Citation
Heavy metals and colloid mobility in soils

Material Information

Title:
Heavy metals and colloid mobility in soils
Creator:
Dong, Yan, 1962-
Publication Date:
Language:
English
Physical Description:
xi, 149 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Dissertations, Academic -- Soil and Water Science -- UF ( lcsh )
Soil and Water Science thesis, Ph.D ( lcsh )
City of Tampa ( local )
Colloids ( jstor )
Soils ( jstor )
Incubation ( jstor )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1999.
Bibliography:
Includes bibliographical references (leaves 136-148).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Yan Dong.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
021556799 ( ALEPH )
43707649 ( OCLC )

Downloads

This item has the following downloads:

heavymetalscollo00dong.pdf

heavymetalscollo00dong_0044.txt

heavymetalscollo00dong_0009.txt

heavymetalscollo00dong_0075.txt

heavymetalscollo00dong_0128.txt

heavymetalscollo00dong_0022.txt

heavymetalscollo00dong_0040.txt

heavymetalscollo00dong_0137.txt

heavymetalscollo00dong_0139.txt

heavymetalscollo00dong_0096.txt

heavymetalscollo00dong_0144.txt

heavymetalscollo00dong_0054.txt

heavymetalscollo00dong_0025.txt

heavymetalscollo00dong_0095.txt

heavymetalscollo00dong_0082.txt

heavymetalscollo00dong_0101.txt

heavymetalscollo00dong_0071.txt

heavymetalscollo00dong_0105.txt

heavymetalscollo00dong_0003.txt

heavymetalscollo00dong_0115.txt

heavymetalscollo00dong_0133.txt

heavymetalscollo00dong_0107.txt

heavymetalscollo00dong_0013.txt

heavymetalscollo00dong_0064.txt

heavymetalscollo00dong_0028.txt

heavymetalscollo00dong_0078.txt

heavymetalscollo00dong_0057.txt

heavymetalscollo00dong_0080.txt

heavymetalscollo00dong_0058.txt

heavymetalscollo00dong_0117.txt

heavymetalscollo00dong_0026.txt

heavymetalscollo00dong_0111.txt

heavymetalscollo00dong_0005.txt

heavymetalscollo00dong_0151.txt

heavymetalscollo00dong_0012.txt

heavymetalscollo00dong_0132.txt

heavymetalscollo00dong_0124.txt

heavymetalscollo00dong_0148.txt

heavymetalscollo00dong_0001.txt

heavymetalscollo00dong_0141.txt

heavymetalscollo00dong_0034.txt

heavymetalscollo00dong_0131.txt

heavymetalscollo00dong_0118.txt

heavymetalscollo00dong_0088.txt

heavymetalscollo00dong_0007.txt

heavymetalscollo00dong_0061.txt

heavymetalscollo00dong_0030.txt

heavymetalscollo00dong_0157.txt

heavymetalscollo00dong_0004.txt

heavymetalscollo00dong_0127.txt

heavymetalscollo00dong_0021.txt

heavymetalscollo00dong_0024.txt

heavymetalscollo00dong_0085.txt

heavymetalscollo00dong_0090.txt

heavymetalscollo00dong_0006.txt

heavymetalscollo00dong_0094.txt

heavymetalscollo00dong_0123.txt

heavymetalscollo00dong_0020.txt

heavymetalscollo00dong_0106.txt

heavymetalscollo00dong_0008.txt

heavymetalscollo00dong_0152.txt

heavymetalscollo00dong_0047.txt

heavymetalscollo00dong_0038.txt

heavymetalscollo00dong_0035.txt

heavymetalscollo00dong_0046.txt

heavymetalscollo00dong_0055.txt

heavymetalscollo00dong_0156.txt

heavymetalscollo00dong_0130.txt

heavymetalscollo00dong_0102.txt

heavymetalscollo00dong_0045.txt

heavymetalscollo00dong_0135.txt

heavymetalscollo00dong_0000.txt

heavymetalscollo00dong_0056.txt

heavymetalscollo00dong_0108.txt

heavymetalscollo00dong_0010.txt

heavymetalscollo00dong_0103.txt

heavymetalscollo00dong_0069.txt

heavymetalscollo00dong_0017.txt

heavymetalscollo00dong_0032.txt

heavymetalscollo00dong_0042.txt

heavymetalscollo00dong_0143.txt

heavymetalscollo00dong_0134.txt

heavymetalscollo00dong_0113.txt

heavymetalscollo00dong_0112.txt

heavymetalscollo00dong_0039.txt

heavymetalscollo00dong_0079.txt

heavymetalscollo00dong_0149.txt

heavymetalscollo00dong_0067.txt

heavymetalscollo00dong_pdf.txt

heavymetalscollo00dong_0083.txt

heavymetalscollo00dong_0154.txt

heavymetalscollo00dong_0119.txt

heavymetalscollo00dong_0147.txt

heavymetalscollo00dong_0070.txt

heavymetalscollo00dong_0011.txt

heavymetalscollo00dong_0050.txt

heavymetalscollo00dong_0093.txt

heavymetalscollo00dong_0120.txt

heavymetalscollo00dong_0150.txt

heavymetalscollo00dong_0100.txt

heavymetalscollo00dong_0125.txt

heavymetalscollo00dong_0019.txt

heavymetalscollo00dong_0041.txt

heavymetalscollo00dong_0016.txt

heavymetalscollo00dong_0027.txt

heavymetalscollo00dong_0074.txt

heavymetalscollo00dong_0063.txt

heavymetalscollo00dong_0089.txt

heavymetalscollo00dong_0153.txt

heavymetalscollo00dong_0051.txt

heavymetalscollo00dong_0059.txt

heavymetalscollo00dong_0062.txt

heavymetalscollo00dong_0043.txt

heavymetalscollo00dong_0014.txt

heavymetalscollo00dong_0098.txt

heavymetalscollo00dong_0068.txt

heavymetalscollo00dong_0159.txt

heavymetalscollo00dong_0104.txt

heavymetalscollo00dong_0110.txt

heavymetalscollo00dong_0099.txt

heavymetalscollo00dong_0023.txt

heavymetalscollo00dong_0109.txt

heavymetalscollo00dong_0072.txt

heavymetalscollo00dong_0031.txt

heavymetalscollo00dong_0002.txt

heavymetalscollo00dong_0114.txt

heavymetalscollo00dong_0142.txt

heavymetalscollo00dong_0049.txt

heavymetalscollo00dong_0145.txt

heavymetalscollo00dong_0029.txt

heavymetalscollo00dong_0037.txt

heavymetalscollo00dong_0077.txt

heavymetalscollo00dong_0122.txt

heavymetalscollo00dong_0146.txt

heavymetalscollo00dong_0053.txt

heavymetalscollo00dong_0033.txt

heavymetalscollo00dong_0087.txt

heavymetalscollo00dong_0084.txt

heavymetalscollo00dong_0121.txt

heavymetalscollo00dong_0060.txt

heavymetalscollo00dong_0091.txt

heavymetalscollo00dong_0097.txt

heavymetalscollo00dong_0129.txt

heavymetalscollo00dong_0136.txt

heavymetalscollo00dong_0138.txt

heavymetalscollo00dong_0048.txt

heavymetalscollo00dong_0076.txt

heavymetalscollo00dong_0158.txt

heavymetalscollo00dong_0066.txt

heavymetalscollo00dong_0073.txt

heavymetalscollo00dong_0036.txt

heavymetalscollo00dong_0116.txt

heavymetalscollo00dong_0092.txt

heavymetalscollo00dong_0126.txt

heavymetalscollo00dong_0052.txt

heavymetalscollo00dong_0081.txt

heavymetalscollo00dong_0086.txt

heavymetalscollo00dong_0018.txt

heavymetalscollo00dong_0140.txt

heavymetalscollo00dong_0065.txt

heavymetalscollo00dong_0155.txt

heavymetalscollo00dong_0015.txt


Full Text










HEAVY METALS AND COLLOID MOBILITY IN SOILS


By

YAN DONG












A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA


1999


























Copyright 1999

by

YAN DONG












ACKNOWLEDGMENTS


First among those deserving credit for their contributions is the committee chairman, Dr. Lena Q. Ma. Her many contributions to the form and content of this dissertation are appreciated, but equally important were the opportunities and directions that she provided for developing skills that are of considerable benefit to my career in research and teaching.

Dr. R. Dean Rhue deserves special thanks for patiently leading me through all the difficulties in finishing Chapter 3. Dr. Willie Harris, Dr. Peter Nkedi-Kizza and Dr. Timothy G. Townsend provided much of the initial direction and a great deal of instructive suggestion to this study. Their efforts were indispensable and are much appreciated.

Special thanks also go to Kennelley, E., Schwandes, L., Thomas, J., Reve, W., Lewis, K., Awuma, K., and Choate, A., for their help with instrumental analysis.













TABLE OF CONTENTS
page

ACKNOW LEDGMENTS ................................................................................................ III

LIST OF TABLES ............................................................................................................ IV

LIST OF FIGURES ........................................................................................................ VII

ABSTRACT ....................................................................................................................... X

CHAPTERS

1 INTRODUCTION .......................................................................................................... 1

2 COLLOID DEPOSITION, RELEASE AND ASSOCIATION WITH HEAVY
METALS IN SOILS ..................................................................................................... 5

Introduction.............................................................................................................. 5
The DLVO Theory................................................................................................10
The Characteristics of Colloids ............................................................................13
Charge development ....................................................................................... 13
Electrical double layer (EDL) in metal oxides-water interface......................14
Charge development in soil particles.............................................................. 21
Charge development in mobile colloids .........................................................22
Hydration................................................................................................ ........ 24
Size development of mobile colloids..............................................................25
The Characteristics of Porous Media....................................................................26
Deposition of Colloids..........................................................................................27
Theoretical background in well-defined porous media ..................................27
Colloid deposition in soil................................................................................29
Release of Colloids in Soil ...................................................................................31
The processes of colloid release .....................................................................33
Mobile colloids and water-dispersible clay ....................................................34
Influenes of exchanegable soldium percentage (ESP) on stability of
water-dispersible clay and mobilization of soil colloids ..........................35
Ion transfer processes during the release of colloids......................................37
The ion transfer during soil colloid mobilization in literature........................38
Ion transfer during soil colloid release in our research..................................41
Association of Colloids with Heavy M etals.........................................................43
Adsorption of heavy metals to surface hydroxyl of colloids..........................44
Effects of heavy metal adsorption on the charges of soil colloids .................47
Partitioning of heavy metals in soil colloids...................................................49









Concluding Remarks ............................................................................................52

3 RELATION OF PB SOLUBILITY TO FE PARTITIONING IN SOILS ...................55

Introduction...........................................................................................................55
M aterials and M ethods .........................................................................................58
Location and characteristics of soil sample....................................................58
Column experiment ........................................................................................58
Sample separation and analysis......................................................................61
Results and Discussion.........................................................................................61
Pb solubility....................................................................................................61
Pb solubility and Fe partitioning.....................................................................64
Pb solubility and Fe partitioning using published data...................................68
Implication of this Research .................................................................................. 72
Conclusion ............................................................................................................73

4 Heavy metal mobility in contaminated soils: Part 1. The role of exchange sites
in controlling solubility and mobility of heavy metals.................................................74

Introduction...........................................................................................................74
M aterials and M ethods .........................................................................................76
Location and characteristics of soil samples ..................................................76
Column Experiment........................................................................................78
Analysis of pore water of soil.........................................................................79
Results and Discussion.........................................................................................79
Changes in heavy metal solubility with incubation........................................79
Heavy metal mobility with incubation ...........................................................86
Implication of this Research.................................................................................90

5 HEAVY METAL MOBILITY IN CONTAMINATED SOILS: PART 2.
COLLOID-FACILITATED METAL MOBILITY IN A PBCONTAM INATED SOIL ............................................................................................ 92

Introduction...........................................................................................................92
M aterials and M ethods .........................................................................................95
Characteristics of soil samples........................................................................95
Column experiment ........................................................................................97
Analysis of Fe(II) and Ca in pore water .........................................................97
Analytical methods.........................................................................................98
Result and discussion............................................................................................99
Colloid mobility and soil redox status.......................... ..................................99
Colloid elution curves...................................................................................105
Colloid-facilitated Pb mobility.....................................................................110
Conclusion.......................................................................................................... 111

6 RELEASE AND DISPERSABILITY OF COLLOIDS IN TWO
CONTAM INATED SOILS ...................................................................................... 113










Introduction.........................................................................................................113
M aterials and M ethods .......................................................................................117
Colum n preparation ......................................................................................117
Colum n leaching test ....................................................................................118
W ater dispersability test ............................................................................... 118
Estim ation of relative colloid stability ratio (RW ) .......................................119
Result and D iscussion.........................................................................................123
Conclusion..........................................................................................................131


7 CON CLU SION ...........................................................................................................132

State of colloid deposition and release in soil and their association with
heavy m etals ...........................................................................................132
Colloidal m etal m obility in contam inated soils............................................132
M etal solubility and m obility in soil.............................................................134


REFEREN CES ................................................................................................................ 136

BIOGRAPH ICA L SKETCH .......................................................................................... 149












LIST OF TABLES


Table page 3-1 Characteristics of the Florida soil used.....................................................................59

4-1 Selected characteristics of the soils used in this study ...........................................77

5-1 Selected properties of the Pb contaminated soil used in this study..........................96

5-2 Minerals in the soil and their Point of zero charge (PZC) .....................................104

6-2 Relative colloid stability ratios (RW) for two soils under different water-flooding
tim e .........................................................................................................................125












LIST OF FIGURES


Figure page

1-1. A road m ap of this dissertation...................................................................................3

2-1. Schematic of potential energy profile of the interaction of surfaces with
inclusion of van der Walls attraction, electrical repulsion, and Born
repulsion, which shows both primary and secondary minima and an energy
barrier as well as the zones in which release and deposition take place 1 ...................6

2-2. Cumulative mass of metals leached from the soil in the three size fractions
after switching salt solution to deionized water. Size separation was by 450nm membrane filter and 1-nm cellulose ultrafiltration membrane (data from
table 4-6 in Am rhein et al, 1993) ............................................................................50

3-1. Relation between Pb solubility and leachate pH, dissolved organic carbon
(DOC), leachate Fe concentration, and ratio of aqueous Fe to sorbed Fe(II) concentrations in a sandy soil. Data from 2.90 mmole kg-1 Pb loading rate
read from left y-axis and data from 0.36 mmole kg-1 Pb loading rate read
from right y-axis .......................................................................................................62

3-2. Relationship between Pb concentrations in pore water and the ratio of
aqueous and sorbed Fe (Fe partition index) in a contaminated sediment. Data
are taken from Lee et al., 1997 .................................................................................69

3-3. Relationship between soluble Pb concentrations and the ratio of soluble to
sorbed Fe in contaminated soils. Data are taken from Karczewska, 1996 ..............70

4-1. Pb and Fe (II) concentrations in pore water of Montreal soil...................................80

4-2. Pb, As, Cu and Fe(II) in pore water of Tampa soil ..................................................81

4-3. Cumulative Pb and Fe leached after 31.8 pore volumes of 0.01 M CaCl2 in
M ontreal soil ............................................................................................................ 84

4-4. Cumulative Pb, As, Cu and Fe leached after 31.8 pore volumes of 0.01 CaCl2
in T am pa soil ............................................................................................................85

5-1. Changes of effluent turbidity with pore volumes under different incubation
times. The insert is a typical breakthrough curve of deionized water
displacing C aCl2 ...................................................... ............................................ 100








5-2. Effects of incubation on aqueous Fe (II) and Ca in the pore waters of soil
columns and the cumulative colloidal Fe and Al in the effluents after 23 pore
volum es................................................................................................................... 102

5-3. Elution curves of colloidal Fe, Al, and Pb concentrations in effluents with
pore volumes under various incubation times. Each point presents the mean
of tw o replicates......................................................................................................106

5-4. Relationship among colloidal Fe and Al, dissolved Ca and pH in effluents
after 3 d of incubation. Each point presents the mean of two replicates ........ : .......107

6-1 Schematic representation of typical absorbency-time curves observed. Linear
regression was used to calculate an apparent flocculation rate indicated by
solid line. The curves are not drawn in the same time and absorbancy scales.
Curve "a" stands for those observed for the Montreal soil in 0.06 NaCl M
solution, and Curve "b" for those for the Tampa soil in 0.06 M NaCl and in
0.01 M CaCl2 solutions, and the Montreal soil in 0.01 M CaCl2 solution..............122

6-2. Absorbency-time curves observed in Ca-saturated Tampa soil suspended in
0.06 M N aC l solution ............................................................................................. 124

6-3. Relation of cumulative colloids and relative stability ratio (RW) .........................128












Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

HEAVY METAL AND COLLOID MOBILITY IN SOILS By

Yan Dong

December 1999

Chairman: Lena Q. Ma
Major Department: Soil and Water Science


Mobility of heavy metals in soils is of environmental significance due to their

toxicity to both humans and animals. In general, heavy metal mobility is low because of its low solubility. However, enhanced heavy metal mobility has been reported under both laboratory and field conditions. It has been attributed to enhanced solubility of heavy metals and colloid-facilitated metal transport.

In this study, the solubility and mobility of heavy metals were examined in a Pbspiked sandy soil and two Pb-contaminated soils. For the Pb-spiked soil, water-flooded and non-water-flooded incubations were used to alter soil solution chemistry in soil columns, which were then leached with de-ionized water. It was found that Pb concentrations in leachates were related to the ratios of Fe concentration in the leachates to Fe concentrations extracted with HC1. Enhanced Pb mobility occurred only when the Fe ratios were lower than a threshold value for a given soil. The two Pb-contaminated soils were incubated for different times under water flooded condition to alter their redox








status. Metal solubility was examined by analyzing the pore water of the incubated soil columns whereas metal mobility was examined by leaching the columns with 0.01 M CaCl2. The data showed that metal solubility in pore water and metal mobility in CaC12 solution were not always directly related. There has been sufficient evidence that a reduction in cation exchange capacity (CEC) occurs with Fe reduction dissolution. However, the consequent redistribution of metals between solution and exchange phases with incubation was determined not only by the magnitude of CEC, but also the characteristics of metal ions, the competing co-ions and counterions (ligands).

Colloid mobility in the two Pb-contaminated soils was examined by switching the influent from 0.01 M CaCl2 to de-ionized water. In addition, a batch dispersability test was conducted by dispersing the incubated soils in 0.01 M CaCl2 and 0.06 M CaCl2, and then a relative stability ratio (W) of an incubated soil was estimated from the absorbencytime curves of a dispersion in the solutions. The results showed that colloid mobility is greatly influenced by incubation and that colloidal metal mobility is enhanced when exchangeble Ca is easily replaced.












CHAPTER 1
INTRODUCTION

Understanding heavy metal migration in order to accurately predict heavy metal mobility in contaminated soils is critical for contaminant risk assessment and costeffective soil remediation (McCarthy and Zachara, 1989). Although much effort has been spent on developing models to predict heavy metal mobility in soils (Cederberg et al., 1985; Sposito, 1984), these models often underestimate metal migration in soils. In most of the models, the interaction between dissolved metals and soil matrix is mainly described using a distribution coefficient (kD), which is generally considered as a constant and usually measured in the laboratory under a particular condition. In a dynamic soil system, however, kD is not a constant. In other words, solubility of metals and related mobility vary with soil conditions.

Besides dissolved species, mobile colloids can also act as vectors to transport

sorbed contaminants such as heavy metals in soils (Seaman et al., 1995). Understanding mobility of potential heavy-metal-bearing colloids is critical for predicting the fates of heavy metals in natural porous media such as soils and aquifers (Swanton, 1995; Ryan and Elimelech, 1996) since colloidal particles are ubiquitous in those systems. In concept, the deposition and release of colloids in porous media can be described with the DLVO (Derjaguin-Landau-Verwey-Overbeek) theory. However, it is difficult to apply to soil because of the heterogeneity and irregular dimensions of soil colloids. A colloid/dissolved metal transport model, COMET (Mills et al., 1991), has been developed








to predict colloidal metal mobility. It is difficult to apply to soil environments because of the difficulties in determining the parameters needed in the model. It has been recognized that colloid mobility in soil is mainly influenced by the physicochemical characteristics of colloids and soils (Seaman et al., 1995). Unfortunately, little is known quantitatively about the relationship between colloidal metal mobility and the characteristics of soil colloids (Amrhein et al., 1993; Seaman et al., 1995; Kaplan et al., 1993).

In this study, the solubility and mobility of heavy metals, colloid mobility and its association with heavy metals are studied. The main objectives of this research are as follows

* To characterize the conditions at which enhanced solubility and mobility of

heavy metals occur with incubation in soils,

* To characterize the conditions at which enhanced colloid mobility oceurs with

incubation in contaminated soils,

* To characterize the conditions at which enhanced mobility of colloidal heavy

metals occurs with incubation in contaminated soils.

Therefore, there are two major subtopics in this dissertation: colloidal metal mobility and dissolved metal mobility (Figure 1-1). The understanding of colloid deposition, release and association with heavy metals, is first reviewed and discussed in the next chapter. In Chapter 5, changes in colloid and colloid-facilitated heavy metal mobility with incubation are examined in two contaminated soils. In Chapter 6, a fundamental approach is taken to relate colloid mobility to colloid stability ratio (W). A relative colloid stability ratio is defined and estimated by the curves of absorbency-time.
















Chapter 1 Introduction


Colloidal Metal Mobility
........... ... ......................... ] ..........................................



Chapter 2
A review on colloid
mobility in soil




Chapter 5,6
Colloidal metal mobility
in Contaminated soils


Dissolved Metal Mobility............................................ ...............................................
Dissolved Metal Mobility


Chapter 3
Pb mobility in a Pb
spiked soil


Chapter 4
Heavy metal mobility in
contaminated soils


Chapter 7 Conclusion


Figure 1-1 A road map of this dissertation


I i


I





4

On the other hand, the relation of Pb solubility in soil and Fe partitioning behavior after incubation is examined in Chapter 3. In Chapter 4, an extension of the study from an uncontaminated soil to two Pb-contaminated soils is made (Figure 1-1).













CHAPTER 2
COLLOID DEPOSITION, RELEASE AND ASSOCIATION WITH HEAVY METALS IN SOILS

Introduction


Colloid-size particles are ubiquitous in soil. It is well known thatcolloids may be mobilized and transported a significant distance (Ryan and Elimelech, 1996; Swanton, 1995) in soil environments. As a result , the mobility of heavy metals associated with the colloids may be enhanced (Kaplan et al. 1995; Amrhein et al., 1993; McCarthy and Zachara, 1989; Newman et al., 1993; Mills et al., 1991; Ouyang et al., 1996). Generally, colloid movement is faster than an inert tracer due to the effect of size exclusion (NocitoGobel et al., 1996). Therefore, colloid-facilitated metal transport has been considered as one of the important mechanisms causing heavy metals to move much faster than expected (Newman et al., 1993), which has attracted much attention in the scientific community.

The term colloid generally applies to suspended particles of 1 nm-2 pm

(McCarthy and Zachara, 1989), whose behaviors are size-dependent. Interactions among large particles (> 1 pm, non-Brownian) are affected by physical forces, usch as gravity and fluid drag, whereas those of submicron particles (1 nm-1 pm, Brownian) are mainly controlled by the interfacial characteristics of particle-solution. Submicron colloids are of particular interest because of their significance in transporting contaminants due to





6













-.2 4�max

Secondary minimum oSeparation distance (h)





1 1
Release Deposition



















Figure 2-1. Schematic of potential energy profile of the interaction of surfaces with inclusion of van der Walls attraction, electrical repulsion, and Born repulsion, which shows both primary and secondary minima and an energy barrier as well as the zones in which release and deposition take place.










their large surface areas and high transportability in porous media. Therefore, this dissertation will mainly focus on submicron colloids excluding microbes.

Since there is a collection of literature published on colloid release, transport and deposition in recent years (Ryan and Elimelech, 1996; Swanton, 1995; Mills et al., 1991; Ouyang et al., 1996; Adamczyk et al., 1983; Kallay et al., 1987; Elimelech et al, 1995; McCarthy and Zachara, 1989; McDowell-Boyer et al., 1986), no attempt is made to present a complete critique on the topics. Instead, focus will be put on several parameters affecting the mobility of colloids and colloid-facilitated heavy metal transport in soils.

At the present time, various processes responsible for colloid deposition and

release in porous media have been well established (Ruckenstein et al., 1976; Adamczyk et al., 1983; Kallay et al., 1987; Elimelech et al., 1995; Ryan and Elimelech, 1996). These processes can be theoretically described by the DLVO (Derjaguin-LandauVerwey-Overbeek; Israelachvili, 1992) theory, which states that the net surface interaction energy is the sum of the interactions of electrical double layer (EDL) and van der Waals-London forces (WL), which vary with separation distance between particles. As particles approach each other, the net interaction energy experiences a secondary minimum first, then a primary minimum (*min) after a maximum energy barrier (#max) (Figure 2-1). After that point, further movement towards each other causes drastic increase in the interaction energy, which is referred to as the Born repulsion (Israelachvili, 1992). In principle, particles may be attached or deposited when the net attractive forces are close to either the primary or secondary energy minimum. Both the magnitude of the energy barrier and the depth of the primary and secondary minimum are








affected by solution chemistry. In order for a deposited particle to be released, repulsive forces have to be generated between the surfaces of particles and stationary grains, as a result of changes in solution chemistry or fluid flow. However, the understanding of colloid release and deposition is far from complete. Colloid deposition under favorable conditions, in the presence of attractive interactions, can be predicted reasonably well, whereas colloid deposition under unfavorable conditions, in the presence of repulsive interactions, can not be predicted by the DLVO theory. In fact, colloid deposition rates observed experimentally are many orders of magnitudes greater than those predicted and are independent of the size of colloid particles (Elimelech et al., 1995). Elimelech et al. (1995) have recently presented an extensive discussion on possible explanations for these discrepancies. However, compared to colloid deposition, colloid release is poorly understood, which will be emphasized in this review.

Transport of mobile colloids is a critical process for evaluating colloid mobility in soils. However, there are several extensive reviews on the topic (Elimelech et al., 1995; Tien, 1989; McDowell-Boyer et al., 1986; Ouyang et al., 1996), so this topic is not emphasized in this review. Association of soil colloids with heavy metals has been an important issue for many years (McBride, 1994; Jones et al., 1975); however, until recently, the attention has not been paid to the effect of such an association on colloid mobility (Kretzschmar et al., 1997a). In fact, in flotation processes widely used in the mining industry, it is well documented that aqueous heavy metals can be potential determining ions or specific adsorbing ions to charged surfaces of minerals (Fuerstennau and Palmer, 1976), therefore, such an association with colloid surfaces may alter the








surface charge significantly. The potential impact of the association of heavy metals with colloids on colloid mobility will thus be emphasized in this review.

More recently, Wan and Wilson (1994a,b) have demonstrated that colloid

particles tend to sorb onto the water-air interface in soil and that the process is almost irreversible. In fact, these findings have sound theoretical and experimental basis in the flotation discipline (Williams and Berg, 1991). They concluded that colloid deposition onto the interface in unsaturated porous media is one of the important mechanisms responsible for the retardation of colloid transport, except under extremely high flow rate that may drag the bubbles with the flow (Wan and Wilson, 1994a). Incorporating those results into a convective-diffusion equation, Corapcioglu and Choi (1996) have developed a model to predict colloid transport in unsaturated porous media. These results have undoubtedly advanced colloid transport in soils significantly, however, they will not be further discussed in this review.

Most of the current reviews focus on either well defined porous systems

(Elimelech et al., 1995; Tien, 1989) or subsurface aquifers (Ryan and Elimelech, 1996; Swanton, 1995; McCarthy and Zachara, 1989). There is apparently much similarity between soil and subsurface or well-defined porous systems. Different from the later two, however, soil is more dynamic in its solution chemistry. In surface soil, pH, ionic strength, and redox potential of soil solution are influenced by rainfall, bioactivity, and changes in land use; whereas in vadose zone they are mainly impacted by the fluctuation in water table (Boul et al., 1997). In addition, as mentioned above, the gas phase is not negligible and the interfaces between colloids and air need to be taken into account along with that between colloids and water. Therefore, colloid transport is more dynamic and








complicated in soils compared to that in subsurface. In this review, we will emphasize how soil environments affect colloid mobility and what are the possible mechanisms.


The DLVO Theory

The classic DLVO theory has been extensively used to describe colloid

interactions and stability (Verwey and Overbeek, 1948). It has been extended to describe colloid deposition and release in porous media by including short-range Born repulsion forces (Ruckenstian et al., 1976). According to the DLVO theory, the total interaction energy between colloids and stationary solid phases (-rotal) is the sum of electrostatic repulsion energy, EL, arising from the overlap of electrical double layers and attractive energy, OLW, due to van der Walls-London dispersion, i.e., Total= OEL+ OWL, which is a function of particle separation distance. At close separation distance, a third repulsive force, the Born repulsion, is also important. Thus, the generalized curve of interaction energy vs. particle separation distance shows one energy maximum (max) & two minimums [primary minimum (0min) and secondary minimum; Figure 2-1]. Aggregation or coagulation occurs when particles collide with sufficient kinetic energy to overcome the energy barrier (max) so that they reach a distance within the omin. On the other hand, if coagulated particles gain enough energy under perturbation to overcome the energy barrier (4max - #min) with increasing separation distance, they tend to be released. The shape, intensities and positions of the maximum and minima, are determined by the interactions between particle surfaces, which change with surface properties of particles and solution, and thus, interfacial characteristics.








In fact, significant discrepancies have been found between experimental

observation and the prediction based on the DLVO theory (Elimelech and O'Melia, 1990b). Therefore, in recent years, various updated theories of colloid stability have been proposed (Shulepov et al., 1995; Adamczyk et al., 1986). The accuracy of prediction has been somewhat increased using more sophisticated models in some systems. Nevertheless, stability and coagulation of clay colloids in solution can be qualitatively described by the DLVO theory in some systems (Verwey and Overbeek, 1948 ); colloid transport in model porous media has also been quantitatively described using the theory plus the Born force (Ruckenstein and Prieve, 1976). Although measured collision efficiency between colloids and porous media surfaces is much greater than the theoretical prediction (Elimelech and O'Melia, 1990b), colloid deposition and release can be qualitatively described by the theory (Roy and Dzombk, 1996a). Therefore, the concept of the DLVO theory is correct even though it is incomplete (Swanton, 1995).

To reasonably predict colloid interactions in a natural system, the classic DLVO theory needs some modification. Direct measurements of colloid interactions indicate other forces exist between particles (Israelachvili, 1992). The interactions between polar electron acceptors (Lewis acids) in solution and polar electron donors (Lewis bases) from colloid particles are termed AB force and are usually dominant in aqueous systems in addition to electrostatic and van der Waals-London force (Oss et al., 1988). It can account for up to 85% of interactions between soil mineral particles in aqueous solution, as such the DLVO theory combined with the AB force adequately describes some anomalous colloid stability (Oss et al., 1990; Wu et al., 1994a,b). One advantage of this approach is the AB force can be estimated simply by the contact angles between isolated








particles and liquids (Oss et al., 1988). Several contact angles are required to calculate the AB force using the formula of Oss et al. (1988). The measurement needs to be made in at least three different liquids, two of which must be polar. In their approach, the WL force (VwL) can be estimated from surface tension instead of using assumptions for natural colloids as done by Ryan and Gschwend (1994a) and Roy and Dzombk (1996a). Here the classic DLVO theory combined with the AB force is termed the extended DLVO theory. Besides the AB force, other non-DLVO forces between particles, such as osmotic and steric interactions, should also be included. The extended DLVO theory should also correct for colloid surface roughness and heterogeneity of colloid compositions, charges and sizes (Elimelech and O'Melia, 1990a,b). Unfortunately, no models yet successfully take all these factors into consideration. Nevertheless, the DLVO theory provides a solid foundation and a useful tool to describe colloid interactions (Swanton, 1995; Ryan and Elimelech, 1996). Therefore, at the present time, it is practical to describe the interactions between colloids using the extended DLVO theory (#Total= EL+ OLW + OAB). The extended DLVO theory has been more successful than the original one in describing colloid behavior (Oss et al., 1990; Wu et al., 1994a,b), which has been discussed to quantify the terms in the equation above by Oss et al. (1988)

As shown in Figure 2-1, in the presence of an energy barrier, the deposition of a particle from solution onto a surface is mainly affected by the magnitude and shape of this barrier extending from its maximum into the solution phase. These are primarily due to electrostatic, van der Waals-London and AB force contributions. In contrast, particle release depends on interaction at separations between the barrier maximum and the surface, which is greatly influenced by additional repulsion at short separation (Born








repulsion). Over this region, the energy profile is very sensitive to the specific characteristics of the interacting surfaces and intervening liquid layer. However, the irregular dimensions of natural soil colloids make the interaction even more complicated, which will be discussed later.


The Characteristics of Colloids

Charge development

Solid particles present in water are often charged. The mechanisms of charge development have been extensively investigated (Parks, 1965; Sumner, 1992). In general, based on the nature of solid particles, charge development can be divided into three categories: 1) lattice substitution is most common in soil clay particles and referred to as permanent charge; 2) specific chemical interactions between surfaces and solution, including hydrolysis of surface functional group (e.g., hydroxyl and carbonyl) and chemical adsorption; and 3) preferential dissolution resulting from preferential hydration of surface atoms. In addition, based on the contribution from electrolytes in solution to the charges of particles, ions can be classified into three categories: potential determining ions, specific adsorbed ions and indifferent ions. Potential determining ions are the constituent ions of solid particles (e.g., H+ and OH" for Fe oxides, Ca2+ and CO32 for CaCO3) and their concentrations determine primarily the surface potential of particles (Parks, 1967). Specific adsorbed ions can change the magnitude and sign of the surface charge (e.g., Ca2+, Pb2+) by adsorbing onto the surfaces and forming a stem plane in electric double layer (EDL) theory (Parks, 1975). Indifferent ions adsorb physically to surfaces and change the magnitude of the surface charge (e.g. Na+, CI), and form a








diffusive layer in EDL resulting from a balance of their electrostatic and osmotic interactions with the surface and bulk solution (Parks, 1975).

In soils, however, colloids are heterogeneous in composition. They generally consist of inorganic and organic constituents. Heterogeneity in the composition and structure of colloidal particles make their charge development a complicated matter. This may be further complicated by the dynamic nature of soil solution chemistry, which may cause colloid composition and structure to vary temporally and spatially. Nevertheless, numerous studies have shown that the charge developing process can still be described using the EDL theory, at least in principle. Electrical double layer (EDL) in metal oxides-water interface

Electrical double layer of a dispersed particle consists of a charged surface and a diffusive layer of counter ions next to the surface based on the double layer theory. Since metal oxides may play an important role in colloid mobility and have been well characterized among various surfaces of soil colloid particles (Schindler and Stumm, 1987; Dzombak and Morel, 1990), we use them as an example. When specific adsorbing ions other than H' and OH'are absent in a system, the charge developing process on the surfaces can be described as follows (Schindler and Stumm, 1987):

SOH2+ = SOH + H+ Kaiint (2-1) SOH = SO + H Ka2int (2-2) Where SOH denotes a surface site and Kalint and Ka2int are the intrinsic surface acidity constants and defined as follows:

Kaiint = [SOH2+]/[SOH][H+] (2-3) Ka2int= [(SO'][H+]/[SOH] (2-4)









Schindler and Stumm (1987) summarized the intrinsic acidity constants (Kalint and Kaint of various metal oxides and found that the constants are generally correlated to those in solution. However, the affinity of protons to the surfaces of metal oxides is much more complicated and influenced by changes in surface composition and structure (Yoon et al.,1979; Sverjensky, 1994; Bleam, 1993). In addition, the values of these constants depend on the particular electrostatic model adopted, such as constant capacitance, diffuse double layer, and triple layer. However, the sum of Eq (2-1) and (2-2) can be expressed as follows:

SO- + 2H+ = SOH2+ (2-5) Eq. 2-5 is simply related to the point of zero charge (PZC or pHo), which is experimentally measured and is independent of electrostatic models for the solid-water interface. When the concentration of positively-charged surface species is equal to that of the negatively-charged, that is, a zero-charge surface, the pHo is related to the intrinsic equilibrium constant

pHo =0.5 log Kaint = 0.5 log Ka (2-6) Eq (2-6) states that intrinsic acidity constant (Kaint) is equal to the apparent acidity constant (Ka) at a surface potential of zero. It not only indicates that pHo is a measure of Kaint, but also raises the question that the bonding of protons at metal oxide surfaces is more analogous to the bonding of similar complex in an aqueous phase (Schindler and Stumm, 1987) or in bulk crystal structure (Sverjensky, 1994). The correlation between the surface acidity constants of metal oxides and those in solution support the former point. However, the scatter on such correlation is substantial for some mineral particles (Schindler and Stumm, 1987). More recently, Blesa et al. (1990) reviewed the








thermodynamic data for the adsorption of protons on metal (hydrous)oxides. They found that the enthalpy values associated with Kaiint and Ka2int are very similar and are much lower than those of the hydrolysis of aqueous metal ions. This is consistent with Lyklema's work (1987a), in which he describes the adsorption of potential-determining ions as a process characterized by a single enthalpy value instead of two. These indicated that it is unrealistic to represent acidic, neutral, and basic surface groups using Eqs. (2-1) and (2-2). The active surface site may be a surface metal complex and notation S

(OH)mo(OH2)n0 (mo and no being the stoichiometric index at pHo) offers a better description of the involved processes; for pHno; for pH>pHo, m>mo and n







1993), Sverjensky (1994) has shown that surface protonation on a wide variety of minerals can be accurately calculated from the dielectric constant of the solid and the ratio of the Pauling bond strength to the cation-hydroxyl bond length of the solid particles. This suggests that much more emphasis should be placed on the analogy between the bonding of the surface protonated species and the bonding in the underlying crystal structure. In Sverjensky's approach (1994), the properties of crystals have been numeralized with its electric constant instead of the qualitative terms such as kinks, edge position, or adatom positions. The details on modeling or calculating pHo can be found in the literature (Parks, 1965; Yoon et al., 1979; Sverjensky, 1994; Bleam, 1993).

Fokkink et al. (1987, 1989) have demonstrated that pH- and temperaturecongruencies exist in the surface charge developing processes of various metal oxides. In a systematic series of experiments in which the surface charge GO of a number of oxides [TiO2 (rutile), RuO2 and a-Fe203 (haematite)] in aqueous solution of KNO3 was measured as a function of pH, ionic strength and some other variables. When ao is plotted as a function of pH-pH0, where pHo is the PZC, the curves coincide for the three oxides at three ionic strengths even though they are different in pHo. In other words, the individual identity of metal oxides makes no difference to the S-shaped plot of co vs. pHpHo. Similarly, in the measurements of the temperature dependence of surface charge development, changes in temperature only affect the positions of PZC, not the trend of the surface charge developing process. They conclude that the electrical double layers in metal oxides can be functionally divided into a specific and a generic part, with respect to the natures of oxide and electrolyte. The specific part, which is determined by the specific interactions of a surface with protons in solution, determines the PZC. This is








consistent with the result of Sverjensky (1994) in concept. The general part, which is controlled by the solution side of the double layer, determines the surface charge development once the surface charge deviates from pHo. It can be described well by the Nemrnst law (y0 =2.303 RT/F (pH-pHo). The significance of this result is, first, that it has provided a thermodynamic basis for double layer model that divides the charged interface into a charged surface and a diffuse layer near the surface; second, that the Nernst law can be used to describe the charge developing process in metal oxides. It has been recognized that the Nernst Equation is valid only when the chemical potential of H+ in interface is a constant (Blesa, 1988). In a practical sense, the potential can be treated as a constant if the amount of H+ in the interfacial layer is much higher than a change caused by the charge developing process (Blesa, 1988). However, for metal oxides only 3-10 sites per 100 A (1 nm2) are usually found (Sigg et al., 1981) and in the experimental pH range up to 60% of the sites may be charged (Blesa, 1988). However, if the notation of the binding site S(OH)m0(OH2)no we discussed above is more representative of the real sites on metal oxides compared with SOH, adsorbed protons can be transformed into neutral species (H20) on surfaces, and thus the adsorbed H+ would not change much with surface charge. This may result in the Nernstian behavior.

In the presence of specific adsorbing ions (SAIs) such as Ca2+ and Mg2+ in

solution, the charge developing process becomes more complicated (Lyklema, 1984; Ardizzone, 1982). Generally, the PZC will be shifted compared with the one in the absence of those SAIs. Lyklema (1984) defined the PZC, even in the presence of SAIs, as a point at which H+ and OH" are just balanced at the surfaces. This is equivalent to the surface charge o = F(FH+-FOH-). The major basis for the definition is the dominant








sensitivity of oxides to H+ and OH'. Consequently, SAIs belong to the Stem charge. Therefore, the PZC in the presence of SAIs remains fully defined. In their definition PZC is still the pH where ao =0. If the cation adsorbs on to the Stem layer PZC is lower than the PZC in the absence of SAIs because cations in the stem layer favor the adsorption of OH- over H+, so that a lower pH is required to restore the H+-OH+ adsorption balance. Similarly, specific adsorption of anions leads to PZC's above the pristine value.

