Dehydration of Flocs by Freezing - Environmental Science

A conceptual model was developed to explain the freeze/thaw conditioning process. This process transforms difficult-to-dewater residuals into a slurry...
0 downloads 0 Views 128KB Size
Environ. Sci. Technol. 1999, 33, 482-488

Dehydration of Flocs by Freezing PHILIP J. PARKER* Civil & Environmental Engineering Department, University of WisconsinsPlatteville, 1 University Plaza, Platteville, Wisconsin 53818 ANTHONY G. COLLINS Department of Engineering, Box 5700, Clarkson University, Potsdam, New York 13699-5700

A conceptual model was developed to explain the freeze/thaw conditioning process. This process transforms difficult-to-dewater residuals into a slurry of granular particles (zots) from which water readily drains. The conceptual model states that the most important effect of freezing and thawing is the removal of surface water from the colloidal particles comprising the floc. In addition to creating a drier floc, this dehydration may also explain the irreversibility of the process: dehydration causes the particles to come into close contact, at which point a new bond forms. Formation of a new bond explains the strength and granularity of the zots.

Introduction Water treatment sludge, or residuals, is a mixture of water and dissolved and colloidal solids. Flocs are aggregates of colloidal particles that are held together by van der Waals forces or polymer bridges. These relatively weak bonds result in a floc with a loose, open structure. Such a structure contains large quantities of bound water since it provides a large surface area for water to adhere to and creates many interstices that retain water. Residuals are generally dewatered prior to disposal, but the success of dewatering is limited since removing the bound water by mechanical dewatering is often impossible. Prior to dewatering, residuals may be conditioned by chemical addition or heating to release some of the bound water. Freezing and thawing also effectively conditions residuals, transforming the flocs into granular particles (zots) (1-12). The term zots was coined by Zolotavin et al. (13), and we use it as a tribute to their thorough work during the 1960s (14-20). Zots look and feel like coffee grounds, are stronger than the original flocs, and do not rehydrate or revert to their original structure upon agitation; upon thawing, water readily drains from the conditioned residuals. Recently Parker et al. (6) showed that, after freezing and thawing, cake solids contents (%) doubled and the “filterability coefficient” (χ; 21), increased by over 2 orders of magnitude for an alum residual. This paper’s objectives are to (i) present a conceptual model explaining the mechanisms responsible for the success of freeze/thaw conditioning; (ii) present experimental results that support the model; and (iii) use some simple mathematical modeling to illustrate and verify portions of the conceptual model. * Corresponding author phone: (608)342-1235; fax: (608)342-1566; e-mail: [email protected]. 482

