control colloid deposition, transport, and mobilization within porous media. In this special section, Ochiai et al. [2006] review recent developments in visible light, magnetic resonance, and X-ray visualization techniques for investigations of pore-scale and mesoscale transport phenomena. As described below, papers by Zevi et al. [2006], DiCarlo et al. [2006], and Baumann and Niessner [2006] document approaches for analyzing visualization experiments, present new techniques for colloid transport visualization, and employ published visualization methods to investigate previously unexplored phenomena.

[12] Although visualization techniques have proliferated, quantification of observations is rarely done and tests of quantification methods are largely unavailable. Zevi et al. [2006] address this issue by testing two procedures (Boolean and region) for quantifying temporal changes in attached and mobile phase colloid concentrations in experiments with a flow chamber packed with partially saturated sand. They find that the region procedure performs better than the Boolean procedure, which underestimates concentrations of fluorescent micropsheres attached near the air/water meniscus/solid interface. With continued development, these quantification procedures should become invaluable in testing rate laws that describe the kinetics of colloid attachment near air-water and solid-water interfaces.

[13] Colloid concentrations in a three-dimensional porous medium can be measured by synchrotron X-ray fluorescence if the colloids are labeled with an ion that fluoresces. DiCarlo et al. [2006] adopt this analytical technique to investigate the transport of cadmium-tagged montmorillonite colloids through variably saturated sand. Observations made during these experiments indicate that colloids were transported along preferred flow pathways formed by fingering flow, with sodium-saturated montmorillonite moving with the wetting front and calcium-saturated montmorillonite being retarded behind the front. Retention of the colloids in the unsaturated and saturated portions of the fingering flow was similar, and colloids retained in the porous medium were readily mobilized by a new wetting front. These findings confirm that transient flow is especially important in colloid mobilization and transport in unsaturated porous media [El-Farhan et al., 2000; Saiers et al., 2003].

[14] Baumann and Niessner [2006] employ a transparent micromodel to analyze the transport of fluorescent latex microspheres in a three-phase system containing water, air, and octanol. These researchers show most of the microspheres were deposited on the interfaces between the water and the hydrophobic octanol. When the microspheres attached to the octanol, they lost their fluorescent dye (dipyrromethene boron difluoride) to the octanol phase, and as the octanol phase dissolved into the water, so did the fluorescent dye. Baumann and Niessner [2006] discuss the relevance of this process to the partitioning of hydrophobic organic contaminants between colloids and nonaqueous liquid phases in aquifers contaminated by chlorinated solvents, coal tar, and fuel products.

[15] Most studies of colloid transport, whether at the pore scale, column scale, or field scale, focus on granular media. Consequently, little is known about the physical and chemical mechanisms that govern colloid transport in fractured media [Wan et al., 1996; Becker et al., 1999]. Fracture flow and transport is of particular concern at sites being considered for radioactive waste disposal in granite and volcanic tuff rocks [Degueldre et al., 1989; Moridis et al., 2003]. As a part of this special section, Zvikelsky and Weisbrod [2006] explore colloid transport through cores extracted from a fractured chalk formation. The authors report that the smallest colloids tested (0.02 μm latex microspheres) were most susceptible to retention within the cores, while maximum breakthrough concentrations measured in the experiments with 0.2 μm and 1.0 μm colloids exceeded those measured for dissolved, conservative tracers. The colloid breakthrough curves were characterized by a lack of tailing, suggesting that colloid exchange between the fracture and matrix was small compared to dissolved solutes. It is likely that interactions between colloid size and properties of the medium, such as fracture aperture width and matrix permeability, will have a profound influence on colloid transport and thus continued progress on the fracture flow-and-transport problem will rely on additional observations that employ approaches similar to those outlined by Zvikelsky and Weisbrod [2006].

[16] Under the right conditions, colloids facilitate the transport of contaminants [McCarthy and Zachara, 1989; Kretzschmar et al., 1999; Elimelech and Ryan, 2002]. If there are sufficient mobile colloids and the contaminants bind strongly to the colloids, then colloid-facilitated transport may play an important role in migration of contaminants. In this special section, two papers focus on colloid-facilitated transport [Pang and Šimůnek, 2006; Turner et al., 2006].

[17] Pang and Šimůnek [2006] model the facilitated transport of cadmium by bacteria in a column containing alluvial gravels, a phenomenon first observed by one of the earliest papers addressing colloid-facilitated transport [Champ and Merritt, 1981]. The Pang and Šimůnek [2006] model accounts for advection, dispersion, cadmium sorption to the gravel and to Bacillus subtilis spores and Escherichia coli cells, bacteria attachment to the gravel, and velocity enhancement of the bacteria. Comparison of the laboratory measurements and model calculations reveals that cadmium adsorption to the bacteria differed by bacteria type and between attached and mobile bacteria. These findings bring to light the formidable challenge associated with parameterizing models that describe the coupled processes governing colloid-facilitated contaminant transport.

[18] Several researchers suggest the strength of the colloid effect on contaminant transport is very sensitive to the kinetics of colloid-contaminant interactions [van de Weerd et al., 1998; Roy and Dzombak, 1998; Bold et al., 2003]. Turner et al. [2006] focus on adsorption and desorption kinetics in a study that integrates mathematical modeling with laboratory experiments on the cotransport of cesium and strontium by illite colloids. The authors observe that desorption of cesium from illite was much slower than desorption of strontium from the clay colloids, presumably because cesium binds strongly to frayed edge sites on illite and strontium does not. Owing to the comparatively slow desorption of cesium from the illite, the colloid effect on cesium transport through the sand columns was greater than the colloid effect on strontium transport. This contrast emphasizes the need to consider the kinetics of desorption when assessing the potential effects of colloids on contaminant transport.

[19] Solving some of the most challenging groundwater quality problems of our time, including the disposal of radioactive waste and the contamination of water supplies by pathogenic microbes, depends on better understanding of colloid transport. The contributions in this special section describe some of the complexities of colloid-transport processes in saturated and unsaturated media and some of the visualization techniques being developed to provide new insight into these processes. We hope that this special section provides readers a useful update on the current state of research on colloid transport, an awareness of the issues still presenting difficulties in understanding and predicting colloid transport, and motivation to find new ways to explore colloid transport.

Acknowledgments

[20] We thank Thomas Torgersen for his recommendations for improving this manuscript.

Ancillary