In fact, in the presence of SAIs, the common intersection point (CIP) of titration curves with various ionic strengths has also been observed. Similar to the situation in the absence of SAIs, the CIP is the equal compensation point (Lyklema, 1984), at which charge-compensating cations and anions have an equal affinity to the surfaces. From this point of view, they explained CIP PZC in the case of specific adsorption as follows: at the PZC adsorption of metal cations are favored over that of anions if cations have a higher affinity to the surface. In order to reach a situation where the intrinsic adsorbability of cations and anions is identical (equal compensation point), a positive charge on the surface has to be developed. This charge should be more positive with higher specific affinity of cation to the surface. Once the equal compensation point has been reached, further increase in concentrations of cations and anions do not lead to further shift, i.e., all successive curves at increasing concentration must pass through the point. Similar reasoning applies to a situation at which anions have a higher affinity over cations. This principle is important to understand the impact of soil solution chemistry on soil colloid charge development. Following this point, it is easy to understand the effects of changes in species of cation and anion in solution on PZC and CIP.








In the discussion above, we have been dealing with well-defined crystal surfaces, in which the dissolution can be neglected using surface complexation approaches. However, solubility of metal oxides has to be taken into account, especially in the case of amorphous (hydr)oxides. Under such conditions, the adsorption and desorption of the hydrolyzed metal complex ions may become an alternative to determine surface charge other than the surface complexation of H and OH'. Based on the minimum solubility theory (Parks et al., 1962), PZC should be identical to the isoelectric point (IEP) of an aqueous solution suspending the particles. The IEP is defined as the pH resulting from an equivalent concentration of positive and negative complexes in aqueous phase (Parks, 1965), which is often found at the pH of minimum solubility of a solid. Blesa et al. (1997) have found that the PZC and IEP for metal oxides do not always coincide; however, the deviation from IEP (PZC - IEP) can be explained well if considering the fact that the complex cations and anions require different dehydration energy when they are transferred from solution to surface. They explained that monomeric cations hydrate more strongly than anions in aqueous solution, so the shift should be small and negative when the charge is determined by monomeric surface complexes and the minimum solubility theory is generally valid. On the other hand, for polymeric species, more specific behaviors are expected. Relative large positive or negative shifts may result if the charge surface complexes are polymeric. Obviously, the underlying assumption of these approaches is that the surface active site is more analogous to their solution species. This may be more suitable to soil environment where various amorphous minerals are abundant and aqueous species are more understood than surface species.








Charge development in soil particles

In general, surface charge of soil particles can be classified into two types:

permanent and pH-dependent charges. Detailed characterization of soil particle charges has been developed by Charlet and Sposito (1987), Anderson et al. (1991) and Chorover and Sposito (1995). Charge developing processes on soil particles are more complicated than those on metal oxides because the surface composition and structure of soil particles varies greatly and PDIs are not limited to just H and OH-. However, similar phenomena such as CIP have been observed, which is referred to as the point of zero salt effect (PZSE). It is often found that PZSE deviates from CIP (Chorover and Sposito, 1995).

Soil is a mixture of various minerals and organic substances. Further, there can be significant amounts of amorphous materials, such as amorphous Fe and Al. It is well documented that the existence of hydrolyzed species of Al and Fe have significant impact on the pH-dependent charge of soil particles (Chorover and Sposito, 1995, Parket et al., 1979), and their complexation with organic matter further complicates the charge developing process (Chorover and Sposito, 1995). In addition, mineral dissolution may be significant for some soils and may affect the charge developing process greatly (Blesa et al., 1997). Combining all these factors with the dynamic nature of solution chemistry, one may conclude at the present time it is impossible to accurately predict proton surface charge developing process.

In principle, however, Blesa et al's approach (1990; 1997) is more promising. They emphasize the role of partitioning of charged species between solid surface and solution during surface charge developing processes, which is driven by the difference in Gibbs free energy between a charged species in solution and on surfaces. The underlying








assumption of this approach is that the charged species in solution and on surface are similar. This approach has relative sound theoretical basis and is consistent with experimental evidence.


Charge development in mobile colloids

Considerable amount of research has been conducted to understand charge

development of water dispersive and mobile colloids (Heil et al., 1993a,b; Kretzschmar, 1993; 1996). It is commonly observed that the stability of colloidal clays in soils is many times greater than that of comparable reference clays. Also, clays isolated from surface soils were found to be more dispersive than clays from the subsurface horizons of kaolinitic soils in the Southeastern USA. Factors contributing to the high dispersibility of soil clays include the presence of adsorbed organic matter, minor quantity of smectite in kaolinitic clay, and larger surface roughness of soil clays compared with unweathered reference clays.

Organic matter is an important source of negative charges in soils. Well

decomposed humus may have a CEC > 300 cmol kg' humus, which is considerably greater than that of clays such as kaolinite (3-15), illite (30-40) and montmorillite (80150). It has been estimated that 20-70 % of the CEC of many soils attributes directly to the soil organic matter alone (Vaughan et al., 1984; Stevenson, 1982). In soils, dissolved organic carbon (DOC) tends to adsorb onto solid particles such as clays and metal oxides, which is driven by ligand exchange, multivalent ion bridging, Van der Waals force and hydrophobic interactions between DOC and the solid minerals (Murphy et al., 1995). In particular, for Fe and Al oxides, when the pH of a system is lower than their PZC, adsorption of DOC onto the minerals increases significantly because of the electrostatic








attraction between the two, resulting in a significant impact on their surface charge. There are numerous studies of the effects of organic coating onto metal oxides (Tipping et al., 1982) and clay minerals (Kretzschmar, 1996) on surface charge. Heil and Sposito (1993a, b) found flocculation of illitic soil colloids with organic coatings increased with pH in Ca solution, which is contrary to their flocculation behavior when organic matter is removed by H202. They concluded that the competition between proton and Ca onto the acidic functional groups of organic molecules is essential to the charge development on coated colloids. A similar result was also found by Kretzschmar (1993;1997) in humiccoated kaolinic soil clay. On the other hand, it has been reported that the electrical properties of the quartz surface dispersed in river water (Whitray Bech, UK) is determined essentially by their interactions with inorganic cations such as Al and Ca instead of organic matter (Findlay et al., 1996). In concept, however, these results are consistent with others, i.e., surface electric properties of minerals are a result of interactions both in the interface of particle-solution and in solution, including particleorganic matter, organic-inorganic ions, particle-inorganic ions, etc. Particularly, types of cations, such as proton and multivalent cations, and their concentrations are important for the charge development on particles in addition to organic matter.

It is unrealistic to model surface charge development on soil particles with reasonable certainty at the present time. However, it is evident soil particles can be treated as assemblages of crystalline and amorphous minerals and organic residuals; Sophisticated models such as the triple layer model do not make much sense in describing the electrified interfaces of such particles considering that the particle surfaces are irregular and the Stern layer often moves inward inside of the physical boundary of








the particles. As a first approximation, on the other hand, the interface of soil particles is more reasonable described by diffuse double layer with a specific part, i.e., a charged surface, and a generic part, i.e., a diffuse layer. The surface charge developing process may be described by the partitioning of charged species between solution and surfaces assuming the charged species are similar between the two phases.


Hydration

Water molecules may be orientated surrounding the immersed particles, forming "Hydration shells" (Clifford, 1975). Once the particles with such shells approach each other, additional forces between the particles arise. The origin of these forces is the interactions between polar electron acceptors (Lewis acids) in water and polar electron donors (Lewis bases) from colloidal particles, and termed AB forces as discussed previously (Oss et al, 1987; 1988). Surface electron donicity of colloid particles is crucial for water molecules to be orientated along the surface. Strong surface ele.ctron donicity causes more ordering of water molecules orientating along the surfaces, resulting in a repulsive force between the particles. Weak surface electron donicity causes less ordering of water orientating, resulting in an attractive force. Oss et al. (1990) have demonstrated between particles dispersed in water, the interaction free energy from AB, whether repulsive or attractive, is commonly as much as 100 times greater than LW energy, and 10 or more times greater than EL energy at close range (1-5 nm). It can account for up to 85% of the interactions between mineral particles in aqueous solution, such that the DLVO theory combined with the AB force can adequately describe some anomalous colloid stability (Oss et al, 1990; Wu et al, 1994a,b). Wu et al. (1994a,b) have experimentally evaluated the interactions between the particles of montmorillonite and








calcite by examining flocculation of their suspensions while increasing Ca2+ concentration. They concluded that Ca2+ not only causes a decrease in particle 4 potential, but it also drastically lowers the electron donicity of the polar surface of the particles, resulting in an AB attractive force, which is more responsible for the flocculation of the suspensions. Obviously, SAIs such as Ca2+ alter not only ( - potential when it adsorbs onto a particle, but also decreases the surface electron donicity. This result means that surface electron donicity and ( - potential of particles are related.


Size development of mobile colloids

The behaviors of colloids are size-dependent. Individual colloidal particles may aggregate depending on the magnitude of the energy barrier of their interaction (Figure 21), which is similar to colloid deposition that will be discussed later. Briefly, if the energy barrier of interactions is low, fast coagulation occurs, in which case the diffusion rate of colloids is the limiting factor; the aggregation rate is otherwise controlled by the interaction of colloidal particles. These two mechanisms result in lower and higher fractal dimensions of aggregates, respectively (Riscovic et al., 1996). However, direct measurement and theoretical simulation of colloid size distribution in natural system have shown that the concentration of colloids < 0.1 [tm should be negligible because of their almost instant coagulation (Filella et al, 1993; Buffle et al., 1995). Kaplan et al. (1997) examined the possibility of aggregation of mobile colloids from a reconstructed soil profile. They analyzed the size distribution of particles in the suspensions with and without sonification, which is supposed to break down aggregation in the suspension. They found that the suspension without sonification exhibits almost the same bimodal particle-size distribution as the one with sonification. This holds for soil suspensions








collected from two different soils: a loamy sand and a sandy soil in Southeastern USA. Their results suggest that colloid aggregation in soil pore space is very limited especially when soil colloid particles are highly charged (Kaplan et al., 1995; 1997). Interestingly, similar bimodal size distribution of mobile colloids has been reported by Ronen et al. (1992), who characterized the suspended particles collected from groundwater in a coastal plain phreatic aquifer of Israel. Kaplan et al. (1996) explained that smaller particles are less likely to be strained during transport in porous media based on the filtration theory (Yao et al., 1971). This suggests that, in the bimodal distribution, colloids of smaller size (< 0.4 jim) may have resulted from long-distant translocation to sampling point, while those of larger size (<1 pim) may come from local dispersion at the sampling point.


The Characteristics of Porous Media

Mobile colloids interact with the surfaces of porous media, which is more

important than those between colloids in limited pore space. Therefore, pore structures and their physical-chemical surface characteristics have tremendous influence on colloid mobility. Extensive studies have been done in the systems consisting of clean surfaces of porous media (Adamczyk et al., 1983; Kallay et al., 1987. Elimelech et al, 1995). However, large discrepancies in colloid deposition rate under unfavorable deposition conditions between theoretical predictions and experimental observations have been found. This has been attributed to the hydrodynamic effect and surface heterogeneity of the porous media, such as surface roughness and local charge heterogeneity. This has been considered to be the most promising approach to understanding the interactions between colloids and porous media in clean systems (Song et al., 1994). There are some








exhaustive reviews on this topic (Swanton, 1995; Ryan and Elimelech, 1996; Elimelech, et al., 1995).

In a soil system, the porous media is an assemblage of a wide range of particles, which vary greatly in mineralogy, surface composition and dimensions. Different from pure systems, where the surface properties of porous media may be far different from those of colloids, the surfaces of soil porous media may at least in part consist of potentially mobile colloids. In sandy soil, a small amount of clay-size particles tends to pack closely along large sand grains driven by energy minimum. This arrangement among a grain and colloidal particles has also been confirmed by fractal analysis (Bartoli et al., 1991). Based on this fact, the interactions between colloids and porous media can be treated as those between colloidal particles, as a first approximation, which has been a long-established concept. Swanton (1995) has suggested that in an undisturbed soil profile, the active site of the porous media may be completely occupied by mobile colloids. Ryan and Gschwend (1994a) have successfully related the observed colloid release rates from an Fe oxide-coated sand columns to EXP (#max-#min)/kBT. Their estimation of the energy barrier (#max-,min) is based on the interactions between colloids rather than those between colloids and porous media surface. The underlying assumption is that porous media surfaces are identical to colloid surfaces.



Deposition of Colloids

Theoretical background in well-defined porous media

Colloid mobility through porous media is mainly determined by the net rate of colloid deposition and release. Generally speaking, deposition of colloids onto porous








media can be viewed as a two-step process: the transport of colloids from bulk solution to the proximity of the surfaces and then their attachment to the surfaces, which depends upon the nature of particle-surface interactions. If the energy barrier (max) is < 0, the flux of colloid transport is equal to that of deposition. In other words, the collection efficiency of the collector (porous media) is 100%. The energy barrier Omax must otherwise be overcome in order for a colloidal particle to attach to the surfaces (Figure 21). Therefore, deposition rate constant (kdep) is related to the energy barrier Omax with kdep oc EXP (-max/kBT), where kB is the Boltzmann constant and T is absolute temperature (Ruckenstein et al., 1976). The major mechanisms of colloid transport to a collector are inertial impaction, interception, sedimentation, electrostatic forces, Brownian diffusion, and straining. Collection efficiency contributed from each of them has been described by Tien (1989), and the details will not be repeated here. In porous media, however, many factors mentioned above may be operative simultaneously. The overall collection efficiency can be theoretically calculated based on the filtration theory (Yao et al., 1971). However, the great discrepancy between prediction and experimental observation under unfavorable deposition conditions has been observed. This has been attributed to the distribution of surface and physical properties, surface charge heterogeneity of solids, surface roughness, interfacial electrodynamics and colloid deposition in secondary minima (Elimelech et al., 1995). In order to take this discrepancy into account, collision efficiency (o=r/1o) is used, where f and rno are observed and calculated collection efficiencies, respectively. As particles deposit on a collector, after an initial stage, the deposition rate will change depending on the nature of particle-particle interaction. If the net interaction is repulsive, the collector surfaces become progressively occluded as








particles accumulate and colloid deposition rate declines accordingly. This surface exclusion phenomenon is termed blocking. There are some excellent recent reviews on this topic (Ryan and Elimelech, 1996; Elimelech et al., 1995; Swanton, 1995).



Colloid deposition in soil

Transient phenomena in colloid deposition. There are no clean or pure surfaces in soils. Therefore, the results obtained in well-defined porous media cannot be directly extrapolated to describe colloid deposition in soils. In fact, in any given time in most soils, the most favorable deposition sites are likely to readily be occupied. Alternatively, if there is influx of mobile colloids as a result of some chemical or mechanical disturbance, there may be limited deposition during the early stages of exposure. This occurs because favorable deposition sites are already filled, which is different from clean porous media. Over a prolonged period of exposure for a colloid flux, on the other hand, soil porous media may develop surfaces that are more similar to those of mobile colloids (Swanton, 1995), which has been discussed before. In soil porous media, therefore, colloid deposition may be a transient phenomenon.

Dispersable fraction of soil clay resulted from vigorously shaking in water can be a relative measure of the amount of potentially mobile colloids. Miler et al. (1986) observed the amount remaining in suspension after 36 h of shaking was highly correlated to surface soil loss in the Southeaster US under high intensity rainfall. Kaplan et al. (1997) have found mobile colloids from two soils are similar in mineralogy to the water dispersible clays and that they are many orders of magnitude lower than the amount of water dispersible clay in the soils. In other words, there is significant amount of








potentially mobile colloids loosely attached to the soil porous media, however, during this period they function as a part of stationary porous media. They may be detached from the porous media (mobilized) once they are exposed to a chemical or hydrological perturbation. I have referred to this stage a transient phenomenon in colloid deposition. In fact, this is not a new concept at all; however, it has to be emphasized to understand the distinctively different features of colloid mobility in soil porous media. Take surface soil for an example. A perturbation such as rainfall occurs on the soil surface and moves downward. Such a perturbation may result in a release of deposited colloids, and then the mobilized colloids move down over the surfaces of porous media consisting of the deposited colloids. Therefore, at the front of the downward-moving perturbation, there are two opposite processes operating simultaneously: deposition and release of soil colloids. The former is mainly influenced by the blocking effect, and the latter will be discussed in the following section.

Size straining in deposition. If the effective size of a colloid particle is larger than the smallest pore through which fluid is flowing, the particle is retained by porous media, which is referred to as size straining. Straining occurs in granular bed filtration if the ratio of the suspended particle diameter to the grain diameter is greater than about 0.2 (Herzig et al., 1970). Straining is one of the most important mechanisms of colloid deposition in soils (Seta et al., 1997; Jacobsen et al., 1997; Kaplan et al., 1997). Seta et al. (1997) have examined the transportability of water dispersible clay through intact soil columns and found that size straining is an important factor in determining the concentrations of colloids in leachates along with pH, total exchangeable bases, cation exchange capacity, organic carbon content, etc. They concluded the relative low colloid








transportability is attributed to larger colloid size, i.e. size straining. The straining may manifest itself in another way. Jacobsen et al. (1997) examined the transportability of illite though soil columns containing macropores, and found illite concentrations moving through soil columns are significantly related to the sizes of active macropores in soil columns. However, no significant difference in mobility through the columns between the illite and that coated with humic acid was found. This suggests size straining is the predominant process in determining the deposition over the zeta potential and steric effect, resulting from the coatings of humic acid. In their study, this is further confirmed by the fact that colloid concentrations in leachates increase with the intensity of infiltration, as they argued, a high intensity of infiltration depresses the effect of straining. However, this is contrary to the results of Kretzschmar et al. (1995). Kretzschmar et al. (1995) found colloids coated with humic acid result in a much lower collection efficiency in columns packed with "clean" saprolite particles. In fact, it is this contradiction that demonstrates the difference of colloid deposition between soil porous media and clean porous media: size straining may be a dominant mechanism if blocking effect is operative in soil.



Release of Colloids in Soil

Different from colloid deposition, which is essentially determined by the interactions at larger separation distance, colloid release depends on interactions at separation between the primary minimum (0min) and surface (Figure 2-1) and is greatly influenced by additional repulsion at short separation. Colloid release is generally expected when the repulsion is generated between porous media and colloids. The








efficiency of colloid release depends on whether the process is caused by diffusion alone or by an applied external force (Kallay et al., 1987). Without external forces, the rate of particle release is proportional to the diffusional escape probability of the deposited colloids through the energy barrier, i.e., the energy well (dmax-O min). It is otherwise determined by the shape of interaction energy profile between surface and the primary energy minimum (Figure 2-1) as well: the smaller the slope (do/dh) of the curve the less force needed for colloid release, where h is the separation distance. Up to this point, it is clear that the two opposite processes, colloid deposition and release, are controlled by different mechanisms and factors. For a given system, mobile colloid concentration is determined by the relative magnitude of colloid deposition and release rates. The residence time of colloid-carrying solution in the system also plays an important role. A longer residence time represents a condition closer to the equilibrium of colloid deposition and release, while a short one may discriminate deposition over release and vice versa. For example, colloid dispersion in batch experiments, in general, is a result of balancing between colloid deposition and release because of the longer residence time in general; whereas colloid release from a short column may be free of redeposition effects. Therefore, one must take extra caution when extrapolating colloid stability obtained from batch experiments to colloid transportability in column experiments where kinetic effects are more pronounced.

In a soil, there are numerous energy profiles for various colloids depicted in Figure 2-1 due to the heterogeneity of colloids and soil porous media. Each energy profile represents the interaction between a individual particle at a specific location on porous media. For simplicity, we imagine that those curves can be lumped into several








groups based on similarity, each of those lumped profiles representing the interaction of a group of similar particles with porous media with a certain magnitude and position of the energy maximum and minimum. Therefore, the deposition and release of soil colloids may manifest themselves in a strong chromatographic manner (Chapter 5). The processes of colloid release

Similar to colloid deposition, colloid release rate is generally determined by both colloid detachment from porous media and transport to bulk solution. Ryan et al (1994a) examined the release rates of colloidal hematite from hematite-coated sand columns under different ionic strengths and flow rates. They found that colloid release rate decreases as ionic strength increases, which was supposed to reduce the size of the energy barriers. In addition, colloid release rate also decreases with increasing flow rate, which is contrary to the expectation that a greater mobilization would occur at a greater flow rate due to greater hydraulic stress on deposited colloids. They suggested the ratelimiting step in colloid release is the transport of detached colloids to the bulk fluid when rapid colloid release corresponds to conditions where the energy barrier has vanished from the potential energy profile. Coincidentally, a similar phenomenon has been observed by Jacobsen et al. (1997) in intact soil-column experiments where natural particles are leached with tap water. Their results showed particle mobilization is not influenced by increase in flow rate. Further, the plot of accumulated amount of mobilized particles versus square root of time shows a fairly linear relation, implying diffusion limited kinetics. One necessity for diffusion kinetics being dominant is the barrier maximum of mobile colloids vanishes, which may be very common in soil because of the transient phenomenon discussed before. The lack of flow rate effect in the








latter study may be caused by the smaller slope of energy profile (Figure 2-1) for potential mobile colloids at the given condition, which may be overcome by increases in hydraulic stress, subsequently, these potential mobile colloids become truly mobile. Therefore, colloid dilution in effluent may be compensated by its enrichment from mobilizing more potential mobile colloids by increased flow rates. This is based on the heterogeneity of the interactions between colloids and media in soil. Mobile colloids and water-dispersible clay

Kaplan et al (1993; 1994; 1997) examined extensively the release of colloids from an Ultisol in the southeastern coast of the US. Mobile colloids were collected during and after mild rain events from the reconstructed pedons. They found the mobile colloids are similar to the water-dispersible clay in mineralogy, which is consistent with the results reported by Seta et al. (1997). However, there were some differences. Kaplan et al (1997) found that the sizes and compositions of the mobile colloids differed from those of the water dispersive clay, with the former being smaller and generally enriched with kaolinite, Fe oxides, gibbsite, and organic carbon. In addition, compared to the total clay fractions in the reconstructed pedon, the mobile colloids are more dilute in quartz and HIV (hydroxy-interlayered-vermiculite). Based on the results from scanning electron microscope (SEM) and photon-correlation spectroscopy, essentially all the mobile colloids (>90 %) have diameters of about 0.23 pm and moved through the soil as discrete, non-aggregated particles. They concluded the particles enriched with mobile colloids are not only readily dispersible but also smaller in size than water dispersive clay. Further, they proposed colloid mobilization in the pedons is a result of two consecutive processes: dispersion of highly charged particles due to changes in soil








chemistry and induced water flow, which is determined by particle composition and can be evaluated by water dispersibility of a soil, and their transport through a pedon, which is size-dependent. However, it has to be realized that water dispersibility of soil particles obtained from a batch experiment is not always equivalent to colloid's ability to stay detached (mobilized) in a soil. Colloid residence time has to be taken into account, which has been discussed above and is supported by Kaplan (1996). Influences of exchangeable sodium percentage (ESP) on stability of water-dispersible clay and mobilization of soil colloids

Kaplan (1996) summarized if particle straining is not a limiting factor for colloid release, the dispersability of soil water-dispersible clay may correlate to colloid transportability in soils. The dispersability of water-dispersable clay has been studied extensively (Miller et al, 1990; Sumner et al., 1993). Even though we cannot completely understand the dispersability of soil colloids, it is undoubtedly affected by factors such as pH, ionic strength and exchangeable sodium percentage.

Kaplan et al. (1997) reported that colloid concentrations in the effluents from soil pedons are highly correlated to the ESP of those soils (R2=0.91-0.93, p< 0.01), but not so to soil pH or total electrolyte concentrations. They explained the dispersive property of the highly hydrated Na+ ion attributes to the positive correlation as the dispersion of the soil colloids appears to be a more important process in colloid release in the sandy soils. In fact, the positive correlation between soil colloid dispersability and exchangeable sodium percentage has been found by other researchers (Frenkle et al., 1978; Suarez, 1984) who investigated the effects of irrigation or rainfall on hydraulic conductivity in a sodic soil.








Seta and Karathanasis (1997) examined the relation between the stability and transportability of water-dispersible soil colloids by pumping water-dispersible clay through an intact soil column. They found colloid recovery, after five pore volumes, depends on not only the type of soil column but also the characteristics of colloids. Among the colloid properties, pH and total exchangeable bases are significantly correlated with colloid recovery. The effect of pH on colloid migration can be explained by its effect on colloid stability: the higher the pH above the pHo (< 4.0), the greater the colloid stability. The strong correlation between total exchangeable bases and colloid recovery is anticipated because of its high correlation with pH. In addition, they stated "contribution of total exchangeable bases to colloid transport may derive from cationexchange reaction between the colloid-saturating cations and the column matrix, which reduces colloid interaction with matrix surfaces and thus enhances colloid transport". Finally, they concluded the best single independent variable predicting colloid recovery is total exchangeable bases (R2=0.97).

Even though there is some inconsistency among the results in different studies

described above, there is sufficient information to show ESP is consistently correlated to the dispersability of soil colloids in a soil. Rengasamy (1982), however, demonstrated that even Ca-saturated clay, i.e., ESP=0, could be dispersed provided the soil is free from electrolyte by dialysis, suggesting exchangeable Na is not essential for dispersing a soil. Further, in a soil system, solution chemistry is dynamic and, therefore, concentrations and types of cations in soil solution and exchange sites may vary greatly. In this sense, parameters such as total exchangeable bases, which takes cations besides sodium into account, may be more reasonable to describe colloid transportability. The effects of








cation exchange reactions on colloid release have been demonstrated by Roy et al. (1996b), which will be-discussed in the following section.



Ion transfer processes during colloid release

Observation of ion transfer during colloid release in well defined systems. Kallay et al. (1986) examined the effects of neutral electrolytes on the detachment of spherical colloidal particles of goethite from glass surface in basic media. They unexpectedly found colloid removal was enhanced with increasing concentration of NaNO3. Such unusual behavior could be predicted if constant potential is assumed. They explained this behavior corresponds to a double layer that undergoes relaxation during particle detachment. In order to keep the potential constant as the distance between the particle and the surface increases, the adsorption of the potential determining ion (OH) must occur with consequent equilibration of all ionic species in the Stem and diffuse layers. During particle detachment, ion transport from solution phase into the interfacial layer is accelerated with increasing distance between the surfaces due to solution influx from bulk. Kallay et al. (1987) summarized this unusual behavior can be observed under the following conditions: colloid redeposition is minimal (short column); particles and media have alike charges; surface potentials are constant during colloid detachment. They argued this apparently unusual behavior can be understood if one takes into consideration the energy profile as a function of distance near the surfaces (Figure 2-1). The probability of colloid detachment depends on the depth of energy well (maxd4min), which is a function of several parameters. On the other hand, the probability of colloid. deposition depends on the height of the energy maximum (max). As the concentration of








NaNO3 in the system increases, the depth of energy well reduces, resulting in an increased rate of colloid detachment; at the same time, the energy maximum decreases, resulting in an increased rate of deposition. The amount of colloidal particles observed in effluents is determined by the net rate of these two opposite processes.

The essence of Kallays (1986, 1987) rationale is that a change in the electrical environment of the interface, resulting from colloid detachment, may induce ion flow between solution and interface (EDL), which in turn may facilitate colloid detachment. This has been known as so-called surface charge- or potential regulation, which has been discussed by Chan et al. (1974). They have shown there may be significant ion flow between solution and interface during EDL interaction if the particles carry pH dependent charge. This indicates ion transfer (He) may also be important during colloid release since pH-dependent charge is generally abundant on soil colloid particles. Ion transfer during soil colloid mobilization in literature

Shainberg et al. (1981 a) examined the effects of electrolyte concentration on hydraulic conductivity (HC) of a sodic soil. One of the major mechanisms of the influence of electrolyte concentrations on hydraulic conductivity is by mobilizing soil colloids and then plugging them into conducting pores. They found both clay dispersibility and hydraulic conductivity of the soil were very sensitive to the level of exchangeable Na in the soil and to the salt concentration of the percolating solution. When salt concentration in the soil solution was 3.0 meq/liter, clay dispersion increased and HC decreased only if ESP >12%. Conversely, when salt concentration was maintained at - 0.5 meq/liter, clay dispersion increased and HC decreased when ESP >12%. These results indicate that ESP itself can not determine soil dispersibility, and salt








concentrations in solution have to be taken into account. A simple principle behind this is colloids flocculate when the critical flocculation concentration of a salt is reached. The higher the ESP of the clay and the lower the salt concentration in solution, the higher the tendency for a soil to disperse, and consequently the HC decreases. Further, they concluded the response of soils to low ESP and leaching with low electrolyte water depends on the concentration of electrolytes in the soil solution that the solid phases of each soil maintains (Shainberg et al., 198 1b). They demonstrated salt concentration in solution was determined by the dissolution rate of soil minerals, and higher mineral dissolution rate resulted in less impact of exchangeable Na on soil dispersion. Obviously, soil dispersion results from equilibrium of colloid deposition and release. However, it is clear that the interaction between cations on exchange sites and bulk solution significantly impacts colloid mobilization.

There is sufficient evidence exchangeable Na has significant influence on colloid release from soil when it is exposed to an influent with low ionic strength. Cummins and Kelley (1923) first demonstrated the hydrolysis of exchangeable Na+ in soil, or the replacement of exchangeable Na+ by H+ from the dissociation of water. They found, in the absence of CaCO3 and CO2, a Na-saturated soil leached with distilled water yields a NaOH solution. In an experiment with Na-montmorillonite, where the hydrolysis products are removed continuously, Bar-on and Shainberg (1970) found a Na+ concentration of 0.1 M in the effluent. Shainberg (1973) demonstrated Namontmorillonite releases Na+ even when the reaction products are not removed. He found the specific conductance of the clay suspension is proportional to the square root of time. These observations are consistent with a hydrolysis mechanism consisting of two








consecutive reactions: a rapid exchange between exchangeable Na+ and IH+, which results in an acidic surface; a slower, first-order transformation of H clay to Mg or Al clay, which increases the amount of exchangeable Na+ release. In soils, similar phenomena have been observed. Oster and Shainberg (1979) demonstrated that, washing three aridzone soils with distilled water, the release of electrolytes can be related to the square root of time exhibiting two linear rates in time sequence. They concluded that the first rate, the more rapid of the two, which occurred right after the washing (<1-2 h), depends on exchangeable Na. Increasing ESP causes increases in the release rate in electrolytes.

In general, the flux of each ion species i across the boundary of the EDL around a particle is governed by the convective-diffusion equation (Van der ven, 1989):

Ni = -z, ul F ci Vy -DiVci +ci v (2-7) Where Ni, Vy, Vci and v are the flux of ion species i, electric field acting on i, concentration gradient of ion species i from bulk solution to interface and the bulk velocity of the fluid motion, respectively. They are all vector quantities. zi F, ui, ci and Di are the charge per mole, mobility, diffusion coefficient, and bulk concentration, respectively, of ion species i. Once a colloidal particle starts to detach from an electrified surface in a given system, the electric field acting on i in the surrounding EDL will change (d (Vy)/dt) in the course of colloid detachment, resulting in changes in Ni because Vci and v do not change much during colloid detachment under stagnant flow during a very short period. With this in mind, the experimental observations we just discussed above can be understood conceptually. Once the electric field acting on a colloid decreases during colloid detachment, exchangeable Na is more readily able to diffuse to bulk solution compared to Ca2+, Mg2+, K+ and etc. This has been confirmed by








the fact that electrolyte concentration in solution increases much faster at higher ESP for a soil (Oster et al., 1979). In essence, this is a process of EDL expansion, which results in increases in repulsive forces between soil colloids or colloids and porous media, and thus mobilization of soil colloids. Therefore, it is easy to understand the observation that colloid mobility are proportional to ESP (Kaplan, 1996) as well as total exchangeable bases (Seta et al., 1997). This is consistent with the study by Roy et al. (1996), which is based on cation exchange reactions.


Ion transfer during soil colloid release in our research

We have examined how colloid mobility in a Pb-contaminated soil, collected from Montreal, Canada, is affected by water-flooding incubation (Chapter 3). The soil packed in a series of short columns was incubated for approximately 3, 20, and 80 d, respectively. After the standing water on the top of soil columns was removed, 0.01 M CaCl2 solution was pumped through the soil columns until the soil was saturated with CaC12. This is designed to eliminate the artifacts resulting from column packing and to use Ca as an index for ion transfer during colloid release. The influent was then switched from CaCl2 to deionized distilled water (DDW) and effluent of -11.5 mL was continuously collected using a fraction collector. Total and dissolved metal concentrations were analyzed. The relation among colloidal Al, Fe, dissolved Ca, and pH with pore volumes of the effluent is presented in Figure 3-2 to demonstrate the proposed mechanisms for Ca release.

After saturating the soil columns with 0.01 CaCl2 solution, Ca was the dominant cation in both bulk solution and electrical double layers surrounding colloidal particles or stationary solid phases in the soil. When the effluent is switched from 0.01 M CaCl2








solution to DDW, the release of Ca from the soil is determined by various mechanisms, which most likely comes to play sequentially. Firstly, at the beginning stage of switching to DDW, Ca release is from the bulk solution, which is confirmed by the fact that Ca concentrations and pH in the effluent at pore volume = 1 are approximately those of the influent ([Ca]= 0.01 M, pH=6.95) (Figure 3-3). Secondly, Ca diffusion from EDL to the bulk solution driven by a concentration gradient occurred from two to 17 pore volumes. When Ca in the bulk solution was depleted, Ca cation moved against electrostatic attraction away from the surfaces, extending EDL surrounding colloids and solid phases to mobilize colloids. Obviously, any significant expending of EDL and resulting colloidal mobilization have to be associated with significant amount of Ca released into bulk solution. In this case, the excess Ca2+ in the diluted bulk solution may be combined with hydroxyl to release proton (Ca2+ + H20 ++ CaOH" + H+) (Oster et al., 1979), resulting in a decrease in pH, which is consistent with our data (Figure 5-3) and consistent with the mechanism we proposed above. Generally, Ca hydrolysis is weak in solution and, thus, not likely to bring the pH down as much as one unit as shown in Figure 5-3. In a heterogeneous system such as this one, however, deficiency of negative charge existed in the bulk solution, hence it is possible that the charge deficiency may further drive Ca hydrolysis. Thirdly, Ca desorption from colloid surfaces and stationary solid phases occurred from 18 to 23 pore volumes (Figure 5-3). This is a slow process and, therefore, occurred at high pore volumes, resulting in an increase in repulsive forces between colloids and stationary solid phases. Increases in Ca concentration and pH were observed in the effluent in pore volumes 17-23 (Figure 5-3). In essence, this process is equivalent to sorbed Ca displaced by H+. However, a direct validation is needed.








Nevertheless, this could be a mechanistic illustration for ion transfer between EDLs and bulk solution as described by Eq (2-7). Significant ion transfer (Ca and OH') during colloid release provides a more promising approach to evaluate colloid mobility in a soil since modem colloid theory has found serious difficulty in soils because of its heterogeneity in both structure and composition of soil particle surfaces.


Association of Colloids with Heavy Metals

There are two major issues involved here: sorption capacity of heavy metals to soil colloidal particles, which has been well documented (Hayes et al., 1990; Dzombak, et al., 1990) and will be briefly described in the following section, and influences of metal sorption on colloid mobility. The association of colloids with heavy metals can be classified into adsorption, precipitation (surface precipitation), and ion exchange. Their influences on surface charge are possibly by virtue of heavy metal ions functioning as specific adsorbing, potential determining, and indifferent ions, respectively. Since most heavy metals are low in solubility in soils and their concentrations are much lower than those of electrolytes in soil solution, it is rare for them to function as indifferent ions. The same reasoning may also apply to the role heavy metal ions play as potential determining ions. However, it has been realized that adsorption is one of the most important mechanism of heavy metal association with colloids (Hayes et al., 1990) and colloid-facilitated heavy metal transport (Kretzschmar et al., 1997), which will be discussed next.

In soils, colloids are heterogeneous, including weathered minerals (clay, metal oxides. etc.), CaCO3, silica, large organic molecules and cellular debris. However, they








may be classified based on surface functional groups: surface hydroxyl, carbonyl, organic complexation group, etc. As an example, we will discuss hydroxyl surfaces in detail. Adsorption of heavy metals to colloids with surface hydroxyl

Surface-hydroxyl-bearing minerals are abundant in soil, such as metal oxides,

quartz, and the edge of kaolinite. Similar with the expression for the de- and protonation processes described previously, surface complexation reactions of surface hydroxyl with heavy metals can be written as follows (Stumm et al., 1987):

SOH + Mz+ = SOMizI) + H Kl, app (2-8) 2 SOH + Mz+= (SO)2M(z-2) + 2 H+ K2, app (2-9) Where SOH denotes a surface site and Mz+ represents a metal cation. Kl, app and K2, app are the apparent surface equilibrium constants and defined as:

KI, app= [SOM'z-i)][H+]/[SOH][Mz+] (2-10) K2, app= [(SO)2M(z-2)] [H+]2/[SOH]2[Mz+] (2-11) where [] indicates concentrations. Because of the charge characteristic of solid surfaces, ion activity at the surface needs to be corrected to obtain the intrinsic equilibrium constants. Accordingly, the intrinsic equilibrium constants can be expressed as:

Ki, int= K1, app exp[(z-1)Fy/RT] (2-12) K2, int= K2, app exp[(z-2)Fy/RT] (2-13) where y is the potential difference between the binding site and bulk solution. The exponential term accounts for the coulombic contribution to the intrinsic equilibrium constants. Because the surface potential cannot be determined experimentally, it is generally formulated based on a variety of models, such as constant capacitance, diffuse layer and triple layer models. It has been found that these models are equivalent in








predicting metal adsorption on surfaces. Applying mass and charge conservation laws to titration data and correcting for the coulombic effect from the charged surface, the intrinsic equilibrium constant for metal ions binding to surface can be determined; however, it is somewhat model-dependent.

It has been found stability constants of surface complexes correlate with those of the hydroxo complexes in aqueous phases (Schindler et al., 1987). This could be supporting evidence for Eqs (2-8) to (2-12), which assume a surface complex is an analogous to aqueous complex and that Gibbs free energy of sorption is the sum of intrinsic and coulombic terms. Recently, Blesa (1990, 1995) pointed out the importance of ion hydrolysis during adsorption, and Sverjensky (1993) has divided the intrinsic terms into two: a solvation contribution and a remaining term. Thus the overall free energy of adsorption of an ion can be written as:

AGads = AGii + AGs + AGcoul (2-14) In this equation, the coulombic term (AGeoul.) represents its contribution to the overall free energy of adsorption owing to the interaction between the ion and surface charge, the solvation term (AGs) accounts for the role of solvation during sorption of the ion, and the remaining term, named an ion-intrinsic term (AGii), is assumed to be a property of the ion alone. With this approach, the experimental results for Pb2+, Cd2+, Cu2+, Mg2+ and Ca2+ can be closely reproduced (Sverjensky, 1993).

If ligands other than H20 and OHf exist in solution, such as Lewis acids or bases ( often the case in soil), competition may occur among dissolved ligands, heavy metals and surface adsorption sites. The competing anions (e.g., sulfate, phosphate or dissolved organic carbon, DOC) can alter surface sorption, and formation of aqueous metal-DOC








complexes stabilizes the metal in solution and effectively decreases the amount of aqueous metal ions available for adsorption. In addition, anions such as sulfate and phosphate will combine with metal cations, reducing the metal adsorption on solids. On the other hand, metal sorption may be enhanced by the formation of ternary surface complexes via sulfate and phosphate and DOC. At the present time, there is no theory that can describe these complicated processes well. However, in principle, the partition of any aqueous species may be described under the framework of Eq N4) (Blesa, 1990; 1995; Sverjensky, 1993).

In metal oxide systems, heavy metal ions interact with surfaces as specific

adsorbed ions (Lyklema, 1984; Ardizzone, 1982) to alter charge development, which is similar with the situation described previously. Effects of heavy metal adsorption on the charges of soil colloids

A number of studies have indicated the importance of particle-solution

interactions in determining surface properties of colloids in natural waters. Electrokinetic measurements made on natural particles dispersed in sea water and natural fresh waters have shown that their electrophoretic mobilities tend to fall within very limited ranges of negative values. The apparent uniformity of this surface electrical property has been attributed to the adsorption of organic materials, particularly humic compounds, at the particle surfaces, which has been supported by a number of studies using synthetic systems. It has been shown that the effects of adsorbed organic compounds on surface properties of minerals present in natural waters are modified, to some extent, by additional cation interactions. In this sense, the interaction of heavy metals with the surfaces of soil colloids cannot be described by the interactions between heavy metals








and pure minerals, rather by those between heavy metal and the complexed surfaces of "real" soil colloids. This complication includes the irregularity of colloid shapes and three-dimensional interfaces between solution and colloids. One of most important examples is organic-coated oxides. It has been proposed as a model: individual colloid particles mostly consist of inner mineral cores with sorbed Fe, Al & Mn (hydroxy)oxides and/or organic matter (OM) (Day et al, 1994), which generally possesses similar composition with soil colloids as well as three dimensional interfaces.