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 3, 1999

Background Bound Water and Freeze/Thaw Conditioning. The nature of water associated with solid surfaces has received much attention in the literature; Pyper (22) and Vesilind (23) have written thorough reviews. Vesilind’s emphasis is on wastewater sludge, and he notes a paucity of information on water in complex mixtures such as sludge. Much of the current understanding is based on investigations of water associated with a variety of surfaces, including soil particles (24-26), wood fibers (27), gels (28-30), proteins (31, 32), and flocs (33-35). As a result, the water associated with solids has been described with a variety of names: water of adhesion, adsorbed, absorbed, bonded, bound, capillary, chemically bound, water of cohesion, combined, coordinated, water of decomposition, film, free, H3O+, hydrogen-bonded, hygroscopic, immobilized, inactive, interfacial, interstitial, lattice, liquid inclusion, osmotically bound, pendular, physically bound, pore angle, water of solvation, stagnant, surface, unfree, vicinal, water of hydration, water of osmotic imbibition, and zeolite water (22, 23, 26, 30, 36-41). The schematic in Figure 1 is based on a classification system proposed by Vesilind (23). Flocs are shown in Figure 1a and are surrounded by free water, which is readily removed by mechanical methods. Figure 1b is a magnified view of a portion of a floc. The floc is composed of individual colloidal particles. The particles within flocs are held together by electrostatics, polymeric bridging, or van der Waals interactions; these are collectively represented by the short dark lines and labeled weak bonds in Figure 1b. Between each particle is interstitial water, which is not attached to a particle surface but is physically confined by the surrounding particles. Interstitial water can be removed from the sludge if the physical confinement is removed, such as by pressure filtration for instance. Several types of water are associated with particle surfaces, and we refer to them collectively as surface water. Surface water differs from vicinal water (23). Surface water is a type of vicinal water found only on the exterior of surfaces, while vicinal water can be attached to any surface, either on the interior or exterior of colloidal particles. Crystalline water is bound within the chemical structure of the particles and is not shown on the diagram. One type of surface water is water of solvation. The solvation layer is the structurally altered liquid layer on colloid surfaces and may be termed the “hydration layer” when the liquid is water (42-45). Researchers have measured the thickness of a structurally altered water layer on surfaces other than sludge particles. Results from Etzler and Fagundus (46) indicated that this water formed a layer between 3 and 5 nm thick on silica gels. The water bound within the amorphous hydroxide precipitate can also be considered surface water since the precipitate is attached to the surface. Although the precipitate may not be physically arranged in uniform layers on the colloid surface like water of hydration, the precipitate can still be envisioned as being equivalent to a thickness of surface water. It is instructive to consider that a slurry of natural colloidal material will dewater to a relatively high solids content despite the fact that it contains some interstices, hydrated ions, and water of solvation. Differences are noted in dewaterability only when amorphous metal hydroxides precipitate in such a slurry. Thus, the major source of bound water within coagulant residuals is associated with the hydroxide precipitate. 10.1021/es980705p CCC: $18.00

 1999 American Chemical Society Published on Web 12/29/1998

FIGURE 1. Floc schematic.

Conceptual Model General. We hypothesize that the improvement in dewaterability by freeze/thaw conditioning can largely be explained by a decrease in surface water; thus freeze/thaw is a dehydration process (11, 12). Similarly, Chen et al. (47) suggested that the water removed by freezing may include vicinal or surface water. In order for frozen and thawed residuals to contain less surface water, freezing must either decrease the surface area of the flocs or decrease the thickness of the surface water layer. We assume that the latter does not occur since the thickness of the layer should primarily depend on surface characteristics, and preliminary unpublished experiments have shown that surface charge and crystalline structure do not change after freezing: the ζ-potential of ferric and aluminum hydroxide precipitates was unaffected, and X-ray diffraction of the same precipitates did not indicate any change in crystallinity as a result of freezing. Therefore, the removal of surface water by freezing and thawing is due primarily to the elimination of surface area. It is important to emphasize that surface water in unfrozen residuals is not only found on the outer surface of the flocs but also on the surface of each colloidal particle. However, after freezing, we hypothesize that surface water is found primarily on the outer surface of zots. To explain this, it is helpful to first review some fundamentals of coagulation. Two colloidal particles are stable if they do not aggregate. Stability occurs when the particle surface attractive forces are less powerful than the repulsive forces. In colloidal systems, attractive forces arise from van der Waals interactions and polymer bridging, and the repulsive forces are provided by electrostatics, steric interactions, and hydration effects (48). When colloids are destabilized, they come into particle-particle contact and are said to have coagulated; they are now part of an aggregate or floc. However, particleparticle contact is often a misnomer: colloid surfaces may not actually be touching in a destabilized aggregate but are separated by the water of hydration and other types of surface water (43, 46). Thus, colloidal particles that have aggregated can still contain large amounts of surface water. As a result of freeze/thaw conditioning, particles move closer to their neighbors, and the intervening water layer may be nonexistent (in the case of true particle-particle contact) or negligibly thin. Those areas of the surface that contact another surface no longer bear water since a new bond is formed (discussed in the following section). Therefore,

