Environ. Sci. Technol. 2001, 35, 2497-2504
Physical and Chemical Determinants of Colloid Transport and Deposition in Water-Unsaturated Sand and Yucca Mountain Tuff Material AMY P. GAMERDINGER* Department of Chemistry and Center for Multiphase Environmental Research, Washington State University, Tri-Cities, Richland, Washington 99352, Applied Geology and Geochemistry, Pacific Northwest National Laboratory, P.O. Box 999, K6-81, Richland, WA 99352 DANIEL I. KAPLAN Savannah River Technology Center, Westinghouse Savannah River Company, Building 773-43a, Room 215, Aiken, South Carolina 29808
Colloid mobility was determined in a system consisting of quartz sand or crushed Yucca Mountain tuff, simulated groundwater (J-13), and hydrophilic latex particles. Water content (θ) and ionic strength (I; DI water, 0.1×, 1×, 10× groundwater dilution) were manipulated to define limiting conditions for colloid transport at the Yucca Mountain site. Colloid transport, measured with a centrifuge method at relatively high θ (saturation >36% for sand, >62% for crushed tuff) in DI water, was equivalent to transport at 100% saturation measured with conventional columns. When variables were isolated, increasing I and decreasing θ resulted in a greater extent of colloid deposition; I was more important at higher θ; physical properties were more important at lower θ. I and θ had an interactive effect on colloid deposition whereby synergism was generally observed, especially for simulated groundwater (1×); antagonism was observed at 10× groundwater dilution. At 19% moisture saturation on the crushed tuff, a decreasing rate of colloid deposition was observed. This corresponded to a hydrodynamic condition of 79% immobile water where solute tracers were excluded from a fraction of the pore volume. This suggests that a portion of the favorable sites for deposition were associated with the excluded or immobile water domain and were not accessible to colloids.
Introduction A major pathway for radionuclide transport to groundwater is through the vadose zone. While the significance in nature is not always clear (1, 2), the potential for colloids to enhance subsurface contaminant transport is well-established (1) and is considered in the design of nuclear waste repositories (3). Issues related to the potential for colloid-facilitated transport of radionuclides at the Yucca Mountain site, the proposed Federal nuclear waste repository, have been extensively reviewed (4, 5). Among the key factors likely to influence the mobility of colloids at the Yucca Mountain site are plume * Corresponding author phone: (509)372-7345; fax: (509)372-7471; e-mail:
[email protected]. 10.1021/es001434x CCC: $20.00 Published on Web 05/10/2001
2001 American Chemical Society
chemistry, specifically ionic strength (I), and the degree of moisture saturation. Several studies have shown a positive correlation between colloid deposition and I (6-11). A limited number of investigations have considered the transport of colloids under conditions of partial water saturation, common at arid region repository sites. These latter studies have considered the influence of flow rate (12, 13) colloid surface properties (14-16), colloid size (13, 16), degree of water unsaturation (13), and pore size (13) on colloid mobility through partially saturated porous systems. The primary objective of this research was to determine the effect of moisture saturation on colloid mobility and deposition in crushed tuff material from Yucca Mountain. Throughout the paper “tuff” refers to crushed tuff material. Particular attention was directed at evaluating the effect on colloid transport of an immobile water domain, which increases in unsaturated systems (17). Immobile water may exist in thin liquid films around soil particles, in dead-end pores, or as relatively isolated regions associated with unsaturated flow (18). Transport in systems with immobile water is described by a “two-region” model where advective transport is limited to the mobile water domain, and transport between the mobile and immobile water domains is diffusionlimited. The implications of mobile-immobile water on colloid transport have not been investigated. For practical reasons, this research was conducted in a porous