There has been a fair amount of research done dealing with how ionic strength or pH affects the mobility of the colloids with the constituent components above (Day et al, 1994). However, little is known about the effects of heavy metals on colloidal charge development and stability of the complex colloids. Only recently, Kretzschmar et al. (1997), have explicitly taken into account the effects of adsorption of Pb2+ and Cu2+ onto humic-coated oxide colloids on colloid charge development and transportability. They investigated the influence of adsorbed heavy metals (Cu2+ and Pb2 ) on the transport and deposition kinetics of submicron size hematite, coated with humic acid suspended in Ca2+ solution, through a natural soil. They found that replacement of Ca2+ by Pb2+ while holding the total concentration (Ca2++ Pb2+) fixed resulted in a slight decrease in electrophoretic mobility of humic-coated hematite colloids. When Ca2+ was completely replaced by Pb2+, the suspensions were destabilized and aggregated within 20 h. In contrast, replacement of Ca2+ by Cu2+ had very little effect on electrophoretic mobility and colloidal stability. Both Pb2+ and Cu2+ are known to bind much more strongly to humic substances and Fe oxide surfaces than Ca2+. At the pH of the suspensions investigated (pH 5.7), Cu2+ has a higher affinity for humic substances than Pb2+, but Pb2+








adsorbs more strongly to hematite than Cu2+. In fact, the interface is three-dimensional since the humic coating can be considered an extension of hematite particles to the solution phase. The bonding difference may direct Pb2+ and Cu2+ to locate preferentially at Stern and diffuse layers, respectively. Therefore, it is expected that the effect is more pronounced by replacing Ca2+ with Cu2+ than with Pb2+ in the colloid suspensions on electrophoretic mobility and colloid deposition. Their results suggested that humiccoated oxide colloids can be stable and mobile in the presence of strongly adsorbing trace metals.


Partitioning of heavy metals in soil colloids

The ability of colloids to facilitate heavy metal transport is largely determined by the distribution of heavy metals among solution, mobile colloids, and stationary grains. Mills et al. (1991) first incorporated partition coefficients into their model in describing colloid-facilitated metal transport in porous media. However, a mechanistic understanding of metal distribution among different phases is still missing. One of the major reasons is the dynamic feature of the distribution of heavy metals, and phase transformations among them, such as from mobile to immobile colloids, once heavy metals are introduced to a system. Kalplan et al. (1995) have found that the mineralogy of mobile colloids from contaminated wells is significantly different from those from the contaminated soil on one site, suggesting the association with heavy metals may alter greatly the mobility of original mobile colloids.

Amrhein et al. (1993) examined the potential of colloid-facilitated heavy metals in roadside soils receiving deicing salts. There are two deicing agents considered, NaCl and calcium magnesium acetate (CMA). In their experiments, a series of 60-mL syringes








were packed with 30 g of soil. The columns were leached on a centurion vacuum extractor with three consecutive 30-mL aliquots of either 0.1 mol L' NaCI or CMA. After initial leaching with either of the salts, the columns were leached with three consecutive 30-mL aliquots of deionized water. This is to simulate the input of salty runoff water to the roadside soil, followed by snowmelt or rainfall. A portion of each leachate was saved without any further filtration and the remainder of each solution was filtered through a 0.45 p.m membrane filter. Then the filtrates were immediately placed into a stirred ultrafiltration cell with a membrane of 1000 MWCO. This procedure separates the particles into three fractions: >0.45 ptm, between 1.0 nm and 0.45 pm and <

1.0 nm. The results have shown that the cumulative leached metals, through the whole process, vary with the initial salt input. In general, NaCl tends to mobilize soil colloids more than CMA ( Amrhein et al., 1993). On the other hand, because of the possibility of artifacts resulting from column packing in their study, it may be more reasonable to examine the leached cumulative metals excluding the first several pore volumes. Figure 2-2 is a plot of leached cumulative metals after switching the influent from salt solution to deionized water. It is well established that reduction of ionic strength may mobilize soil colloids. As shown in Figure 2-2, the concentrations of each metal (Cu, Pb, Fe, Ni





50







100000
Cu MNi 10000
) Pb Cr
1000 Fe
" 100
__ Fe

100
CU

10




0.1
CMA NaCI CMA NaCI CMA NaCI > 450 nm 1.0 - 450 nm < 1.0 nm














Figure 2-2. Cumulative concentrations of metals leached from the soil in the three size fractions after switching salt solution to deionized water. Size separation was by 0.45 gm membrane filter and 1-nm cellulose ultrafiltration membrane (adapted from Amrhein et al., 1993)








and Cr) are consistently higher in the leachates from soil columns pre-leached with NaCI than that with CMA in all three size fractions. The fraction <1 nm will hereafter not be emphasized since it is generally considered as dissolved. Among those size fractions, colloids ranged from 1 -450 nm is most important for colloid transport (Kaplan et al., 1995), and their concentrations of each metal are much higher for the columns preleached with NaCl than those with CMA. This is consistent with normal intuition, that is, Na promotes colloid mobility, and so does colloid-facilitated metal transport if one assumes association of metals with mobile colloids are not significantly affected by changes in solution chemistry. Furthermore, metals in fraction 1.0 -450 nm are much higher than those in fractions > 450 nm for columns pre-leached with CMA, whereas there are no significant differences for the column with NaCl except for Fe and Cu. This suggests CMA is more selective in mobilizing smaller colloidal particle than NaCl. Along with this fact, the authors pointed out that colloidal Cu did not correlate well with DOC between the two fractions for the columns pre-leached with NaC1, nor does colloidal Fe with other colloidal metals pre-leached with CMA (Figure 2-2). This suggests that heavy-metal-bearing mobile colloids vary with both size and solution chemistry. In other words, phase transformation among heavy-metal-bearing colloids, such as mobile to immobile colloids or among mobile colloids with various sizes, has to be taken into consideration.

We have examined the effects of water flooding incubation on colloidal Pb

mobility in soil columns. Detailed descriptions about the experiment procedure can be found in Chapter 5. As shown in Figure 5-2, colloidal Pb concentrations varied with incubation. Interestingly, there was no significant difference in concentrations of








colloidal Fe and Al in the first several pore volumes when the columns were incubated from 3 to 20 d, while concentrations of colloidal Pb decreased approximately 4 times after 20-d than 3-d incubation. This suggested that Pb-bearing colloids tended to be immobilized or the tendency of association of Pb with mobile colloids decreased with water-flooded incubation.


Concluding Remarks

Soil has distinctively different features in colloid mobility compared with welldefined porous media or subsurface systems. At the present time, it is not realistic to predict colloid mobility and colloid-facilitated heavy metal mobility in soil with certainty.

Soil colloids are exposed to solution where they take part in chemical processes. Therefore it is difficult to predict charge development of soil colloids since it is determined by the interaction of surfaces and solution. Point of zero charge (PZC) of mineral particles suspended in simple solution can be theoretically predicted, while little success has been achieved in soil. In general, soil particles are assemblages of crystalline and amorphous minerals, and organic residuals; for such particles sophisticated models such as triple layer model do not make much sense in describing the electrified interfaces since particle surfaces are irregular and the Stem layer is often inside of the physical boundary of the particles. As a first approximation, however, It may be more pratical to describ soil particle interfaces by diffuse double layer with a specific part (a charged surface), and a generic part (a diffuse layer). The surface charge development process may be described by the partitioning of charged species between solution and surfaces assuming the charged species are similar between the two phases. Similarly, the association of heavy metals with colloids may have significant influences on colloid








mobility, which is subject to changes in dynamic solution chemistry. However, very limited information is available at the present time.

Colloids are ubiquitous in soils. Different from clean porous media, the surfaces of soil stationary phases consist of deposited colloids. Consequently, the blocking effect of colloid deposition in soil is obvious, which may greatly enhance colloid mobility in soil compared to that in a comparable clean porous medium. Under such a condition, size straining is a dominant mechanism in soil colloid deposition. The deposited colloids may release depending on the nature of interactions with the stationary phase, that is, the interaction energy profiles (Figure 2-1), which vary because of heterogeneity of the surfaces. Phenomenally, this process is time-dependent or condition-dependent, and ,therefore, is referred to as transient phenomena in colloid deposition. At the present time, water dispersible clay extracted from a soil is widely used to indicate colloid mobility. However, it has to be realized that in both cases there are two simultaneous, opposite processes: colloid deposition and release. In a given system, the relative effect of each on either soil clay dispersion in batch tests or colloid mobility in a column experiments depends on mobile colloid residence time in the systems.

Despite the shortcomings of the DLVO (Derjaguin-Landau-Verwey-Overbeek) theory, we believe it provides a basic theoretical framework in describing soil colloid dispersion and colloid deposition and release, which is simple and applicable in concept. It has to be pointed out that all modem colloid theories (not only the DLVO theory) are difficult to apply to soil because of its heterogeneity in both structure and composition of soil particle surfaces. In order to determine the potential mobility of soil colloids, a phenomenological approach may be more practical. Soil colloids possess high





54


concentrations of exchangeable ions in their EDLs (high CEC), which does not always equilibrate with those in bulk solution. It has long been recognized that colloid release in soil is coupled by significant ion transfer between EDLs and bulk solution. By evaluating the possibility of the ion transfer, potential colloid mobility may be estimated qualitatively.












CHAPTER 3
RELATION OF PB SOLUBILITY TO FE PARTITIONING IN SOILS Introduction

Lead contamination in soils is of environmental significance due to its toxicity to both humans and animals (Ma et al., 1995). In general, Pb mobility is low because of its low solubility. Lead solubility may be further reduced as a result of its interactions with soil solid phase via sorption and ion exchange. However, enhanced Pb solubility has been found under both laboratory and field conditions (Amrhein, et al., 1994; Glazovskaya, 1994). As such, Pb may migrate through a soil profile to contaminate groundwater.

Although much effort has been spent to model heavy metal solubility (Cederberg et al., 1985; Sposito, 1984), such prediction under field conditions involves large uncertainty. It is partially because of the difficulty in assessing the effects of dynamic soil solution chemistry on heavy metal speciation. However, changes in solution chemistry, such as pH, redox potential and ionic strength, may shift Pb retention processes significantly. These impacts may be further complicated by aqueous Pb competition with other cations for ligands, which may enhance Pb mobility under certain conditions (Amrhein, et al., 1994).

In natural aquatic environments, Pb (II) and Fe (II) have similar affinity to

complex with ligands, and thus they show similar patterns of species distribution (Turner et al., 1981). Theoretically, both Pb (II) and Fe (II) are in " transient class" in terms of








metal classification based on electronegativity and covalent index (Nieboer and Richardson, 1980). This evidence implies that it is important to examine Fe chemistry when studying Pb solubility. In fact, good correlation (r2 = 0.71-0.91, p<0.05) between concentrations of aqueous Pb and Fe in sediments has been reported (Lee et al., 1997; Routh and Ikramuddin, 1996). In soil environment, data published by Karczewska (1996) showed that concentrations of mobile Fe and Pb were related (r2= 0.54, p<0.05) in several polluted soils near a copper smelter. The theoretical evidence along with the published experimental data suggests that Pb and Fe solubility may be related in a soil.

In addition, there are two other possible factors for correlation between Pb and Fe in solution. First, Fe (oxy) hydroxides are good sorbents for aqueous Pb (Ainsworth et al., 1994; Laxen, 1985; Benjamin and Leckie, 1981), and their dissolution or precipitation may release or sorb Pb. In soils, dissolution of Fe (oxy) hydroxides is generally promoted by reducing Fe (III) to Fe (II), which is sensitive to soil redox status (Lindsay, 1979; Gotoh and Patrick, 1974). Under oxidizing condition, Fe (oxy) hydroxides tend to immobilize Pb (Gambrell et al., 1980); whereas under reducing condition they dissolve to release Pb (Gambrell, 1994). Secondly, soil organic carbon, especially dissolved organic C (DOC), is an important factor controlling Pb and Fe solubility in soil (Dorr and Munnich, 1991; Davis. 1984; Laxen, 1985). DOC functions as a ligand to complex with Pb and Fe to increase their solubility (Davis and Leckie, 1978). On the other hand, DOC may also be sorbed by Fe (oxy) hydroxides under certain conditions, via ligand exchange or hydrophobic interaction in soils (Murphy and Zachara, 1995), thus decreasing Pb and Fe solubility.








Despite all the circumstantial evidence that Fe solubility impacts Pb solubility in soil, there has been little effort to quantify the relationship between the two. However, understanding the relationship between heavy metal solubility and the behavior of Fe may provide important information for assessing potential mobility of heavy metals in soil environments, which cannot be modeled successfully at present time.

Soil redox status varies temporally. The intensity of redox status can be described by redox potential in general. In surface soil it is influenced by rainfall, bioactivity, and changes in land use, whereas in vadose zone by fluctuation in water table (Buol et al., 1997). The redox potentials of sediments that remain saturated can be measured readily, whereas characterization of the redox status in a soil is still a challenge. As mentioned previously, however, soil redox status affects metal solubility greatly under certain conditions. To evaluate the effects of varying redox status, saturated/unsaturated (aerobic/ anaerobic) incubation techniques have recently been developed (Amrhein et al., 1994; Karczewska, 1996). Water-flooded incubation of soil was used to effectively decrease soil redox potential to study the speciation and fate of heavy metals in contaminated soils (Karczewska, 1996). Even though saturation rarely happens in surface soils, some valuable information may be obtained from such a study. In this paper, two redox statuses in soils will be obtained under water-flooding (anaerobic) and no flooding (aerobic) incubations.

The major objective of this chapter is to examine the relationship between Pb

solubility and Fe partitioning in soils with different solution chemistry. Throughout this paper, Fe partition will be expressed as a concentration ratio of aqueous to sorbed Fe (II)








unless otherwise specified. The results showed that Pb solubility was related to this ratio in the soil.



Materials and Methods

Location and characteristics of soil sample

The soil samples used for this study were collected in March 1996 from a research site in Hawthorne, Florida. The soil is an acidic fine sand (typic quartzipsamment) with a spodic horizon below 2 m. The samples were collected from 2-20 cm below the surface after removing organic residue. The samples were air-dried, sieved through 2-mm screen, and stored at 4�C prior to use. Some characteristics of the soil are listed in Table 3-1.


Column experiment

A series of 60-mL syringe (columns, d=2.6 cm), which were prepacked with a layer (- 3mm) of acid-washed Ottawa sand (20-30 mesh) in the bottom, were packed with 40 g of soil sample. The columns were then packed with another layer of the acidwashed sand on the top to minimize disturbance on the soil from influents. All the syringes were set on a Centurion Vacuum Extractor (Centurion International, Lincoln, NE) for incubation and leaching. The experiment procedure was as follows:

Prewetting soil columns. 30 mL of deionized distilled water (DDW) was slowly poured into each column and the soil columns were saturated for 48 h before the water was extracted.









Table 3-1. Characteristics of the Florida soil used


pH Organic C Extractable Fe* Total concentrations** Texture***
mgg' mgkg- Al Fe Ca Na I Pb Sand Silt Clay mg kg-' %
5.4 6.0 810 2067 976 87 27 1.3 94.6 4.5 0.9
*Heron, et al., 1994.
**Ma et al., 1996.
***Day, 1965.








Addition of aqueous Pb. 30 mL of 0.48 mM Pb(N03)2 was added to 12 soil columns and 30 mL of 3.68 mM Pb(N03)2 to 16 soil columns. The solutions were extracted after 24 h at 2.5 mL h-'. This resulted in lead-loading rates of 0.36 (low Pb loading) and 2.90 mmoles kg-' (high Pb loading) in two groups of soil columns.

Addition of electrolyte solutions. 30 mL of different electrolyte solutions was added to different columns to vary solution chemistry. DDW of pH 5.5 and 2.17 and

4.35 mM NaCl solution were added to the first 12 soil columns, and 0.52, 1.08, and 2.17 mM NaCl solutions, 0.25, 0.55, and 1.10 mM CaCl2 solutions, and DDW of pH 4.5 and 5.5 to the other 16 soil columns. The solution was extracted out after 24 h at the rte of

2.5 mL h-1. For the first 12 soil columns with low Pb loading, each treatment was replicated four times (three treatments) and for the 16 soil columns with high Pb loading, each treatment was duplicated (six treatments). Soil columns in each treatment (a total of nine treatments) varied not only in Pb concentrations but also in pH and composition of incubation solutions.

Incubation of soil columns. The above 28 soil columns (with the same Pb loading rate and electrolyte solution) were further divided into two subgroups (14 soil columns each) for incubation. One subgroup was filled with 30 mL of DDW (2 cm above the soil surface) for water-flooded incubation. The other subgroup was incubated as it is for nonwater-flooded incubation. All soil columns were incubated for 40 d at room temperature.

Leaching. DDW was added to each soil column to make 30 mL of standing water above the soil for all columns before leaching. The leaching was conducted at 60 mL h-' and approximately 25-50 mL leachates were collected. The soil samples with moisture content - 25% in the columns were sealed and stored in refrigerator for further analysis.








Sample separation and analysis

Separation and analysis were conducted for both leachates and leached soil

samples to examine the relationship between Pb solubility and Fe partitioning in the soil. The pH was measured immediately after the leachates were collected. The leachates were then filtered through 0.22-pm membrane filters. The filtrate was analyzed for DOC and total Pb and Fe concentrations. The frozen soil columns were immersed in warm water to push the soil out of the syringes with minimum disturbance. The intact soils from the columns were then divided into three equal sections (top, middle and bottom) for further analysis. The data presented in this paper for soil, however, was the average over the three sections since there were no significant differences in Fe concentrations in the three sections. Fe(II) content in the leached soil was analyzed following the same method described in Chapter 3


Result and Discussion

Pb solubility

Aqueous Pb concentrations in leachates showed strong pH-dependence with r2

0.92 for soils with Pb loading of 2.90 mM kg-' (high Pb loading, Figure 3-1A). Aqueous Pb concentrations decreased as pH increased from 4 to 6. It has long been recognized that pH is an important factor affecting metal solubility, with aqueous metal concentration increasing as pH decreases (Chuan et al., 1996). Leachate Pb concentrations in soils with Pb loading of 0.36 mM kg-' (low Pb loading) were also correlated with pH with correlation coefficient of r2= 0.51 (p<0.05). However, they

















7UU600
I
500
400" 3001, 301

20


A
Pb loading of2.90 mmole kg'
4 water-flooded
- non-water-flooded


3.5 4 4.5 5 5.5 6 6.5 7
Leachate pH


0 I 2 3 4 5 6 7 Fe concentration in leachates (p.M)


- 00
-- - = -50 =


1.5 2 2.5 3 3.5 DOC concentration in leachate (mg/L-) D 400 350


0 2 4 6 8 10 12 14 16 Ratio of aqueous Fe to sorbed Fe (II) concentrations (kgL')


Figure 3-1 Relation between Pb solubility and leachate pH, dissolved organic carbon (DOC), leachate Fe concentration, and ratio of aqueous Fe to sorbed Fe(II) concentrations in a sandy soil. Data from 2.90 mmole kg'' Pb loading rate read from left y-axis and data from 0.36 mmole kg'1 Pb loading rate read from right y-axis.









increased as pH increased from 6.1 to 6.8 (Figure 3-1A). Obviously, mechanisms that determined aqueous Pb concentrations in soils of low Pb loading was different from that of high Pb loading. It seemed that DOC was a primary factor in determining Pb concentrations in soils with low Pb loading as reflected by their correlation coefficient of r2= 0.47, p<0.05 (Figure 3-1B), i.e. Pb concentrations increased as DOC concentrations increased. However, such correlation did not exist for soils with high Pb loading (r2=

0.24) possibly due to the fact that Pb solubility was more strongly correlated to pH.

Beside pH, redox status is another factor affecting metal solubility in soil (Chuan et al., 1996). Stumm (1984) suggested that it is reasonable to measure important redox species instead of redox potential to indicate redox status in natural systems. We used aqueous Fe as a measure of redox status for soils. As expected, aqueous Fe concentrations were much higher in soils under water-flooded incubation than non-waterflooded incubation (Figure 3-1C). In soils with low Pb loading, aqueous Fe concentrations was over three times greater under water-flooded incubation (averaged 2.12 nM) than under non-water-flooded incubation (averaged 0.57 nM). Similarly, the average aqueous Fe concentrations in soils with high Pb loading were 4.46 and 1.74 jiM, respectively. Greater aqueous Fe concentrations in soils with high Pb loading than with low Pb loading was mainly due to pH difference in the soils (Figure 3-1 A). However, there was no relationship shown between aqueous concentrations of Fe and Pb for soils of either Pb loading rates (Figure 3-1C). It was reported that aqueous Pb and Fe concentrations are linearly related due to the fact that heavy metals are released from Fe (oxy)hydroxide surface when it dissolves (Chuan et al., 1996). In our study, however, this was not the case (Figure 3-1C), suggesting that aqueous Pb may not be controlled by








Fe (oxy)hydroxide surface alone. Similar results have also been reported by other researchers (Gambrell et al., 1991). These data suggested that, in general, aqueous Fe concentration alone could not describe Pb solubility well. Nevertheless, as expected, aqueous concentrations in soils under reduced conditions were much higher than those under less reduced conditions. Aqueous Pb concentrations in soils with high Pb loading were 52 times greater under water-flooded incubation (averaged 490 pM) than under non-water-flooded incubation (averaged 9.4 pM) (Figure 3-1). Similarly, aqueous Pb concentrations were 91 and 22 nM, respectively, for soils with low Pb loading.


Pb solubility and Fe partitioning

As discussed previously, pH was related to Pb solubility in our study. However, even for the same soils, the relation between pH and aqueous Pb differed with Pb loading rates (Figure 3-lA). In other words, there is no simple relation between pH and Pb solubility when soil solution chemistry varies greatly. There were no simple relationships between Pb and DOC concentrations and Pb and Fe concentrations either (Figure 3-lA,C)

To better describe Pb solubility, we defined the ratio of aqueous Fe to sorbed Fe

(II) concentrations as a Fe partitioning index. In this study, sorbed Fe (II) was operationally defined as Fe(II) that was extracted by 0.5 M HCI (Lovely and Philip, 1987; Heron et al., 1994). By definition, Fe partitioning index represents the relative affinity of aqueous Fe to solid phases in a soil. A larger number indicates a lower Fe affinity to solid phases. There was no simple relationship that existed between aqueous Pb concentrations and Fe partitioning index (Figure 3-1D). For soils with low Pb loading








and incubated under water-flooded condition, the relation between aqueous Pb and Fe partitioning index was approximately parabolic. In general, aqueous Pb concentrations increased as the index decreased. However; as the index decreased to below 2 kg L', increases in Pb concentrations were substantial (Figure 3-1D). A similar trend was also observed for soils with high Pb loading and incubated under non-water-flooded condition, i.e. substantial increase in Pb concentrations were observed when the index was < 2 kg L'. However, no enhanced Pb solubility was found in soils with low Pb loading and incubated under non-water-flooded condition since the Fe index was > 2 kg L '. However, for soils with high Pb loading rate and incubated under water-flooded condition, extremely high Pb concentrations were observed since the Fe index was <2 kg L-1. Obviously, compared to pH, DOC and Fe concentrations, a more consistent relation between aqueous Pb concentrations and the Fe partitioning index in soils was observed.

For a pH-dependent process, Kurbatov equation (Kurbatov et al., 1951) has been successfully used to describe the macroscopic behaviors of metal partitioning in natural aquatic systems (Fuller et al., 1996; Tessier et al., 1985, Balistrieri et al., 1983). In our system, the Fe index in soils with both high (r2=0.61, p<0.05) and low (r2=0.55, p<0.05) Pb loading rates was correlated to pH. Thus, the Kurbatov equation can be applied to our system as follows:

log(Fe index) z -XpH - log KP + log Ns, 3-1 where z is a macroscopic proton coefficient, Kp a Kurbatov partition coefficient and Ns the total number of exchangeable sites (Fuller et al., 1996, Balistrieri et al., 1983). Assume that the above equation is correct, then the Fe index is related to pH in addition to X, Kp and Ns. The number of exchangeable sites Ns in the equation for a given soil is








generally a constant. On the other hand, X varies with metal species in solution and on surfaces and Kp does not vary much unless the adsorption density exceeds its maximum (Balistrieri et al., 1983). Therefore, for a given system, the Fe index changes mainly with pH and X. The relation between pH and the Fe index in the Kurbatov equation is as follows. For soils with high Pb loading, pH was inversely related to the Fe index as expected with r2=0.55, i.e. the index decreased as pH increased from 6.1 to 6.8. However, for soils with low Pb loading, pH was positively related to the Fe index with r2=0.61, p<0.05 (Figure 3-1A&D). The impact of on the index is unclear at the present time.

As shown in the Kurbatov equation (Eq. 3-1), the Fe index is a lumped parameter that presents the properties of both soil solution (pH and X) and solid (Kp and Ns) that determine Fe partitioning. To evaluate the relationship between Fe index and other metal solubility, it is expected that the more Fe-like a metal acts, the more related its solubility is to the Fe index. As mentioned previously, Pb(II) shows some similarity to Fe (II) in both metal classification and speciation distribution in aquatic systems (Turner et al., 1981; Nieboer and Richard, 1980). Therefore, it is expected that aqueous Pb concentration in soil is related to the Fe index (Figure 3-1D).

Turner et al. (1981) reported that in natural waters Pb (II) and Fe (II) exhibit some similarity, meaning that Fe (II) may be a major competitive cation to Pb (II) for ligands. Therefore, it is important to evaluate the competition between Fe (II) and Pb (II) for ligands. It should be pointed out that, in this chapter, we did not emphasize the competition between Fe (II) and Pb (II) for adsorption sites on solid phases. The reason is that some researches have shown that the competition may be insignificant (Rose and








Bianchi-Moaquera, 1993; Coughlin and Stone, 1995). In our study, soil solution chemistry varied in different soil columns. These variations may shift the competition. At the present time, it is difficult to fully evaluate these effects on metal solubility in the soil, especially when DOC is present. The Fe index presented in this paper, however, may serve as a simple measure of competing ability of Fe with Pb for ligands. At this stage it is more important to look at the conceptual nature of the index than the detailed mechanistic interpretation, which need to be explored further in future.

Assuming that Fe and Pb compete for ligands in a soil, then the following can be inferred. As Fe partition index increases, there is more aqueous Fe and/or less sorbed Fe, which means there is more competition for ligands from aqueous Fe in the soil and thus results in less aqueous Pb in the soil. On the other hand, as Fe partition index decreases, there is less aqueous Fe and/or more sorbed Fe, which means there is less competition for ligands from aqueous Fe and thus results in more aqueous Pb in the soil as shown in Figure 3-1D.

In theory, there should be a minimum for the Fe index, i.e. the ratio of the

minimum of aqueous Fe concentration to the maximum sorbed Fe concentrations. If aqueous Fe concentrations approach zero, i.e. all aqueous Fe was transferred onto sorbed form, the Fe index is zero. At this situation, Pb dominates the ligand competition in the solution, so that enhanced Pb solubility would be expected. The overall trend shown in Figure 3-1D confirmed this rationale in principle. In reality, however, aqueous Fe concentrations cannot be zero. Therefore, the Fe index must be larger than zero. Theoretically, in a soil the minimum of aqueous Fe concentration is mainly determined by the solubility of Fe minerals and the maximum sorbed Fe concentration is soil-








dependent. Therefore, the Fe index is generally insensitive to solution chemistry in a soil. Since it is impossible to accurately measure the maximum sorbed Fe concentrations in a soil, no attempt was made to evaluate the real value of the minimum Fe index in the soil we used from the definition. However, the minimum Fe index when enhanced Pb solubility occurred was approximately 2 kg L"' in the soil studied, which was estimated from the relationship between Pb solubility and Fe index (Figure 3-1D). As expected, the number did not vary much with the incubation conditions and Pb loadings (Figure 3-1D). It has been noticed that Pb Concentrations in some soil columns still stayed low even when the Fe index was < -2 kg L-1 (Figure 3-1D), which is unclear at the present time. Nevertheless, the significance between Pb solubility and Fe index was that significantly enhanced Pb solubility does not occur unless the Fe index was < ~2 kg L' (Figure 3-1D).


Relation of Pb solubility and Fe partitioning in published data

In this research, sorbed Fe was defined as the Fe (II) fraction extracted by 0.5 M HC1, which is supposed to extract exchangeable, some adsorbed and freshly-precipitated Fe(II) (Heron et al., 1994). In literature, however, such data are not always available. In order to conceptually test the relation between Pb solubility and Fe partitioning index we defined in this paper using available data in the literature, other extractants that may be equivalent to 0.5 M HCl have to be used. Two published data sets were used. In both data set, the Fe index was calculated using the total sorbed Fe instead of Fe (II) since the quantity of the later is unavailable. However, Fe index calculated from either total Fe or total Fe (II) should be consistent conceptually since Fe (III) is even more competitive compared with Fe (II).





























0.55


0.65


0.75


0.85


Fe partition index (kg L-1 )


















Figure 3-2. Relationship between Pb concentrations in pore water and the ratio of aqueous and sorbed Fe (Fe partition index) in a contaminated sediment. Data are adapted from Lee et al. (1997)


[E



[[]
[














160
140 120 100
80 60
40 20 0(


0.05
0.05


0.15


Fe partition index (kg L-)




















Figure 3-3 Relationship between soluble Pb concentrations and the ratio of soluble to sorbed Fe in contaminated soils. Data are adapted from Karczewska (1996).


)A
-f ~ t








The first data set was from Lee et al. (1997) who characterized a sediment

contaminated by heavy metals in a retention pond in France. In the study, concentrations of aqueous Pb and Fe in pore water of the sediment were analyzed, and Fe concentrations in the sediment was determined by sequentially extracting into five fractions: exchangeable extracted by 1 M MgC12, bound to carbonate by 1 M sodium acetate, bound to amorphous Fe and Mn hydroxides by 0.04M hydroxylamine hydrochloric acid in 25% acetic acid, bound to organic matters and sulfides by 30% H202 and 0.02 M nitric acid, and residual by concentrated HNO3 and HCO104. Assuming that the sum of Fe concentrations in the first two fractions (exchangeable and bound to carbonate) was equivalent to that extracted by 0.5 M HC1, a plot of aqueous Pb concentrations to the Fe index (Fe concentration in pore water/the sum of Fe concentrations of exchangeable and bound to carbonate) is shown in Figure 3-2. A similar trend shown in Figure 3-1D was observed: enhanced Pb solubility occurred only when the Fe index was low. However, low Pb solubility was also observed at low Fe index as did in our data (Figure 3-1D). This again suggests that high Pb solubility will not occur unless the Fe index was below a certain value and low Fe index will not necessarily guarantee high Pb solubility. The second data set was taken from Karczewska (1996) who determined concentrations of heavy metals via a sequential extraction in soils polluted by a c er smelter. In the study, heavy metals were fractionated into 7 fractions, i.e., soluble (1 M NH4NO3), exchangeable (lM NH4OAc), bound in MnOx (IM NH2OH-HCl/1M NH4OAc) and etc. We assumed that Pb extracted by IM NH4NO3 is equivalent to the aqueous Pb concentration and sum of exchangeable Fe and bound to MnOx was equivalent to Fe(II) extracted by 0.5 M HCl. Again, a similar trend shown in Figure 3-1D was observed








between soluble Pb and Fe index (Figure 3-3), i.e. highest Pb solubility was observed at the lowest Fe partition index and in this particular case the Fe index was zero. Like the data of Lee et al. (1997), the soil samples were collected from various locations near a smelter site in Poland and there were large variations in soil properties among the samples (CEC: 10-84 meq kg''; organic C: 0.05-1.42 %; clay content: 1-6 %) (Karczewska, 1996). This suggested that the relationship between Pb and Fe partition index may be applicable to field data even with large spatial variability in soil properties.



Implication of this Research

Partition coefficients (kd) are important parameters in assessing the potential

impacts from metal contaminated soils. However, kd is not a constant in a dynamic soil environment. It is determined not only by the characteristics of solid phases but also solution chemistry. In a given soil system, variation is mainly determined by metal solubility. There are several speciation models available to describe the interactions of metals and soil components to predict their solubility. Unfortunately, such predictions generally lack certainty primarily because of soil dynamic nature.

At the present time, assessing changes in solution chemistry in field condition is difficult if not impossible. This information, however, is critical to estimate the solubility of heavy metals in speciation models. In contrast, the approach presented in this chapter provides a simple relation between the probability of enhanced Pb solubility (or kd) and Fe partition index, which is generally not sensitive to changes in solution chemistry. In this approach, we simply determine the minimum Fe index in a field condition to predict the possibility of enhanced Pb mobility (the lower kd). Of course, more work is fieeded to








further test this concept in different soils and determine the ranges of Fe index for enhanced Pb mobility.


Conclusion

Our data using a synthetically contaminated soil demonstrated that enhanced Pb solubility was only observed at low Fe partition index. Such a trend was also observed using two published data sets in the literature, thus further validating the relationship conceptually. Since data used in this paper were from various soil environments (roadside sediment in France, Pb contaminated soil in Poland, a sandy soil from Florida), it is reasonable to conclude that Pb solubility may be indeed related to Fe partition index in soils. This concept may also be applied to solubility of other heavy metals such as Cd, Cu, Ni and Zn.













CHAPTER 4
HEAVY METAL MOBILITY IN CONTAMINATED SOILS: PART 1. ROLE OF
EXCHANGE SITES IN CONTROLLING SOLUBILITY AND MOBILITY OF HEAVY METALS

Introduction

Heavy metal mobility in soils is of environmental significance due to its toxicity to both humans and animals (Ma et al., 1995). At a constant water flow condition in a soil, heavy metal mobility is determined by its solubility. It has been well recognized that the solubility of heavy metals in soil is mainly regulated by adsorption, precipitation and ion exchange reactions. Although much effort has been spent to model heavy metal solubility (Cederberg et al., 1985; Sposito, 1984), such a prediction under field conditions contains large uncertainty. It is partially because of the difficulty in assessing the effects of dynamic soil solution chemistry on heavy metal speciation. However, changes in solution chemistry, such as pH, redox potential and ionic strength, may shift the retention processes of heavy metals significantly. These impacts may be further complicated by the competition of cations for ligands, which may enhance heavy metal mobility under certain conditions (Amrhein, et al., 1994).

Soil redox status varies temporally and spatially. In a surface soil it is influenced by rainfall, bioactivity, and changes in land use, whereas in vadose zone mostly by fluctuation in water table (Buol et al., 1997). A reduction in redox potential may cause changes in metal oxidation state, formation of new low soluble precipitates, and Fe








dissolution resulting in release of metals (Amrhein et al., 1994; Chuan et al., 1996; Masscheleyn et al., 1991). To experimentally evaluate the effects of different redox status on metal solubility, various techniques have been developed: redox-potentialcontrolled batch experiment using a suspension (Chuan et al., 1996; Masscheleyti et al., 1991), saturated/unsaturated (aerobic/ anaerobic) incubation (Amrhein et al., 1994; Karczewska, 1996, Ma et al., 1995) and water-flooded incubation (Karczewska, 1996). However, some contradictory results have been found when the different techniques were employed. In redox-controlled suspension experiments, Chuan et al. (1996) reported that aqueous concentrations of Pb, Cd and Zn increased with Fe(II) as redox potential decreases, suggesting metal sorption onto surfaces of Fe (hydr)oxides is a dominant process in controlling aqueous metal concentrations. On the other hand, in experiments using saturated paste, Amrhein et al. (1994) found that concentrations of Cu and Cd decreased whereas those of Fe(II) increased in solution as the redox potential was decreased, however, no precipitation of any known Cu or Cd mineral can be proved under these conditions. Similar controversial results are found in other studies (Masscheleyn et al., 1991; Ma et al., 1995). However, it should be pointed out that among the studies, the samples analyzed for metals were different. They can be classified into two groups: 1) pore water and leachates from soil leached with DDW, and 2) filtrates separated from the suspension and leachates from soil leached with electrolytes. Obviously, metals determined in samples of the first group are water soluble whereas those of the second group include the exchangeable in addition to water soluble in soils. Therefore, a natural question is raised: with incubation, do metal concentrations in solution change proportionally with those in exchangeable phases? To our knowledge,








the role of exchange phase in controlling the solubility and mobility of heavy metals in incubated soils has not been reported. However, this knowledge may improve our understanding of the controversy discussed above.

In Chapter 3, Pb solubility was examined in a sandy soil spiked with Pb and incubated for 40 d under water-flooded or non-water-flooded conditions. Solution chemistry in soil columns was adjusted using different concentrations of NaCl and CaC12 and deionized water of varying pH before incubation. The results show that Pb solubility in the incubated soil can be related to the ratio of aqueous Fe to 0.5 M HCI extracted Fe

(II) from solid phases instead of soluble Fe alone. If we assume 0.5 M HCI extractable Fe(II) from the solid phases is proportional to exchangeable Fe (II), it is suggested that exchange sites have significant effect on heavy metal solubility in soils.

In this paper, we will extend our study from a synthetically contaminated soil to two naturally contaminated soils. In addition to Pb, concentrations of Cu and As will be presented for comparison. The major objective of this paper was to examine the role of exchange sites on the solubility and mobility of Pb, Cu, and As in soils during incubation.


Materials and Methods

Location and characteristics of soil samples

The two Pb-contaminated soils used in this study were from Montreal, Canada

and Tampa, Florida, which were exposed to Pb-battery recycling and concurrent smelting operation in the past. Selected characteristics of the soils are listed in Table 4-1.










TABLE 4-1


Selected characteristics of the soils used in this study


pH CEC Organic C Particle size distribution (%) Total element analysis (mg kg ) Soil location (Day, 1965) (Ma et al., 1996)

(1:1) cmol/kg (%) Clay Silt Sand Pb Fe Al Mn Ca Montreal, 7.67 29.0 2.24 49 40 11 1600 133300 169500 800 15700

Canada

Tampa, 3.80 2.71 8.28 4.0 28 68 2500 5200 1200 23 2500 Florida










Column Experiment

Packing soil columns and Incubation process. 5.0 g acid-washed sand (20-30 mesh) was packed in the bottom of 60-mL columns (13 cm in length x 2.6 cm in diameter), then 30 g air-dried soil (moisture content = 9.0%) was packed on top of the sand layer. Preliminary data showed that the sand layer helps to prevent the outlet clogging in a column during incubation and leaching ad has no detectable impact on soil colloid mobility. During each step, the columns were gently shaken horizontally for a few minutes to minimize the packing effects. This resulted in 2.6x5.2 and x7.8-cm soil columns for the Montreal and Tampa soils, respectively. The soil columns were set vertically and prewetted by pumping deionized distilled water (DDW) into the bottom of the columns until the water level in the columns was above the soil. These columns were then incubated under water-flooded condition for approximately 3, 20 and 80 d for the Tampa soil and 3, 20, 30, 40, 50, and 60 d for the Montreal soil.

Leaching. Once the redox potential monitored had reached to a certain level, the water on top of the soil column was removed and a rubber stopper was put on top to seal the syringes. The syringes were then turned upside down. The lower side of the column was connected to a needle as an inlet for the influent, 0.01 M CaCl2. The pumping rate was 1.20 mL min". The effluent from the top of the column was collected with a fractionation collector.

Analysis of leachates. Each fraction of effluents was filtered through 0.22 ptm membrane filter, then acidified to pH<2 to yield the soluble metal concentrations. The details can be found in Chapter 3.








Analysis of pore water of soil

To analyze metal concentrations in pore water, an incubated but unleached soil

column for each incubation time was used after removing the standing water on top. Pore water was separated from the soil by the centrifugation method described by Nkedi-Kizza et al. (1982) and Dao and Levy (1978). Then the pore water was filtered with 0.22-jm membrane filter for metal analysis. For detailed procedure of Fe(II) analysis, please see Chapter 3.

Result and Discussion

Changes in heavy metal solubility with incubation

Concentrations of heavy metals and Fe (II) in pore water varied with incubation

time (Figures 4-1 and 4-2). Fe behaved differently in the two soils with incubation: in the Montreal soil, Fe(II) decreased with incubation (Figure 4-1) whereas in the Tampa soil it increased(Figure 4-2). Similarly, Pb increased in the Montreal soil and decreased in the Tampa soil with incubation. However, in both soils the concentrations of Fe (II) and Pb were inversely correlated with incubation (Figure 4-1 and 4-2). Concentrations of As and Cu increased first then decreased with incubation in the Tampa soil.

In similar experiments by Amrhein et al. (1994) using soils incubated in waterflooding condition, they found that Fe (II) concentration in pore water increases with incubation time. This is consistent with the result we found in the Tampa soil; however it is contrary to that in the Montreal soil. With incubation, redox potential reduced and thus





80










70 -1000
0 O- Pb O Fe 900 c 60 -C
O 80050 -700 600 0
0 40 .- 500
S30 -400 20 - L 300 C " 200 c c 0 0 10 0
3100

0 I .... .... .... 0 u
0 10 20 30 40 50 60 70 80
Incubation time (d)


Figure 4-1. Pb and Fe (II) concentrations in pore water of the Montreal soil.

















E Pb O Cu A As * Fe(II)


245 40 935




C

5
I.

o
0
C
C
C
0


05

0



0


Figure 4-2 Concentrations of Pb, As, Cu and Fe(II) in pore water of the Tampa soil


10 20 30 40 50 60

Incubation time (d)


1.2


E

a)


0.8
0

C
0.6 c
o Cu

0.4 c
8
C 0


a) L-










dissolved Fe(II) increased due to Fe(III) reduction to Fe(II) (Amrhein et al. , 1994). However the unexpected behavior of Fe (II) in the Montreal soil may be caused by formation of FeCO3, which will be discussed in Chapter 5 in details.