after freezing, surface water will primarily be found on the surface of zots. Proposed Mechanism: Cryosuction. A possible mechanism by which surface water is removed is cryosuction (4956). The formation of ice lenses in frozen soils and the resulting frost heave are evidence of cryosuction (49, 51, 57, 58). If a closed container of soil is frozen directionally downward, negligible heave occurs. However, if the bottom of the container is connected to a supply of water, heaving occurs. Heaving is due to the movement of the water upward through the soil to a horizontal layer of ice called an ice lens. [Work by Taber (59) and Williams and Smith (55) contain excellent photographs of ice lenses and the effects of heaving.] Williams and Smith (55) explain that the transport of unfrozen water in soil will occur along free energy gradients. Unfrozen water adjacent to the ice will have a relatively low free energy due to the effects of capillarity and adsorption; thus water farther away from the ice surface will tend to migrate toward the ice. Moreover, as the temperature of the system lowers, the free energy of the remaining unfrozen water will decrease further, creating larger suctions. Unfortunately, neither the suction forces during freezing of residuals or the hydration forces between natural colloids have been measured. If such measurements were available, the suction pressures could be compared to the hydration forces, and the extent of the cryosuction effect could be directly verified. The best estimate available is to compare measured solvation forces and cryosuction values, albeit for different materials, as outlined in Table 1. Table 1 shows that measurements of the suction exerted by freezing are much greater than those measured for solvation forces; this fact supports, but does not prove, the hypothesis that dehydration occurs via cryosuction. Very simple particle/solvent systems are used for measurement of solvation forces and are much different from a complex system such as sludge. The change in floc density and structure noted following freezing (2-4) also supports the fact that freezing initiates dehydration by bringing the floc particles into particleparticle contact. Zots are denser then flocs, probably because the space occupied by water has been replaced by a solid with a higher density. New Bond Formed. We propose that relatively strong bonds form between the dehydrated surfaces in contrast to the weak bonds shown in Figure 1. These bonds form between particles whose surfaces are touching or perhaps separated VOL. 33, NO. 3, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

483

TABLE 1. Comparison between Solvation and Cryosuction Pressures terminology

suction pore water suction suction at the frost front pore water pressure solvation force per unit radius, F/R F*, dimensionless solvation force F /R F /R a

ref

magnitude of suction

Experimentally Measured Cryosuction Values 76 10 kg/cm2 (986 kPa) 95 500 kPa 71 6.2-11.0 kPa 96 30-55 kPa Experimentally Measured Solvation Forces 65 20 mN/ma (2.4 kPa)b 64 20 63 106 µN/m (80 kPa)b 97 104 µN/m (0.8 kPa)b

Force is oscillatory; the maximum value is reported in the table.

by a very thin layer of water. Particles in flocs are held together by hydrogen bonds or van der Waals interactions, and a liquid layer that cannot withstand shear exists between the particles. We propose that colloidal particles in zots are in true particleparticle contact and that no water layer exists between them, or that particles are close enough such that van der Waals interactions can dominate. Although the exact nature of the new bond formed between surfaces is not known, there is some evidence for its existence: flocs are weak and loosely bound, while zots are much stronger; flocs are easily broken apart by agitation, while zots retain their granular, particulate nature even after extensive agitation. Both the newfound strength and the irreversibility indicate that a new bond has formed within the floc. However, the outer surface of the floc does not form bonds with any other surface, and thus the outer surface is proposed to contain the same quantity of surface water as before freezing. A similar phenomenon may develop strength in soil aggregates after freezing. Hohmann (49) notes that suction induced by freezing pulls particles together to produce a denser soil, and cryogenical consolidation or a volume decrease in the soil solids results; moreover, the process is partly irreversible since the soil does not expand to its original volume upon thawing. Blank (60) subjected soil colloidal material to freeze/thaw cycles, which resulted in the formation of silt- and sand-sized particles. Also, Williams and Smith (55) state that the volume reduction of clays after freezing and thawing is similar to the reduction upon drying. Zot formation and cryogenical consolidation are similar to aggregation of soils that occurs during drying. Indeed, during drying, soils develop strength (61). Unfortunately, the mechanisms by which soil colloids are held together are not well understood (62); the phenomenon has been named however: interaggregate bridges (63), cementing agents (64), or cementing bonds (65). Mielke et al. (66), Mullins et al. (67), and Dexter (65) suggested that stronger bonds replaced the water between soil particles, and therefore the strength of soil aggregates was a function of water content. Additional Mechanisms. In addition to dehydration, other processes may be occurring during freezing. As water freezes, it attempts to reject all impurities in its path. Thus, flocs can be rejected by the ice/water interface, most noticeably at low freezing rates and low initial solids content. The solids are said to migrate, and the unfrozen residuals are thickened (1, 6, 8). Upon entrapment, thickened residuals are entrapped as larger agglomerates of several flocs. The total surface area of a collection of particles will decrease as the particles comprise larger and larger zots. Thus, thickening and the concomitant formation of larger zots will enhance the effects of freezing and thawing. Parker et al. (5) showed that, when an ice/water interface interacted with flocs, it could fragment the flocs. This behavior was most noticeable at very high freezing rates. In contrast 484