matrix; the importance of fracture flow at Yucca Mountain is recognized but was beyond the scope of this investigation. Throughout this paper, the term “unsaturated” refers to water-unsaturated sediments. Theory and Modeling. The importance of I and the theoretical basis for colloid deposition during miscible displacement experiments to determine breakthrough curves [BTCs (which depict the dimensionless effluent concentration, C/Co, as a function of cumulative pore volumes eluted)] are addressed in detail elsewhere (11 and references therein). When examining reactive processes during transport, conservative tracers are used to independently determine hydrodynamic processes. Regions of immobile or stagnant water develop with decreasing water content and can be described with a two-region transport model (19, 20). The continuum between a single and a two-region flow domain, governing equations, and approach for modeling two-region transport are discussed in detail elsewhere (17). For BTCs where the conventional single-flow domain form of the convection-dispersion equation (CDE) was adequate for characterizing the BTC, dispersion (D) was determined by a single-parameter curve fit. At lower water contents, when the conventional CDE failed to capture the early breakthrough and tailing phenomena that is characteristic of mobileimmobile water, a two-region model was applied to describe the BTC. This corresponded to cases where the Peclet number determined with the CDE was less than 2 and the fraction of immobile water determined with the two-region model was greater than 10%. The dimensionless parameters (β and ω) were determined by curve-fitting. β defines the fraction of mobile water (m) (β ) θm/θ, where θ is the volumetric water content); ω is the coefficient for mass transfer between mobile and immobile water domains [ω ) RL/q, where R is the mass transfer coefficient (h-1), L is the column length, and q is the water flux]. For these cases, the Peclet number [P (P ) vL/D, where v is the average pore water velocity] was calculated directly from the slope of the BTC where C/Co was equal to 0.5 (21). Due to the low water content, small pore volume, and early breakthrough, the initial points of some BTCs are at C/Co > 0.5, requiring extrapolation through the VOL. 35, NO. 12, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2497
origin. This can result in low estimates of P that approach the theoretical minimum of 2 (maximum dispersion, as determined using the Einstein-Smoluchowski equation); this bias decreases the estimated fraction of immobile water. Colloid BTCs were modeled to determine pseudo-firstorder rate constants for deposition (µ) (6, 10, 13), which can be readily incorporated into larger scale transport models that are used for performance assessments such as that for Yucca Mountain. When determining colloid deposition, hydrodynamic parameters were fixed to the values determined in separate conservative tracer experiments on the same column for similar flow conditions. Curve-fitting was accomplished using the CXTFIT2 program (22). Wan and Tokunaga (13) presented a film-straining theory highlighting two related factors, water discontinuity (specifically pendular rings) and film thickness, as determinants of colloid deposition in unsaturated sediments. Colloids are expected to be mobile at water contents greater than the critical saturation value (pendular rings are connected) and water film thickness greater than the colloid diameter. Theoretical determinations of critical values for water content and film thickness are based on assumptions of ideal particle geometries, packing, and surface properties. These simplifying assumptions are compromised as the matrix material geometry deviates from the ideal. Because of surface roughness, irregular shape, and internal porosity of the tuff, theoretical determinations of film thickness are not expected to be representative of the tuff (discussed further below). However, application of film-straining theory to the smoother surfaced and rounder sand particles is more appropriate and provides a useful conceptual model for evaluating the effect of water content on physical properties of the sediment related to water continuity and film thickness.