Chuan et al. (1996) examined the release of heavy metals from a contaminated soil, which was suspended in water (soil:water ratio = 1:7), and found that aqueous concentrations of Pb, Cd and Zn are positively correlated with those of Fe during incubation because heavy metals are released from Fe (oxy)hydroxide surface as it dissolves (Chuan et al., 1996). Obviously, it is contrary to what we observed for Pb concentrations in both the Tampa and Montreal soils. On the other hand, Amrhein et al. (1994) reported that, under water-saturated (soil:water ratio = 1:0.2) incubation, Cd and Cu concentrations in pore water decrease whereas Fe(II) increases with time. This is consistent with our results. It is possible that the different relations between Fe(II) and heavy metals in solution result from the different soil-water ratios used in different studies. In the studies discussed above, contributions of exchangeable metals to aqueous metal concentrations during incubation varied with the soil:water ratios. When the ratio is low, i.e., small quantity of soil is suspended in large amount of water, ions on the exchange sites tend to diffuse into the water to maintain their chemical potentials between solution and exchange sites. Consequently, the importance of exchange sites in holding metal ions reduces in a relative sense. On the other hand, as the soil:water ratio increases, the importance of exchangeable metals in contributing to soluble metals increases since the amount of exchange sites increases relative to solution volume. Therefore, metal distribution between solution and exchange sites has to be taken into








consideration, which is mainly controlled by metal competition between exchange sites and aqueous ligands in solution. Therefore, the observation of Amrhein et al. (1994) and the one presented here may be due to the fact that, with incubation, more heavy metals go to exchange sites as Fe (II) concentrations increase in solution. We have demonstrated that Pb solubility in soil columns is inversely related to aqueous Fe concentration in Chapter 3. Therefore, in this sense, all the studies discussed above along with the one presented here are consistent, i.e., with Fe dissolution upon reduction, Pb is released into solution and exchange sites.

Similarly, arsenic concentrations in pore water and suspension varied differently with incubation. Masscheleyn et al., (1991) examined the-arsenic release from a contaminated soil in a suspension (soil:water = 1:6) during incubation, and found that soluble arsenic increases as redox potential decreases (+500 -200 my). It was attributed to As(V) reduction to As(III) and Fe reduction dissolution to release adsorbed As. However, analysis of pore water in soils after water flooded incubation showed a different trend (Onken and Hossner, 1996). They found that soluble arsenic concentrations increased first and then decreased with incubation, and a maximum occurred at 20-30 d after flooding. Further analysis of the solid phase showed the loss in soluble arsenic could be accounted by surface bound when the soil was incubated longer than 20-30 d. The result of the later study is consistent with our data that showed As concentration increased and then decreased with Fe reduction dissolution (Figure 4-2), which will be discussed in the following section.















100
90


0 10 20 30 40


50 60 70 80


incubation time (d)














Figure 4-3. Cumulative Pb and Fe leached after 31.8 pore volumes of 0.01 M CaC12 in Montreal soil


Pb


O]


- X
















35 5
* Fe A Cu
30
"o "
O As O Pb 4 3 c 25 C

E 3- 3 S20

" -2
a) .
e15
. -2

w 10 (Dw I2
E 5 EE

0 - 0 . O
0 10 20 30 40 50 60 70 Incubation time (d)














Figure 4-4. Cumulative Pb, As, Cu and Fe leached after 31.8 pore volumes of 0.01 CaCl2 in Tampa soil.









Heavy metal mobility with incubation

Metal mobility in this paper is examined using cumulative metals leached with 32 pore volumes of 0.01 M CaC12 solution. It is well understood that the metals that can be leached out with CaCl2 solution include both the soluble and the exchangeable. For the Montreal soil, the cumulative leached Pb concentrations increased with incubation (Figure 4-3), which is consistent with the overall trend shown in the pore waters (Figure 4-1). However, the cumulative leached Fe concentrations showed a different trend from Fe(II) concentrations in pore waters (Figure 4-3), suggesting that Fe(II) in exchange site is not proportional to that in aqueous phase assuming that Fe(II) - soluble Fe. For the Tampa soil, changes in overall trends for metals with incubation can be divided into two groups: 1.) Pb and Cu decreased with incubation; 2.) As and Fe increased with incubation, which is not consistent with the case in pore water.

In fact, the reduction of cation exchange capacity (CEC) caused by the association of hydroiron polymers or iron hydroxides with soil mineral has been long recognized (Coleman and Thomas, 1964; Carstea et al., 1970). Hendershot and Lavkulich (1983) reported that Fe coatings on illite lower CEC and the ones on kaolinite increase anion exchange capacities (AEC) in comparison to the uncoated ones at pH< 7. They suggested that the Fe coatings reduce the CEC and increase AEC measured at low pH by either physically blocking or electrostatically canceling the permanent negative charge carried by the crystalline minerals. Similarly, Arias et al., (1995) found that the point of zero charge (PZC) of Fe-coated kaolinite increases linearly with the amount of the coating Fe. In the experiments by Shainberg, et al. (1987), they demonstrated that the CEC of the mixture Fe (FeC13) and soil decreases by 52% in the 10 mmol Fe kg"








treatment to 72% in the 40 mmol Fe kg"' treatment, and correspondingly the exchangeable sodium percentage from 23.8 to 15.4. Rengasamy and Oades (1977) reported that with addition of ferric hydroxide the electrophoretic mobility of iron-clay increase from negative to positive values. All these data surpport that with the reduction dissolution of Fe, Fe coating fraction decreases, and then CEC increases for a soil. Even though it is supported by sufficient experimental and theoretical evidence, it has, to our knowledge, not been stated for incubated soils. However, the reduction in CEC, along with solution chemistry, can have significant effects on heavy metal retention.

With the reduction dissolution of Fe, therefore, the blocked exchangeable sites will be freed and the concentration of Fe in pore water increases, and thus Fe (II) concentration in exchangeable phases may increase as well. It was evident that, for the Tampa soil, changes in Fe (II) concentrations in pore water and its cumulative leached amount with incubation time were similar: they both increased with incubation (Figure 44). However, for the Montreal soil, a sudden increase in cumulative leached Fe was found at 20 day incubation compared to Fe(II) concentrations in pore water, suggesting that exchangeable Fe (II) became dominant in cumulative leached Fe. In fact, there was

5.55 p.moles Fe leached during 7-16 pore volumes (which can be considered as exchangeable) among 7.11 [tmoles of the total cumulative leached Fe after 20-d incubation.

In the Tampa soil, changes in Pb concentrations in pore water and the cumulative leached Pb with incubation time were similar: they both decreased with incubation. This is contrary to Fe behavior, suggesting that Pb concentrations in solution may be controlled by competition between aqueous Fe and Pb for ligands and exchange sites. It








is well documented that Fe(II) and Pb(II) have similar affinity for ligands (Turner et al., 1981; Nieboer and Richard, 1980). As aqueous Fe (II) concentrations increase with incubation due to Fe reduction dissolution, more aqueous Pb (II) may be forced onto exchange sites. If this is simply the case, however, the cumulative leached Pb was not supposed to decrease with incubation (Figure 4-4). However, it was difficult to imagine that there was so much Pb present in the exchange sites without some of it adsorbing chemically onto the surfaces. However, it should be pointed out that there was significant difference between the relations of Pb concentrations in pore water and cumulative leached Pb with incubation: while the Pb concentration in pore water decreased exponentially the cumulative leached Pb decreased only linearly (incubation time < 30 day). It is suggested that less reduction of the cumulative leached Pb than that of Pb concentrations in pore water could be caused by more exchangeable Pb with incubation.

In the Tampa soil, cumulative leached Cu decreased with incubation (Figure 4-4), which is inconsistent with changes in Cu concentrations in pore water (Figure 4-2). The significantly greater cumulative leached Cu after 3-d incubation can be attributed to more Cu coming from exchange sites when leached with CaCl2. In fact, Cu (II) has greater affinity for exchange sites than Pb (II) (McBride, 1994).

Interestingly, As concentrations in pore water increased first and then decreased with incubation, whereas the amount of cumulative leached As showed no significant drop with incubation (Figure 4-4). As discussed above, as Fe "coatings" was removed with incubation via reduction dissolution, soil CEC increased as well. This caused more Pb (II) to transfer from solution to exchange sites, as a result, local charge reversal might








occur, therefore positively-charged particles held more H2AsO3" and /or H3AsO4". It has been realized that at slightly reducing condition and pH = 4 (for the Tampa soil in this study) both H2AsO3- and H3AsO4" are not stable thermodynamically (Cullen and Reimer, 1989). However, local pH near positively charged surface may be significantly greater than in bulk solution, therefore H2AsO3" may be stable near highly positively-charged surfaces. On the other hand, there has been evidence showing that even under reducing condition, only H3AsO4 is associated with natural particle surfaces (Belzile and Tessier, 1989). The existence of H3AsO4" under reducing condition has been attributed to the oxidation of arsenite to arsenate by Mn and Fe oxyhydroxides (Peterson and Carpenter, 1983).

As discussed above, CEC increases with Fe reduction dissolution, especially at pH < PZC of Fe minerals. The consequent redistribution of metals between solution and exchange sites is probably determined not only by the magnitude of CEC, but also the characteristics of metal ions, the competing co-ions and counter-ions (ligands). Therefore, changes in metal concentrations in a soil solution with time under incubation can be complicated. However, the following interactions have to be taken into consideration:

1. Fe reduction dissolution releases Fe(II) and adsorbed heavy metals into solution at the

same time frees exchange sites,

2. Fe (II) and heavy metal cations compete for ligands to stay in solution,

3. Fe (II) and heavy metal cations compete for the exchange sites, transferring from

solution to solid phase,




Full Text

PAGE 1

HEAVY METALS AND COLLOID MOBILITY IN SOILS By YAN DONG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1999

PAGE 2

Copyright 1999 by YAN DONG

PAGE 3

ACKNOWLEDGMENTS First among those deserving credit for their contributions is the committee chairman, Dr. Lena Q. Ma. Her many contributions to the form and content of this dissertation are appreciated, but equally important were the opportunities and directions that she provided for developing skills that are of considerable benefit to my career in research and teaching. Dr. R. Dean Rhue deserves special thanks for patiently leading me through all the difficulties in finishing Chapter 3. Dr. Willie Harris, Dr. Peter Nkedi-Kizza and Dr. Timothy G. Townsend provided much of the initial direction and a great deal of instructive suggestion to this study. Their efforts were indispensable and are much appreciated. Special thanks also go to Kennelley, E., Schwandes, L., Thomas, J., Reve, W., Lewis, K., Awuma, K., and Choate, A., for their help with instrumental analysis. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS HI LIST OF TABLES IV LIST OF FIGURES VII ABSTRACT X CHAPTERS 1 INTRODUCTION 1 2 COLLOID DEPOSITION, RELEASE AND ASSOCIATION WITH HEAVY METALS IN SOILS 1 5 Introduction 5 The DLVO Theory 10 The Characteristics of Colloids 13 Charge development 13 Electrical double layer (EDL) in metal oxideswater interface 14 Charge development in soil particles 21 Charge development in mobile colloids 22 Hydration 24 Size development of mobile colloids 25 The Characteristics of Porous Media 26 Deposition of Colloids 27 Theoretical background in well-defined porous media 27 Colloid deposition in soil 29 Release of Colloids in Soil 31 The processes of colloid release 33 Mobile colloids and water-dispersible clay 34 Influenes of exchanegable soldium percentage (ESP) on stability of water-dispersible clay and mobilization of soil colloids 35 Ion transfer processes during the release of colloids 37 The ion transfer during soil colloid mobilization in literature 38 Ion transfer during soil colloid release in our research 41 Association of Colloids with Heavy Metals 43 Adsorption of heavy metals to surface hydroxyl of colloids 44 Effects of heavy metal adsorption on the charges of soil colloids 47 Partitioning of heavy metals in soil colloids 49 iv

PAGE 5

Concluding Remarks 52 3 RELATION OF PB SOLUBILITY TO FE PARTITIONING IN SOILS 55 Introduction 55 Materials and Methods 58 Location and characteristics of soil sample 58 Column experiment 58 Sample separation and analysis 61 Results and Discussion 61 Pb solubility 61 Pb solubility and Fe partitioning 64 Pb solubility and Fe partitioning using published data 68 Implication of this Research • 72 Conclusion 73 4 Heavy metal mobility in contaminated soils: Part 1. The role of exchange sites in controlling solubility and mobility of heavy metals 74 Introduction 74 Materials and Methods 76 Location and characteristics of soil samples 76 Column Experiment 78 Analysis of pore water of soil 79 Results and Discussion 79 Changes in heavy metal solubility with incubation 79 Heavy metal mobility with incubation 86 Implication of this Research 90 5 HEAVY METAL MOBILITY IN CONTAMINATED SOILS: PART 2. COLLOID-FACILITATED METAL MOBILITY IN A PBCONTAMINATED SOIL 92 Introduction 92 Materials and Methods 95 Characteristics of soil samples 95 Column experiment 97 Analysis of Fe(II) and Ca in pore water 97 Analytical methods 98 Result and discussion 99 Colloid mobility and soil redox status 99 Colloid elution curves 105 Colloid-facilitated Pb mobility 110 Conclusion Ill 6 RELEASE AND DISPERS ABILITY OF COLLOIDS IN TWO CONTAMINATED SOILS 1 1 3 v

PAGE 6

Introduction Materials and Methods 1 1? Column preparation 1 1 ' Column leaching test Water dispersability test Estimation of relative colloid stability ratio (RW) 119 Result and Discussion : 123 Conclusion 131 7 CONCLUSION 132 State of colloid deposition and release in soil and their association with heavy metals 132 Colloidal metal mobility in contaminated soils 132 Metal solubility and mobility in soil 134 REFERENCES 136 BIOGRAPHICAL SKETCH 149 vi

PAGE 7

LIST OF TABLES Table page 31 Characteristics of the Florida soil used 59 41 Selected characteristics of the soils used in this study 77 51 Selected properties of the Pb contaminated soil used in this study 96 52 Minerals in the soil and their Point of zero charge (PZC) 104 62 Relative colloid stability ratios (RW) for two soils under different water-flooding time 125 vii

PAGE 8

LIST OF FIGURES Figure page 11. A road map of this dissertation 3 21. Schematic of potential energy profile of the interaction of surfaces with inclusion of van der Walls attraction, electrical repulsion, and Born repulsion, which shows both primary and secondary minima and an energy barrier as well as the zones in which release and deposition take place 1 6 22. Cumulative mass of metals leached from the soil in the three size fractions . after switching salt solution to deionized water. Size separation was by 450nm membrane filter and 1 -nm cellulose ultrafiltration membrane (data from table 4-6 in Amrhein et al , 1993) 50 31. Relation between Pb solubility and leachate pH, dissolved organic carbon (DOC), leachate Fe concentration, and ratio of aqueous Fe to sorbed Fe(II) concentrations in a sandy soil. Data from 2.90 mmole kg-1 Pb loading rate read from left y-axis and data from 0.36 mmole kg-1 Pb loading rate read from right y-axis 62 3-2. Relationship between Pb concentrations in pore water and the ratio of aqueous and sorbed Fe (Fe partition index) in a contaminated sediment. Data are taken from Lee et al., 1997 69 33. Relationship between soluble Pb concentrations and the ratio of soluble to sorbed Fe in contaminated soils. Data are taken from Karczewska, 1996 70 41 . Pb and Fe (II) concentrations in pore water of Montreal soil 80 4-2. Pb, As, Cu and Fe(II) in pore water of Tampa soil 81 4-3. Cumulative Pb and Fe leached after 3 1 .8 pore volumes of 0.01 M CaC12 in Montreal soil 84 44. Cumulative Pb, As, Cu and Fe leached after 31.8 pore volumes of 0.01 CaCb in Tampa soil 85 51. Changes of effluent turbidity with pore volumes under different incubation times. The insert is a typical breakthrough curve of deionized water displacing CaCl 2 100 viii

PAGE 9

5-2. Effects of incubation on aqueous Fe (II) and Ca in the pore waters of soil columns and the cumulative colloidal Fe and Al in the effluents after 23 pore volumes 10* 5-3. Elution curves of colloidal Fe, Al, and Pb concentrations in effluents with pore volumes under various incubation times. Each point presents the mean of two replicates 106 54. Relationship among colloidal Fe and Al, dissolved Ca and pH in effluents after 3 d of incubation. Each point presents the mean of two replicates : 107 61 Schematic representation of typical absorbency-time curves observed. Linear regression was used to calculate an apparent flocculation rate indicated by solid line. The curves are not drawn in the same time and absorbancy scales. Curve "a" stands for those observed for the Montreal soil in 0.06 NaCl M solution, and Curve "b" for those for the Tampa soil in 0.06 M NaCl and in 0.01 M CaCl 2 solutions, and the Montreal soil in 0.01 M CaCl 2 solution 122 6-2. Absorbency-time curves observed in Ca-saturated Tampa soil suspended in 0.06 M NaCl solution 124 6-3. Relation of cumulative colloids and relative stability ratio (RW) 128 ix

PAGE 10

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy HEAVY METAL AND COLLOID MOBILITY IN SOILS By Yan Dong December 1999 Chairman: Lena Q. Ma Major Department: Soil and Water Science Mobility of heavy metals in soils is of environmental significance due to their toxicity to both humans and animals. In general, heavy metal mobility is low because of its low solubility. However, enhanced heavy metal mobility has been reported under both laboratory and field conditions. It has been attributed to enhanced solubility of heavy metals and colloid-facilitated metal transport. In this study, the solubility and mobility of heavy metals were examined in a Pbspiked sandy soil and two Pb-contaminated soils. For the Pb-spiked soil, water-flooded and non-water-flooded incubations were used to alter soil solution chemistry in soil columns, which were then leached with de-ionized water. It was found that Pb concentrations in leachates were related to the ratios of Fe concentration in the leachates to Fe concentrations extracted with HC1. Enhanced Pb mobility occurred only when the Fe ratios were lower than a threshold value for a given soil. The two Pb-contaminated soils were incubated for different times under water flooded condition to alter their redox x

PAGE 11

status. Metal solubility was examined by analyzing the pore water of the incubated soil columns whereas metal mobility was examined by leaching the columns with 0.01 M CaCl 2 . The data showed that metal solubility in pore water and metal mobility in CaCl 2 solution were not always directly related. There has been sufficient evidence that a reduction in cation exchange capacity (CEC) occurs with Fe reduction dissolution. However, the consequent redistribution of metals between solution and exchange phases with incubation was determined not only by the magnitude of CEC, but also the characteristics of metal ions, the competing co-ions and counterions (ligands). Colloid mobility in the two Pb-contaminated soils was examined by switching the influent from 0.01 M CaCl 2 to de-ionized water. In addition, a batch dispersability test was conducted by dispersing the incubated soils in 0.01 M CaCl 2 and 0.06 M CaCl 2 , and then a relative stability ratio (W) of an incubated soil was estimated from the absorbencytime curves of a dispersion in the solutions. The results showed that colloid mobility is greatly influenced by incubation and that colloidal metal mobility is enhanced when exchangeble Ca is easily replaced. « xi

PAGE 12

CHAPTER 1 INTRODUCTION Understanding heavy metal migration in order to accurately predict heavy metal mobility in contaminated soils is critical for contaminant risk assessment and costeffective soil remediation (McCarthy and Zachara, 1989). Although much effort' has been spent on developing models to predict heavy metal mobility in soils (Cederberg et al., 1985; Sposito, 1984), these models often underestimate metal migration in soils. In most of the models, the interaction between dissolved metals and soil matrix is mainly described using a distribution coefficient (k D ), which is generally considered as a constant and usually measured in the laboratory under a particular condition. In a dynamic soil system, however, k D is not a constant. In other words, solubility of metals and related mobility vary with soil conditions. Besides dissolved species, mobile colloids can also act as vectors to transport sorbed contaminants such as heavy metals in soils (Seaman et al., 1995). Understanding mobility of potential heavy-metal-bearing colloids is critical for predicting the fates of heavy metals in natural porous media such as soils and aquifers (Swanton, 1995; Ryan and Elimelech, 1996) since colloidal particles are ubiquitous in those systems. In concept, the deposition and release of colloids in porous media can be described with the DLVO (Derjaguin-Landau-Verwey-Overbeek) theory. However, it is difficult to apply to soil because of the heterogeneity and irregular dimensions of soil colloids. A colloid/dissolved metal transport model, COMET (Mills et al., 1991), has been developed 1

PAGE 13

to predict colloidal metal mobility. It is difficult to apply to soil environments because of the difficulties in determining the parameters needed in the model. It has been recognized that colloid mobility in soil is mainly influenced by the physicochemical characteristics of colloids and soils (Seaman et al., 1995). Unfortunately, little is known quantitatively about the relationship between colloidal metal mobility and the characteristics of soil colloids (Amrhein et al., 1993; Seaman et al., 1995; Kaplan et al., 1993). In this study, the solubility and mobility of heavy metals, colloid mobility and its association with heavy metals are studied. The main objectives of this research are as follows • To characterize the conditions at which enhanced solubility and mobility of heavy metals occur with incubation in soils, • To characterize the conditions at which enhanced colloid mobility occurs with incubation in contaminated soils, • To characterize the conditions at which enhanced mobility of colloidal heavy metals occurs with incubation in contaminated soils. Therefore, there are two major subtopics in this dissertation: colloidal metal mobility and dissolved metal mobility (Figure 1-1). The understanding of colloid deposition, release and association with heavy metals, is first reviewed and discussed in the next chapter. In Chapter 5, changes in colloid and colloid-facilitated heavy metal mobility with incubation are examined in two contaminated soils. In Chapter 6, a fundamental approach is taken to relate colloid mobility to colloid stability ratio (W). A relative colloid stability ratio is defined and estimated by the curves of absorbency-time.

PAGE 14

Chapter 1 Introduction Colloidal Metal Mobility Dissolved Metal Mobility V Chapter 2 A review on colloid mobility in soil V Chapter 3 Pb mobility in a Pb spiked soil V V Chapter 5,6 Colloidal metal mobility in Contaminated soils Chapter 4 Heavy metal mobility in contaminated soils Chapter 7 Conclusion Figure 1-1 A road map of this dissertation

PAGE 15

On the other hand, the relation of Pb solubility in soil and Fe partitioning behavior after incubation is examined in Chapter 3. In Chapter 4, an extension of the study from an uncontaminated soil to two Pb-contaminated soils is made (Figure 1-1).

PAGE 16

CHAPTER 2 COLLOID DEPOSITION, RELEASE AND ASSOCIATION WITH HEAVY METALS IN SOILS Introduction Colloid-size particles are ubiquitous in soil. It is well known thatcolloids may be mobilized and transported a significant distance (Ryan and Elimelech, 1996; Swanton, 1995) in soil environments. As a result , the mobility of heavy metals associated with the colloids may be enhanced (Kaplan et al. 1995; Amrhein et al., 1993; McCarthy and Zachara, 1989; Newman et al., 1993; Mills et al., 1991; Ouyang et al., 1996). Generally, colloid movement is faster than an inert tracer due to the effect of size exclusion (NocitoGobel et al., 1996). Therefore, colloid-facilitated metal transport has been considered as one of the important mechanisms causing heavy metals to move much faster than expected (Newman et al, 1993), which has attracted much attention in the scientific community. The term colloid generally applies to suspended particles of 1 nm-2 um (McCarthy and Zachara, 1989), whose behaviors are size-dependent. Interactions among large particles (> 1 um, non-Brownian) are affected by physical forces, usch as gravity and fluid drag, whereas those of submicron particles (1 nm-l^m, Brownian) are mainly controlled by the interfacial characteristics of particle-solution. Submicron colloids are of particular interest because of their significance in transporting contaminants due to 5

PAGE 17

6 Figure 2-1. Schematic of potential energy profile of the interaction of surfaces with inclusion of van der Walls attraction, electrical repulsion, and Born repulsion, which shows both primary and secondary minima and an energy barrier as well as the zones in which release and deposition take place.

PAGE 18

7 their large surface areas and high transportability in porous media. Therefore, this dissertation will mainly focus on submicron colloids excluding microbes. Since there is a collection of literature published on colloid release, transport and deposition in recent years (Ryan and Elimelech, 1996; Swanton, 1995; Mills et al, 1991; Ouyang et al., 1996; Adamczyk et al., 1983; Kallay et al., 1987; Elimelech et al, 1995; McCarthy and Zachara, 1989; McDowell-Boyer et al., 1986), no attempt is made to present a complete critique on the topics. Instead, focus will be put on several parameters affecting the mobility of colloids and colloid-facilitated heavy metal transport in soils. At the present time, various processes responsible for colloid deposition and release in porous media have been well established (Ruckenstein et al., 1976; Adamczyk et al, 1983; Kallay et al., 1987; Elimelech et al., 1995; Ryan and Elimelech, 1996). These processes can be theoretically described by the DLVO (Derjaguin-LandauVerwey-Overbeek; Israelachvili, 1992) theory, which states that the net surface interaction energy is the sum of the interactions of electrical double layer (EDL) and van der Waals-London forces (WL), which vary with separation distance between particles. As particles approach each other, the net interaction energy experiences a secondary minimum first, then a primary minimum (§ mm ) after a maximum energy barrier (<)>max) (Figure 2-1). After that point, further movement towards each other causes drastic increase in the interaction energy, which is referred to as the Born repulsion (Israelachvili, 1992). In principle, particles may be attached or deposited when the net attractive forces are close to either the primary or secondary energy minimum. Both the magnitude of the energy barrier and the depth of the primary and secondary minimum are

PAGE 19

8 affected by solution chemistry. In order for a deposited particle to be released, repulsive forces have to be generated between the surfaces of particles and stationary grains, as a result of changes in solution chemistry or fluid flow. However, the understanding of colloid release and deposition is far from complete. Colloid deposition under favorable conditions, in the presence of attractive interactions, can be predicted reasonably well, whereas colloid deposition under unfavorable conditions, in the presence of repulsive interactions, can not be predicted by the DLVO theory. In fact, colloid deposition rates observed experimentally are many orders of magnitudes greater than those predicted and are independent of the size of colloid particles (Elimelech et al., 1995). Elimelech et al. (1995) have recently presented an extensive discussion on possible explanations for these discrepancies. However, compared to colloid deposition, colloid release is poorly understood, which will be emphasized in this review. Transport of mobile colloids is a critical process for evaluating colloid mobility in soils. However, there are several extensive reviews on the topic (Elimelech et al.j 1995; Tien, 1989; McDowell-Boyer et al., 1986; Ouyang et al., 1996), so this topic is not emphasized in this review. Association of soil colloids with heavy metals has been an important issue for many years (McBride, 1994; Jones et al., 1975); however, until recently, the attention has not been paid to the effect of such an association on colloid mobility (Kretzschmar et al., 1997a). In fact, in flotation processes widely used in the mining industry, it is well documented that aqueous heavy metals can be potential determining ions or specific adsorbing ions to charged surfaces of minerals (Fuerstennau and Palmer, 1 976), therefore, such an association with colloid surfaces may alter the

PAGE 20

surface charge significantly. The potential impact of the association of heavy metals with colloids on colloid mobility will thus be emphasized in this review. More recently, Wan and Wilson (1994a,b) have demonstrated that colloid particles tend to sorb onto the water-air interface in soil and that the process is almost irreversible. In fact, these findings have sound theoretical and experimental basis in the flotation discipline (Williams and Berg, 1991). They concluded that colloid deposition onto the interface in unsaturated porous media is one of the important mechanisms responsible for the retardation of colloid transport, except under extremely high flow rate that may drag the bubbles with the flow (Wan and Wilson, 1994a). Incorporating those results into a convective-diffusion equation, Corapcioglu and Choi (1996) have developed a model to predict colloid transport in unsaturated porous media. These results have undoubtedly advanced colloid transport in soils significantly, however, they will not be further discussed in this review. Most of the current reviews focus on either well defined porous systems (Elimelech et al., 1995; Tien, 1989) or subsurface aquifers (Ryan and Elimelech, 1996; Swanton, 1995; McCarthy and Zachara, 1989). There is apparently much similarity between soil and subsurface or well-defined porous systems. Different from the later two, however, soil is more dynamic in its solution chemistry. In surface soil, pH, ionic strength, and redox potential of soil solution are influenced by rainfall, bioactivity, and changes in land use; whereas in vadose zone they are mainly impacted by the fluctuation in water table (Boul et al., 1997). In addition, as mentioned above, the gas phase is not negligible and the interfaces between colloids and air need to be taken into account along with that between colloids and water. Therefore, colloid transport is more dynamic and

PAGE 21

complicated in soils compared to that in subsurface. In this review, we will emphasize how soil environments affect colloid mobility and what are the possible mechanisms. The DLVO Theory The classic DLVO theory has been extensively used to describe colloid interactions and stability (Verwey and Overbeek, 1948). It has been extended to describe colloid deposition and release in porous media by including short-range Born repulsion forces (Ruckenstian et al., 1976). According to the DLVO theory, the total interaction energy between colloids and stationary solid phases (<|)Totai) is the sum of electrostatic repulsion energy, <(>el, arising from the overlap of electrical double layers and attractive energy, LW , due to van der Walls-London dispersion, i.e., Totai = Ei+ <(>wl 5 which is a function of particle separation distance. At close separation distance, a third repulsive force, the Born repulsion, is also important. Thus, the generalized curve of interaction energy vs. particle separation distance shows one energy maximum (<|>max) & two minimums [primary minimum Cmin) and secondary minimum; Figure 2-1]. Aggregation or coagulation occurs when particles collide with sufficient kinetic energy to overcome the energy barrier ( m j n . On the other hand, if coagulated particles gain enough energy under perturbation to overcome the energy barrier (
PAGE 22

11 In fact, significant discrepancies have been found between experimental observation and the prediction based on the DLVO theory (Elimelech and OMelia, 1990b). Therefore, in recent years, various updated theories of colloid stability have been proposed (Shulepov et al., 1995; Adamczyk et al., 1986). The accuracy of prediction has been somewhat increased using more sophisticated models in some systems. Nevertheless, stability and coagulation of clay colloids in solution can be qualitatively described by the DLVO theory in some systems (Verwey and Overbeek, 1948 ); colloid transport in model porous media has also been quantitatively described using the theory plus the Born force (Ruckenstein and Prieve, 1976). Although measured collision efficiency between colloids and porous media surfaces is much greater than the theoretical prediction (Elimelech and O Melia, 1990b), colloid deposition and release can be qualitatively described by the theory (Roy and Dzombk, 1996a). Therefore, the concept of the DLVO theory is correct even though it is incomplete (Swanton, 1995). To reasonably predict colloid interactions in a natural system, the classic DLVO theory needs some modification. Direct measurements of colloid interactions indicate other forces exist between particles (Israelachvili, 1992). The interactions between polar electron acceptors (Lewis acids) in solution and polar electron donors (Lewis bases) from colloid particles are termed AB force and are usually dominant in aqueous systems in addition to electrostatic and van der Waals-London force (Oss et al., 1988). It can account for up to 85% of interactions between soil mineral particles in aqueous solution, as such the DLVO theory combined with the AB force adequately describes some anomalous colloid stability (Oss et al., 1990; Wu et al., 1994a,b). One advantage of this approach is the AB force can be estimated simply by the contact angles between isolated

PAGE 23

12 particles and liquids (Oss et al., 1988). Several contact angles are required to calculate the AB force using the formula of Oss et al. (1988). The measurement needs to be made in at least three different liquids, two of which must be polar. In their approach, the WL force (V W l) can be estimated from surface tension instead of using assumptions for natural colloids as done by Ryan and Gschwend (1994a) and Roy and Dzombk (1996a). Here the classic DLVO theory combined with the AB force is termed the extended DLVO theory. Besides the AB force, other non-DLVO forces between particles, such as osmotic and steric interactions, should also be included. The extended DLVO theory should also correct for colloid surface roughness and heterogeneity of colloid compositions, charges and sizes (Elimelech and O'Melia, 1990a,b). Unfortunately, no models yet successfully take all these factors into consideration. Nevertheless, the DLVO theory provides a solid foundation and a useful tool to describe colloid interactions (Swanton, 1995; Ryan and Elimelech, 1996). Therefore, at the present time, it is practical to describe the interactions between colloids using the extended DLVO theory ((j)Totar <|>el+
PAGE 24

repulsion). Over this region, the energy profile is very sensitive to the specific characteristics of the interacting surfaces and intervening liquid layer. However, the irregular dimensions of natural soil colloids make the interaction even more complicated, which will be discussed later. The Characteristics of Colloids Charge development Solid particles present in water are often charged. The mechanisms of charge development have been extensively investigated (Parks, 1965; Sumner, 1992). In general, based on the nature of solid particles, charge development can be divided into three categories: 1) lattice substitution is most common in soil clay particles and referred to as permanent charge; 2) specific chemical interactions between surfaces and solution, including hydrolysis of surface functional group (e.g., hydroxyl and carbonyl) and chemical adsorption; and 3) preferential dissolution resulting from preferential hydration of surface atoms. In addition, based on the contribution from electrolytes in solution to the charges of particles, ions can be classified into three categories: potential determining ions, specific adsorbed ions and indifferent ions. Potential determining ions are the constituent ions of solid particles (e.g., H + and OH" for Fe oxides, Ca 2+ and CO3 2 " for CaCOs) and their concentrations determine primarily the surface potential of particles (Parks, 1967). Specific adsorbed ions can change the magnitude and sign of the surface charge (e.g., Ca 2+ , Pb 2+ ) by adsorbing onto the surfaces and forming a stern plane in electric double layer (EDL) theory (Parks, 1975). Indifferent ions adsorb physically to surfaces and change the magnitude of the surface charge (e.g. Na + , CI"), and form a

PAGE 25

diffusive layer in EDL resulting from a balance of their electrostatic and osmotic interactions with the surface and bulk solution (Parks, 1975). In soils, however, colloids are heterogeneous in composition. They generally consist of inorganic and organic constituents. Heterogeneity in the composition and structure of colloidal particles make their charge development a complicated matter. This may be further complicated by the dynamic nature of soil solution chemistry, which may cause colloid composition and structure to vary temporally and spatially. Nevertheless, numerous studies have shown that the charge developing process can still be described using the EDL theory, at least in principle. Electrical double layer (EDL) in metal oxides-water interface Electrical double layer of a dispersed particle consists of a charged surface and a diffusive layer of counter ions next to the surface based on the double layer theory. Since metal oxides may play an important role in colloid mobility and have been well characterized among various surfaces of soil colloid particles (Schindler and Stumm, 1987; Dzombak and Morel, 1990), we use them as an example. When specific adsorbing ions other than H + and OH'are absent in a system, the charge developing process on the surfaces can be described as follows (Schindler and Stumm, 1987): SOH 2 + = SOH +H + K a i int (2-1) SOH SO + H + Ka2 int (2-2) Where SOH denotes a surface site and K a i' nt and Ka2 int are the intrinsic surface acidity constants and defined as follows: K a i int [SOH 2 + ]/[SOH] [H + ] (2-3) Ka2 int = [(SO"] [H + ]/[SOH] (2-4)

PAGE 26

15 Schindler and Stumm (1987) summarized the intrinsic acidity constants (K a i' n and Ka2 in ) of various metal oxides and found that the constants are generally correlated to those in solution. However, the affinity of protons to the surfaces of metal oxides is much more complicated and influenced by changes in surface composition and structure (Yoon et al.,1979; Sverjensky, 1994; Bleam, 1993). In addition, the values of these constants depend on the particular electrostatic model adopted, such as constant capacitance, diffuse double layer, and triple layer . However, the sum of Eq (2-1) and (2-2) can be expressed as follows: SCr + 2H + =SOH 2 + (2-5) Eq. 2-5 is simply related to the point of zero charge (PZC or pH 0 ), which is experimentally measured and is independent of electrostatic models for the solid-water interface. When the concentration of positively-charged surface species is equal to that of the negatively-charged, that is, a zero-charge surface, the pHo is related to the intrinsic equilibrium constant pH 0 =0.5 log K a int = 0.5 log K a (2-6) Eq (2-6) states that intrinsic acidity constant (K a int ) is equal to the apparent acidity constant (K a ) at a surface potential of zero. It not only indicates that pHo is a measure of K a mt , but also raises the question that the bonding of protons at metal oxide surfaces is more analogous to the bonding of similar complex in an aqueous phase (Schindler and Stumm, 1987) or in bulk crystal structure (Sverjensky, 1994). The correlation between the surface acidity constants of metal oxides and those in solution support the former point. However, the scatter on such correlation is substantial for some mineral particles (Schindler and Stumm, 1987). More recently, Blesa et al. (1990) reviewed the

PAGE 27

16 thermodynamic data for the adsorption of protons on metal (hydrous)oxides. They found that the enthalpy values associated with K a i' m and Ka2 int are very similar and are much lower than those of the hydrolysis of aqueous metal ions. This is consistent with Lyklemas work (1987a), in which he describes the adsorption of potential-determining ions as a process characterized by a single enthalpy value instead of two. These indicated that it is unrealistic to represent acidic, neutral, and basic surface groups using Eqs. (2-1) and (2-2). The active surface site may be a surface metal complex and notation S (OH) m0 (OH2) n0 (m 0 and n 0 being the stoichiometric index at pH 0 ) offers a better description of the involved processes; for pHno; for pH>pHo, m>mo and n
PAGE 28

17 1993), Sverjensky (1994) has shown that surface protonation on a wide variety of minerals can be accurately calculated from the dielectric constant of the solid and the ratio of the Pauling bond strength to the cation-hydroxyl bond length of the solid particles. This suggests that much more emphasis should be placed on the analogy between the bonding of the surface protonated species and the bonding in the underlying crystal structure. In Sverjensky s approach (1994), the properties of crystals have been numeralized with its electric constant instead of the qualitative terms such as kinks, edge position, or adatom positions. The details on modeling or calculating pH 0 can be found in the literature (Parks, 1965; Yoon et al., 1979; Sverjensky, 1994; Bleam, 1993). Fokkink et al. (1987, 1989) have demonstrated that pHand temperaturecongruencies exist in the surface charge developing processes of various metal oxides. In a systematic series of experiments in which the surface charge ao of a number of oxides [Ti0 2 (rutile), Ru0 2 and a-Fe 2 0 3 (haematite)] in aqueous solution of KN0 3 was measured as a function of pH, ionic strength and some other variables. When ao is plotted as a function of pH-pH 0 , where pH 0 is the PZC, the curves coincide for the three oxides at three ionic strengths even though they are different in pH 0 . In other words, the individual identity of metal oxides makes no difference to the S-shaped plot of ao vs. pHpHo. Similarly, in the measurements of the temperature dependence of surface charge development, changes in temperature only affect the positions of PZC, not the trend of the surface charge developing process. They conclude that the electrical double layers in metal oxides can be functionally divided into a specific and a generic part, with respect to the natures of oxide and electrolyte. The specific part, which is determined by the specific interactions of a surface with protons in solution, determines the PZC. This is

PAGE 29

18 consistent with the result of Sverjensky (1994) in concept. The general part, which is controlled by the solution side of the double layer, determines the surface charge development once the surface charge deviates from pH 0 . It can be described well by the Nernst law (\\i 0 =2.303 RT/F (pH-pH 0 ). The significance of this result is, first, that it has provided a thermodynamic basis for double layer model that divides the charged interface into a charged surface and a diffuse layer near the surface; second, that the Nernst law can be used to describe the charge developing process in metal oxides. It has been recognized that the Nernst Equation is valid only when the chemical potential of H + in interface is a constant (Blesa, 1988). In a practical sense, the potential can be treated as a constant if the amount of H + in the interfacial layer is much higher than a change caused by the charge developing process (Blesa, 1988). However, for metal oxides only 3-10 sites per 100 A (1 ran 2 ) are usually found (Sigg et al., 1981) and in the experimental pH range up to 60% of the sites may be charged (Blesa, 1988). However, if the notation of the binding site S(OH) m o(OH2) n0 we discussed above is more representative of the real sites on metal oxides compared with SOH, adsorbed protons can be transformed into neutral species (H2O) on surfaces, and thus the adsorbed H + would not change much with surface charge. This may result in the Nernstian behavior. 2+ 2+ In the presence of specific adsorbing ions (SAIs) such as Ca and Mg in solution, the charge developing process becomes more complicated (Lyklema, 1984; Ardizzone, 1982). Generally, the PZC will be shifted compared with the one in the absence of those SAIs. Lyklema (1984) defined the PZC, even in the presence of SAIs, as a point at which H + and OH' are just balanced at the surfaces. This is equivalent to the surface charge
PAGE 30

sensitivity of oxides to H + and OH'. Consequently, SAIs belong to the Stern charge. Therefore, the PZC in the presence of SAIs remains fully defined. In their definition PZC is still the pH where ct 0 =0. If the cation adsorbs on to the Stern layer PZC is lower than the PZC in the absence of SAIs because cations in the stern layer favor the adsorption of OH" over H + , so that a lower pH is required to restore the H + -OH + adsorption balance. Similarly, specific adsorption of anions leads to PZC's above the pristine value. In fact, in the presence of SAIs, the common intersection point (CIP) of titration curves with various ionic strengths has also been observed. Similar to the situation in the absence of SAIs, the CIP is the equal compensation point (Lyklema, 1984), at which charge-compensating cations and anions have an equal affinity to the surfaces. From this point of view, they explained CIP *PZC in the case of specific adsorption as follows: at the PZC adsorption of metal cations are favored over that of anions if cations have a higher affinity to the surface. In order to reach a situation where the intrinsic adsorbability of cations and anions is identical (equal compensation point), a positive charge on the surface has to be developed. This charge should be more positive with higher specific affinity of cation to the surface. Once the equal compensation point has been reached, further increase in concentrations of cations and anions do not lead to further shift, i.e., all successive curves at increasing concentration must pass through the point. Similar reasoning applies to a situation at which anions have a higher affinity over cations. This principle is important to understand the impact of soil solution chemistry on soil colloid charge development. Following this point, it is easy to understand the effects of changes in species of cation and anion in solution on PZC and CIP.