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 3, 1999

b

system

soil/water soil/water soil/water soil/water mica surface and [(CH3)2SiO]4 model results mica in aqueous KCl (10-3 M) muscovite in aqueous KBr (0.5 M)

Calculated assuming R ) 1 µm.

to the beneficial effects of thickening, this mechanism will create smaller zots, perhaps consisting of tens of particles. By the same reasoning as stated previously, production of smaller zots will lessen the effectiveness of the process.

Experimental Study Procedure. An experimental procedure was designed to test our primary hypothesis: the principle difference between frozen and unfrozen residuals is the quantity of surface area. Four water treatment residuals were frozen under different conditions, producing zots with varying surface areas. The quantity of surface water was measured for each sample and correlated to the surface area. Sludge samples came from Lake Gaillard (CT), Potsdam (NY), Shenango (PA), and Swimming River (NJ). These residuals are characterized by Parker et al. (9). Samples were frozen in a manner similar to that described previously (6) at three freezing rates (5, 15, and 150 mm/h), three initial solids contents (1%, 3%, and 10%), and cured for 96 h at -10 °C. The fastest freezing rates were obtained by freezing the samples on liquid nitrogen and correspond to ultra-rapid freezing rates (7). These freezing conditions created a wide range of zot sizes. Both frozen and thawed and unfrozen samples were analyzed for particle size and surface area using a Malvern Mastersizer/E laser particle size analyzer. The Mastersizer uses Fourier optics to calculate the particle size distribution using the Fraunhafer diffraction principles. Since we believe the surface water completely covers the particles, it was important to ensure that the Mastersizer was analyzing individual particles. Therefore, unconditioned residuals were broken down by stirring or sonication so that the primary particle size could be measured. Frozen samples were stirred but not sonicated, since sonication destroyed the zots and reverted them to their original form. Several methods are utilized in the literature to determine bound water content, but as Knocke and Trahern (2) state, the measurement is inherently operationally defined. Bound water content was determined in this study by placing 10 g of residuals on a Gelman GN-6 filter paper (0.45 µm openings) and exerting a suction of 38 cmHg. Thus, most of the colloidal material and its attached surface water remained on the filter paper. The filter cake was then analyzed for water content by drying at 103 °C, which was assumed to remove bound water. We further assumed that the water of hydration content remained the same before and after freezing; thus differences in water content were due only to changes in the quantity of surface water and interstitial water. Discussion of Experimental Results. Figure 2 shows that the experimental procedure was successful in creating a wide range of surface areas on which to test our hypothesis. The surface area of the unfrozen residuals is enclosed in

FIGURE 3. Effect of surface area on bound water content.