Experimental Section Materials. The solutions used in these experiments were deionized (DI) water and simulated J-13 groundwater (0.368 mM Na2CO3 and 10.6 mM NaHCO3, pH 7.8, I ) 1.16 × 10-2, termed 1×) at 0× (DI water), 0.1×, 1×, and 10× dilutions. J-13 designates a groundwater well at the Yucca Mountain site. The immobile solid matrix consisted of either quartz sand (0.590-0.850 mm; Accusand, Unimin Corp.) or crushed Yucca Mountain tuff (0.250-0.841 mm) collected from an outcropping of the Topopah Spring member of the Paintbrush Tuff (likely of the Tptp mn lithophysal). The colloids used in these experiments were fluorescent carboxyl-modified polystyrene latex (CML) microspheres that had a 0.280-µm diameter; the rationale for selection is described elsewhere (11). Average sand and tuff particle size was determined from approximately 20 SEM images of matrix particles. Specific surface area of sand and tuff was determined by the BET method using Kr and N2, respectively (ASAP 2010, Micromeritics Instrument Corporation, Norcross, GA). The initial colloid and bromide concentrations (Co) for transport experiments were ∼4.5 and ∼5000 mg/L, respectively. Bromide, used as a conservative tracer, and colloid BTC experiments were conducted separately. Additional details on materials and analytical methods are provided elsewhere (11). Transport Method. Colloid and tracer transport in unsaturated sediments were determined using a centrifuge method; the experimental system and method are described in detail elsewhere (17). A brief description of the approach follows and is similar to that used for saturated column (miscible displacement) experiments. Columns [length (L) ) 6.0 cm; radius (r) ) 2.25 cm; bulk volume (Vbulk) 95.43 cm3] had mesh at the inflow and outflow and were packed with sand or tuff. On the basis of previous studies (17), compaction resulting from centrifugal force was not a concern for the course-textured sand and tuff materials used here. Columns 2498
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 12, 2001
were saturated with s-J-13 solution; sediment bulk density [F (g cm-3)] and volumetric water content [θ (cm3 cm-3)] were determined from the mass of the sediment and/or water. The percent moisture saturation was calculated from the ratio of θ (water-filled porosity) to the total porosity (φ), which was calculated from the bulk density and particle density. An average water film thickness (w) for the sand was estimated from the bulk density, specific surface area (As), and average volumetric water content (w ) θ/[AsF]). This provides an average thickness and does not account for the presence of pendular rings. With the centrifuge method (model L8-UFA, Beckman Coulter, Inc., Fullerton, CA), water content is influenced by centrifugal force and the fluid flux [q (cm h-1)]. Columns were first saturated and then placed in the centrifuge. Unsaturated transport experiments were initiated when the columns had reached a steady-state average water content. Experiments at each ionic strength were executed consecutively (increasing I) on a column for a particular water content; a new column was used for each series. It was shown previously that conservative tracer and colloid transport were not affected by prior experiments on the same column (11). The most common effect of continued colloid application is “filter ripening” or an increase in colloid removal with time. This would result in a negative slope of the BTC after initial breakthrough and was not observed. Tracers and colloids were applied as step inputs where the pulse length (to) is defined as the volume of the step input, expressed as the number of column pore volumes. Practical considerations resulted in some variation in to. A pore volume (Vw) is defined as the volume of water retained within the pore space of the sediment matrix. Effluent was collected manually and analyzed for colloid or bromide concentration. For each effluent sampling, rotation of the centrifuge and flow were stopped. Prior evaluation indicates no significant effect of stopping flow on the transport of nonretarded, conservative, solute tracers (17).