PAGE 31

20 In the discussion above, we have been dealing with well-defined crystal surfaces, in which the dissolution can be neglected using surface complexation approaches. However, solubility of metal oxides has to be taken into account, especially in the case of amorphous (hydr)oxides. Under such conditions, the adsorption and desorption of the hydrolyzed metal complex ions may become an alternative to determine surface charge other than the surface complexation of H + and OH'. Based on the minimum solubility theory (Parks et al., 1962), PZC should be identical to the isoelectric point (IEP) of an aqueous solution suspending the particles. The IEP is defined as the pH resulting from an equivalent concentration of positive and negative complexes in aqueous phase (Parks, 1965), which is often found at the pH of minimum solubility of a solid. Blesa et al. (1997) have found that the PZC and IEP for metal oxides do not always coincide; however, the deviation from IEP (PZC IEP) can be explained well if considering the fact that the complex cations and anions require different dehydration energy when they are transferred from solution to surface. They explained that monomeric cations hydrate more strongly than anions in aqueous solution, so the shift should be small and negative when the charge is determined by monomeric surface complexes and the minimum solubility theory is generally valid. On the other hand, for polymeric species, more specific behaviors are expected. Relative large positive or negative shifts may result if the charge surface complexes are polymeric. Obviously, the underlying assumption of these approaches is that the surface active site is more analogous to their solution species. This may be more suitable to soil environment where various amorphous minerals are abundant and aqueous species are more understood than surface species.

PAGE 32

21 Charge development in soil particles In general, surface charge of soil particles can be classified into two types: permanent and pH-dependent charges. Detailed characterization of soil particle charges has been developed by Charlet and Sposito (1987), Anderson et al. (1991) and Chorover and Sposito (1995). Charge developing processes on soil particles are more complicated than those on metal oxides because the surface composition and structure of soil particles varies greatly and PDIs are not limited to just H + and OH'. However, similar phenomena such as CIP have been observed, which is referred to as the point of zero salt effect (PZSE). It is often found that PZSE deviates from CIP (Chorover and Sposito, 1995). Soil is a mixture of various minerals and organic substances. Further, there can be significant amounts of amorphous materials, such as amorphous Fe and Al. It is well documented that the existence of hydrolyzed species of Al and Fe have significant impact on the pH-dependent charge of soil particles (Chorover and Sposito, 1995, Parker et al., 1 979), and their complexation with organic matter further complicates the charge developing process (Chorover and Sposito, 1995). In addition, mineral dissolution may be significant for some soils and may affect the charge developing process greatly (Blesa et al., 1997). Combining all these factors with the dynamic nature of solution chemistry, one may conclude at the present time it is impossible to accurately predict proton surface charge developing process. In principle, however, Blesa et al's approach (1990; 1997) is more promising. They emphasize the role of partitioning of charged species between solid surface and solution during surface charge developing processes, which is driven by the difference in Gibbs free energy between a charged species in solution and on surfaces. The underlying

PAGE 33

22 assumption of this approach is that the charged species in solution and on surface are similar. This approach has relative sound theoretical basis and is consistent with experimental evidence. Charge development in mobile colloids Considerable amount of research has been conducted to understand charge development of water dispersive and mobile colloids (Heil et al., 1993a,b; Kretzschmar, 1993; 1996). It is commonly observed that the stability of colloidal clays in soils is many times greater than that of comparable reference clays. Also, clays isolated from surface soils were found to be more dispersive than clays from the subsurface horizons of kaolinitic soils in the Southeastern USA. Factors contributing to the high dispersibility of soil clays include the presence of adsorbed organic matter, minor quantity of smectite in kaolinitic clay, and larger surface roughness of soil clays compared with unweathered reference clays. Organic matter is an important source of negative charges in soils. Well decomposed humus may have a CEC > 300 cmol kg" 1 humus, which is considerably greater than that of clays such as kaolinite (3-15), illite (30-40) and montmorillite (80150). It has been estimated that 20-70 % of the CEC of many soils attributes directly to the soil organic matter alone (Vaughan et al., 1984; Stevenson, 1982). In soils, dissolved organic carbon (DOC) tends to adsorb onto solid particles such as clays and metal oxides, which is driven by ligand exchange, multivalent ion bridging, Van der Waals force and hydrophobic interactions between DOC and the solid minerals (Murphy et al., 1995). In particular, for Fe and Al oxides, when the pH of a system is lower than their PZC, adsorption of DOC onto the minerals increases significantly because of the electrostatic

PAGE 34

23 attraction between the two, resulting in a significant impact on their surface charge. There are numerous studies of the effects of organic coating onto metal oxides (Tipping et al., 1982) and clay minerals (Kretzschmar, 1996) on surface charge. Heil and Sposito (1993a, b) found flocculation of illitic soil colloids with organic coatings increased with pH in Ca solution, which is contrary to their flocculation behavior when organic matter is removed by H2O2. They concluded that the competition between proton and Ca onto the acidic functional groups of organic molecules is essential to the charge development on coated colloids. A similar result was also found by Kretzschmar (1993; 1997) in humiccoated kaolinic soil clay. On the other hand, it has been reported that the electrical properties of the quartz surface dispersed in river water (Whitray Bech, UK) is determined essentially by their interactions with inorganic cations such as Al and Ca instead of organic matter (Findlay et al., 1996). In concept, however, these results are consistent with others, i.e., surface electric properties of minerals are a result of interactions both in the interface of particle-solution and in solution, including particleorganic matter, organic-inorganic ions, particle-inorganic ions, etc. Particularly, types of cations, such as proton and multivalent cations, and their concentrations are important for the charge development on particles in addition to organic matter. It is unrealistic to model surface charge development on soil particles with reasonable certainty at the present time. However, it is evident soil particles can be treated as assemblages of crystalline and amorphous minerals and organic residuals; Sophisticated models such as the triple layer model do not make much sense in describing the electrified interfaces of such particles considering that the particle surfaces are irregular and the Stern layer often moves inward inside of the physical boundary of

PAGE 35

the particles. As a first approximation, on the other hand, the interface of soil particles is more reasonable described by diffuse double layer with a specific part, i.e., a charged surface, and a generic part, i.e., a diffuse layer. The surface charge developing process may be described by the partitioning of charged species between solution and surfaces assuming the charged species are similar between the two phases. Hydration Water molecules may be orientated surrounding the immersed particles, forming "Hydration shells" (Clifford, 1975). Once the particles with such shells approach each other, additional forces between the particles arise. The origin of these forces is the interactions between polar electron acceptors (Lewis acids) in water and polar electron donors (Lewis bases) from colloidal particles, and termed AB forces as discussed previously (Oss et al, 1987; 1988). Surface electron donicity of colloid particles is crucial for water molecules to be orientated along the surface. Strong surface electron donicity causes more ordering of water molecules orientating along the surfaces, resulting in a repulsive force between the particles. Weak surface electron donicity causes less ordering of water orientating, resulting in an attractive force. Oss et al. (1990) have demonstrated between particles dispersed in water, the interaction free energy from AB, whether repulsive or attractive, is commonly as much as 100 times greater than LW energy, and 10 or more times greater than EL energy at close range (1-5 nm). It can account for up to 85% of the interactions between mineral particles in aqueous solution, such that the DLVO theory combined with the AB force can adequately describe some anomalous colloid stability (Oss et al, 1990; Wu et al, 1994a,b). Wu et al. (1994a,b) have experimentally evaluated the interactions between the particles of montmorillonite and

PAGE 36

25 calcite by examining flocculation of their suspensions while increasing Ca 2+ concentration. They concluded that Ca 2+ not only causes a decrease in particle % potential, but it also drastically lowers the electron donicity of the polar surface of the particles, resulting in an AB attractive force, which is more responsible for the flocculation of the suspensions. Obviously, SAIs such as Ca 2+ alter not only % potential when it adsorbs onto a particle, but also decreases the surface electron donicity. This result means that surface electron donicity and \ potential of particles are related. Size development of mobile colloids The behaviors of colloids are size-dependent. Individual colloidal particles may aggregate depending on the magnitude of the energy barrier of their interaction (Figure 21), which is similar to colloid deposition that will be discussed later. Briefly, if the energy barrier of interactions is low, fast coagulation occurs, in which case the diffusion rate of colloids is the limiting factor; the aggregation rate is otherwise controlled by the interaction of colloidal particles. These two mechanisms result in lower and higher fractal dimensions of aggregates, respectively (Riscovic et al., 1996). However, direct measurement and theoretical simulation of colloid size distribution in natural system have shown that the concentration of colloids < 0.1 um should be negligible because of their almost instant coagulation (Filella et al, 1993; Buffle et al., 1995). Kaplan et al. (1997) examined the possibility of aggregation of mobile colloids from a reconstructed soil profile. They analyzed the size distribution of particles in the suspensions with and without sonification, which is supposed to break down aggregation in the suspension. They found that the suspension without sonification exhibits almost the same bimodal particle-size distribution as the one with sonification. This holds for soil suspensions

PAGE 37

26 collected from two different soils: a loamy sand and a sandy soil in Southeastern USA. Their results suggest that colloid aggregation in soil pore space is very limited especially when soil colloid particles are highly charged (Kaplan et al., 1995; 1997). Interestingly, similar bimodal size distribution of mobile colloids has been reported by Ronen et al. (1992), who characterized the suspended particles collected from groundwater in a coastal plain phreatic aquifer of Israel. Kaplan et al. (1996) explained that smaller particles are less likely to be strained during transport in porous media based on the filtration theory (Yao et al., 1971). This suggests that, in the bimodal distribution, colloids of smaller size (< 0.4 urn) may have resulted from long-distant translocation to sampling point, while those of larger size (<1 urn) may come from local dispersion at the sampling point. The Characteristics of Porous Media Mobile colloids interact with the surfaces of porous media, which is more important than those between colloids in limited pore space. Therefore, pore structures and their physical-chemical surface characteristics have tremendous influence on colloid mobility. Extensive studies have been done in the systems consisting of clean surfaces of porous media (Adamczyk et al., 1983; Kallay et al., 1987. Elimelech et al, 1995). However, large discrepancies in colloid deposition rate under unfavorable deposition conditions between theoretical predictions and experimental observations have been found. This has been attributed to the hydrodynamic effect and surface heterogeneity of the porous media, such as surface roughness and local charge heterogeneity. This has been considered to be the most promising approach to understanding the interactions between colloids and porous media in clean systems (Song et al., 1994). There are some

PAGE 38

exhaustive reviews on this topic (Swanton, 1995; Ryan and Elimelech, 1996; Elimelech, etal., 1995). In a soil system, the porous media is an assemblage of a wide range of particles, which vary greatly in mineralogy, surface composition and dimensions. Different from pure systems, where the surface properties of porous media may be far different from those of colloids, the surfaces of soil porous media may at least in part consist of potentially mobile colloids. In sandy soil, a small amount of clay-size particles tends to pack closely along large sand grains driven by energy minimum. This arrangement among a grain and colloidal particles has also been confirmed by fractal analysis (Bartoli et al., 1991). Based on this fact, the interactions between colloids and porous media can be treated as those between colloidal particles, as a first approximation, which has been a long-established concept. Swanton (1995) has suggested that in an undisturbed soil profile, the active site of the porous media may be completely occupied by mobile colloids. Ryan and Gschwend (1994a) have successfully related the observed colloid release rates from an Fe oxide-coated sand columns to EXP (min)/kBT. Their estimation of the energy barrier (min) is based on the interactions between colloids rather than those between colloids and porous media surface. The underlying assumption is that porous media surfaces are identical to colloid surfaces. Deposition of Colloids Theoretical background in well-defined porous media Colloid mobility through porous media is mainly determined by the net rate of colloid deposition and release. Generally speaking, deposition of colloids onto porous

PAGE 39

28 media can be viewed as a two-step process: the transport of colloids from bulk solution to the proximity of the surfaces and then their attachment to the surfaces, which depends upon the nature of particle-surface interactions. If the energy barrier ( max ) is < 0, the flux of colloid transport is equal to that of deposition. In other words, the collection efficiency of the collector (porous media) is 100%. The energy barrier <|> max must otherwise be overcome in order for a colloidal particle to attach to the surfaces (Figure 21). Therefore, deposition rate constant (k dep ) is related to the energy barrier max with k dep oc EXP H>max/kBT), where k B is the Boltzmann constant and T is absolute temperature (Ruckenstein et al., 1976). The major mechanisms of colloid transport to a collector are inertial impaction, interception, sedimentation, electrostatic forces, Brownian diffusion, and straining. Collection efficiency contributed from each of them has been described by Tien (1989), and the details will not be repeated here. In porous media, however, many factors mentioned above may be operative simultaneously. The overall collection efficiency can be theoretically calculated based on the filtration theory (Yao et al., 1971). However, the great discrepancy between prediction and experimental observation under unfavorable deposition conditions has been observed. This has been attributed to the distribution of surface and physical properties, surface charge heterogeneity of solids, surface roughness, interfacial electrodynamics and colloid deposition in secondary minima (Elimelech et al., 1995). In order to take this discrepancy into account, collision efficiency (a=r|/r|o) is used, where r\ and r|oare observed and calculated collection efficiencies, respectively. As particles deposit on a collector, after an initial stage, the deposition rate will change depending on the nature of particle-particle interaction. If the net interaction is repulsive, the collector surfaces become progressively occluded as

PAGE 40

29 particles accumulate and colloid deposition rate declines accordingly. This surface exclusion phenomenon is termed blocking. There are some excellent recent reviews on this topic (Ryan and Elimelech, 1996; Elimelech et al., 1995; Swanton, 1995). Colloid deposition in soil Transient phenomena in colloid deposition . There are no clean or pure surfaces in soils. Therefore, the results obtained in well-defined porous media cannot be directly extrapolated to describe colloid deposition in soils. In fact, in any given time in most soils, the most favorable deposition sites are likely to readily be occupied. Alternatively, if there is influx of mobile colloids as a result of some chemical or mechanical disturbance, there may be limited deposition during the early stages of exposure. This occurs because favorable deposition sites are already filled, which is different from clean porous media. Over a prolonged period of exposure for a colloid flux, on the other hand, soil porous media may develop surfaces that are more similar to those of mobile colloids (Swanton, 1995), which has been discussed before. In soil porous media, therefore, colloid deposition may be a transient phenomenon. Dispersable fraction of soil clay resulted from vigorously shaking in water can be a relative measure of the amount of potentially mobile colloids. Miler et al. (1986) observed the amount remaining in suspension after 36 h of shaking was highly correlated to surface soil loss in the Southeaster US under high intensity rainfall. Kaplan et al. (1997) have found mobile colloids from two soils are similar in mineralogy to the water dispersible clays and that they are many orders of magnitude lower than the amount of water dispersible clay in the soils. In other words, there is significant amount of

PAGE 41

30 potentially mobile colloids loosely attached to the soil porous media, however, during this period they function as a part of stationary porous media. They may be detached from the porous media (mobilized) once they are exposed to a chemical or hydrological perturbation. I have referred to this stage a transient phenomenon in colloid deposition. In fact, this is not a new concept at all; however, it has to be emphasized to understand the distinctively different features of colloid mobility in soil porous media. Take surface soil for an example. A perturbation such as rainfall occurs on the soil surface and moves downward. Such a perturbation may result in a release of deposited colloids, and then the mobilized colloids move down over the surfaces of porous media consisting of the deposited colloids. Therefore, at the front of the downward-moving perturbation, there are two opposite processes operating simultaneously: deposition and release of soil colloids. The former is mainly influenced by the blocking effect, and the latter will be discussed in the following section. Size straining in deposition . If the effective size of a colloid particle is larger than the smallest pore through which fluid is flowing, the particle is retained by porous media, which is referred to as size straining. Straining occurs in granular bed filtration if the ratio of the suspended particle diameter to the grain diameter is greater than about 0.2 (Herzig et al., 1970). Straining is one of the most important mechanisms of colloid deposition in soils (Seta et al., 1997; Jacobsen et al., 1997; Kaplan et al., 1997). Seta et al. (1997) have examined the transportability of water dispersible clay through intact soil columns and found that size straining is an important factor in determining the concentrations of colloids in leachates along with pH, total exchangeable bases, cation exchange capacity, organic carbon content, etc. They concluded the relative low colloid

PAGE 42

31 transportability is attributed to larger colloid size, i.e. size straining. The straining may manifest itself in another way. Jacobsen et al. (1997) examined the transportability of illite though soil columns containing macropores, and found illite concentrations moving through soil columns are significantly related to the sizes of active macropores in soil columns. However, no significant difference in mobility through the columns between the illite and that coated with humic acid was found. This suggests size straining is the predominant process in determining the deposition over the zeta potential and steric effect, resulting from the coatings of humic acid. In their study, this is further confirmed by the fact that colloid concentrations in leachates increase with the intensity of infiltration, as they argued, a high intensity of infiltration depresses the effect of straining. However, this is contrary to the results of Kretzschmar et al. (1995). Kretzschmar et al. (1995) found colloids coated with humic acid result in a much lower collection efficiency in columns packed with "clean" saprolite particles. In fact, it is this contradiction that demonstrates the difference of colloid deposition between soil porous media and clean porous media: size straining may be a dominant mechanism if blocking effect is operative in soil. Release of Colloids in Soil Different from colloid deposition, which is essentially determined by the interactions at larger separation distance, colloid release depends on interactions at separation between the primary minimum (§ mm ) and surface (Figure 2-1) and is greatly influenced by additional repulsion at short separation. Colloid release is generally expected when the repulsion is generated between porous media and colloids. The

PAGE 43

32 efficiency of colloid release depends on whether the process is caused by diffusion alone or by an applied external force (Kallay et al., 1987). Without external forces, the rate of particle release is proportional to the diffusional escape probability of the deposited colloids through the energy barrier, i.e., the energy well ( ma x-<|>min). It is otherwise determined by the shape of interaction energy profile between surface and the primary energy minimum (Figure 2-1) as well: the smaller the slope (d(j)/dh) of the curve the less force needed for colloid release, where h is the separation distance. Up to this point, it is clear that the two opposite processes, colloid deposition and release, are controlled by different mechanisms and factors. For a given system, mobile colloid concentration is determined by the relative magnitude of colloid deposition and release rates. The residence time of colloid-carrying solution in the system also plays an important role. A longer residence time represents a condition closer to the equilibrium of colloid deposition and release, while a short one may discriminate deposition over release and vice versa. For example, colloid dispersion in batch experiments, in general, is a result of balancing between colloid deposition and release because of the longer residence time in general; whereas colloid release from a short column may be free of redeposition effects. Therefore, one must take extra caution when extrapolating colloid stability obtained from batch experiments to colloid transportability in column experiments where kinetic effects are more pronounced. In a soil, there are numerous energy profiles for various colloids depicted in Figure 2-1 due to the heterogeneity of colloids and soil porous media. Each energy profile represents the interaction between a individual particle at a specific location on porous media. For simplicity, we imagine that those curves can be lumped into several

PAGE 44

33 groups based on similarity, each of those lumped profiles representing the interaction of a group of similar particles with porous media with a certain magnitude and position of the energy maximum and minimum. Therefore, the deposition and release of soil colloids may manifest themselves in a strong chromatographic manner (Chapter 5). The processes of colloid release Similar to colloid deposition, colloid release rate is generally determined by both colloid detachment from porous media and transport to bulk solution. Ryan et al (1994a) examined the release rates of colloidal hematite from hematite-coated sand columns under different ionic strengths and flow rates. They found that colloid release rate decreases as ionic strength increases, which was supposed to reduce the size of the energy barriers. In addition, colloid release rate also decreases with increasing flow rate, which is contrary to the expectation that a greater mobilization would occur at a greater flow rate due to greater hydraulic stress on deposited colloids. They suggested the ratelimiting step in colloid release is the transport of detached colloids to the bulk fluid when rapid colloid release corresponds to conditions where the energy barrier has vanished from the potential energy profile. Coincidentally, a similar phenomenon has been observed by Jacobsen et al. (1997) in intact soil-column experiments where natural particles are leached with tap water. Their results showed particle mobilization is not influenced by increase in flow rate. Further, the plot of accumulated amount of mobilized particles versus square root of time shows a fairly linear relation, implying diffusion limited kinetics. One necessity for diffusion kinetics being dominant is the barrier maximum of mobile colloids vanishes, which may be very common in soil because of the transient phenomenon discussed before. The lack of flow rate effect in the

PAGE 45

34 latter study may be caused by the smaller slope of energy profile (Figure 2-1) for potential mobile colloids at the given condition, which may be overcome by increases in hydraulic stress, subsequently, these potential mobile colloids become truly mobile. Therefore, colloid dilution in effluent may be compensated by its enrichment from mobilizing more potential mobile colloids by increased flow rates. This is based on the heterogeneity of the interactions between colloids and media in soil. Mobile colloids and water-dispersible clay Kaplan et al (1993; 1994; 1997) examined extensively the release of colloids from an Ultisol in the southeastern coast of the US. Mobile colloids were collected during and after mild rain events from the reconstructed pedons. They found the mobile colloids are similar to the water-dispersible clay in mineralogy, which is consistent with the results reported by Seta et al. (1997). However, there were some differences. Kaplan et al (1997) found that the sizes and compositions of the mobile colloids differed from those of the water dispersive clay, with the former being smaller and generally enriched with kaolinite, Fe oxides, gibbsite, and organic carbon. In addition, compared to the total clay fractions in the reconstructed pedon, the mobile colloids are more dilute in quartz and HIV (hydroxy-interlayered-vermiculite). Based on the results from scanning electron microscope (SEM) and photon-correlation spectroscopy, essentially all the mobile colloids (>90 %) have diameters of about 0.23 um and moved through the soil as discrete, non-aggregated particles. They concluded the particles enriched with mobile colloids are not only readily dispersible but also smaller in size than water dispersive clay. Further, they proposed colloid mobilization in the pedons is a result of two consecutive processes: dispersion of highly charged particles due to changes in soil

PAGE 46

35 chemistry and induced water flow, which is determined by particle composition and can be evaluated by water dispersibility of a soil, and their transport through a pedon, which is size-dependent. However, it has to be realized that water dispersibility of soil particles obtained from a batch experiment is not always equivalent to colloid's ability to stay detached (mobilized) in a soil. Colloid residence time has to be taken into account, which has been discussed above and is supported by Kaplan (1996). Influences of exchangeable sodium percentage (ESP) on stability of water-dispersible clay and mobilization of soil colloids Kaplan (1996) summarized if particle straining is not a limiting factor for colloid release, the dispersability of soil water-dispersible clay may correlate to colloid transportability in soils. The dispersability of water-dispersable clay has been studied extensively (Miller et al, 1990; Sumner et al., 1993). Even though we cannot completely understand the dispersability of soil colloids, it is undoubtedly affected by factors such as pH, ionic strength and exchangeable sodium percentage. Kaplan et al. (1997) reported that colloid concentrations in the effluents from soil pedons are highly correlated to the ESP of those soils (R 2 =0.9 1-0.93, p< 0.01), but not so to soil pH or total electrolyte concentrations. They explained the dispersive property of the highly hydrated Na + ion attributes to the positive correlation as the dispersion of the soil colloids appears to be a more important process in colloid release in the sandy soils. In fact, the positive correlation between soil colloid dispersability and exchangeable sodium percentage has been found by other researchers (Frenkle et al., 1978; Suarez, 1984) who investigated the effects of irrigation or rainfall on hydraulic conductivity in a sodic soil.

PAGE 47

36 Seta and Karathanasis (1997) examined the relation between the stability and transportability of water-dispersible soil colloids by pumping water-dispersible clay through an intact soil column. They found colloid recovery, after five pore volumes, depends on not only the type of soil column but also the characteristics of colloids. Among the colloid properties, pH and total exchangeable bases are significantly correlated with colloid recovery. The effect of pH on colloid migration can be explained by its effect on colloid stability: the higher the pH above the pH 0 (< 4.0), the greater the colloid stability. The strong correlation between total exchangeable bases and colloid recovery is anticipated because of its high correlation with pH. In addition, they stated "contribution of total exchangeable bases to colloid transport may derive from cationexchange reaction between the colloid-saturating cations and the column matrix, which reduces colloid interaction with matrix surfaces and thus enhances colloid transport". Finally, they concluded the best single independent variable predicting colloid recovery is total exchangeable bases (R 2 =0.97). Even though there is some inconsistency among the results in different studies described above, there is sufficient information to show ESP is consistently correlated to the dispersability of soil colloids in a soil. Rengasamy (1982), however, demonstrated that even Ca-saturated clay, i.e., ESP=0, could be dispersed provided the soil is free from electrolyte by dialysis, suggesting exchangeable Na is not essential for dispersing a soil. Further, in a soil system, solution chemistry is dynamic and, therefore, concentrations and types of cations in soil solution and exchange sites may vary greatly. In this sense, parameters such as total exchangeable bases, which takes cations besides sodium into account, may be more reasonable to describe colloid transportability. The effects of

PAGE 48

37 cation exchange reactions on colloid release have been demonstrated by Roy et al. (1996b), which will be-discussed in the following section. Ion transfer processes during colloid release Observation of ion transfer during colloid release in well defined systems . Kallay et al. (1986) examined the effects of neutral electrolytes on the detachment of spherical colloidal particles of goethite from glass surface in basic media. They unexpectedly found colloid removal was enhanced with increasing concentration of NaNCh. Such unusual behavior could be predicted if constant potential is assumed. They explained this behavior corresponds to a double layer that undergoes relaxation during particle detachment. In order to keep the potential constant as the distance between the particle and the surface increases, the adsorption of the potential determining ion (OH") must occur with consequent equilibration of all ionic species in the Stern and diffuse layers. During particle detachment, ion transport from solution phase into the interfacial layer is accelerated with increasing distance between the surfaces due to solution influx from bulk. Kallay et al. (1987) summarized this unusual behavior can be observed under the following conditions: colloid redeposition is minimal (short column); particles and media have alike charges; surface potentials are constant during colloid detachment. They argued this apparently unusual behavior can be understood if one takes into consideration the energy profile as a function of distance near the surfaces (Figure 2-1). The probability of colloid detachment depends on the depth of energy well ( max -<|>min), which is a function of several parameters. On the other hand, the probability of colloid, deposition depends on the height of the energy maximum ( ma x). As the concentration of

PAGE 49

38 NaN03 in the system increases, the depth of energy well reduces, resulting in an increased rate of colloid detachment; at the same time, the energy maximum decreases, resulting in an increased rate of deposition. The amount of colloidal particles observed in effluents is determined by the net rate of these two opposite processes. The essence of Kallay s (1986, 1987) rationale is that a change in the electrical environment of the interface, resulting from colloid detachment, may induce ion flow between solution and interface (EDL), which in turn may facilitate colloid detachment. This has been known as so-called surface chargeor potential regulation, which has been discussed by Chan et al. (1974). They have shown there may be significant ion flow between solution and interface during EDL interaction if the particles carry pH dependent charge. This indicates ion transfer (H + ) may also be important during colloid release since pH-dependent charge is generally abundant on soil colloid particles. Ion transfer during soil colloid mobilization in literature Shainberg et al. (1981a) examined the effects of electrolyte concentration on hydraulic conductivity (HC) of a sodic soil. One of the major mechanisms of the influence of electrolyte concentrations on hydraulic conductivity is by mobilizing soil colloids and then plugging them into conducting pores. They found both clay dispersibility and hydraulic conductivity of the soil were very sensitive to the level of exchangeable Na in the soil and to the salt concentration of the percolating solution. When salt concentration in the soil solution was 3.0 meq/liter, clay dispersion increased and HC decreased only if ESP >12%. Conversely, when salt concentration was maintained at ~ 0.5 meq/liter, clay dispersion increased and HC decreased when ESP >12%. These results indicate that ESP itself can not determine soil dispersibility, and salt

PAGE 50

39 concentrations in solution have to be taken into account. A simple principle behind this is colloids flocculate when the critical flocculation concentration of a salt is reached. The higher the ESP of the clay and the lower the salt concentration in solution, the higher the tendency for a soil to disperse, and consequently the HC decreases. Further, they concluded the response of soils to low ESP and leaching with low electrolyte water depends on the concentration of electrolytes in the soil solution that the solid phases of each soil maintains (Shainberg et al., 1981b). They demonstrated salt concentration in solution was determined by the dissolution rate of soil minerals, and higher mineral dissolution rate resulted in less impact of exchangeable Na on soil dispersion. Obviously, soil dispersion results from equilibrium of colloid deposition and release. However, it is clear that the interaction between cations on exchange sites and bulk solution significantly impacts colloid mobilization. There is sufficient evidence exchangeable Na has significant influence on colloid release from soil when it is exposed to an influent with low ionic strength. Cummins and Kelley (1923) first demonstrated the hydrolysis of exchangeable Na + in soil, or the replacement of exchangeable Na + by H + from the dissociation of water. They found, in the absence of CaC03 and CO2, a Na-saturated soil leached with distilled water yields a NaOH solution. In an experiment with Na-montmorillonite, where the hydrolysis products are removed continuously, Bar-on and Shainberg (1970) found aNa + concentration of 0.1 M in the effluent. Shainberg (1973) demonstrated Namontmorillonite releases Na + even when the reaction products are not removed. He found the specific conductance of the clay suspension is proportional to the square root of time. These observations are consistent with a hydrolysis mechanism consisting of two

PAGE 51

40 consecutive reactions: a rapid exchange between exchangeable Na + and H + , which results in an acidic surface; a slower, first-order transformation of H clay to Mg or Al clay, which increases the amount of exchangeable Na + release. In soils, similar phenomena have been observed. Oster and Shainberg (1979) demonstrated that, washing three aridzone soils with distilled water, the release of electrolytes can be related to the square root of time exhibiting two linear rates in time sequence. They concluded that the first rate, the more rapid of the two, which occurred right after the washing (<1~2 h), depends on exchangeable Na. Increasing ESP causes increases in the release rate in electrolytes. In general, the flux of each ion species i across the boundary of the EDL around a particle is governed by the convective-diffusion equation (Van der ven, 1989): Nj = -zi ui F ci Vi|/ -DjVcj +Cj v (2-7) Where Nj , Viy, Vcj and v are the flux of ion species i, electric field acting on i, concentration gradient of ion species i from bulk solution to interface and the bulk velocity of the fluid motion, respectively. They are all vector quantities. Zj F, Uj , Cj and Dj are the charge per mole, mobility, diffusion coefficient, and bulk concentration, respectively, of ion species i. Once a colloidal particle starts to detach from an electrified surface in a given system, the electric field acting on i in the surrounding EDL will change (d (Vi|/)/dt) in the course of colloid detachment, resulting in changes in Nj because Vcj and v do not change much during colloid detachment under stagnant flow during a very short period. With this in mind, the experimental observations we just discussed above can be understood conceptually. Once the electric field acting on a colloid decreases during colloid detachment, exchangeable Na is more readily able to diffuse to bulk solution compared to Ca 2+ , Mg 2+ , K + and etc. This has been confirmed by

PAGE 52

41 the fact that electrolyte concentration in solution increases much faster at higher ESP for a soil (Oster et al., 1979). In essence, this is a process of EDL expansion, which results in increases in repulsive forces between soil colloids or colloids and porous media, and thus mobilization of soil colloids. Therefore, it is easy to understand the observation that colloid mobility are proportional to ESP (Kaplan, 1996) as well as total exchangeable bases (Seta et al., 1997). This is consistent with the study by Roy et al. (1996), which is based on cation exchange reactions. Ion transfer during soil colloid release in our research We have examined how colloid mobility in a Pb-contaminated soil, collected from Montreal, Canada, is affected by water-flooding incubation (Chapter 3). The soil packed in a series of short columns was incubated for approximately 3, 20, and 80 d, respectively. After the standing water on the top of soil columns was removed, 0.01 M CaCh solution was pumped through the soil columns until the soil was saturated with CaCl 2 . This is designed to eliminate the artifacts resulting from column packing and to use Ca as an index for ion transfer during colloid release. The influent was then switched from CaCl 2 to deionized distilled water (DDW) and effluent of —11.5 mL was continuously collected using a fraction collector. Total and dissolved metal concentrations were analyzed. The relation among colloidal Al, Fe, dissolved Ca, and pH with pore volumes of the effluent is presented in Figure 3-2 to demonstrate the proposed mechanisms for Ca release. After saturating the soil columns with 0.01 CaCl 2 solution, Ca was the dominant cation in both bulk solution and electrical double layers surrounding colloidal particles or stationary solid phases in the soil. When the effluent is switched from 0.01 M CaCl 2

PAGE 53

42 solution to DDW, the release of Ca from the soil is determined by various mechanisms, which most likely comes to play sequentially. Firstly, at the beginning stage of switching to DDW, Ca release is from the bulk solution, which is confirmed by the fact that Ca concentrations and pH in the effluent at pore volume =1 are approximately those of the influent ([Ca]= 0.01 M, pH=6.95) (Figure 3-3). Secondly, Ca diffusion from EDL to the bulk solution driven by a concentration gradient occurred from two to 17 pore volumes. When Ca in the bulk solution was depleted, Ca cation moved against electrostatic attraction away from the surfaces, extending EDL surrounding colloids and solid phases to mobilize colloids. Obviously, any significant expending of EDL and resulting colloidal mobilization have to be associated with significant amount of Ca released into bulk solution. In this case, the excess Ca 2+ in the diluted bulk solution may be combined with hydroxyl to release proton (Ca 2+ + H 2 0 <-> CaOH" + H + ) (Oster et al, 1979), resulting in a decrease in pH, which is consistent with our data (Figure 5-3) and consistent with the mechanism we proposed above. Generally, Ca hydrolysis is weak in solution and , thus, not likely to bring the pH down as much as one unit as shown in Figure 5-3. In a heterogeneous system such as this one, however, deficiency of negative charge existed in the bulk solution, hence it is possible that the charge deficiency may further drive Ca hydrolysis. Thirdly, Ca desorption from colloid surfaces and stationary solid phases occurred from 18 to 23 pore volumes (Figure 5-3). This is a slow process and, therefore, occurred at high pore volumes, resulting in an increase in repulsive forces between colloids and stationary solid phases. Increases in Ca concentration and pH were observed in the effluent in pore volumes 17-23 (Figure 5-3). In essence, this process is equivalent to sorbed Ca displaced by H + . However, a direct validation is needed.

PAGE 54

43 Nevertheless, this could be a mechanistic illustration for ion transfer between EDLs and bulk solution as described by Eq (2-7). Significant ion transfer (Ca and OH') during colloid release provides a more promising approach to evaluate colloid mobility in a soil since modern colloid theory has found serious difficulty in soils because of its heterogeneity in both structure and composition of soil particle surfaces. Association of Colloids with Heavy Metals There are two major issues involved here: sorption capacity of heavy metals to soil colloidal particles, which has been well documented (Hayes et al., 1990; Dzombak, et al., 1990) and will be briefly described in the following section, and influences of metal sorption on colloid mobility. The association of colloids with heavy metals can be classified into adsorption, precipitation (surface precipitation), and ion exchange. Their influences on surface charge are possibly by virtue of heavy metal ions functioning as specific adsorbing, potential determining, and indifferent ions, respectively. Since most heavy metals are low in solubility in soils and their concentrations are much lower than those of electrolytes in soil solution, it is rare for them to function as indifferent ions. The same reasoning may also apply to the role heavy metal ions play as potential determining ions. However, it has been realized that adsorption is one of the most important mechanism of heavy metal association with colloids (Hayes et al., 1990) and colloid-facilitated heavy metal transport (Kretzschmar et al., 1997), which will be discussed next. In soils, colloids are heterogeneous, including weathered minerals (clay, metal oxides, etc.), CaC0 3 , silica, large organic molecules and cellular debris. However, they

PAGE 55

44 may be classified based on surface functional groups: surface hydroxyl, carbonyl, organic complexation group, etc. As an example, we will discuss hydroxyl surfaces in detail. Adsorption of heavy metals to colloids with surface hydroxyl Surface-hydroxyl-bearing minerals are abundant in soil, such as metal oxides, quartz, and the edge of kaolinite. Similar with the expression for the deand protonation processes described previously, surface complexation reactions of surface hydroxyl with heavy metals can be written as follows (Stumm et al., 1987): SOH + M z+ SOM^ 0 + H + Ki. apP (2-8) 2 SOH + M z+ = (SO) 2 M (z 2) + 2 H + K 2 , app (2-9) Where SOH denotes a surface site and M z+ represents a metal cation. Ki ; app and K 2; apP are the apparent surface equilibrium constants and defined as: K,, app = [SOM^tH+MSOHHM 2 *] (2-10) K 2 , app = [(SO) 2 M (z 2) ][H + ] 2 /[SOH] 2 [M z+ ] (2-1 1) where [] indicates concentrations. Because of the charge characteristic of solid surfaces, ion activity at the surface needs to be corrected to obtain the intrinsic equilibrium constants. Accordingly, the intrinsic equilibrium constants can be expressed as: K U nt= K,,app exp[(z-l)Fv|//RT] (2-12) K 2 , in ,= K 2 . app exp[(z-2)Fv|//RT] (2-13) where \\i is the potential difference between the binding site and bulk solution. The exponential term accounts for the coulombic contribution to the intrinsic equilibrium constants. Because the surface potential cannot be determined experimentally, it is generally formulated based on a variety of models, such as constant capacitance, diffuse layer and triple layer models. It has been found that these models are equivalent in

PAGE 56

45 predicting metal adsorption on surfaces. Applying mass and charge conservation laws to titration data and correcting for the coulombic effect from the charged surface, the intrinsic equilibrium constant for metal ions binding to surface can be determined; however, it is somewhat model-dependent. It has been found stability constants of surface complexes correlate with those of the hydroxo complexes in aqueous phases (Schindler et al., 1987). This could be supporting evidence for Eqs (2-8) to (2-12), which assume a surface complex is an analogous to aqueous complex and that Gibbs free energy of sorption is the sum of intrinsic and coulombic terms. Recently, Blesa (1990, 1995) pointed out the importance of ion hydrolysis during adsorption, and Sverjensky (1993) has divided the intrinsic terms into two: a solvation contribution and a remaining term. Thus the overall free energy of adsorption of an ion can be written as: AGads = AGu + AGs + AGcoui (21 4) In this equation, the coulombic term (AG C0U i.) represents its contribution to the overall free energy of adsorption owing to the interaction between the ion and surface charge, the solvation term (AGs) accounts for the role of solvation during sorption of the ion, and the remaining term, named an ion-intrinsic term (AGu), is assumed to be a property of the ion alone. With this approach, the experimental results for Pb 2+ , Cd 2+ , Cu 2+ , Mg 2+ and Ca 2+ can be closely reproduced (Sverjensky, 1993). If ligands other than H 2 0 and OH" exist in solution, such as Lewis acids or bases ( often the case in soil), competition may occur among dissolved ligands, heavy metals and surface adsorption sites. The competing anions (e.g., sulfate, phosphate or dissolved organic carbon, DOC) can alter surface sorption, and formation of aqueous metal-DOC

PAGE 57

46 complexes stabilizes the metal in solution and effectively decreases the amount of aqueous metal ions available for adsorption. In addition, anions such as sulfate and phosphate will combine with metal cations, reducing the metal adsorption on solids. On the other hand, metal sorption may be enhanced by the formation of ternary surface complexes via sulfate and phosphate and DOC. At the present time, there is no theory that can describe these complicated processes well. However, in principle, the partition of any aqueous species may be described under the framework of Eq (2M4) (Blesa, 1990; 1995; Sverjensky, 1993). In metal oxide systems, heavy metal ions interact with surfaces as specific adsorbed ions (Lyklema, 1984; Ardizzone, 1982) to alter charge development, which is similar with the situation described previously. Effects of heavy metal adsorption on the charges of soil colloids A number of studies have indicated the importance of particle-solution interactions in determining surface properties of colloids in natural waters. Electrokinetic measurements made on natural particles dispersed in sea water and natural fresh waters have shown that their electrophoretic mobilities tend to fall within very limited ranges of negative values. The apparent uniformity of this surface electrical property has been attributed to the adsorption of organic materials, particularly humic compounds, at the particle surfaces, which has been supported by a number of studies using synthetic systems. It has been shown that the effects of adsorbed organic compounds on surface properties of minerals present in natural waters are modified, to some extent, by additional cation interactions. In this sense, the interaction of heavy metals with the surfaces of soil colloids cannot be described by the interactions between heavy metals