Surface Area Model

FIGURE 2. Effect of freezing speed on surface area. parentheses following the name of the residuals. For each of the residuals tested, an increase in surface area is noted as the freezing rate increases. The variation in bound water content as a function of unit surface area is shown in Figure 3. The results support our hypothesis: bound water content is strongly correlated with surface area for all residuals tested. As total surface area decreases, the bound water content decreases. This suggests that the changes in bound water content are primarily due to changes in water associated with the surface. These results also support our hypothesis that increasing freezing rates decrease the quantity of bound water in zots: Figure 2 shows that increasing rate generally yields the most surface area, and Figure 3 illustrates that the most bound water is associated with the most total surface area. Moreover, the residuals frozen at high rates have more surface area and contain more bound water than residuals frozen at lower rates. This and other principles are illustrated in the model discussed below.

We have developed a surface area model to illustrate how the reduction in surface area can explain the success of freeze/ thaw conditioning. The model floc was defined as a monodisperse collection of spherical particles with a diameter of 10 µm, a void ratio of 0.54 (68), and surface roughness corresponding to 25 m2/g. The conceptual model proposes that bound water is found in unfrozen flocs in two locations: on the surface of each individual particle and in the interstices between particles. This is illustrated in Figure 4. Residuals containing only bound water are assumed to act as a continuum of particles: i.e., flocs are not distinct entities. The bound water content of the unfrozen residuals can be expressed as

Wc,o ) {mass of water in interstices + mass of water on colloid surface}/mass of solids 4π F n e (r [ 3 ) w p

)

[ 4π3 (r

Fw e

p

p

]

+ do)3 + [FwnpApdo] npVpFs

+ do)3 + Apdo

]

(1)

VpFs

where Wc,o is the water content before freezing and thawing (g/g); Fw is the density of water, assumed to be equal to bulk VOL. 33, NO. 3, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

485

FIGURE 4. Effect of freezing on surface water. water density (106g/m3); Fs is the density of solids (2.5 × 106 g/m3); np is the total number of particles (colloids) in the system (5 × 107); e is the void ratio (0.54); rp is the radius of particle (5 × 10-6 m); do is the thickness of surface water layer in unfrozen residuals (m); Ap is the surface area of particles ) 4πr2pR (m2); Vp is the volume of particle (solids only) ) (4πr3p)/3; and R is the roughness coefficient (a value of 100 yielded a surface area of 25 m2/g for the model system, which is typical for natural colloids). The value of np was arbitrary and is only used to define the size of the system. Equation 1 was solved for do (using Mathcad) given a value of Wc,o. Wc,o was set equal to 10 g/g; this is a typical experimental value obtained for the residuals in this paper. Given these values, the surface water thickness on particles before freezing, do, was calculated to be 0.41 µm. This value can be interpreted as an average thickness of a layer of water of hydration and a layer of the water-binding amorphous hydroxide precipitate. Our conceptual model proposed that the surfaces of the particles within a zot are extensively dehydrated, while the outer surface of the zots are not dehydrated since no new bond is formed at that surface (Figure 4). The extent and nature of particle surface dehydration has been taken into account by modeling the remaining water as a continuous, uniform layer of thickness df. The thickness of the surface water layer on the outer zot surface is assumed to be equal to the surface water layer thickness on the original particles, namely do. Therefore, as a result of freezing, a much smaller fraction of the total surface area in the system will be covered with a layer of water of thickness df rather than the original, relatively thick layer of surface water do. Using this conceptualization and the calculated value of do, the model can predict the bound water content in conditioned residuals. The total bound water in zots is a summation of water found on zot surfaces, on particle surfaces, in interstices within the zots, and in interstices between the zots. The latter two classes we refer to as intra486

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 3, 1999

zot interstitial water and inter-zot interstitial water, respectively. The intra-zot void ratio was assumed to be unchanged after freezing. However, unlike the unfrozen residuals, frozen and thawed residuals contain inter-zot void ratio; residuals do not behave as a continuum, and the zots act as individual particles. We assumed a self-similar system for which the value of the inter-zot void ratio equaled that for the intra-zot void ratio. The bound water content in zots is expressed by