Results and Discussion Particle Characterization. SEM images of representative sand and tuff particles are shown in Figure 1. The primary morphological differences are rounded edges for the sand and a coarse surface for the tuff. The average width and length of the sand particles was 710 ( 106 and 723 ( 198 µm (n ) 20); that for the tuff was 634 ( 134 and 514 ( 117 µm (n ) 20). Specific surface areas for the sand and tuff were 0.075 and 1.18 m2 g-1, respectively. The higher value for the tuff reflects surface roughness and includes internal porosity, where 43% of the total surface area was associated with pores of 0× groundwater dilution (Figures 4B, 5B, and 6B; T-64-0.1, T-63-1, and T-62-10; Table 2), deposition was greater than observed in saturated sediments (Table 3 columns T,100 and T,62-66), and deposition was greater than the sum of the isolated contributions of water content and ionic strength (Table 3 and Figure 7). While this effect was more variable for the sand, results for the tuff consistently suggest a synergistic effect between ionic strength and film straining. The more pronounced effect of physical properties on colloid deposition on the tuff can be attributed to the greater surface area and surface roughness. Rather than a flat plateau, BTCs for 1× and 10× at ∼63% moisture saturation (Figures 5B and 6B) show a gradual increase in the colloid concentration after initial breakthrough. As noted previously for 0× and 19% saturation on the tuff, this BTC characteristic indicates a decreasing rate of deposition. The absence of immobile water (determined in conservative tracer experiments) at this water content range (62-66% saturation) VOL. 35, NO. 12, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2503
suggests that this is due to ionic strength or simply a decreasing availability of favorable sites as the experiment progresses. This BTC characteristic is also apparent for the sand at 10× (Figure 6A). At the lower water content range of 17-19% saturation on the tuff, deposition at 0.1× and 1× again is greater than expected based on the isolated contributions of ionic strength and surface properties. As noted for the sand, deposition at 10× on the tuff is less than at the higher water content and less than predicted by the sum of the isolated contributions of ionic strength and film straining. There is more variability in the BTCs and retardation factors at lower water contents where estimated rate constants for deposition are higher (Table 2). This is illustrated in Figure 4B for T-19-0.1; the decreasing rate of deposition that was noted previously for 0× (Figure 3B, T-19-0) is distinguishable for 10× (Figure 6B, T-17-0) but not 0.1× or 1× (Figure 5B, T-19-1). The results of this research have several implications for understanding colloid transport and deposition in water unsaturated sediments. Close agreement between results at low ionic strength for 100% saturation and the higher water content ranges investigated here (saturation >36% for sand and >62% for tuff) suggests that colloid transport at high unsaturated water contents and low ionic strength might be predicted from behavior in saturated systems. At these higher water contents, the results were in agreement with DLVO theory that predicts greater colloid deposition with increasing ionic strength. At low ionic strength, the results were in agreement with film straining theory that predicts greater colloid deposition with lower water content. When simultaneously considering the contributions of physical and chemical properties to colloid deposition, there were deviations from trends that were established when each variable was considered in isolation. A simple relationship for the dependence of colloid deposition on ionic strength that was developed for saturated conditions did not hold for unsaturated conditions. At the ionic strength approximating groundwater at Yucca Mountain (1× J-13, I ) 1.16 × 10-2) and several water content ranges, the combined effects of physical and chemical processes were synergistic, resulting in greater colloid deposition (Table 3 and Figure 7). At the maximum ionic strength and minimum water content conditions that were investigated, the combined effects were antagonistic, resulting in greater colloid mobility. Results for 0× groundwater dilution reflect deposition on a clean bed. The observed decreasing rate of deposition during the experiment at 19% saturation on the tuff (Figure 3B, T-19-0) corresponded to a significant fraction of immobile water where conservative tracers were excluded from a fraction of the water-filled pore volume. This suggests that the change in water distribution and hydrodynamics associated with two-region flow at this low water content decreased colloid accessibility to favorable sites for deposition. Furthermore, it suggests that these sites were associated with the immobile water regime.
Acknowledgments This research was supported by EPRI; the authors greatly appreciate the technical support of the project manager, Dr. John Kessler. Student assistants, Kathleen A. David, Jonathan R. Ferris, and Lauren J. Webb assisted with sample analysis.
2504
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 12, 2001
Kent E. Parker, PNNL, performed the surface area analysis. A major portion of this research was conducted with the Applied Geology and Geochemistry group at Pacific Northwest National Laboratory, which is operated for the U.S. DOE by Battelle Memorial Institute, under Contract DE-AC0676RLO 1831. Westinghouse Savannah River Co. is operated for the U.S. DOE under Contract DE-AC09-89SR-18035.