PAGE 58

and pure minerals, rather by those between heavy metal and the complexed surfaces of "real" soil colloids. This complication includes the irregularity of colloid shapes and three-dimensional interfaces between solution and colloids. One of most important examples is organic-coated oxides. It has been proposed as a model: individual colloid particles mostly consist of inner mineral cores with sorbed Fe, Al & Mn (hydroxy )oxides and/or organic matter (OM) (Day et al, 1994), which generally possesses similar composition with soil colloids as well as three dimensional interfaces. There has been a fair amount of research done dealing with how ionic strength or pH affects the mobility of the colloids with the constituent components above (Day et al, 1994). However, little is known about the effects of heavy metals on colloidal charge development and stability of the complex colloids. Only recently, Kretzschmar et al. (1997), have explicitly taken into account the effects of adsorption of Pb 2+ and Cu 2+ onto humic-coated oxide colloids on colloid charge development and transportability. They investigated the influence of adsorbed heavy metals (Cu 2+ and Pb 2+ ) on the transport and deposition kinetics of submicron size hematite, coated with humic acid suspended in Ca 2+ solution, through a natural soil. They found that replacement of Ca 2+ by Pb 2+ while holding the total concentration (Ca 2+ + Pb 2+ ) fixed resulted in a slight decrease in electrophoretic mobility of humic-coated hematite colloids. When Ca 2+ was completely replaced by Pb 2+ , the suspensions were destabilized and aggregated within 20 h. In contrast, replacement of Ca 2+ by Cu 2+ had very little effect on electrophoretic mobility and colloidal stability. Both Pb 2+ and Cu 2+ are known to bind much more strongly to humic substances and Fe oxide surfaces than Ca 2+ . At the pH of the suspensions investigated (pH 5.7), Cu 2+ has a higher affinity for humic substances than Pb 2+ , but Pb 2+

PAGE 59

48 adsorbs more strongly to hematite than Cu + . In fact, the interface is three-dimensional since the humic coating can be considered an extension of hematite particles to the solution phase. The bonding difference may direct Pb 2+ and Cu 2+ to locate preferentially at Stern and diffuse layers, respectively. Therefore, it is expected that the effect is more pronounced by replacing Ca 2+ with Cu 2+ than with Pb 2+ in the colloid suspensions on electrophoretic mobility and colloid deposition. Their results suggested that humiccoated oxide colloids can be stable and mobile in the presence of strongly adsorbing trace metals. Partitioning of heavy metals in soil colloids The ability of colloids to facilitate heavy metal transport is largely determined by the distribution of heavy metals among solution, mobile colloids, and stationary grains. Mills et al. (1991) first incorporated partition coefficients into their model in describing colloid-facilitated metal transport in porous media. However, a mechanistic understanding of metal distribution among different phases is still missing. One of the major reasons is the dynamic feature of the distribution of heavy metals, and phase transformations among them, such as from mobile to immobile colloids, once heavy metals are introduced to a system. Kalplan et al. (1995) have found that the mineralogy of mobile colloids from contaminated wells is significantly different from those from the contaminated soil on one site, suggesting the association with heavy metals may alter greatly the mobility of original mobile colloids. Amrhein et al. (1993) examined the potential of colloid-facilitated heavy metals in roadside soils receiving deicing salts. There are two deicing agents considered, NaCl and calcium magnesium acetate (CMA). In their experiments, a series of 60-mL syringes

PAGE 60

were packed with 30 g of soil. The columns were leached on a centurion vacuum extractor with three consecutive 30-mL aliquots of either 0.1 mol L" 1 NaCl or CMA. After initial leaching with either of the salts, the columns were leached with three consecutive 30-mL aliquots of deionized water. This is to simulate the input of salty runoff water to the roadside soil, followed by snowmelt or rainfall. A portion of each leachate was saved without any further filtration and the remainder of each solution was filtered through a 0.45 um membrane filter. Then the filtrates were immediately placed into a stirred ultrafiltration cell with a membrane of 1000 MWCO. This procedure separates the particles into three fractions: >0.45 um, between 1.0 nm and 0.45 um and < 1.0 nm. The results have shown that the cumulative leached metals, through the whole process, vary with the initial salt input. In general, NaCl tends to mobilize soil colloids more than CMA ( Amrhein et al., 1993). On the other hand, because of the possibility of artifacts resulting from column packing in their study, it may be more reasonable to examine the leached cumulative metals excluding the first several pore volumes. Figure 2-2 is a plot of leached cumulative metals after switching the influent from salt solution to deionized water. It is well established that reduction of ionic strength may mobilize soil colloids. As shown in Figure 2-2, the concentrations of each metal (Cu, Pb, Fe, Ni

PAGE 61

50 100000 f CMA NaCI CMA NaCI CMA NaCI > 450 nm 1.0450 nm < 1 .0 nm Figure 2-2. Cumulative concentrations of metals leached from the soil in the three size fractions after switching salt solution to deionized water. Size separation was by 0.45 (am membrane filter and 1-nm cellulose ultrafiltration membrane (adapted from Amrhein et al., 1993)

PAGE 62

51 and Cr) are consistently higher in the leachates from soil columns pre-leached with NaCl than that with CMA in all three size fractions. The fraction <1 nm will hereafter not be emphasized since it is generally considered as dissolved. Among those size fractions, colloids ranged from 1 -450 nm is most important for colloid transport (Kaplan et al., 1995), and their concentrations of each metal are much higher for the columns preleached with NaCl than those with CMA. This is consistent with normal intuition, that is, Na promotes colloid mobility, and so does colloid-facilitated metal transport if one assumes association of metals with mobile colloids are not significantly affected by changes in solution chemistry. Furthermore, metals in fraction 1 .0 -450 nm are much higher than those in fractions > 450 nm for columns pre-leached with CMA, whereas there are no significant differences for the column with NaCl except for Fe and Cu. This suggests CMA is more selective in mobilizing smaller colloidal particle than NaCl. Along with this fact, the authors pointed out that colloidal Cu did not correlate well with DOC between the two fractions for the columns pre-leached with NaCl, nor does colloidal Fe with other colloidal metals pre-leached with CMA (Figure 2-2). This suggests that heavy-metal-bearing mobile colloids vary with both size and solution chemistry. In other words, phase transformation among heavy-metal-bearing colloids, such as mobile to immobile colloids or among mobile colloids with various sizes., has to be taken into consideration. We have examined the effects of water flooding incubation on colloidal Pb mobility in soil columns. Detailed descriptions about the experiment procedure can be found in Chapter 5. As shown in Figure 5-2, colloidal Pb concentrations varied with incubation. Interestingly, there was no significant difference in concentrations of

PAGE 63

52 colloidal Fe and Al in the first several pore volumes when the columns were incubated from 3 to 20 d, while concentrations of colloidal Pb decreased approximately 4 times after 20-d than 3-d incubation. This suggested that Pb-bearing colloids tended to be immobilized or the tendency of association of Pb with mobile colloids decreased with water-flooded incubation. Concluding Remarks Soil has distinctively different features in colloid mobility compared with welldefined porous media or subsurface systems. At the present time, it is not realistic to predict colloid mobility and colloid-facilitated heavy metal mobility in soil with certainty. Soil colloids are exposed to solution where they take part in chemical processes. Therefore it is difficult to predict charge development of soil colloids since it is determined by the interaction of surfaces and solution. Point of zero charge (PZC) of mineral particles suspended in simple solution can be theoretically predicted, while little success has been achieved in soil. In general, soil particles are assemblages of crystalline and amorphous minerals, and organic residuals; for such particles sophisticated models such as triple layer model do not make much sense in describing the electrified interfaces since particle surfaces are irregular and the Stern layer is often inside of the physical boundary of the particles. As a first approximation, however, It may be more pratical to describ soil particle interfaces by diffuse double layer with a specific part (a charged surface), and a generic part (a diffuse layer). The surface charge development process may be described by the partitioning of charged species between solution and surfaces assuming the charged species are similar between the two phases. Similarly, the association of heavy metals with colloids may have significant influences on colloid

PAGE 64

53 mobility, which is subject to changes in dynamic solution chemistry. However, very limited information is available at the present time. Colloids are ubiquitous in soils. Different from clean porous media, the surfaces of soil stationary phases consist of deposited colloids. Consequently, the blocking effect of colloid deposition in soil is obvious, which may greatly enhance colloid mobility in soil compared to that in a comparable clean porous medium. Under such a condition, size straining is a dominant mechanism in soil colloid deposition. The deposited colloids may release depending on the nature of interactions with the stationary phase, that is, the interaction energy profiles (Figure 2-1), which vary because of heterogeneity of the surfaces. Phenomenally, this process is time-dependent or condition-dependent, and therefore, is referred to as transient phenomena in colloid deposition. At the present time, water dispersible clay extracted from a soil is widely used to indicate colloid mobility. However, it has to be realized that in both cases there are two simultaneous, opposite processes: colloid deposition and release. In a given system, the relative effect of each on either soil clay dispersion in batch tests or colloid mobility in a column experiments depends on mobile colloid residence time in the systems. Despite the shortcomings of the DLVO (Derjaguin-Landau-Verwey-Overbeek) theory, we believe it provides a basic theoretical framework in describing soil colloid dispersion and colloid deposition and release, which is simple and applicable in concept. It has to be pointed out that all modern colloid theories (not only the DLVO theory) are difficult to apply to soil because of its heterogeneity in both structure and composition of soil particle surfaces. In order to determine the potential mobility of soil colloids, a phenomenological approach may be more practical. Soil colloids possess high

PAGE 65

54 concentrations of exchangeable ions in their EDLs (high CEC), which does not always equilibrate with those in bulk solution. It has long been recognized that colloid release in soil is coupled by significant ion transfer between EDLs and bulk solution. By evaluating the possibility of the ion transfer, potential colloid mobility may be estimated qualitatively.

PAGE 66

CHAPTER 3 RELATION OF PB SOLUBILITY TO FE PARTITIONING IN SOILS Introduction Lead contamination in soils is of environmental significance due to its toxicity to both humans and animals (Ma et al., 1995). In general, Pb mobility is low because of its low solubility. Lead solubility may be further reduced as a result of its interactions with soil solid phase via sorption and ion exchange. However, enhanced Pb solubility has been found under both laboratory and field conditions (Amrhein, et al., 1994; Glazovskaya, 1994). As such, Pb may migrate through a soil profile to contaminate groundwater. Although much effort has been spent to model heavy metal solubility (Cederberg et al., 1985; Sposito, 1984), such prediction under field conditions involves large uncertainty. It is partially because of the difficulty in assessing the effects of dynamic soil solution chemistry on heavy metal speciation. However, changes in solution chemistry, such as pH, redox potential and ionic strength, may shift Pb retention processes significantly. These impacts may be further complicated by aqueous Pb competition with other cations for ligands, which may enhance Pb mobility under certain conditions (Amrhein, et al., 1994). In natural aquatic environments, Pb (II) and Fe (II) have similar affinity to complex with ligands, and thus they show similar patterns of species distribution (Turner et al., 1981). Theoretically, both Pb (II) and Fe (II) are in " transient class" in terms of 55

PAGE 67

56 metal classification based on electronegativity and covalent index (Nieboer and Richardson, 1980). This evidence implies that it is important to examine Fe chemistry when studying Pb solubility. In fact, good correlation (r 2 = 0.71-0.91, p<0.05) between concentrations of aqueous Pb and Fe in sediments has been reported (Lee et al., 1997; Routh and Ikramuddin, 1996). In soil environment, data published by Karczewska (1996) showed that concentrations of mobile Fe and Pb were related (r 2 = 0.54, p<0.05) in several polluted soils near a copper smelter. The theoretical evidence along with the published experimental data suggests that Pb and Fe solubility may be related in a soil. In addition, there are two other possible factors for correlation between Pb and Fe in solution. First, Fe (oxy) hydroxides are good sorbents for aqueous Pb (Ainsworth et al., 1994; Laxen, 1985; Benjamin and Leckie, 1981), and their dissolution or precipitation may release or sorb Pb. In soils, dissolution of Fe (oxy) hydroxides is generally promoted by reducing Fe (III) to Fe (II), which is sensitive to soil redox status (Lindsay, 1979; Gotoh and Patrick, 1974). Under oxidizing condition, Fe (oxy) hydroxides tend to immobilize Pb (Gambrell et al., 1980); whereas under reducing condition they dissolve to release Pb (Gambrell, 1994). Secondly, soil organic carbon, especially dissolved organic C (DOC), is an important factor controlling Pb and Fe solubility in soil (Dorr and Munnich, 1991; Davis. 1984; Laxen, 1985). DOC functions as a ligand to complex with Pb and Fe to increase their solubility (Davis and Leckie, 1978). On the other hand, DOC may also be sorbed by Fe (oxy) hydroxides under certain conditions, via ligand exchange or hydrophobic interaction in soils (Murphy and Zachara, 1995), thus decreasing Pb and Fe solubility.

PAGE 68

57 Despite all the circumstantial evidence that Fe solubility impacts Pb solubility in soil, there has been little effort to quantify the relationship between the two. However, understanding the relationship between heavy metal solubility and the behavior of Fe may provide important information for assessing potential mobility of heavy metals in soil environments, which cannot be modeled successfully at present time. Soil redox status varies temporally. The intensity of redox status can be described by redox potential in general. In surface soil it is influenced by rainfall, bioactivity, and changes in land use, whereas in vadose zone by fluctuation in water table (Buol et al., 1997). The redox potentials of sediments that remain saturated can be measured readily, whereas characterization of the redox status in a soil is still a challenge. As mentioned previously, however, soil redox status affects metal solubility greatly under certain conditions. To evaluate the effects of varying redox status, saturated/unsaturated (aerobic/ anaerobic) incubation techniques have recently been developed (Amrhein et al., 1994; Karczewska, 1996). Waterflooded incubation of soil was used to effectively decrease soil redox potential to study the speciation and fate of heavy metals in contaminated soils (Karczewska, 1996). Even though saturation rarely happens in surface soils, some valuable information may be obtained from such a study. In this paper, two redox statuses in soils will be obtained under water-flooding (anaerobic) and no flooding (aerobic) incubations. The major objective of this chapter is to examine the relationship between Pb solubility and Fe partitioning in soils with different solution chemistry. Throughout this paper, Fe partition will be expressed as a concentration ratio of aqueous to sorbed Fe (II)

PAGE 69

58 unless otherwise specified. The results showed that Pb solubility was related to this ratio in the soil. Materials and Methods Location and characteristics of soil sample The soil samples used for this study were collected in March 1996 from a research site in Hawthorne, Florida. The soil is an acidic fine sand (typic quartzipsamment) with a spodic horizon below 2 m. The samples were collected from 2-20 cm below the surface after removing organic residue. The samples were air-dried, sieved through 2-mm screen, and stored at 4°C prior to use. Some characteristics of the soil are listed in Table 3-1. Column experiment A series of 60-mL syringe (columns, d=2.6 cm), which were prepacked with a layer (~ 3 mm) of acidwashed Ottawa sand (20-30 mesh) in the bottom, were packed with 40 g of soil sample. The columns were then packed with another layer of the acidwashed sand on the top to minimize disturbance on the soil from influents. All the syringes were set on a Centurion Vacuum Extractor (Centurion International, Lincoln, NE) for incubation and leaching. The experiment procedure was as follows: Prewetting soil columns . 30 mL of deionized distilled water (DDW) was slowly poured into each column and the soil columns were saturated for 48 h before the water was extracted.

PAGE 70

59 Table 3-1. Characteristics of the Florida soil used pH Organic C mg g"' Extractable Fe* mg kg" 1 Total concentrations** Texture*** Al Fe Ca Na Pb Sand Silt Clay mg kg" 1 % 5.4 6.0 810 2067 976 87 27 1.3 94.6 4.5 0.9 *Heron, etal., 1994. **Maetal., 1996. ***Day, 1965.

PAGE 71

60 Addition of aqueous Pb . 30 mL of 0.48 mM Pb(N0 3 ) 2 was added to 12 soil columns and 30 mL of 3.68 mM Pb(N0 3 ) 2 to 16 soil columns. The solutions were extracted after 24 h at 2.5 mL h' 1 . This resulted in lead-loading rates of 0.36 (low Pb loading) and 2.90 mmoles kg' 1 (high Pb loading) in two groups of soil columns. Addition of electrolyte solutions. 30 mL of different electrolyte solutions was added to different columns to vary solution chemistry. DDW of pH 5.5 and 2.17 and 4.35 mM NaCl solution were added to the first 12 soil columns, and 0.52, 1.08, and 2.17 mM NaCl solutions, 0.25, 0.55, and 1.10 mM CaCl 2 solutions, and DDW of pH 4.5 and 5.5 to the other 16 soil columns. The solution was extracted out after 24 h at the rte of 2.5 mL h . For the first 12 soil columns with low Pb loading, each treatment was replicated four times (three treatments) and for the 16 soil columns with high Pb loading, each treatment was duplicated (six treatments). Soil columns in each treatment (a total of nine treatments) varied not only in Pb concentrations but also in pH and composition of incubation solutions. Incubation of soil columns. The above 28 soil columns (with the same Pb loading rate and electrolyte solution) were further divided into two subgroups (14 soil columns each) for incubation. One subgroup was filled with 30 mL of DDW (2 cm above the soil surface) for water-flooded incubation. The other subgroup was incubated as it is for nonwater-flooded incubation. All soil columns were incubated for 40 d at room temperature. Leaching . DDW was added to each soil column to make 30 mL of standing water above the soil for all columns before leaching. The leaching was conducted at 60 mL h" 1 and approximately 25-50 mL leachates were collected. The soil samples with moisture content ~ 25% in the columns were sealed and stored in refrigerator for further analysis.

PAGE 72

61 Sample separation and analysis Separation and analysis were conducted for both leachates and leached soil samples to examine the relationship between Pb solubility and Fe partitioning in the soil. The pH was measured immediately after the leachates were collected. The leachates were then filtered through 0.22-um membrane filters. The filtrate was analyzed for DOC and total Pb and Fe concentrations. The frozen soil columns were immersed in warm water to push the soil out of the syringes with minimum disturbance. The intact soils from the columns were then divided into three equal sections (top, middle and bottom) for further analysis. The data presented in this paper for soil, however, was the average over the three sections since there were no significant differences in Fe concentrations in the three sections. Fe(II) content in the leached soil was analyzed following the same method described in Chapter 3 Result and Discussion Pb solubility Aqueous Pb concentrations in leachates showed strong pH-dependence with 0.92 for soils with Pb loading of 2.90 mM kg" 1 (high Pb loading, Figure 3-1 A). Aqueous Pb concentrations decreased as pH increased from 4 to 6. It has long been recognized that pH is an important factor affecting metal solubility, with aqueous metal concentration increasing as pH decreases (Chuan et al., 1996). Leachate Pb concentrations in soils with Pb loading of 0.36 mM kg" 1 (low Pb loading) were also correlated with pH with correlation coefficient of r 2 = 0.51 (p<0.05). However, they

PAGE 73

62 700600500 4003003020ioH o i A Pb loading of2.90 mmole kg ^ water-flooded — non-water-flooded 3.5 4 4.5 5 5.5 6 6.5 Leachate pH 0 12 3 4 5 6 7 Fe concentration in leachates (uM) 4 Pb loading of0.36 mmolekg water-flooded non-water-flooded 1.5 2 2.5 400 350 300 -250 -200 150 100 50 0 3.5 K DOC concentration in leachate (mg/L' ) 400 I 1 1 1 I 10 12 14 16 Ratio of aqueous Fe to sorbed Fe (II) concentrations (kgL" 1 ) Figure 3-1 Relation between Pb solubility and leachate pH, dissolved organic carbon (DOC), leachate Fe concentration, and ratio of aqueous Fe to sorbed Fe(II) concentrations in a sandy soil. Data from 2.90 mmole kg' 1 Pb loading rate read from left y-axis and data from 0.36 mmole kg" 1 Pb loading rate read from right y-axis.

PAGE 74

63 increased as pH increased from 6.1 to 6.8 (Figure 3-1A). Obviously, mechanism's that determined aqueous Pb concentrations in soils of low Pb loading was different from that of high Pb loading. It seemed that DOC was a primary factor in determining Pb concentrations in soils with low Pb loading as reflected by their correlation coefficient of r 2 = 0.47, p<0.05 (Figure 3-1 B), i.e. Pb concentrations increased as DOC concentrations increased. However, such correlation did not exist for soils with high Pb loading (r= 0.24) possibly due to the fact that Pb solubility was more strongly correlated to pH. Beside pH, redox status is another factor affecting metal solubility in soil (Chuan et al., 1996). Stumm (1984) suggested that it is reasonable to measure important redox species instead of redox potential to indicate redox status in natural systems. We used aqueous Fe as a measure of redox status for soils. As expected, aqueous Fe concentrations were much higher in soils under water-flooded incubation than non-waterflooded incubation (Figure 3-1 C). In soils with low Pb loading, aqueous Fe concentrations was over three times greater under water-flooded incubation (averaged 2.12 nM) than under nonwater-flooded incubation (averaged 0.57 nM). Similarly, the average aqueous Fe concentrations in soils with high Pb loading were 4.46 and 1.74 uM, respectively. Greater aqueous Fe concentrations in soils with high Pb loading than with low Pb loading was mainly due to pH difference in the soils (Figure 3-1 A). However, there was no relationship shown between aqueous concentrations of Fe and Pb for soils of either Pb loading rates (Figure 3-1C). It was reported that aqueous Pb and Fe concentrations are linearly related due to the fact that heavy metals are released from Fe (oxy)hydroxide surface when it dissolves (Chuan et al., 1996). In our study, however, this was not the case (Figure 3-1C), suggesting that aqueous Pb may not be controlled by

PAGE 75

64 Fe (oxy)hydroxide surface alone. Similar results have also been reported by other researchers (Gambrell et al., 1991). These data suggested that, in general, aqueous Fe concentration alone could not describe Pb solubility well. Nevertheless, as expected, aqueous concentrations in soils under reduced conditions were much higher than those under less reduced conditions. Aqueous Pb concentrations in soils with high Pb loading were 52 times greater under water-flooded incubation (averaged 490 uM) than under non-water-flooded incubation (averaged 9.4 uM) (Figure 3-1). Similarly, aqueous Pb concentrations were 91 and 22 nM, respectively, for soils with low Pb loading. Pb solubility and Fe partitioning As discussed previously, pH was related to Pb solubility in our study. However, even for the same soils, the relation between pH and aqueous Pb differed with Pb loading rates (Figure 3-1 A). In other words, there is no simple relation between pH and Pb solubility when soil solution chemistry varies greatly. There were no simple relationships between Pb and DOC concentrations and Pb and Fe concentrations either (Figure 31A,C) To better describe Pb solubility, we defined the ratio of aqueous Fe to sorbed Fe (II) concentrations as a Fe partitioning index. In this study, sorbed Fe (II) was operationally defined as Fe(II) that was extracted by 0.5 M HC1 (Lovely and Philip, 1987; Heron et al., 1994). By definition, Fe partitioning index represents the relative affinity of aqueous Fe to solid phases in a soil. A larger number indicates a lower Fe affinity to solid phases. There was no simple relationship that existed between aqueous Pb concentrations and Fe partitioning index (Figure 3ID). For soils with low Pb loading

PAGE 76

65 and incubated under water-flooded condition, the relation between aqueous Pb and Fe partitioning index was approximately parabolic. In general, aqueous Pb concentrations increased as the index decreased. However; as the index decreased to below 2 kg L" , increases in Pb concentrations were substantial (Figure 3ID). A similar trend was also observed for soils with high Pb loading and incubated under non-water-flooded condition, i.e. substantial increase in Pb concentrations were observed when the index was < 2 kg L' 1 . However, no enhanced Pb solubility was found in soils with low Pb loading and incubated under non-water-flooded condition since the Fe index was > 2 kg L" 1 . However, for soils with high Pb loading rate and incubated under water-flooded condition, extremely high Pb concentrations were observed since the Fe index was <2 kg L . Obviously, compared to pH, DOC and Fe concentrations, a more consistent relation between aqueous Pb concentrations and the Fe partitioning index in soils was observed. For a pH-dependent process, Kurbatov equation (Kurbatov et al., 1951) has been successfully used to describe the macroscopic behaviors of metal partitioning in natural aquatic systems (Fuller et al., 1996; Tessier et al., 1985, Balistrieri et al., 1983). In our system, the Fe index in soils with both high (r 2 =0.61, p<0.05) and low (r 2 =0.55, p<0.05) Pb loading rates was correlated to pH. Thus, the Kurbatov equation can be applied to our system as follows: log(Fe index) « -xpH log K p + log N s , 3-1 where x is a macroscopic proton coefficient, K p a Kurbatov partition coefficient and N s the total number of exchangeable sites (Fuller et al., 1996, Balistrieri et al, 1983). Assume that the above equation is correct, then the Fe index is related to pH in addition to x, K p and N s . The number of exchangeable sites N s in the equation for a given soil is

PAGE 77

66 generally a constant. On the other hand, % varies with metal species in solution and on surfaces and K p does not vary much unless the adsorption density exceeds its maximum (Balistrieri et al., 1983). Therefore, for a given system, the Fe index changes mainly with pH and xThe relation between pH and the Fe index in the Kurbatov equation is as follows. For soils with high Pb loading, pH was inversely related to the Fe index as expected with r 2 =0.55, i.e. the index decreased as pH increased from 6.1 to 6.8. However, for soils with low Pb loading, pH was positively related to the Fe index with r 2 =0.61, p<0.05 (Figure 3-1A&D). The impact of % on the index is unclear at the present time. As shown in the Kurbatov equation (Eq. 3-1), the Fe index is a lumped parameter that presents the properties of both soil solution (pH and x) and solid (K p and N s ) that determine Fe partitioning. To evaluate the relationship between Fe index and other metal solubility, it is expected that the more Fe-like a metal acts, the more related its solubility is to the Fe index. As mentioned previously, Pb(II) shows some similarity to Fe (II) in both metal classification and speciation distribution in aquatic systems (Turner et al., 1981; Nieboer and Richard, 1 980). Therefore, it is expected that aqueous Pb concentration in soil is related to the Fe index (Figure 3-1 D). Turner et al. (1981) reported that in natural waters Pb (II) and Fe (II) exhibit some similarity, meaning that Fe (II) may be a major competitive cation to Pb (II) for ligands. Therefore, it is important to evaluate the competition between Fe (II) and Pb (II) for ligands. It should be pointed out that, in this chapter, we did not emphasize the competition between Fe (II) and Pb (II) for adsorption sites on solid phases. The reason is that some researches have shown that the competition may be insignificant (Rose and

PAGE 78

67 Bianchi-Moaquera, 1993; Coughlin and Stone, 1995). In our study, soil solution chemistry varied in different soil columns. These variations may shift the competition. At the present time, it is difficult to fully evaluate these effects on metal solubility in the soil, especially when DOC is present. The Fe index presented in this paper, however, may serve as a simple measure of competing ability of Fe with Pb for ligands. At this stage it is more important to look at the conceptual nature of the index than the detailed mechanistic interpretation, which need to be explored further in future. Assuming that Fe and Pb compete for ligands in a soil, then the following can be inferred. As Fe partition index increases, there is more aqueous Fe and/or less sorbed Fe, which means there is more competition for ligands from aqueous Fe in the soil and thus results in less aqueous Pb in the soil. On the other hand, as Fe partition index decreases, there is less aqueous Fe and/or more sorbed Fe, which means there is less competition for ligands from aqueous Fe and thus results in more aqueous Pb in the soil as shown in Figure 3ID. In theory, there should be a minimum for the Fe index, i.e. the ratio of the minimum of aqueous Fe concentration to the maximum sorbed Fe concentrations. If aqueous Fe concentrations approach zero, i.e. all aqueous Fe was transferred onto sorbed form, the Fe index is zero. At this situation, Pb dominates the ligand competition in the solution, so that enhanced Pb solubility would be expected. The overall trend shown in Figure 3ID confirmed this rationale in principle. In reality, however, aqueous Fe concentrations cannot be zero. Therefore, the Fe index must be larger than zero. Theoretically, in a soil the minimum of aqueous Fe concentration is mainly determined by the solubility of Fe minerals and the maximum sorbed Fe concentration is soil-

PAGE 79

68 dependent. Therefore, the Fe index is generally insensitive to solution chemistry in a soil. Since it is impossible to accurately measure the maximum sorbed Fe concentrations in a soil, no attempt was made to evaluate the real value of the minimum Fe index in the soil we used from the definition. However, the minimum Fe index when enhanced Pb solubility occurred was approximately 2 kg L" 1 in the soil studied, which was estimated from the relationship between Pb solubility and Fe index (Figure 3ID). As expected, the number did not vary much with the incubation conditions and Pb loadings (Figure 3ID). It has been noticed that Pb Concentrations in some soil columns still stayed low even when the Fe index was < ~2 kg L" 1 (Figure 3-1 D), which is unclear at the present time. Nevertheless, the significance between Pb solubility and Fe index was that significantly enhanced Pb solubility does not occur unless the Fe index was < ~2 kg L' 1 (Figure 3ID). Relation of Pb solubility and Fe partitioning in published data In this research, sorbed Fe was defined as the Fe (II) fraction extracted by 0.5 M HC1, which is supposed to extract exchangeable, some adsorbed and freshly-precipitated Fe(II) (Heron et al., 1994). In literature, however, such data are not always available. In order to conceptually test the relation between Pb solubility and Fe partitioning index we defined in this paper using available data in the literature, other extractants that may be equivalent to 0.5 M HC1 have to be used. Two published data sets were used. In both data set, the Fe index was calculated using the total sorbed Fe instead of Fe (II) since the quantity of the later is unavailable. However, Fe index calculated from either total Fe or total Fe (II) should be consistent conceptually since Fe (III) is even more competitive compared with Fe (II).

PAGE 80

69 o 03 CD i_ O CL c g I c CD O c o o _Q CL 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 Fe partition index (kg L" 1 ) Figure 3-2. Relationship between Pb concentrations in pore water and the ratio of aqueous and sorbed Fe (Fe partition index) in a contaminated sediment. Data are adapted from Lee etal. (1997)

PAGE 81

70 21 3 -Q CL
PAGE 82

71 The first data set was from Lee et al. (1997) who characterized a sediment contaminated by heavy metals in a retention pond in France. In the study, concentrations of aqueous Pb and Fe in pore water of the sediment were analyzed, and Fe concentrations in the sediment was determined by sequentially extracting into five fractions: exchangeable extracted by 1 M MgCl 2 , bound to carbonate by 1M sodium acetate, bound to amorphous Fe and Mn hydroxides by 0.04M hydroxylamine hydrochloric acid in 25% acetic acid, bound to organic matters and sulfides by 30% H 2 0 2 and 0.02 M nitric acid, and residual by concentrated HNO3 and HCIO4. Assuming that the sum of Fe concentrations in the first two fractions (exchangeable and bound to carbonate) was equivalent to that extracted by 0.5 M HC1, a plot of aqueous Pb concentrations to the Fe index (Fe concentration in pore water/the sum of Fe concentrations of exchangeable and bound to carbonate) is shown in Figure 3-2. A similar trend shown in Figure 3-1D was observed: enhanced Pb solubility occurred only when the Fe index was low. However, low Pb solubility was also observed at low Fe index as did in our data (Figure 3ID). This again suggests that high Pb solubility will not occur unless the Fe index was below a certain value and low Fe index will not necessarily guarantee high Pb solubility. The second data set was taken from Karczewska (1996) who determined concentrations of heavy metals via a sequential extraction in soils polluted by a copper smelter. In the study, heavy metals were fractionated into 7 fractions, i.e., soluble (1M NH4NO3), exchangeable (1M NH 4 OAc), bound in MnO x (1M NH 2 0H-HC1/1M NHjOAc) and etc. We assumed that Pb extracted by 1M NH4NO3 is equivalent to the aqueous Pb concentration and sum of exchangeable Fe and bound to MnO x was equivalent to Fe(II) extracted by 0.5 M HC1. Again, a similar trend shown in Figure 3-1 D was observed

PAGE 83

72 between soluble Pb and Fe index (Figure 3-3), i.e. highest Pb solubility was observed at the lowest Fe partition index and in this particular case the Fe index was zero. Like the data of Lee et al. (1997), the soil samples were collected from various locations near a smelter site in Poland and there were large variations in soil properties among the samples (CEC: 10-84 meq kg' 1 ; organic C: 0.05-1.42 %; clay content: 1-6 %) (Karczewska, 1996). This suggested that the relationship between Pb and Fe partition index may be applicable to field data even with large spatial variability in soil properties. Implication of this Research Partition coefficients (kd) are important parameters in assessing the potential impacts from metal contaminated soils. However, kd is not a constant in a dynamic soil environment. It is determined not only by the characteristics of solid phases but also solution chemistry. In a given soil system, variation is mainly determined by metal solubility. There are several speciation models available to describe the interactions of metals and soil components to predict their solubility. Unfortunately, such predictions generally lack certainty primarily because of soil dynamic nature. At the present time, assessing changes in solution chemistry in field condition is difficult if not impossible. This information, however, is critical to estimate the solubility of heavy metals in speciation models. In contrast, the approach presented in this chapter provides a simple relation between the probability of enhanced Pb solubility (or k d ) and Fe partition index, which is generally not sensitive to changes in solution chemistry. In this approach, we simply determine the minimum Fe index in a field condition to predict the possibility of enhanced Pb mobility (the lower k d ). Of course, more work is heeded to

PAGE 84

73 further test this concept in different soils and determine the ranges of Fe index for enhanced Pb mobility. Conclusion Our data using a synthetically contaminated soil demonstrated that enhanced Pb solubility was only observed at low Fe partition index. Such a trend was also observed using two published data sets in the literature, thus further validating the relationship conceptually. Since data used in this paper were from various soil environments (roadside sediment in France, Pb contaminated soil in Poland, a sandy soil from Florida), it is reasonable to conclude that Pb solubility may be indeed related to Fe partition index in soils. This concept may also be applied to solubility of other heavy metals such as Cd, Cu, Ni and Zn.

PAGE 85

CHAPTER 4 HEAVY METAL MOBILITY IN CONTAMINATED SOILS: PART 1. ROLE OF EXCHANGE SITES IN CONTROLLING SOLUBILITY AND MOBILITY OF HEAVY METALS Introduction Heavy metal mobility in soils is of environmental significance due to its toxicity to both humans and animals (Ma et al., 1995). At a constant water flow condition in a soil, heavy metal mobility is determined by its solubility. It has been well recognized that the solubility of heavy metals in soil is mainly regulated by adsorption, precipitation and ion exchange reactions. Although much effort has been spent to model heavy metal solubility (Cederberg et al., 1985; Sposito, 1984), such a prediction under field conditions contains large uncertainty. It is partially because of the difficulty in assessing the effects of dynamic soil solution chemistry on heavy metal speciation. However, changes in solution chemistry, such as pH, redox potential and ionic strength, may shift the retention processes of heavy metals significantly. These impacts may be further complicated by the competition of cations for ligands, which may enhance heavy metal mobility under certain conditions (Amrhein, et al., 1994). Soil redox status varies temporally and spatially. In a surface soil it is influenced by rainfall, bioactivity, and changes in land use, whereas in vadose zone mostly by fluctuation in water table (Buol et al., 1997). A reduction in redox potential may cause changes in metal oxidation state, formation of new low soluble precipitates, and Fe 74

PAGE 86

75 dissolution resulting in release of metals (Amrhein et al., 1994; Chuan et al., 1996; Masscheleyn et al., 1991). To experimentally evaluate the effects of different redox status on metal solubility, various techniques have been developed: redox-potentialcontrolled batch experiment using a suspension (Chuan et al., 1996; Masscheleyn et al., 1991), saturated/unsaturated (aerobic/ anaerobic) incubation (Amrhein et al., 1994; Karczewska, 1996, Ma et al., 1995) and waterflooded incubation (Karczewska, 1996). However, some contradictory results have been found when the different techniques were employed. In redox-controlled suspension experiments, Chuan et al. (1996) reported that aqueous concentrations of Pb, Cd and Zn increased with Fe(II) as redox potential decreases, suggesting metal sorption onto surfaces of Fe (hydr)oxides is a dominant process in controlling aqueous metal concentrations. On the other hand, in experiments using saturated paste, Amrhein et al. (1994) found that concentrations of Cu and Cd decreased whereas those of Fe(II) increased in solution as the redox potential was decreased, however, no precipitation of any known Cu or Cd mineral can be proved under these conditions. Similar controversial results are found in other studies (Masscheleyn et al., 1991 ; Ma et al., 1995). However, it should be pointed out that among the studies, the samples analyzed for metals were different. They can be classified into two groups: 1) pore water and leachates from soil leached with DDW, and 2) filtrates separated from the suspension and leachates from soil leached with electrolytes. Obviously, metals determined in samples of the first group are water soluble whereas those of the second group include the exchangeable in addition to water soluble in soils. Therefore, a natural question is raised: with incubation, do metal concentrations in solution change proportionally with those in exchangeable phases? To our knowledge,

PAGE 87

76 the role of exchange phase in controlling the solubility and mobility of heavy metals in incubated soils has not been reported. However, this knowledge may improve our understanding of the controversy discussed above. In Chapter 3, Pb solubility was examined in a sandy soil spiked with Pb and incubated for 40 d under water-flooded or non-water-flooded conditions. Solution chemistry in soil columns was adjusted using different concentrations of NaCl and CaCb and deionized water of varying pH before incubation. The results show that Pb solubility in the incubated soil can be related to the ratio of aqueous Fe to 0.5 M HQ extracted Fe (II) from solid phases instead of soluble Fe alone. If we assume 0.5 M HC1 extractable Fe(II) from the solid phases is proportional to exchangeable Fe (II), it is suggested that exchange sites have significant effect on heavy metal solubility in soils. In this paper, we will extend our study from a synthetically contaminated soil to two naturally contaminated soils. In addition to Pb, concentrations of Cu and As will be presented for comparison. The major objective of this paper was to examine the role of exchange sites on the solubility and mobility of Pb, Cu, and As in soils during incubation. Materials and Methods Location and characteristics of soil samples The two Pb-contaminated soils used in this study were from Montreal, Canada and Tampa, Florida, which were exposed to Pb-battery recycling and concurrent smelting operation in the past. Selected characteristics of the soils are listed in Table 4-1.

PAGE 88

E VI — — _o u 15 o H ON ea C3 a 0 o O O o GO o © o o rn m O o o o o o o ri o o n IT) o o CN 5? o x B _C 'w 3 X) c -3 J) 5 n Oh T3 Q oo O a — © oo 1 u it o U zu ri ri oo CN 00 u w u 0JJ o E u o CN X a. o 00 rn

PAGE 89

78 Column Experiment Packing soil columns and Incubation process . 5.0 g acidwashed sand (20-30 mesh) was packed in the bottom of 60-mL columns (13 cm in length x 2.6 cm in' diameter), then 30 g air-dried soil (moisture content = 9.0%) was packed on top of the sand layer. Preliminary data showed that the sand layer helps to prevent the outlet clogging in a column during incubation and leaching ad has no detectable impact on soil colloid mobility. During each step, the columns were gently shaken horizontally for a few minutes to minimize the packing effects. This resulted in 2.6x5.2 and x7.8-cm soil columns for the Montreal and Tampa soils, respectively. The soil columns were set vertically and prewetted by pumping deionized distilled water (DDW) into the bottom of the columns until the water level in the columns was above the soil. These columns were then incubated under water-flooded condition for approximately 3, 20 and 80 d for the Tampa soil and 3, 20, 30, 40, 50, and 60 d for the Montreal soil. Leaching . Once the redox potential monitored had reached to a certain level, the water on top of the soil column was removed and a rubber stopper was put on top to seal the syringes. The syringes were then turned upside down. The lower side of the column was connected to a needle as an inlet for the influent, 0.01 M CaCl 2 . The pumping rate was 1.20 mL min" 1 . The effluent from the top of the column was collected with a fractionation collector. Analysis of leachates . Each fraction of effluents was filtered through 0.22 urn membrane filter, then acidified to pH<2 to yield the soluble metal concentrations. The details can be found in Chapter 3.

PAGE 90

79 Analysis of pore water of soil To analyze metal concentrations in pore water, an incubated but unleached soil column for each incubation time was used after removing the standing water on top. Pore water was separated from the soil by the centrifugation method described by Nkedi-Kizza et al. (1982) and Dao and Levy (1978). Then the pore water was filtered with 0.22-um membrane filter for metal analysis. For detailed procedure of Fe(II) analysis, please see Chapter 3. Result and Discussion Changes in heavy metal solubility with incubation Concentrations of heavy metals and Fe (II) in pore water varied with incubation time (Figures 4-1 and 4-2). Fe behaved differently in the two soils with incubation: in the Montreal soil, Fe(II) decreased with incubation (Figure 4-1) whereas in the Tampa soil it increased(Figure 4-2). Similarly, Pb increased in the Montreal soil and decreased in the Tampa soil with incubation. However, in both soils the concentrations of Fe (II) and Pb were inversely correlated with incubation (Figure 4-1 and 4-2). Concentrations of As and Cu increased first then decreased with incubation in the Tampa soil. In similar experiments by Amrhein et al. (1994) using soils incubated in waterflooding condition, they found that Fe (II) concentration in pore water increases with incubation time. This is consistent with the result we found in the Tampa soil; however it is contrary to that in the Montreal soil. With incubation, redox potential reduced and thus

PAGE 91

80 10 20 30 40 50 60 70 80 Incubation time (d) Figure 4-1. Pb and Fe (II) concentrations in pore water of the Montreal soil.