Wc,f ) {[(zot surface water) + (particle surface water) + (intra-zot interstitial water) + (inter-zot interstitial water)]}/mass of solids )

(nzFwAzdo) + (npFwApdf) + (npFweVp) + (nzFweVz) FsnpVp (2)

where Wc,f is the water content in the residuals following freezing and thawing; Vp is the volume of a particle within a zot ) 4π/3(rp + df)3; Vz is the volume of zots ) (number of particles per zot)[(volume of particle) + (volume of interstices per floc)] ) PVp(1 + e); nz is the number of zots ) np/P; P is the number of particles per zot; Az ) 4πr2z R; and 3 rz is the radius of zot ) x3Vz/4π. When P is defined, eq 2 contains two unknowns, df and Wc,f. Varying the value of P is equivalent to varying the size of the zot. Thus modelderived results showing the variation of bound water content with the quantity of surface area can be compared to the experimental results in Figure 3. To illustrate the effect of zot size on dewaterability, P was varied from 101 to 106, which yielded zot diameters from 10-5 to 10-3 m. The square of the mean floc diameter was used as an indicator of the surface area and termed unit surface area. Information relating floc density to the number of primary particles is not available, thus the units are m2/floc and not m2/g. The variation in bound water content as a function of unit surface area is plotted as solid lines in Figure

spheres (R ) 0), which is highly unlikely in natural systems. Thus, df does not strongly depend on the various model parameters.

Acknowledgments This research was funded by the American Water Works Association Research Foundation (AWWARF) Project 386. The support of the project manager, Albert Ilges, is gratefully acknowledged.

Literature Cited

FIGURE 5. Surface area model results.

TABLE 2. Sensitivity Analysis variable

e R

R (µm)

value of variable

df (nm)

0.5 0.54 0 50 100 300 2 5 10 25

37 4.2 2700 9.5 4.2 1.3 0.35 4.2 11 31

5, where each line represents a different value of df. The symbols in Figure 5 correspond to the data presented in Figure 3. Figure 5 supports the conceptual model and emphasizes the importance of incorporating agglomerates of flocs when freezing, whether by slow freezing or by freezing high solids content residuals: the smallest zots, consisting of only a few colloidal particles, are essentially unaffected by conditioning since a large fraction of the total surface area is not dehydrated but contains a layer of surface water with thickness do. Note that the surface area of the unfrozen residuals are not contained within the model envelope because the model was calibrated assuming an initial bound water content of 10 g/g. Moreover, the values of df that yield the envelope shown in Figure 5 are of the same order of magnitude as the thickness of the layer of water that is influenced by surface forces. Typical values for this layer thickness are 30 nm (69), 20-40 nm (70), and 2-10 nm (45, 71). Also, for a given value of bound water content, the model can calculate df. The sensitivity of df to e, rp, and R was determined and is shown in Table 2. P was chosen equal to 1000 for the sensitivity analysis and corresponds to a floc radius of 1.01 × 10-4 m; a bound water content of approximately 2 g/g was chosen from Figure 3. Table 2 shows that df is generally less than 100 nm for a wide range in values of e, rp, and R. The only exception is for the case of smooth