Literature Cited (1) McCarthy, J. F.; Zachara, J. M. Environ. Sci. Technol. 1989, 23, 496-502. (2) Honeyman, B. D. Nature 1999, 397, 23-24. (3) Ramsay, J. D. F. Radiochim. Acta 1988, 44/45, 165-170. (4) Triay, I.; Simmons, A.; Levy, S.; Nelson, S.; Nuttall, H.; Robinson, B.; Steinkampf, W.; Viani, B. Colloid-Facilitated Radionuclide Transport at Yucca Mountain; LA-12779-MS; Los Alamos National Laboratory: Los Alamos, NM, 1995. (5) Bennett, D. G.; Crowford, M. G.; Wickham, S. M. Colloids in Saturated and Partially-Saturated Porous Media: Approaches to the Treatment of Colloids in Yucca Mountain Total System Performance Assessment; TR-112135; EPRI: Palo Alto, CA, 1999. (6) Tobiason, J. E. Colloids Surf. 1989, 39, 53-77. (7) Chang, J. S.; Vigneswaran, S. Water Res. 1990, 24, 1425-1430. (8) Elimelech, M.; O’Melia, C. R. Environ. Sci. Technol. 1990, 24, 1528-1536. (9) McDowell-Boyer, L. Environ. Sci. Technol. 1992, 26, 586-593. (10) Saiers, J. E.; Hornberger, G. M. Water Resour. Res. 1999, 35, 1713-1727. (11) Gamerdinger, A. P.; Kaplan, D. I. Environ. Sci. Technol. (in press). (12) Goldenberg, L. C.; Hutcheon, I.; Wardlaw, N. Transp. Porous Media 1989, 4, 129-145. (13) Wan, J.; Tokunaga, T. Environ. Sci. Technol. 1997, 31, 24132420. (14) Wan, J.; Wilson, J. L. Water Resour. Res. 1994, 30, 11-23. (15) Wan, J.; Wilson, J. L. Water Resour. Res. 1994, 30, 857-864. (16) McGraw, M. Ph.D. Dissertation, University of California, Berkeley, CA, 1996. (17) Gamerdinger, A. P.; Kaplan, D. I. Water Resour. Res. 2000, 36, 1747-1755. (18) Nielsen. D. R.; van Genuchten, M. Th.; Biggar, J. W. Water Resour. Res. 1986, 22, 89S-108S. (19) van Genuchten, M. Th.; Wierenga, P. J. Soil Sci. Soc. Am. J., 1976, 40, 473-480. (20) Griffioen, J. W.; Barry, D. A.; Parlange, J.-Y. Water Resour. Res. 1998, 34, 373-384. (21) van Genuchten, M. Th.; Wierenga, P. J. In Methods of Soil Analysis, Part I; Klute, A., Ed.; Soil Science Society of America and American Society of Agronomy: Madison, WI, 1986; pp 1025-1054. (22) Toride, N.; Leij, F. J.; van Genuchten, M. Th. The CXTFIT Code for Estimating Transport Parameters from Laboratory or Field Tracer Experiments, Version 2.1; Research Report 137; U.S. Salinity Laboratory, Agricultural Research Service, U.S. Department of Agriculture: Riverside, CA, 1995. (23) Valocchi, A. L. Water Resour. Res. 1985, 21, 808-820. (24) Rao, P. S. C.; Jessup, R. E.; Rolston, D. E.; Davidson, J. M.; Kilcrease, D. P. Soil Sci. Soc. Am. J. 1980, 44, 684-688. (25) Rao, P. S. C.; Rolston, D. E.; Jessup, R. E.; Davidson, J. M. Soil Sci. Soc. Am. J. 1980, 44, 1139-1146. (26) Liu, D.; Johnson, P. R.; Elimelech, M. Environ. Sci. Technol. 1995, 29, 2963-2973. (27) Johnson, P. R.; Sun, N.; Elimelech, M. Environ. Sci. Technol. 1996, 30, 3284-3293.
Received for review June 29, 2000. Revised manuscript received December 11, 2000. Accepted December 12, 2000. ES001434X