PAGE 92

81 0 Pb O Cu A As Fe (II) 10 20 30 40 Incubation time (d) 50 60 Figure 4-2 Concentrations of Pb, As, Cu and Fe(II) in pore water of the Tampa soil

PAGE 93

82 dissolved Fe(II) increased due to Fe(III) reduction to Fe(II) (Amrhein et al. , 1994). However the unexpected behavior of Fe (II) in the Montreal soil may be caused by formation of FeC0 3 , which will be discussed in Chapter 5 in details. Chuan et al. (1996) examined the release of heavy metals from a contaminated soil, which was suspended in water (soihwater ratio = 1 :7), and found that aqueous concentrations of Pb, Cd and Zn are positively correlated with those of Fe during incubation because heavy metals are released from Fe (oxy)hydroxide surface as it dissolves (Chuan et al., 1996). Obviously, it is contrary to what we observed for Pb concentrations in both the Tampa and Montreal soils. On the other hand, Amrhein et al. (1994) reported that, under water-saturated (soihwater ratio 1:0.2) incubation, Cd and Cu concentrations in pore water decrease whereas Fe(II) increases with time. This is consistent with our results. It is possible that the different relations between Fe(II) and heavy metals in solution result from the different soil-water ratios used in different studies. In the studies discussed above, contributions of exchangeable metals to aqueous metal concentrations during incubation varied with the soihwater ratios. When the ratio is low, i.e., small quantity of soil is suspended in large amount of water, ions on the exchange sites tend to diffuse into the water to maintain their chemical potentials between solution and exchange sites. Consequently, the importance of exchange sites in holding metal ions reduces in a relative sense. On the other hand, as the soihwater ratio increases, the importance of exchangeable metals in contributing to soluble metals increases since the amount of exchange sites increases relative to solution volume. Therefore, metal distribution between solution and exchange sites has to be taken into

PAGE 94

83 consideration, which is mainly controlled by metal competition between exchange sites and aqueous ligands in solution. Therefore, the observation of Amrhein et al. (1994) and the one presented here may be due to the fact that, with incubation, more heavy metals go to exchange sites as Fe (II) concentrations increase in solution. We have demonstrated that Pb solubility in soil columns is inversely related to aqueous Fe concentration in Chapter 3. Therefore, in this sense, all the studies discussed above along with the one presented here are consistent, i.e., with Fe dissolution upon reduction, Pb is released into solution and exchange sites. Similarly, arsenic concentrations in pore water and suspension varied differently with incubation. Masscheleyn et al., (1991) examined the-arsenic release from a contaminated soil in a suspension (soihwater = 1:6) during incubation, and found that soluble arsenic increases as redox potential decreases (+500 -200 mv). It was attributed to As(V) reduction to As(III) and Fe reduction dissolution to release adsorbed As. However, analysis of pore water in soils after water flooded incubation showed a different trend (Onken and Hossner, 1996). They found that soluble arsenic concentrations increased first and then decreased with incubation, and a maximum occurred at 20-30 d after flooding. Further analysis of the solid phase showed the loss in soluble arsenic could be accounted by surface bound when the soil was incubated longer than 20-30 d. The result of the later study is consistent with our data that showed As concentration increased and then decreased with Fe reduction dissolution (Figure 4-2), which will be discussed in the following section.

PAGE 95

0 10 20 30 40 50 60 70 80 incubation time (d) Figure 4-3. Cumulative Pb and Fe leached after 3 1 .8 pore volumes of 0.01 M CaC12 Montreal soil

PAGE 96

85 • Fe 330 n < O As a e o A Cu Pb s i B 13 O I 10 20 30 40 Incubation time (d) 50 60 70 Figure 4-4. Cumulative Pb, As, Cu and Fe leached after 31.8 pore volumes of 0.01 CaCl 2 in Tampa soil.

PAGE 97

86 Heavy metal mobility with incubation Metal mobility in this paper is examined using cumulative metals leached with 32 pore volumes of 0.01 M CaCl 2 solution. It is well understood that the metals that can be leached out with CaCl 2 solution include both the soluble and the exchangeable. For the Montreal soil, the cumulative leached Pb concentrations increased with incubation (Figure 4-3), which is consistent with the overall trend shown in the pore waters (Figure 4-1). However, the cumulative leached Fe concentrations showed a different trend from Fe(II) concentrations in pore waters (Figure 4-3), suggesting that Fe(II) in exchange site is not proportional to that in aqueous phase assuming that Fe(II) « soluble Fe. For the Tampa soil, changes in overall trends for metals with incubation can be divided into two groups: 1.) Pb and Cu decreased with incubation; 2.) As and Fe increased with incubation, which is not consistent with the case in pore water. In fact, the reduction of cation exchange capacity (CEC) caused by the association of hydroiron polymers or iron hydroxides with soil mineral has been long recognized (Coleman and Thomas, 1964; Carstea et al., 1970). Hendershot and Lavkulich (1983) reported that Fe coatings on illite lower CEC and the ones on kaolinite increase anion exchange capacities (AEC) in comparison to the uncoated ones at pH< 7. They suggested that the Fe coatings reduce the CEC and increase AEC measured at low pH by either physically blocking or electrostatically canceling the permanent negative charge carried by the crystalline minerals. Similarly, Arias et al., (1995) found that the point of zero charge (PZC) of Fe-coated kaolinite increases linearly with the amount of the coating Fe. In the experiments by Shainberg, et al. (1987), they demonstrated that the CEC of the mixture Fe (FeCl 3 ) and soil decreases by 52% in the 10 mmol Fe kg" 1

PAGE 98

87 treatment to 72% in the 40 mmol Fe kg" 1 treatment, and correspondingly the exchangeable sodium percentage from 23.8 to 15.4. Rengasamy and Oades (1977) reported that with addition of ferric hydroxide the electrophoretic mobility of iron-clay increase from negative to positive values. All these data surpport that with the reduction dissolution of Fe, Fe coating fraction decreases, and then CEC increases for a soil. Even though it is supported by sufficient experimental and theoretical evidence, it has, to our knowledge, not been stated for incubated soils. However, the reduction in CEC, along with solution chemistry, can have significant effects on heavy metal retention. With the reduction dissolution of Fe, therefore, the blocked exchangeable sites will be freed and the concentration of Fe in pore water increases, and thus Fe (II) concentration in exchangeable phases may increase as well. It was evident that, for the Tampa soil, changes in Fe (II) concentrations in pore water and its cumulative leached amount with incubation time were similar: they both increased with incubation (Figure 44). However, for the Montreal soil, a sudden increase in cumulative leached Fe was found at 20 day incubation compared to Fe(II) concentrations in pore water, suggesting that exchangeable Fe (II) became dominant in cumulative leached Fe. In fact, there was 5.55 umoles Fe leached during 7-16 pore volumes (which can be considered as exchangeable) among 7.1 1 umoles of the total cumulative leached Fe after 20-d incubation. In the Tampa soil, changes in Pb concentrations in pore water and the cumulative leached Pb with incubation time were similar: they both decreased with incubation. This is contrary to Fe behavior, suggesting that Pb concentrations in solution may be controlled by competition between aqueous Fe and Pb for ligands and exchange sites. It

PAGE 99

88 is well documented that Fe(II) and Pb(II) have similar affinity for ligands (Turner et al., 1981; Nieboer and Richard, 1980). As aqueous Fe (II) concentrations increase with incubation due to Fe reduction dissolution, more aqueous Pb (II) may be forced onto exchange sites. If this is simply the case, however, the cumulative leached Pb was not supposed to decrease with incubation (Figure 4-4). However, it was difficult to imagine that there was so much Pb present in the exchange sites without some of it adsorbing chemically onto the surfaces. However, it should be pointed out that there was significant difference between the relations of Pb concentrations in pore water and cumulative leached Pb with incubation: while the Pb concentration in pore water decreased exponentially the cumulative leached Pb decreased only linearly (incubation time < 30 day). It is suggested that less reduction of the cumulative leached Pb than that of Pb concentrations in pore water could be caused by more exchangeable Pb with incubation. In the Tampa soil, cumulative leached Cu decreased with incubation (Figure 4-4), which is inconsistent with changes in Cu concentrations in pore water (Figure 4-2). The significantly greater cumulative leached Cu after 3-d incubation can be attributed to more Cu coming from exchange sites when leached with CaCl 2 . In fact, Cu (II) has greater affinity for exchange sites than Pb (II) (McBride, 1994). Interestingly, As concentrations in pore water increased first and then decreased with incubation, whereas the amount of cumulative leached As showed no significant drop with incubation (Figure 4-4). As discussed above, as Fe "coatings" was removed with incubation via reduction dissolution, soil CEC increased as well. This caused more Pb (II) to transfer from solution to exchange sites, as a result, local charge reversal might

PAGE 100

occur, therefore positively-charged particles held more H2ASO3" and /or H3ASO4". It has been realized that at slightly reducing condition and pH = 4 (for the Tampa soil in this study) both H2ASO3' and H3ASO4' are not stable thermodynamically (Cullen and Reimer, 1989). However, local pH near positively charged surface may be significantly greater than in bulk solution, therefore H2ASO3" may be stable near highly positively-charged surfaces. On the other hand, there has been evidence showing that even under reducing condition, only H3ASO4" is associated with natural particle surfaces (Belzile and Tessier , 1989). The existence of H3ASO4" under reducing condition has been attributed to the oxidation of arsenite to arsenate by Mn and Fe oxyhydroxides (Peterson and Carpenter, 1983). As discussed above, CEC increases with Fe reduction dissolution, especially at pH < PZC of Fe minerals. The consequent redistribution of metals between solution and exchange sites is probably determined not only by the magnitude of CEC, but also the characteristics of metal ions, the competing co-ions and counter-ions (ligands). Therefore, changes in metal concentrations in a soil solution with time under incubation can be complicated. However, the following interactions have to be taken into consideration: 1 . Fe reduction dissolution releases Fe(II) and adsorbed heavy metals into solution at the same time frees exchange sites, 2. Fe (II) and heavy metal cations compete for ligands to stay in solution, 3. Fe (II) and heavy metal cations compete for the exchange sites, transferring from solution to solid phase,

PAGE 101

90 4. Charge reversal on exchange sites results in negatively changed heavy metal to transfer from solution to exchange sites. In a particular soil system, different processes may be dominant in controlling metal concentrations in soil solution. There is an alternative explanation for the As behavior with incubation. It has been well established that P0 4 3 ' may precipitate with Fe 2+ to form minerals such as vivianite, Fe 3 (P0 4 ) 2 8H 2 0 at low soil redox potential (Holford and Patrick, 1979; Patrick and Khalid, 1974). Because of the great similarity between H2PO4' and H 2 As0 4 "or H2ASO3", it can be inferred that formation of a new solid phase by Fe 2+ and H2ASO4" or H2ASO3 limited the solubility of As in pore water as the incubation time approached to 60 days (Figure 4-4). Implication of this research In literature, there has been a fair amount of papers focusing on changes in solubility and mobility of heavy metals with incubation. However, reported data have often been contradictory partially because of differences in soil/water ratios used in different experiments such as metal concentrations determined in pore water, filtrates separated from soil suspension, and leachates from a soil leached with DDW or an electrolyte. It has been noticed that concentration changes of heavy metals with incubation are often confusing. For example, metal concentration in pore water or metal mobility leached by DDW suddenly decreases with incubation when occurrence of precipitation can not be proved. As discussed in this chapter, the roles of exchange sites in controlling metal concentrations in these systems have to be taken into consideration. With sufficient evidence, we showed that CEC of a soil change with incubation, and thus

PAGE 102

may greatly affect metal solubility and mobility. The role of exchange sites on the solubility and mobility of heavy metals is crucial to understand various behaviors of heavy metal upon incubation. Furthermore, incorporation of these results into future metal transport models may improve their accuracy.

PAGE 103

CHAPTER 5 HEAVY METAL MOBILITY IN CONTAMINATED SOILS: PART 2. COLLOIDFACILITATED METAL MOBILITY IN A PB-CONTAMINATED SOIL Introduction Understanding metal mobility in soils is of significant environmental interest. In general, it is widely accepted that metal mobility, such as Pb, is practically negligible in soils due to their low solubility. However, enhanced Pb mobility occurs in soils under certain conditions (Newman et al, 1993). Colloid-facilitated metal transport in soils has been proposed as one of the mechanisms (McCarthy and Zachara, 1989). Colloids, e.g. clay, organic matter, and quartz, are ubiquitous in soils. Several processes responsible for colloid release and deposition have been well established. Based on the DLVO (Derjaguin-Landau-Verwey-Overbeek) theory (Israelachvili, 1992), the net surface interaction energy is equal to the sum of the interactions between electrical double layer (EDL) and van der Walls forces, which varies with separation distance. As particles move close to each other, the net interaction energy experiences a secondary minimum first, then a primary minimum after an energy barrier. After that point, further particle move towards each other causes drastic increases in the interaction energy, which is referred to as Born repulsion. In principle, particles are attached (deposited) due to the net attractive forces at either the primary or secondary energy minimum. On the other hand, both the magnitude of the energy barrier and the separation distances for the primary and secondary minima are affected by solution chemistry. In order for a deposited particle to be released, repulsive forces have to be 92

PAGE 104

93 generated between surfaces of colloid and stationary solid phases such as resulting from changes in solution chemistry. Changes in solution chemistry may reduce or eliminate the primary or secondary minimum so that the release of deposited particles become favored (McDowell-Boyer, 1992; Ryan and Gschwend, 1994b). Numerous studies have shown that solution chemistry is one of the primary factors affecting particle deposition and release in natural porous media (Elemelech and O'Melia, 1990a; Elimelech, 1992; McDowell-Boyer, 1992; Ryan and Gschwend, 1994; Seaman et aL, 1995). However influence of solution chemistry on the interaction between EDLs of particles is a rather complicated issue. In addition, solution chemistry in soil varies spatially and temporally, therefore, it is still a challenge to model the interaction of EDLs between colloids and stationary solid phases in soils. Nevertheless, effects of solution chemistry on colloid mobility are closely related to potential determining ions (Parks, 1967), specific adsorbed ions (Parks, 1975) and indifferent ions (Park, 1 967) present in soil solution. Potential determining ions are constituent ions of solid particles (e.g. H + and OH* of Fe oxides, Ca 2+ and CO3 + of CaCCb) and their concentrations primarily determine particle surface potentials. Specific adsorbed ions are those adsorbed onto solid surfaces and may change the magnitude and even sign of the surface charge (e.g. Ca 2+ and Pb 2+ ), whereas indifferent ions are those adsorbed physically to surfaces and change only the magnitude of the surface charge (e.g. Na + and CI"). It is important to consider all three types of ions when studying the impacts of dynamic solution chemistry on colloid mobility in soils. In surface soils, solution pH, ionic strength, and redox potential are influenced by rainfall, bioactivity, or change of land use; whereas in vadose zone they are mostly influenced by the fluctuation of water

PAGE 105

94 table (Boul et al, 1997). In other words, in soils all three kinds of ions (potential determining ions, specifically adsorbed ions and indifferent ions) vary drastically with time and space in terms of species and concentrations. Inevitably, those changes will significantly affect the interactions between colloid surfaces and stationery solid phases, and thus colloid release and deposition rate. In particular, solution chemistry has great effect on surface charge heterogeneity of soil. Oxides of Fe, Al and Mn are the most common sources of surface charge heterogeneity in natural aqueous environments (Ryan and Elimelech, 1996). At neutral pH, these compounds may carry a positive surface charge, which are called positivelycharged patches in matrix minerals such as quartz and feldspar that carries negative surface charge (Song et al., 1994). In natural aquatic systems, the range and magnitude of the existence of these patches may affect colloid deposition and release significantly. Recently, Johnson et al ( 1 996) have demonstrated that colloid deposition rate is controlled by the extent of positively-charged patches developed in stationary solid phases in porous media. In dynamic soil environments, however, these patches may vary temporally and spatially since solution chemistry (e.g. pH and redox potential) greatly affects oxide solubility. However, the effects of the dynamic feature of surface charge heterogeneity in soil on colloid mobility are not considered in most studies. Metals of low solubility tend to associate with colloids (Mills et al, 1991). They may adsorb or precipitate onto colloid surfaces, or even form their own colloids. Take Pb 2+ for example, in the first case, Pb 2+ may affect the surface potential between solution and solid drastically by acting as a specific adsorbed ion (Kretzschmar et al., 1997). In the second case, Pb 2+ is a potential determining ion that may change the surface potential

PAGE 106

95 greatly. In other words, the mobility of these colloids may be different from the original ones during charge developments. The mobility of metal-bearing colloids thus needs to be examined (Temminghoff et al, 1997; Kretzschmar et al., 1997). The objective of this study was to evaluate the effects of soil solution chemistry on colloid-facilitated metal mobility, such as Fe, Al, and Pb, in a Pb-contaminated. Such effects were limited to those rising from surface properties of colloids and stationery solid phases, which are impacted by soil solution chemistry. In our experiment, soil solution chemistry was altered through incubation under saturation condition. After prewashing the soil columns with 0.01 M CaCb solution, the influent was switched to deionized water (DDW) to mobilize colloids. Short columns were chosen to avoid colloid redeposition. Our results showed that colloid mobility varied significantly with incubation, which may be caused by changes in surface charge heterogeneity on colloids and stationery solid phases. Materials and Methods Characteristics of soil samples The lead-contaminated soil used in this study was collected from Montreal, Canada. The site was contaminated via Pb-battery recycling and concurrent smelting operation in the past. The soil was air-dried and passed through 2-mm sieve. Total elemental concentrations of the soil was determined with a ThermoJarrellAsh 61 E ICP after digestion in a CEM MDS-2000 microwave using EPA method 3051. The soil was also analyzed using x-ray diffraction and thermal analytical techniques, i.e.

PAGE 107

96 Table 5-1 . Selected properties of the Pb contaminated soil used in this study Soil location Montreal, Canada pH (1 : 1 with distilled water) 7.7 Fraction of organic carbon 2.24 Total Elemental Analysis (Ma, 1996) gkg 1 Pb 1.60 Fe 133.30 Al 169.50 Mn 0.80 Ca 15.70 Particle size distribution (Day, 1965) % sand 11 Silt 40 clay 49

PAGE 108

97 thermogravimetry and differential thermal analysis. The major minerals present in this soil include Fe oxides (amorphous), quartz, calcite, dolomite, vermiculite, mica, and feldspar. Selected soil properties are shown in Table 5-1. Column Experiment This experimentation is a continuation of Chapter 4 Leaching . After displacing the original soil solution as described In Chapter 4, the influent was switched to DDW. The effluent from the top of the column was collected with a fractionation collector. Turbidity and pH in the effluent were determined immediately. Analysis of effluent . Each fraction of effluents was split into two parts. One was acidified with HNO3 to pH < 2 without filtration. The metal concentrations determined this way was termed as total metal concentrations (sum of colloidal and soluble Pb). The second part was acidified to pH<2 after filtering through 0. 1 um membrane filter to obtain dissolved metal concentrations. Analysis of Fe(II) and Ca in pore water Redox potential normally decreases with incubation under water-flooded condition. In our study, the electrode installed in soil columns measured soil redox potential only in a relative sense as discussed previously. In addition, Stumm (1984) suggested that it is more reasonable to measure important redox species than redox potential to indicate redox status in natural systems. Fe(II) concentration in pore water was thus used as a measure of soil redox status.

PAGE 109

98 Pore water was separated from the soil by centrifugation (Nkeda-Kizza et al, 1982; Dao and Lavy, 1978) after removing standing water on top of the soil columns. The pore water was then filtered with 0.1 -um membrane filter for metal analysis. Aliquot of 0.1 mL of the filtrate was transferred to 10 mL of ferrozine (lg L" 1 ) in 50 mM HEPES (N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid) buffer at pH=7 (Lovely and Philips, 1987; Heron et al., 1994). Concentration of Fe (II) was determined by measuring the absorbance of the filtrates at 562 nm (Stookey, 1970). In addition, Ca concentration was determined by atomic absorption spectrophotometer. Analytical methods All experiments were conducted in duplicate in acid-washed (5% HNO3) labware. All chemicals used in this study were of analytical grade or better. Double deionized water from a Barnstead NANOpure water system was used. Total metal concentrations of the filtrates were analyzed with an atomic absorption spectrophotometer (Perkin-Elmer 2380) equipped with a graphite furnace atomizer. Flame atomic absorption was used to analyze metal concentrations > 1 mg L" 1 and graphite furnace atomizer was used to measure metal concentrations < 1 mg L" . Multilevel standards (Fisher Scientific) for all metals were prepared in the same matrix as the extracting reagents to minimize matrix effects. Blanks were used for background correction and other sources of error. At least one duplicate and one spike sample were run with every 20 samples to verify method precision. The spike recovery and precision were found to be within 100% + 10%.

PAGE 110

99 Result and Discussion Colloid mobility and incubation Effluent turbidity, a measure of relative colloid concentrations, generally decreased with incubation time, with longer incubation resulting in lower turbidity (Figure 5-1). Maximum turbidity was reached at 3-4 pore volumes for all soil columns, suggesting that colloids were mobilized by switching the influent from 0.01 M CaCb solution to DDW. Effluent turbidity stayed more or less constant as pore volume increased from 3-4 to 24. Chloride concentration, which was 20 raM in the influent before switch to DDW, decreased significantly from 20 to 0.2 raM as pore volume increased from 0 to 1.6. Meanwhile, the turbidity increased drastically from 0.2 to 0.8 NTU. It is suggested that colloids were mobilized by DDW immediately while 0.01 M CaCb in the soil columns was displacing by DDW. This fast colloid release has been reported by Nocito-Gobel et al. (1996) who observed that polystyrene latex was released from a silica-sand column due to reduction of ionic strength in the influent. They explained that reduction in ionic strength may eliminate prior net attractive force minima (primary or secondary) so that colloid release rate is fast relative to the advective solution flow through the column. As indicated by turbidity, colloid mobility decreased with incubation or with decrease in aqueous Fe(II) concentrations from 0.91 to 0.17 uM in pore water (Figure 52A). The cumulative leached colloidal Fe or Al in 23 pore volumes after 3-d incubation (aqueous Fe (II) = 0.91 uM ) were significantly greater than they were after 20or 80-d incubation (Fe (II)=0.27 or 0.17 uM ) (Figure 5-2A). It has long been recognized that

PAGE 111

100 2.0 0 5 10 15 20 25 Pore volumes Figure 5-1 Changes of effluent turbidity with pore volumes under different incubation times. The insert is a typical breakthrough curve of CI" when deionized water displaced CaCl 2 .

PAGE 112

101 redox status may affect colloid mobility in soils (McCarthy and Zachara, 1989; Grolimund et al., 1996). However, there is little information available on this topic in the literature. In Fe oxide-coated sand, Ryan and Gschwend (1994a) reduced goethite with ascorbic acid to remove its cement effect, and found that the amount of colloids in effluents is directly related to the concentrations of dissolved Fe. This is seemingly consistent with our result in that cumulative colloidal Al and Fe were proportional to aqueous Fe (II) concentrations in pore water (Figure 5-2A, r 2 = 0.92-0.97). In this study, soil redox status decreased with incubation, which was confirmed by decrease in redox potentials monitored (Figure 5-2). In general, a more reducing condition produces a greater aqueous Fe concentration (Chuan et al., 1996). In our system, however, the opposite was observed, i.e. aqueous Fe (II) concentration in pore water decreased from 0.91 to 0.17 uM as incubation increased from 3 to 80 d, suggesting that decementing by reductive dissolution of Fe-oxide may not be the only factor controlling colloid mobility in our system. Though Fe (II) concentration in pore water decreased with incubation, it did not change much when the incubation time was increased from 20 to 80 d (Figure 3-2), suggesting that Fe concentrations may be controlled by mineral solubility in the period. From a thermodynamic point of view, siderite (FeC03) can control Fe 2+ solubility in soil depending on redox conditions and C0 3 2 " concentrations (Lindsay, 1979). In this study, soil redox status became more reducing with incubation, which was confirmed by redox electrode measurement (Figure 5-2). Therefore, aqueous Fe (II) was released by Fe reduction dissolution. The soil used in this study contained -8% of calcite (CaCC»3, log K = 9.74), which is much more soluble than siderite (log K = 7.92, Lindsay, 1979). In

PAGE 113

102 Figure 5-2 Relation of aqueous Fe (II) and Ca concentrations in the pore water of soil columns on the cumulative colloidal Fe and Al concentrations in the effluent after 23 pore volumes during incubation.

PAGE 114

103 this case, formation of siderite was favored as a stable iron mineral, which was confirmed by MINTEQA2 calculation (Allison et al., 1991). This may explain why aqueous Fe(II) decreased as incubation time increased. Most minerals present in the soil used in this study carried mostly negative charges as expected (Table 5-2). For minerals carrying pH dependent-charge, their point of zero charge (PZC) was generally lower than pH =7.7 of the soil (Table 5-2). The mobile colloids in the effluents has ^-potential = -13 mv in 0.1 M NaCl. These evidences supported that the soil was negatively-charged, however, the positively-charged patches may also exist. It is commonly believed that Fe oxides are major sources of positive charge near neutral pH. This led us to expect that as more Fe oxides dissolved with incubation, positively-charged patches decreased, colloid mobility would increase. However, the opposite relation between incubation and Fe(II) concentrations were observed in this study (Figure 5-2A). As discussed previously, Fe oxides may be transformed to siderite. This transformation promoted aqueous Ca concentrations in pore water, which was consistent with the fact that Ca concentrations increased with incubation (Figure 5-2B). On the surfaces of carbonate minerals, there are sufficient examples (Charlet et al., 1990; Cappellen et al., 1993.) showing that the positive-charge development can be mainly described as >C0 3 H° + Me 2+ <-> >C0 3 Me + + H + (51 ) where >C03H° and >CC>3Me + stand for bicarbonate sites and its complex with free metal cation (Me) at the surface of carbonates, respectively. In our system, Ca 2+ was dominant among all the Me in pore water, for example, Ca concentrations was much greater

PAGE 115

104 Table 5-2 Minerals in the soil and their Point of zero charge (PZC) Minerals PZC References Vermiculite Mica Feldspar Fe oxides (amorphous) 7.9-8.2 Dzombak et al., 1990 Quartz 2-3 Parks, 1967 Calcite 5.5-6 Parks 1967 Dolomite 8-9 Parks 1967

PAGE 116

105 than Fe(II) (Figure 5-2). In addition, no changes in pH among soil columns were found (pH=~7.7), which may be due to the buffering capacity of carbonates. Nevertheless, This gave us a basis to assume Ca 2+ concentrations were proportional to that of aqueous Ca in the pore water. Obviously, Eq (5-1) suggested that the positive-charged patches on calcite increased with aqueous Ca concentrations and thus that colloid mobility decreased because more mobile colloids had deposited on the stationary grains with a greater bonding (primary minimum deposition). This is consistent with the observation that colloidal Al and Fe were inversely related to aqueous Ca (Figure 5-2, R 2 =0.43 and 0.86, respectively). Colloid elution curves Concentrations of colloidal Fe and Al changed with pore volumes are demonstrated in Figure 5-3. Different from the observations made in well defined systems (Nocito-Gobel et al., 1996) that the elution curves consists of one major narrow peaks, the colloidal metal concentrations peaked over a wide range of pore volumes or repeatedly at several numbers of pore volume. It has confirmed that the patterns of the elution curves are reproducable. The velocity difference between fluid (tracer) and colloidal particles have been long recognized (Small, 1974). Size exclusion of colloidal particles could cause them to move faster than the tracer. On the other hand, colloid deposition in a secondary minimum may retard colloids significantly (Nocito-Gobel et al., 1996). To our knowledge, however, no observation has been reported that soil colloids moved in the manner as shown in Figure 5-3.

PAGE 117

106 0 5 10 15 20 25 Pore volumes Figure 5-3 Elution curves of colloidal Fe, Al, and Pb concentrations in effluents with pore volumes under various incubation times. Each point represents the mean of two replicates

PAGE 118

107 Figure 5-4 Relationship among colloidal Fe and Al concentrations, dissolved Ca concentrations and pH in effluents after 3 d of incubation. Each point represents the mean of two replicates. The trend is reproducable. The arrows indicate the boundary Beyond which the Ca release mechanism from the soil may change.

PAGE 119

108 In fact, after the displacement of the original soil solution in a soil column with 0.01 M CaCl 2 solution, Ca was the dominant cation in both bulk solution and electrical double layers (EDL) surrounding colloidal particles and stationary solid phases in the soil. When the effluent was switched from 0.01 M CaCl 2 solution to DDW, we expect that the release of Ca from the soil be determined by various mechanisms, which came to play probably in a time sequence. At the beginning stage of switching to DDW, the release of Ca was from the bulk solution, which was confirmed by the fact that Ca concentrations and pH in the effluent at pore volume =1 were approximately equal to those of the influent ([Ca]= 0.01 M, pH=6.95) (Figure 5-4). Secondly, Ca diffusion from EDL to the bulk solution driven by a concentration gradient may become dominant as the number of pore volumes increased (between the two arrows in Figure 5-4).. When Ca in the bulk solution was depleted, Ca cations moved against electrostatic attraction away from the surfaces driven by osmotic pressure, extending EDLs surrounding colloids and solid phases to mobilize colloids. Obviously, any significant expanding of EDL and resultant colloidal mobilization had to be associated with significant amount of Ca released into bulk solution. In this case, the excess Ca 2+ in the diluted bulk solution may be combined with hydroxyl to release proton (Ca 2+ + H2O <-» CaOH" + H + ) (Oste'r and Shainberg, 1979), resulting in a decrease in pH. Between pore volumes 6-17, significant increase in Ca concentration and decrease in pH accompanied significant increase in colloid concentrations (Figure 5-4). This is consistent with the mechanism we proposed above. Generally Ca hydrolysis is weak in solution, it is thus not likely for it to bring the pH down as much as 1 unit as shown in Figure 5-4. In a heterogeneous system such as in

PAGE 120

109 this study, however, the deficiency of negative charge existed in the bulk solution, hence it was possible that the charge deficiency may further drive dissociation of H 2 0, releasing protons. Thirdly, desorption of Ca from colloid surfaces and stationary solid phases occurred from 1 8 to 23 pore volumes (after the second arrow in Figure 5-4). This was a slow process and therefore occurred at high pore volumes, resulting in an increase in repulsive forces between colloids and stationary solid phases. An increase in Ca concentration and in pH were observed in the effluent in pore volumes between 17-23 (Figure 5-4). In essence, this process is equivalent to sorbed Ca displaced by H + . However, a direct validation is needed. Obviously, dissolution of Ca minerals in the soil may also play a role in regulating the release of Ca from soil columns, especially during pore volumes of 1 8 to 23. During pore volumes of 3 to 23, however, it was impossible for dissolution to play a significant role in Ca releasing, because the pulse-like feature of Ca releasing cannot be explained by dissolution of Ca minerals. Of course, overlapping between the different stages may occur. However, each individual mechanism may dominate a particular range of leaching (Figure 5-4). As discussed above, the colloid elution curves can be mainly divided into two regions between pore volumes of 3-23, which were regulated by Ca diffusion or desorption (Figure 5-3). This suggested that a wide range of surface heterogeneity existed in soil colloids and stationary solid phases. In theory, different from the colloids mobilized by Ca release via desorption, those mobilized by diffusion must contain less amount of reactive groups such as >C0 3 H, -OH on the surfaces. However, there was no direct evidence to support this conclusion in this study. In the range of Ca diffusion,

PAGE 121

110 colloid release showed a clear elution pattern. In a well-controlled system, Nocito-Gobel et al. (1996) have demonstrated that polystyrene colloids are released from the column packed with silica sand in a similar manner, i.e. as pore volumes increased colloid concentrations appeared in two pulse-like peaks. They attributed these to the inherent heterogeneity of the silica sand surface, and suggested the two separate peaks are most likely related to the primary and secondary deposition of colloids. In our study, the interactions between colloids and stationary solid phases are much more complicated because of the heterogeneity of both colloids and stationary solid phases. Therefore, one may expect that soil colloids release would manifest themselves in a chromatographic manner since each group of colloids, which possess similar surface properties, would appear in two peaks in effluents, which is caused by colloids releasing from the primary and secondary deposition. This is consistent with the trend at pore volumes less than 1 7 shown in Figure 5-3. As shown in Figure 5-3, colloid mobility changed with incubation time. However, the peak position of colloid concentrations under different incubation basically matched each other, even though some shifted slightly (Figure 5-3). This suggested that incubation did not significantly change the major types of mobile colloids and their mobility rather it affected the mobile population of each type colloid when they were exposed to a certain perturbation. Colloid-facilitated Pb mobility Similar to colloidal Fe and Al, colloidal Pb concentrations varied with incubation (Figure 5-3). Interestingly, there was no significant difference in the colloidal Fe and Al in the first several pore volumes when the columns were incubated from 3 to 20 d (Figure

PAGE 122

Ill 5-3), while the concentrations of colloidal Pb decreased approximately 4 times after 20-d than 3-d incubation (Figure 5-3). This suggested that Pb-bearing colloids tended to be immobilized or the tendency of association of Pb with mobile colloids decreased with waterflooded incubation. However, a recent research (Kretzschmar et al., 1997) suggests that the association of Pb with colloids does not affect colloid mobility significantly, especially at low surface coverage. This may exclude the first possibility. It is well known that Pb is most likely sorbed onto Fe oxides (Benjamin and Leckie, 1981). However, when the environment becomes reducing, Pb is released as Fe oxides dissolve (Gambrell et al., 1980). This suggests that Pb carried by mobile colloids may redistribute with incubation. One possibility is that the released Pb resorbed onto the surface sites bicarbonate (>C03H) of carbonate minerals (Zachara et al., 1991), and the particles bearing carbonate mineral may carry positive charge and not tend to be mobilized as discussed before. Therefore, colloidal Pb decreased with incubation. However, further study is needed to verify this mechanism. Conclusion In this Pb-contaminated soil, colloid mobility decreased with incubation, and it was positively and inversely correlated to the concentrations of dissolved Fe and Ca in the pore water, respectively. This cannot be understood by either the decementing effect from reducing dissolution of Fe oxides and charge heterogeneity rising from them. Instead, it is possible that the decrease of colloid mobility with incubation was caused by an increase of positive-charged patches of carbonate surfaces, which is determined by Ca concentration in pore water.

PAGE 123

112 A elution pattern of colloidal Fe and Al concentration changes with pore volumes was observed. In general, a peak of colloid release accompanied a higher dissolved Ca concentration. Three possible mechanisms, i.e., bulk solution displacement, diffusion out of EDL and desorption from surfaces, were proposed and played a major role in a time sequence in Ca release. These mechanisms may explain the chromatographic pattern of colloidal Al and Fe concentration changes with incubation time. Colloid particles can facilitate Pb transport, however, the ability decreased with water-flooded incubation for this soil. It is possible that Pb redistribution from Fe oxides to carbonates occurred during incubation.

PAGE 124

CHAPTER 6 RELEASE AND DISPERS ABILITY OF COLLOIDS IN TWO CONTAMINATED SOILS Introduction Colloids, which are defined as particles < 2 um, are ubiquitous in soils. Under certain conditions they may be mobilized and facilitate transport of heavy metals in soils (McCarthy and Zachara, 1989; Newman et al., 1993), possibly imposing a threat on the environment. Net release of colloids from a soil is mainly determined by rates of two opposite processes: colloid release (Kallay et al., 1987) and deposition (Adamczyk et al., 1983). These processes are mainly controlled by the net surface interaction energy between colloids and stationary solid phases, which can be described by the DLVO theory (Derjaguin-Landau-Verwey-Overbeek ) (Israelachvili, 1992). Based on the DLVO theory, the net surface interaction energy is equal to the sum of the interactions between electrical double layer (EDL) and van der Waals forces, which vary with the separation distance between colloids and stationary solid phases (Figure 2-1). As colloids move close to stationary solid phases, the net surface interaction energy experiences a secondary minimum first, and then a primary minimum (§ m i n ) after a maximum energy barrier ( ma x). After that point ( m i n ), further movement of colloids towards the stationary solid phase causes drastic increases in the interaction energy, which is referred to as the Born repulsion. Theoretically, particles are deposited when the net attractive forces are at either the primary or secondary energy minimum. Colloid deposition is considered as a 113

PAGE 125

114 two-step process: transport of colloids from bulk solution to the stationary solid surfaces and then their attachment to the surfaces, which depends upon the nature of particlesurface interaction. Colloid attachment rate is exponentially related to the maximum energy barrier max-<|>min) (Figure 2-1). To validate the DLVO theory in describing colloid dispersion and coagulation in a suspension, colloid stability test using batch experiments was developed (Reerink and Overbeek, 1954a and b). Such a test has been partially adopted to evaluate the interactions among soil colloids, i.e. water dispersability of soil colloids (Miller et al., 1990; Shainberg et al., 1981). In general, water dispersability of soil colloids decreases with ionic strength at a given pH. There is a critical flocculation concentration (CFC) of electrolyte for a given soil suspension, above which flocculation occurs. There has been effort to relate this CFC-based dispersability of soil colloids to their mobility in soil (Kaplan et al., 1996; Seta and Karathanasis, 1997, Shainberg et al., 1981a and b), i.e. applying batch-experiment based results to predict colloid behaviors in column experiment since colloid stability tests are normally conducted in batch experiments whereas colloid mobility tests are done in column experiments. Soils with higher CFC may have greater colloid mobility since colloids can be more stable in a suspended state at a given condition. However, no consistent result was achieved when the CFC of soil dispersive clay was used to predict soil colloid mobility (Kaplan et al., 1996). They summarized that CFC predicts colloid dispersability, but not mobility when size straining

PAGE 126

115 is dominant in controlling colloid deposition, which occurs when colloid particles are larger than pore opening in porous media. Different from the colloid stability test, in which colloid deposition (aggregation) is dominant, colloid mobility in a soil is determined by the net release rate of colloids at a given flow condition as shown in Eq. (6-1). Release Transport Immobile colloids > Mobile colloids < =, Colloids in leachate (6-1) Deposition Colloid release can be induced by a rapid reduction of influent ionic strength in well-defined porous media (Ruckenstein and Prieve, 1976; Roy and Dzomk, 1996a; Nocito-Gobel and Tobiason, 1996) and soil (Seaman et al., 1995; Jacobsen et al., 1997). However, this may not always be the case since the opposite result was also observed by Kallay et al. (1987). They attributed to the elimination of colloid redeposition in short columns as a possible cause for their observation. This suggests that the net colloid released often is a result of both colloid deposition and release. The effect of ionic strength reduction on colloid release has been attributed to the repulsion between colloids and porous media arising from EDLs expansion, resulting in a reduction in the depth of the energy well (Kallay et al., 1987). Different from colloid deposition, however, colloid release is more difficult to evaluate both theoretically and experimentally. Using hematite colloids and quartz columns, Ryan and Gschwend( 1994a) have shown that the transport of detached colloids to the bulk solution is the rate-limiting step during rapid colloid release when the maximum energy barrier (c|) ma x) approaches zero (Figure 2-1). In soil, a similar phenomenon has been observed by Jacobsen et al. (1997) in an intact column experiment in which the colloids were leached with tap water. By combining this

PAGE 127

116 with the discussion that the net colloid release rate is determined by the relative rates of colloid release and deposition as shown in Eq. (6-1), one may conclude that, at a rapid release condition, the net colloid release rate is determined by the rates of the net colloid diffusion to bulk solution and attachment onto stationary solid surfaces. Based on the above discussion, we hypothesize that, under a given water flow condition in a soil, the net colloid release rate may be controlled primarily by colloid attachment rate, which can be evaluated using the colloid stability ratio (W) in suspensions. Colloid stability ratio is the ratio of colloid flocculation rate of a suspension at normal observation condition (slow coagulation) to that under rapid flocculation condition when the energy barrier ( ma x) = 0 (Reerink and Overbeek, 1954). Under slow flocculation condition, only a fraction (W) of colloid collisions leads to coagulation. In fact, CFC represents only a specific case of flocculation kinetic where W =1. Therefore, W instead of CFC will be used to evaluate disperabililty of soil colloids in this paper. Assume that the rate of flocculation is related to the changes in light absorbency of solution, W can be obtained by dividing the rate of absorbency change observed at rapid coagulation condition by that at slow coagulation condition (Reerink and Overbeek, 1954). In soil, measurement of light absorbency has been widely used to evaluate CFC of colloids (Heil and Sposito, 1993; Kretzschmar et al., 1997). We have demonstrated that colloid release from soil columns is influenced by soil redox status (Chapter 5), which cannot be simply related to the decementing effect of Fe oxides. Instead, colloid release can be related to changes in major potential determining ions with changes in soil redox potential. In this paper, a different approach will be taken. We will 1) evaluate colloid dispersability by comparing relative colloid stability ratios;

PAGE 128

117 and 2) determine the relation between the relative colloid stability ratio and colloid release in two selected contaminated soils. Materials and Methods Column preparation The two Pb-contaminated soils used in this study were from Montreal, Canada and Tampa, Florida, which were exposed to Pb-battery recycling and concurrent smelting operation in the past. Selected characteristics of the soils are listed in Table 4-1. Soil columns were prepared as previously described (Chapter 4) and were briefly outlined here. 5.0 g of acid-washed sand (20-30 mesh) was placed in the bottom of 60mL columns (13 cm in length x 2.6 cm in diameter), then 30 g air-dried soil was placed on top of the sand layer. This resulted in 2.6x5.2 and x7.8-cm soil columns for the Montreal and Tampa soils, respectively. The soil columns were set vertically and prewetted by pumping deionized distilled water (DDW) into the bottom of the columns until the water level in the columns was above the soil. These columns were then incubated under water-flooded condition for approximately 3, 20 and 80 d for the Tampa soil and 3, 20, 30, 40, 50, and 60 d for the Montreal soil. Then the standing water on top of the soil columns was removed and the columns were inverted after sealing their tops with rubber stoppers. The rubber stoppers on bottom of the columns were then connected to a needle as inlet for influent. Then, 0.01 M CaCl 2 solution was pumped through the columns until Ca concentration in the effluent was approximately equal to that in the influent, i.e. when CaCl 2 completely displaced the soil solution in the column. The soil at this stage was referred to as Ca-saturated. A total of 27 Ca saturated soil columns were