(1) Logsdon, G. S.; Edgerley, E. J. Am. Water Works Assoc. 1971, 63, 740. (2) Knocke, W. R.; Trahern, P. Water Res. 1989, 23, 35. (3) Lee, D. J.; Hsu, Y. H. Environ. Sci. Technol. 1994, 28, 1444. (4) Hung, W. T.; Chang, I. L.; Lin, W. W.; Lee, D. J. Environ. Sci. Technol. 1996, 30, 2391. (5) Parker, P. J.; Collins, A. G.; Dempsey, J. P. J. Environ. Eng. 1998, 124, 249. (6) Parker, P. J.; Collins, A. G.; Dempsey, J. P. Environ. Sci. Technol. 1998, 32, 383. (7) Parker, P. J.; Collins, A. G.; Dempsey, J. P. Water Res. In press. (8) Parker, P. J.; Collins, A. G. J. Am. Water Works Assoc. (in review). (9) Parker, P. J.; Collins, A. G.; DeWolfe, J. R. J. Am. Water Works Assoc. (in review). (10) Clements, G. S.; Stephenson, R. J.; Regan, C. J. In The Public Works and Municipal Services Congress; London, 1950; pp 318337. (11) Katz, W. J.; Mason, D. G. Water Sewage Works 1970, 117, 110. (12) Vesilind, P. A.; Martel, C. J. J. Environ. Eng. 1990, 116, 854. (13) Zolotavin, V. L.; Vol’khin, V. V.; Rezvushkin, V. V. Colloid J. USSR 1960, 22, 317. (14) Zolotavin, V. L.; Vol’khin, V. V. Colloid J. USSR 1960, 333, 2110. (15) Zolotavin, V. L.; Vol’khin, V. V. Colloid J. USSR 1961, 23, 3. (16) Vol’khin, V. V.; Zolotavin, V. L. J. Appl. Chem. USSR 1961, 34, 1163. (17) Vol’khin, V. V.; Zolotavin, V. L.; Tipikin, S. A. Colloid J. USSR 1961, 23, 339. (18) Vol’khin, V. V.; Koblova, A. A.; Ponomarev, E. I. J. Appl. Chem. USSR 1963, 36, 196. (19) Vol’khin, V. V.; Zolotavin, V. L. Colloid J. USSR 1961, 23, 113. (20) Vol’khin, V. V.; Ponomarev, E. I. Colloid J. USSR 1965, 27, 10. (21) Vesilind, P. A. J. Water Pollut. Control Fed. 1988, 60, 215. (22) Pyper, J. W. Anal. Chim. Acta 1985, 170, 159. (23) Vesilind, P. A. Water Environ. Res. 1994, 66, 4. (24) Anderson, D. M.; Tice, A. R. In Symposium on Soil-Water Physics and Technology; Springer-Verlag: Rehovot, Israel, 1973; pp 107124. (25) Macey, H. H. Trans. Brit. Ceram. Soc. 1942, 41, 73. (26) Mattson, S. Soil Sci. 1932, 32, 301. (27) Nelson, R. M. Wood Sci. Technol. 1991, 25, 193. (28) Lee, H. B.; Jhon, M. S.; Andrade, J. D. J. Colloid Interface Sci. 1975, 51, 225. (29) Lele, A. K.; Hirve, M. M.; Badiger, M. V.; Masheklar, R. A. Macromolecules 1997, 30, 157. (30) Jones, R. L.; King, C. J. AIChE Symp. Ser. 1977, 73, 113. (31) Miura, N.; Asaka, N.; Shinyashiki, N.; Mashimo, S. Biopolymers 1994, 34, 357. (32) Forman-Kay, J. D.; Gronenborn, A. M.; Wingfield, P. T.; Clore, G. M. J. Mol. Biol. 1991, 220, 209. (33) Forster, C. F. J. Chem. Technol. Biotechnol. 1983, 33B, 76. (34) Heukelekian, H.; Weisberg, E. J. Water Pollut. Control Fed. 1956, 28, 558. (35) Tsang, K. R.; Vesilind, P. A. Water Sci. Technol. 1990, 22, 135. (36) Bouyoucos, G. J. J. Agric. Res. 1917, 8, 195. (37) Bouyoucos, G. Soil Sci. 1921, 11, 33. (38) Luikov, A. V. Heat and Mass Transfer in Capillary-porous Bodies; Pergamon Press: Oxford, 1966; pp 1-523. (39) Smollen, M. Water SA 1988, 14, 25. (40) Smith, J. K. Masters Thesis, Duke University, Durham, NC, 1992. (41) Etzler, F. M.; Drost-Hansen, W. Croat. Chem. Acta 1983, 56, 563. (42) Pashley, R. M.; Israelachvili, J. N. J. Colloid Interface Sci. 1984, 101, 511. (43) vanMegen, W.; Snook, I. J. Chem. Soc., Faraday Trans. 1979, 75, 1095. (44) Israelachvili, J. N. Acc. Chem. Res. 1987, 20, 415. (45) Israelachvili, J. N.; Kott, S. J. J. Colloid Interface Sci. 1989, 129, 461. (46) Etzler, F. M.; Fagundus, D. M. J. Colloid Interface Sci. 1987, 115, 513. VOL. 33, NO. 3, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

487

(47) Chen, G. W.; Hung, W. T.; Chang, I. L.; Lee, S. F.; Lee, D. J. J. Environ. Eng. 1997, 123, 253. (48) Gregory, J. Crit. Rev. Environ. Control 1989, 19, 185. (49) Hohmann, M. Cold Reg. Sci. Technol. 1997, 25, 101. (50) Lunardini, V. J.; et al. in Army Sci. Conf. 1982. (51) Konrad, J.-M.; Morgenstern, N. R. Can. Geotech. J. 1981, 18, 482. (52) Mageau, D. W.; Morgenstern, N. R. Can. Geotech. J. 1980, 17, 54. (53) Miller, R. D. Cold Reg. Sci. Technol. 1980, 3, 83. (54) Takagi, S. Cold Reg. Sci. Technol. 1980, 3, 233. (55) Williams, P. J.; Smith, M. W. The Frozen Earth; Cambridge University Press: Cambridge, 1989; pp 1-306. (56) WIlliams, P. J. Norwegian Geotechnical Institute, Oslo, 1967. (57) Cary, J. W. Water Resour. Res. 1987, 23, 1620. (58) Takagi, S. AIChE Symp. Ser. Heat Transfer Appl. 1978, 174, 235. (59) Taber, S. J. Geol. 1929, 37, 428. (60) Blank, R. R. In International Symposium on Physics, Chemistry, and Ecology of Seasonally Frozen Soils; Iskandar, I. K., Ed.; U.S. Army Cold Regions Research and Engineering Laboratory: Fairbanks, AK, 1997; pp 212-217. (61) Bresson, L. M.; Moran, C. J. Eur. J. Soil Sci. 1995, 46, 205. (62) Emerson, W. W. In Soils: an Australian viewpoint; Academic Press: London, 1983.

488

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 3, 1999

(63) Mullins, C. E.; MacLeod, D. A.; Northcote, K. H.; Tisdall, J. M.; Young, I. M. In Soil Degradation; Lal, R., Stewart, B. A., Eds.; Springer-Verlag: New York, 1990; p 343. (64) Chartres, C. J. In Advances in Soil Science; Lewis Publishers: Boca Raton, LA, 1992; pp 339-365. (65) Dexter, A. R. Catena Suppl. 1988, 11, 35. (66) Mielke, L. N.; Powers, W. L.; Badri, S.; Jones, A. J. Soil Tillage Res. 1994, 31, 199. (67) Mullins, C. E.; Young, I. M.; Bengough, A. G.; Ley, G. J. Soil Use Manage. 1987, 3, 79. (68) Smith, W. O. Physics 1932, 3, 139. (69) Butt, H.-J.; Jaschke, M.; Ducker, W. Bioelectrochem. Bioenerg. 1995, 38, 191. (70) Ducker, W. A.; Xu, Z.; Clarke, D. R.; Israelachvili, J. N. J. Am. Ceram. Soc. 1994, 77, 437. (71) Christenson, H. K. J. Dispersion Sci. Technol. 1988, 9, 171.

Received for review July 13, 1998. Revised manuscript received November 9, 1998. Accepted November 18, 1998. ES980705P