PAGE 129

118 prepared (three replicates for each treatment), 18 of which were used for the column leaching test (two replicates) and nine of which for the water dispersability test. Column leaching test After the soil columns were saturated with Ca, the influent in 18 of the soil columns was switched from 0.01 M CaCl 2 solution to DDW at a pumping rate of 1.20 mL per minute. The effluent from the top of each of the column was collected with a fractionation collector. Concentrations of Fe, Al, and organic C were determined in the unfiltered and filtered (0.2 um) solutions. The differences between the two fractions were considered as colloidal Fe, Al, and organic C. Based on preliminary experiment, colloidal Fe and Al are the major components of colloids for the Tampa soil and organic C are the major component of colloids for the Montreal soil. Thus, colloidal Fe and Al, and colloidal organic C were used to approximate colloid concentrations in these two soils. Cumulative amount of colloids leached after the first 10 pore volumes was calculated. Water dispersability test After the soil columns were saturated with Ca, 9 soil columns (each for each treatment) was unpacked for the water dispersability test. The soil core inside the column was taken out and then thoroughly mixed. 1 .0 g of soil for each treatment was weighed into 16 polycarbonate centrifuge tubes (40-mL) with half containing 40-mL of 0.01 M CaCl 2 and half containing 40-mL of 0.06 M NaCl solution. All tubes were placed in an ultrasonic system (FS28H, Fisher Scientific) for 3 min at room temperature. Preliminary research showed that soil dispersion increased with time initially then leveled off after

PAGE 130

119 reaching a maximum. After 3 minutes of sonification, 90 % of dispersible particles in both soils were dispersed (data not shown). The resulted suspensions were centrifuged (Beckman, Model J2-21) at 1000 rpm for 2 minutes at 20 °C. This process supposedly removes all particles > 2 urn with a particle density of 2.4 g cm" 3 (Tanner and Jackson, 1947). The top portion of the supernatant (-20 mL) in each tube was removed carefully from the tube using a pipette. Those containing the same electrolyte solution (CaCL; or NaCl) were then combined to obtain two stock suspensions (~ 160 mL) for the dispersability test. A procedure to measure absorbency change of a suspension with time was developed by modifying those of Thellier and Sposito (1989), Kretzschmar et al. (1997) and Seta and Karathanasis (1997). A total of 14 polystyrene test tubes (13-mL) were used for each stock suspension. After filled with the stock suspension, these tubes were capped. A subsample of 2.0 mL was removed at 2 cm below the surface from two test tubes (duplicate) after 2, 4, 6, 8 , 10, 20 and 24 h. The light absorbency of the subsamples was then read immediately at 525 nm on a Shimadzu UV-160 Spectrophotometer. All procedures were conducted under N2 atmosphere. Estimation of relative colloid stability ratio (RW) Reerink and Overbeek (1954) determined the rate of coagulation of silver iodide particles from light absorbency. In a short period of time (-15 min), light scattering increases as particle sizes increases due to aggregation, and so does the reading of absorbency. This process can be described as follows based on the Rayleigh Law (Overbeek, 1952):

PAGE 131

120 I 0 =eV 0 2 Co (6-2) It = eZVi (t) 2 Cj(t) (6-3) Where I 0 , V 0 and C 0 are initial absorbency, average volume per particle and particle concentration, respectively; I t , Vj(t) and Q (t) are absorbency, volume and concentration of an aggregated particle consisting of i particles at time t, respectively; and e is an optical constant. For soil colloids, however, their aggregation kinetics is much more complicated because of the heterogeneity in colloid composition, mineralogy and size. To our knowledge, using colloid flocculation rate to estimate stability ratio has not been done in literature. To estimate the overall rate of soil colloid flocculation, it is necessary to minimize the complicated details of soil colloid aggregation with time. In our experiment, longer settling time (> 2 h) resulted in large, aggregated particles settling out of the sampling zone in the test tubes via gravity, and thus the absorbency decreases with time. Obviously, the mechanism resulting in the changes in light scattering (the reading of light absorbency) with time in our method is different from the one established by Reerink and Overbeek (1954). However, assuming the size and optical constant of the suspending particles in the sampling zone of the test tube are constant (settling time > 2 h), the basic rationale for the method of Reerink and Overbeek (1954) is valid in our system. In our system, Vj(t) could be treated as a constant V, then Eq (6-3) can be written as Eq (6-4) states that flocculation rate can be obtained by observing the rate change of absorbency with time in a linear stage of flocculation. I t =sV 2 C(t) (6-4)

PAGE 132

121 However, we found that colloid flocculation rate varied with time in a complicated manner, which is similar to those reported by Ferretti et al. (1997). To minimize uncertainty in calculating colloid flocculation rate, we choose a linear region of absorbency. Two types of absorbency-time curves were observed in our study: 1) Absorbency decreased somewhat exponentially with time at initial measurement. The rate was calculated using the slope of the linear region (6-10 h) after an initial sharp decrease in absorbency (Figure 6la); and 2) No distinct "initial drop" period was detectable. The rate was determined by linear regression of the data obtained in the first 8 h of measurement (Figure 6lb). The above procedures yield an apparent flocculation rate in the linear stage. Although it is subject to bias, it is useful to calculate and compare such apparent flocculation rates based on absorbency-time curves of a soil under different conditions. Based on the definition of colloid stability ratio (W) of Reerink and Overbeek (1954), W can be expressed as follows: W oc Co 2 L/( dl,/dt) (6-5) Where dl t /dt is the slope of the absorbency-time curve in a linear stage and L is the length of the light path through the suspension. In addition, W can be estimated using the ratio of the rates of rapid flocculation to slow flocculation. Rapid colloid flocculation rate, in this paper, was measured using a Ca-saturated soil in 0.01 M CaCb electrolyte background solution. Colloids in the Ca-saturated soils were mobilized when the influent was switched from CaCh solution to DDW. However, in the absence of background electrolytes the measurement of the light absorbency in a soil suspension is subject to a

PAGE 133

122 Time Figure 6-1 Schematic representation of typical absorbency-time curves observed. Linear regression was used to calculate an apparent flocculation rate indicated by solid line. The curves are not drawn in the same time and absorbancy scales. Curve "a" stands for those observed for the Montreal soil in 0.06 NaCl M solution, and Curve "b" for those for the Tampa soil in 0.06 M NaCl and in 0.01 M CaCl 2 solutions, and the Montreal soil in 0.01 M CaCl 2 solution.

PAGE 134

123 great uncertainty. Therefore, 0.06 M NaCl solution was used as background electrolytes. Slow colloid flocculation rate was thus measured using Ca-saturated soil in 0.06 M NaCl background electrolyte solution. Same ionic strength in both background solutions was designed to minimize the effect of soil mineral dissolution caused by electrostatic interaction between solution and mineral on colloid dispersion. This effect has to be considered here since the residence time of solution is much longer than that in column leaching. Therefore, for a soil incubated after a given time, its relative W (RW) is defined as RW = W Na AV Ca oc {(dI,/dt)ca(C 0 ) 2 Na}/{(dIt/dt) Na (C 0 ) 2 Ca } (6-6) Where the subscripts Na and Ca are the measurement taken in 0.06 NaCl and 0.01 CaCb M solutions, respectively. Combining Eq (4-2) and (4-6) yields RW oc {(dI t /dt) Ca (Io) 2 Na}/{(dI,/dt) Na (I 0 ) 2 C a (6-7) Eq (6-7) was used in this paper to estimate RW of a soil incubated after a given time in 0. 06 M NaCl solution relative to in 0.01 M CaCb solution. Based on the definition above, RW's relative magnitude represents the colloid's tendency to disperse in a soil, 1. e. a larger number indicates greater tendency to be dispersed. Result and Discussion Figure 6-2 is a typical example of absorbency-time curves observed for the Ca-saturated Tampa soil in 0.06 M NaCl background electrolyte solution. The overall trend of these curves is similar to Figure 6lb. Therefore, the slope, (dI,/dt)Na and initial absorbency (Io) Na were determined from the initial linear stage (2-8 h). Table 6-2 lists all the parameters

PAGE 135

124 77 0 o § o A 9 3d 20 d • 30 d A 40 d 50 d O 60 d A A J I L. -I I I I I i I , 10 15 Time (h) 20 25 Figure 6-2 Absorbency-time curves observed in Ca-saturated Tampa soil suspended in 0.06MNaCl solution.

PAGE 136

125 Table 6-2. Relative colloid stability ratios (RW) for two soils under different waterflooding time Incubation time (d) and redox potential monitored (d) Tampa Soil Montreal Soil RW & 3 20 30 40 50 60 3 20 80 parameters 240 180 100 40 -20 -240 230 170 90 -(dI/dt) Ca 0.008 0.036 0.047 0.020 0.061 0.025 0.014 0.016 0.011 (It)Na 0.40 0.39 0.58 0.46 0.62 0.79 0.06 0.09 0.10 -(dI/dt) Na 0.021 0.009 0.020 0.026 0.041 0.05 0.001 0.008 0.015 RW 0.60 17.6 2.98 0.57 1.34 2.70 6.93 1.24 0.86

PAGE 137

needed to calculate RW via Eq (6-7) and RW values. Clearly, RW varied from 0.5 to 17.7 and 0.86 to 6.9 for the Tampa and Montreal soils with incubation time, and reached a maximum at 20 d and 3 d, respectively (Table 6-2). This is consistent with our earlier findings (Chapter 5) that incubation significantly influences colloid release from soils. The basic mechanism for dispersion of the Ca-saturated soil in the Na solution was that Ca in the Stern layer and in the diffuse layer was displaced by Na in bulk solution, leading to electrical repulsion as a result of the EDL expansion (Roy and Dzombk, 1996b). As more Na replaced Ca on the exchangeable sites, the soil became more dispersed, i.e. more colloids were released. As a result, RW, a measure of relative colloid stability, will increase with an increase in Na concentrations on the exchangeable sites. This ion exchange process can be described by the Gaines-Thomas equation: K Na -ca ={[Na + ] 2 E C aX2}/{[Ca 2+ ]E 2 NaX } (6-8) Where Kwa-ca is Gaines-Thomas selectivity coefficient, [Na + ] and [Ca 2+ ] represent Na and Ca molar concentrations, and EcaX2 and En 3 x are the equivalent fractions of Ca 2+ and Na + in the exchange phase of soil particles which is denoted here as X. Larger Kwa-ca value indicates greater selectivity of soil exchangeable sites for Ca. Since only 1 g of Casaturated soil was mixed with 40 mL of 0.06 M NaCl solution, it is expected that [Na + ] »[Ca 2+ ], Eq (4-8) can be rewritten as (McBride, 1994): KNa-Ca = {EcaX2/E 2 Nax}/{mNa/mc a }M T . (6-9) Where m Na and m Ca are moles of Ca 2+ and Na + in the solution phase and M T is the total molarity of the solution. In our experiment, M T can be considered a constant. Thus, significant increase in RW (Table 6-2) at 20 d and 3 d incubation for the two soils was 113

PAGE 138

127 most likely caused by the decrease in E Ca x2/E 2 NaxTherefore, Eq. (6-9) suggests that K Na . ca cannot be a constant under different incubation times especially when there is significant change in the ratio of E Ca x2/ E 2 Na xThis is inferred by the fact that the decrease in the ratio of E Ca x2/ E 2 n 3 x cannot be compensated by the increase in the ratio of m Na /m Ca on the right side of Eq. (6-9). It is possible that significant decrease in the selectivity coefficient K Na -c a led to increase in RW for the two soils under different incubation times. Thus, it seemed that RW and KN a -c a were inversely related in this experiment. In fact, K.N a -c a is influenced by many factors such as pH, ionic strength, and exchangeable site geometry (McBride, 1994). However, to our knowledge, there is no information available about how incubation time affects KN a -c a in a soil. In addition, the heterogeneity of exchangeable sites in soil colloids make it difficult to describe the characteristics of soil particles when K.N a -c a was low. In the stability test we actually examined only a small portion of the soil colloids in a test tube, the heterogeneity of the soil colloids thus must be taken into account. However, with the current knowledge it is difficult to further discuss this topic with confidence. Nevertheless, high value of RW for both soils may be due to the low value of KiM a -c a , which resulted from the water-flooding incubation. In Chapter 5, we suggested that the surface properties of colloid particles may be altered with incubation, and thus changing colloid mobility. This is consistent with the result we present here conceptually. Cumulative mobile colloids in the first 10 pore volumes in the two contaminated soils under different incubation times were plotted against its corresponding relative colloid stability ratios (Figure 6-3). Since organic matter, and Al and Fe are the major

PAGE 139

350 300 — 250 CD E ^200 "D SB i 3 CD 3 2.5 I * 1.5 1 0.5 0 _ G Al +Fe (Montreal) • OM (Tampa) 0 2 4 6 8 10 12 14 16 18 Relative stability ratio (RW) Figure 4-3. Relation of cumulative colloids and relative stability ratio (RW)

PAGE 140

129 components of the mobile colloids in the Tampa and Montreal soil, respectively, they were used to approximate the mobile colloids released from these two soils. Generally, RW was positively related to the amount of cumulative colloids (~ 10 pore volumes) in , leachates for both soils (Figure 6-3). Interestingly, though the characteristics of the two soils were significantly different (Table 6-1) and colloid composition in leachates were different (inorganic Al /Fe vs. organic matter), the relation between cumulative colloids and R W for the two soils was similar. It is possible that the decreases in the amount of cumulative colloid with WR for the two soils in Figure 6-3 was caused by transition from dispersability control to size straining control of existing mobile colloid as RW increased. Our results suggest that RW may be used to predict overall colloid release in soils. However, theoretically, colloid release is determined by the magnitude of the energy barrier (max-<|)min) (Figure 2-1). Using hematite colloids and quartz grains, Ryan and Gschwend (1994a) found that both colloid release rates and the magnitude of the energy barriers decreased as ionic strength increased under rapid colloid release condition. They suggested that rapid colloid release is controlled by transport of detached colloids to the bulk solution instead of detachment itself since the energy barrier has vanished from the potential energy profile due to changes in solution chemistry. A similar phenomenon has been observed by Jacobsen et al. (1997) using intact soil column experiments in which the natural particles were leached with tap water. Their results showed that particle release from the columns is not affected by an increase in flow rate. Further, their plot of accumulated amount of released particles versus square root of time show a fairly linear

PAGE 141

130 relation, implying diffusion-limited kinetics is a dominant process. One condition for this to occur is that the maximum energy barrier of mobile colloids must vanish. This may be very common in soil, which is consistent with the transient phenomenon we discussed in Chapter 2. Under such a condition, the existing colloid concentration at a given residence time and water flow condition in a soil is controlled by colloid attachment rate, i.e., the size of the energy barrier (§ max ). Therefore from this point of view, it can be understood that the cumulative amount of released colloids in leachates can be well related to RW, even though RW is only a measure of the magnitude of energy barrier not the depth of the energy well. In fact, there has been considerable effort to use colloid dispersability to predict colloid release in soil. Kaplan et al. (1996) has shown that CFC of soil colloids predicts colloid mobility reasonably well unless size straining is dominant in controlling colloid deposition, which are consistent with those reported by Seta and Karathanasis (1997) and in the result presented here. However, different from CFC, the RW we defined in this paper included kinetic information of soil colloid flocculation when the system is subject to a perturbation in solution chemistry. In general, higher CFC indicates greater colloid stability and thus greater absorbency of suspension at a given condition. However, CFC failed to explain the colloid release in our leaching test. For example, the Tampa soil incubated after 60 d has the highest readings of absorbency, i.e. higher tendency for dispersion (Figure 4-2), however, its cumulative colloids was almost the lowest (Table 62).

PAGE 142

131 Conclusion The release of soil colloids into groundwater has attracted considerable attention because of the possibility of their association with insoluble contaminants. Much effort was spent to use the parameters derived from colloid deposition to predict the overall release of colloids in a soil. However, based on modern colloid theory, colloid deposition and release are different processes and thus have different dependence on the interaction between surfaces. In this chapter, we defined and estimated relative colloid stability ratio of a Ca-saturated soil from absorbency-time curves in 0.01 CaCb and 0.06 NaCl M solutions. The rapid and slow flocculation were defined as those in 0.01 CaCh and 0.06 NaCl M solutions, respectively. These are consistent with the condition we used in the leaching test to some extent. The dispersion of a Ca-saturated soil in 0.06 NaCl M solution was mainly caused by cation exchange between Ca in exchangeable phase and Na in bulk solution. We inferred that the significant increase in stability ratio for a soil with incubation was probably caused by a decrease in Kwa-caHowever, the mechanistic understanding is missing. Interestingly, a very obvious relation was observed that overall release of colloids from the soils increased as the stability ratio increased, and the overall trend is similar for Tampa and Montreal soils. This result suggests that the overall release of colloids from these soils could be evaluated using the relative stability ratio qualitatively, even though they are different conceptually.

PAGE 143

CHAPTER 7 CONCLUSION Metal solubility and mobility in soil Our data showed that enhanced Pb solubility was only observed at low Fe partition index for a Pb-spiked sandy soil. Such a trend was also observed using two published data sets in the literature. Since data used in this paper were from various soil environments (roadside sediment in France, Pb contaminated soil in Poland, a sandy soil from Florida), it may be reasonable to conclude that Pb solubility may be related to Fe partition index in soils. For the two Pb-contaminated soils, metal solubility was examined by analyzing the pore waters of the incubated soil columns whereas metal mobility was examined by leaching the columns with 0.01 M CaCb solution. The data showed that metal solubility in pore water and mobility with CaCb solution was not always directly related. There are sufficient evidence that increases in CEC occur with Fe reduction dissolution. However metal redistribution between solution and exchange sites with incubation is determined not only by the magnitude of CEC, but also the characteristics of metal ions, competing co-ions, and counter-ions (ligands). State of Colloid Deposition and Release in Soil and Their Association with Heavy Metals Soil has distinctively different features in colloid mobility compared with welldefined porous media or subsurface systems. Soil colloids are also exposed to more dynamic solution chemistry, it is therefore more difficult to predict the charges of soil 132

PAGE 144

133 colloids since it is determined by the interactions between solid surfaces and solution. In general, soil particles are assemblages of crystalline and amorphous minerals, and organic residuals. For such particles, the interface of soil particle-solution may be more practical to be described by diffuse double layer with a specific part, i.e., a charged surface, and a generic part, i.e., a diffuse layer. The surface charge development process may be described by partitioning of charged species between solution and solid surfaces assuming the charged species are similar. Similarly, association of heavy metals with colloids may have significant influences on colloid mobility, which is subject to dynamic changes in solution chemistry. However, very limited information is available at the present time. Different from clean porous media, the surfaces of soil stationary phase consist of deposited colloids. Consequently, the blocking effect of colloid deposition in soil is significant, which may greatly enhance colloid mobility in soil compared to that in a comparable clean porous medium. Under such a condition, size straining is dominant mechanism in soil colloid deposition. Phenomenologically, this process is timedependent or condition-dependent, which is referred to as transient phenomena in colloid deposition. At the present time, water dispersible clay extracted from a soil is widely used to indicate colloid mobility. However, it has to be realized that there are two simultaneous, opposite processes: colloid deposition and release. In a given system, the relative effect of each on either soil clay dispersion in batch test or colloid mobility in a column experiment depends-on colloid residence time. Despite shortcomings of the DLVO (Derjaguin-Landau-Verwey-Overbeek) theory, it provides a basic theoretical framework in describing soil colloid dispersion,

PAGE 145

134 deposition and release, which is simple and applicable conceptually. It has to be admitted that the whole modern colloid theory (not only the DLVO theory) has found serious difficulty in applying it to soil because of heterogeneity in both structure and composition of soil particle surfaces. To determine the potential mobility of soil colloid in a soil, a phenomenological approach may be more practical at the present time. Soil colloids possess high concentrations of exchangeable ions in their EDLs (high CEC). It has been long recognized that colloid release in soil is coupled by significant ion transfer between EDLs and bulk solution. Potential colloid mobility may be estimated qualitatively by evaluating ion transfer. Colloidal metal mobility in contaminated soils Our results showed that colloid mobility varied greatly with incubation. For the Montreal soil, it decreased with incubation, and was directly correlated to dissolved Fe concentrations and inversely to Ca concentrations in pore water. This cannot be understood by either the decementing effect from reducing dissolution of Fe oxides and charge heterogeneity rising from Fe oxides. Instead, it is possible that the decrease in colloid mobility with incubation was probably caused by an increase in positivelycharged patches of carbonate surfaces, which is determined by Ca 2+ concentration. In the Montreal soil, a pattern of colloidal Fe and Al elution was observed. In general, a peak colloid release was accompanied by a greater dissolved Ca concentration. Three possible mechanisms, i.e., bulk solution displacement, diffusion out of EDL and desorption from surfaces, were proposed to have played a major role in a time sequence in Ca release. These mechanisms may explain the pattern in changes of colloidal Al and Fe concentration. Colloids particles can facilitate Pb transport; however, this more

PAGE 146

135 reduced with water-flooded incubation for this soil. It is possible that Pb redistribution from Fe oxides to carbonates occurred during incubation. In addition, we defined and estimated the relative colloid stability ratio of a Casaturated, incubated soil from the absorbency-time curves in 0.01 M CaCb and 0.06 M NaCl solutions. -Rapid and slow flocculations were defined as those in 0.01 M CaCb and 0.06 M NaCl solutions, respectively. These are consistent with the condition used in the leaching test. The dispersion of a Ca-saturated soil in 0.06 M NaCl was mainly caused by cation exchange between Ca in exchange sites and Na in bulk solution. -Significant increase in the stability ratio with incubation in a soil was probably caused by a decrease in the selectivity coefficient Kwa-caInterestingly, the overall release of colloid from the soils increased as the stability ratio increased, and the overall trend is similar for both Tampa and Montreal soils. These data suggest that the overall release of colloids from these soils may be evaluated qualitatively using the relative stability ratio, even though they are different conceptually.

PAGE 147

REFERENCES Adamczyk, Z, T. Dabros, J. Czanechi, and T.G.M. Van De Ven. 1983. Particle transfer to solid surfaces. Adv. Coll. Int. Sci. 19: 183-252. Adamczyk, Z., J. Czarnecki, and P. Warszynski. 1986. The effect of fluctuations of the energy barrier on colloid stability. J. Coll. Int. Sci. 106: 299-304. Ainsworth,C.C, L.P. James, L.G. Paul, and G.V.D.S. William. 1994. Cobalt, cadmium, and lead sorption to hydrous iron oxides: residence time effect. Soil Sci. Soc. Am. J. 58: 1615-1623. Allison, J.D., D.S. Brown, and K.J. Novo-Gradac. 1991. Minteqa/Prodefa2, A Geochemical Assessment Model For Environmental Systems: Version 3.0 User's Manual. Amrhein, C, P.A. Mosher, and J.E. Strong. 1993. Colloid-assisted transport of trace metals in roadside soils receiving deicing salts. Soil Sci. Soc. Am. J. 57: 12121217. Amrhein, C, P. A. Mosher, J. E., Strong, and P.G. Pacheco. 1994. Trace metal solubility in soils and waters receiving deicing salts. J. Environ. Qual. 23: 219-227. Anderson, S.J., and G. Sposito. 1991. Cesium adsorption methods for measuring accessible structural surface charge. Soil Sci. Soc. Am. J. 55: 1569-1576. Ardizzone, S., L. Formaro, and J. Lyklema. 1982. Adsorption from mixtures containing monoand bivalent cations on insoluble oxides and a reversible oxides and a reversion of the interpretation of points of zero charge obtained by titration. J. Electroanal.chem.. 133: 147-156. Arias,M., M.T. Barral, and F. Diaz-Fierros. 1995. Effects of iron and aluminium oxides on the colloidal and surface properties of kaolin. Clay and clay minerals. 43: 406-416. Balistrier, L., P. G. Brewer and J. W. Murray. 1981. Scavenging residence times of trace metals and surface chemistry of sinking particles in the deep ocean. Deep-Sea Res. 28A: 101-121. Balistrieri, L.S., and J.W. Murry. 1983. Metal-solid interactions in the marine environment: estimating apparent equilibrium binding constants. Geochim. Cosm. Acta. 47: 1091-1098. 136

PAGE 148

137 Bartoli, F., R. Philippy, M. Moirisse, S. Niquet, and M. Dubuit. 1991. Structure and self-similarity in silty and sandy soils: the fractal approach. J. Soil. Sci. 42: 167185. Belzile, N., and A. Tessier. 1989. Interractions between arsenic and iron oxyhydroxides in lacustrine sediments. Geochim. Cosm. Acta. 54:103-109. Benjamin, MM, and J.O. Leckie. 1981. Multiple-site adsorption of Cd, Cu, Zn, and Pb on amorphous iron oxyhydroxide. J. Colloid Interface Sci. 79:209-221. Bleam, W.F. 1993. On modeling proton affinity at the oxide/water interface. J. Coll. Int. Sci. 159:312-318. Blesa, M.A. 1988. The metal oxide-electrolyte solution interface revised. Adv. Coll. Int. Sci. 28: 111-134. Blesa, M.A., G. Magaz, J. A. Salfity, and A.D. Weisz. 1997. Structure and reactivity of colloidal metal oxide particles immersed in water. Solid state ionics. 101-103: 1235-1241. Blesa, M.A., A.J. Maroto, and A.E. Regazzoni. 1990. Surface acidity of metal oxides immersed in water: a critical analysis of thermodynamic data. J. Coll. Int. Sci. 140: 287-290. Boul, S. W., F.D. Hole, R.J. McCracken, and R.J. Southard. 1997. Soil genesis and classification. PI 16-1 18. Iowa State University Press, Ames, Iowa. Buffle, J., and G.G. Leppard. 1995. Characterization of aquatic colloids and macromolecules. 1 . Structure and behavior of colloidal material. Environ. Sci. Technol. 29:2169-2175. Cappellen, P.V., L.Charlet, L., W. Stumm, and P. Wersin. 1993. A surface complexation model of the carbonate mineral-aqueous solution interface. Geochim. Cosmochim. Acta. 57: 3505-3518. Carstea, D.D., M.E. Harward, and E.G. Knox. 1970. Comparison of iron and aluminiumhydroxy interlayers in montmorillonite and vermiculite: I. Formation. Proc. Soil Sci. Soc. Am. 34:517-521. Castiho, P., W.J. Chardon, and W. Salomons. 1993. Influence of cattle-manure slurry application on the solubility of cadmium, copper, and zinc in a manured acidic, loamy-sand soil. J. Environ. Qual. 22: 689-697. Cederberg, G.A., R.L. Street, and J.O. Leckie. 1985. A groundwater mass transport & equilibrium chemistry model for multicomponent systems. Water Resources Res. 21: 1095-1104.

PAGE 149

138 Chan, D., J.W. Perram, L.R. White, and T.W. Healy. 1976. Regulation of surface potential at amphoteric surfaces during particle-particle interaction. J. Chem. Soc. Faraday. 71: 1046-1057. Charlet, L., and G. Sposito. 1987. Monovalent adsorption by an Oxisol. Soil Sci. Soc. Am. J. 51: 1155-1160. Charlet, L., P. Wersin, and W. Stumm. 1990. Surface charge of MnC03 and FeC03. Geochim. Cosmochim. Acta. 54: 2329-2336. Chorover, J., and G. Sposito. 1995. Surface charge characteristics of kaolinitic tropical soils. Geochim. Cosm. Acta. 59: 875-884. Chuan, M.C., G.Y. Shu, and J.C. Liu. 1996. Solubility of heavy metals in a contaminated soil: effects of redox potential and pH. Water Air Soil Pollut. 91 : 543-556. Clifford, J. 1975. Properties of water in capillaries and thin film. In "Water: A Comprehensive Treatise" (ed. F. Franks), V 5, Plenum, New York. P75-131. Coleman, N.T., and G.W. Thomas. 1964. Buffer curves of acid clays as affected by the presence of ferric iorn and aluminium. Proc. Soil Sci. Soc. Am. 28: 187-190. Corapcioglu, M.Y., and H. Choi. 1996. Modeling colloid transport in unsaturated porous media and validation with laboratory column data. Water Res. Res. 32: 34373449. Coughlin, B.R., and A.T Stone. 1995. Nonreversible Adsorption of Divalent metal ions (Mnll, Coll, Cull, and Pbll) onto goethite: effects of acidification, Fell addition, and picolinic acid addition. Environ. Sci. Technol. 29: 2445-2455. Cullen, W., and K.J. Reimer. 1989. Arsenic speciation in the environment. Chem. Rev. 89: 713-764. Cummins, A.B., and W.P. Kelley. 1923. The formation of sodium carbonate in soils. Cali. Agric. Exp. Stn. Tech. Paper. 3: 1-35. Davis, J. A. 1984. Complexation of trace metals by adsorbed natural organic material. Geochim. Cosm. Acta. 48:679-691. Davis, J.A., and J.O. Leckie. 1978. Effect of adsorbed complexing ligands on trace metal uptake by hydrous oxides. Environ. Sci. Technol. 12:1309-1315. Day, G.M., and B.T. Hart. 1994. Adsorption of natural organic matter onto goethite. Colloids Surfaces A. 89:1-13. Day, P.R. 1965. Particle fractionation and particle-size analysis. In "Methods of Soil Analysis" (ed. C.A. Black), part 1, P 377-381.

PAGE 150

139 Dao, T.H., and T.I. Lavy. 1978. Extraction of soil solution using a simple centrifugation method for pesticide adsorption-desorption studies. Soil Sci. Soc. Am. J. 42: 375-377. Dorr, H., and K. O. Munnich. 1991. Lead and cesium transport in European forest soils. Water Air Soil Pollut. 57-58: 809-818. Dzombak, D.A., and F.M.M. Morel. 1990. Surface complexation modeling: hydrous ferric oxide. John Wiley & Sons, New York. Elimelech, M. 1992. Predicting collision efficiencies of colloidal particles in porous media. Wat. Res. 26: 1-8. Elimelech, M., J. Gregory, X Jia, and R. Williams. 1995. Particle deposition and aggregation: measurement, modeling and stimulation. Butterworth-Heinemann Ltd. Oxford. Elimelech,M., and C.R. 0,Melia. 1990a. Kinetics of deposition of colloidal particles in porous media. Environ. Sci. Technol. 24:1528-1536. Elimelech, M., and C.R. O'Melia. 1990b. Effect of particle size on collision efficiency in the deposition of Brownia particles with electrostatic energy barriers. Langmuir. 6: 1153-1163. Ferretii, R., J. Zhang, and J. Buffle. 1997. Kinetics of hematite aggregation by polyacrylic acid: effect of polymer molecular weights. Colloids and surfaces A. 121:203-216. Filella, M., and J. Buffle. 1993. Factors controlling the stability of submicron colloids in natural waters. Colloids Surfaces A. 73:255-273. Findlay, A.D., D.W. Thompson, and E. Tipping. 1996. Electrokinetic properties of oxide particles in natural water. Colloids Surfaces A. 1 1 1 :203-212. Fokkink, L.G.J., A. De Keizer, J.M. Kleijn, and J. Lyklema. 1987. Uniformity of electrical double layer on oxides. J. Electroanal.Chem.. 208: 401-403. Fokkink, L.G.J., A. De Keizer, and J. Lyklema. 1989. Temerature dependence of the electrical double layer on oxides: rutile and hematite. J. Coll. Int. Sci. 127: 116131. Frenkle, H., J.O. Goertzen, and J.D. Rhoades. 1978. Effects of clay type and content, exchangeable soldium percentage and electrolyte concentration on clay dispersion and soil hydraulic conductivity. Soil Sci. Soc. Am. J. 42: 32-39. Fuerstanau, M.C., and B.R. Palmer. 1976. Anionic flotation of oxides and silicates. In "Flotation" V 1. (ed. M.C. Fuerstanau), American Institute of Mining, Metallurgical, and Petroleum Engineers, Inc. New York. PI 48196.

PAGE 151

140 Fuller, C.C., J. A. Davis, J.A. Coston, and E. Dixon. 1996. Characterization of metal adsorption variability in a sand and gravel aquifer, Cape Code, Massachusetts, U.S.A. J. Contain. Hydrol. 22:165-187. Gambrell, R.P. 1994. Trace & toxic metals in wetlands-a review. J. Environ. Qual. 23: 883-891. Gambrell, R.P., R.A Khalid,. and W.H. Patrick, Jr. 1991. The effects of pH, redox potential , salinity on metal release from a contaminated sediment. Water Air Soil Pollut. 57-58: 359-367. Gambrell, R.P., R.A. Khalid, and W.H. Patrick, Jr. 1980. Chemical availability of mercury, lead, and zinc in mobile bay sediment suspensions as affected by pH and oxidation-reduction conditions. Environ. Sci. Technol. 1980,14,431-436. Glazovskaya, M.A. 1994. Criteria for classification of soils according to lead-pollution risk. Eurasian Soil Sci. 26: 58-74. Gotoh, H., and W. H. Patrick, Jr. 1974. Transformation of iron in a waterlogged soil as influenced by redox potential and pH. Soil Sci. Soc. Am. J. 38: 66-71. Heil, H., and G. Sposito. 1993. Organic matter role in illite soil colloids flocculation: I. Counter ions and pH. Soil Sci. Soc. Am. J. 57: 1241-1246. Hendershot, W.H., and L.M. Lavkulich. 1983. Effect of sesquioxide coatings on surface charge of standard mineral and soil samples. Soil Sci. Soc. Am. J. 47: 12521260. Heron, G., T.H. Christensen, and J.C. Tjell. 1994. Oxidation capacity of aquifer sediments. Environ. Sci. Technol. 28: 153-158. Heron, G, C. Crouzet, A.C. M. Bourg, and T. H. Christensen. 1994. Speciation of Fe(II) and Fe(III) in contaminated aquifer sediments using chemical extraction techniques. Environ. Sci. Technol. 28: 1698-1705. Herzig, J.P., LeCleerc, D.M., and LeGoff, P. 1970. Flow of suspensions through porous media — applications to deep filtration. Ind. Eng. Chem. 62: 8-35. Holford, I.C.R., and W.H. Patrick Jr. 1979 Effects of reduction and pH changes on phosphate sorption and mobility in an acid soil. Soil Sci. Soc. Am. J. 43: 292297. Honeyman, B.D., and J.O. Lecke. 1997. Macroscopic partitioning coefficients for metal ion adsorption: Proton stoichiometry at pH and adsorption density. In "Geochemical Processes at Mineral Surfaces" (ed. J.A. Davis and K.F. Hayes),. Am. Chem. Soc, Symp. Ser., 323: 162-190. Israelachvili, J. 1992. Intermolecular and surface forces. Academic Press. New York.

PAGE 152

141 Jacobsen, O.H., P. Molddrup, C. Larsen, L. Konnerup, and L.W. Petersen. 1997. J. Hydro. 196: 185-203. Johnson, P.R., N. Sun, and M. Elimlech. 1996. Colloid transport in geochemical heterogeneous porous media: modeling and measurements. Environ. Sci. Technol. 30:3284-3293. Jones, L.H.P., and S.C. Jarvis. 1975. The fate of heavy metals. In "The Chemistry of Soils Processes" (ed. D.J. Greenlans and M.H.B. Hayes), John Wiley & Sons, New York. P593-620. Kallay, N., E. Barouch, and E. Matijevic. 1987. Diffusional detachment of colloidal particles from solid/solution interfaces. Adv. Coll. Int. Sci. 27: 1-42. Kallay, N., B. Biskup, M. Tomic, and E. Matijevic. 1986. Particle adhesion and removal in model systems, X. The effect of electrolytes on particle detachment. J. Coll. Int. Sci. 114:357-362. Kaplan, D.I., P.M. Bertsch, and D.C. Adriano. 1995. Facilitated transport of contaminat metals through an acidified aquifer. Groundwater. 33:708-717. Kaplan, D.I., P.M. Bertsch, and D.C. Adriano. 1996. Mineralogical and physicochemical differences between mobile and nonmobile colloidal phases in reconstructed pedons. Soil Sci. Soc. Am. J. 61: 641-649. Kaplan, D.I., P.M. Bertsch. 1993. Soil-borne mobile colloids as influenced by water flow and organic carbon. Environ. Sci. Technol. 27: 1 193-1200. Kaplan, D.I., P.M. Bertch, and D.C. Adriano. 1997. Mineralogical and physicochemical differences between mobile and nonmobile colloidal phases in reconstructed pedons. Soil Sci. Soc. Am. J. 61:641-649. Kaplan, D.I, M.E. Sumner, P.M. Bertsch, and D.C. Adriano. 1996. Chemical conditions conducive to the release of mobile colloids from ultisol profiles. Soil Sci. Soc. Am. J. 60:269-274. Karczewska, A. 1996. Chemical speciation and fate of selected heavy metals in.soils strongly polluted by copper smelters. In "Geochemical Approaches to Environmental Engineering of Metals" (ed. R. Reuther), Springer, London. P5679. Kretzschmar R., D. Hesterberg, and H. Sticher. 1997. Effects of adsorbed humic acid on surface charge and flocculation of kaolinite. Soil Sci. Soc. Am. J. 61: 101-108. Kretzschmar, R., W.P. Robarge, and A. Amoozegar. 1995. Influence of natural organic matter on colloid transport through saprolite. Water Resour. Res., 3 1 : 435-445.

PAGE 153

142 Kretzschmar, R., and H. Sticher. 1997. Transport of humic-coated iron oxide colloids in a sandy soil: influence of Ca 2+ and trace metals. Environ. Sci. Technol. 3 1 :34973504. Kurbatov, M.H., G.B. Wood, and J.D. Kurbatov. 1951. Isothermal adsorption of cobalt from dilute solutions. J. Phys. Chem. 55: 1170-1182. Laxen,D.P.H. 1985. Trace metal adsorption/coprecipitation on hydrous ferric oxide under realistic conditions. Water Res. 19: 1229-1236. Lee, P., B. Patrick, and J. Touray. 1997. Geochemical behavior and relative mobility of metals (Mn, Cd, Zn and Pb) in recent sediments of a retention pond along the A71 motorway in Solonge, France. Environ. Geology. 32: 142-152. Lindsay, W.L. 1979. Chemical equilibria in soils. John Wiley & Sons, Inc.: New York. Lovely, D.R., and E.J.P. Philips. 1987. Rapid assay for microbial reducible ferric iron in aquatic sediments. App. Environ. Microbiol. 53:1536-1540. Lyklema, J. 1987a. Electrical double layers on oxides: disparate observations and unifying principles. Chem. Ind. 2:741-748. Lyklema, J. 1987b. Points of zero charge in the presence of specific adsorption. J. Coll. Int. Sci. 99: 109-117. Ma, Q.Y., T.J. Logan, and S.J. Traina. 1995. Lead immobilization from aqueous solutions and contaminated soils using phosphate rocks. Environ. Sci. Technol. 29: 1118-1126. Ma, Q.Y., F. Tan, and W.G Harris. 1997. Concentrations and distributions of eleven metals in Florida soils. J. Environ. Qual. 26: 769-775. Masscheleyn, P.H., R.D. Delaune, and W.H. Patrick, Jr. 1991. Effects of redox potential and pH on arsenic speciation and solubility in contaminated soil. Environ. Sci. Technol. 25: 1414-1419. McBride, M.B. 1994. Environmental chemistry of soils. Oxford university press. New York. McCarthy, J.F., and J.M. Zachara. 1989. Subsurface transport of contaminants-mobile colloids in the subsurface environment may alter the transport of contaminants. Environ. Sci. Technol. 23:496-502. McDowell-Boyer, L.M. 1992. Chemical mobilization of micro-sized particles in saturated porous media under steady flow conditions. Environ. Sci. Technol. 26: 586-593.

PAGE 154

143 McDowell-Boyer, L.M., J.R. Hunt, and N. Sitar. 1986. Particle transport through porous media. Water Res. Res. 22:1901-1921. Miller, W.P., and M.K. Baharuddin. 1986. Relationship of soil dispersibility to infiltration and erosion of southeaster soils. Soil Sci. 142: 235-140. Miller, W.P., H. Frenkle, and K.D. Newman. 1990. Flocculation concentration and sodium/calcium exchange of kaolinitic soil clays. Soil Sci. Soc. Am. J. 54: 346351. Mills, W.B., S. Liu, and F.K. Fong. 1991. Literature review and model (COMET) for colloids/metals transport in porous media. Groundwater. 29: 199-208. Murphy, E.M., and J.M. Zachara. 1995. The role of sorbed humic substances on the distribution of organic and inorganic contaminants in groundwater. Geoderma. 67: 103-124. Newman M.E., A.W. Elzerman and B.B. Looney. 1993. Facilitated transport of selected metals in aquifer material packed columns. J. Contam. Hydrol. 14: 233-246. Nieboer, E., and D.H. S. Richardson. 1980. The replacement of the nondescript term " heavy metal by a biologically and chemically significant classification of metal ions. Environ. Pollut. (Series B). 1: 3-26. Nkedi-kizza, P., P.S. C. Roa, R.E.Jessup, and J.M. Davidson. 1982. Ion exchange and diffusive mass transfer during miscible displacement through and aggregated oxisol. Soil Sci. Soc. Am. J. 46: 471-476. Nocito-Gobel, J., and J.E.Tobiason. 1996. Effects of ionic strength on colloid deposition and release. Colloids and Surfaces A. 107:223-231. Onken, B.M., and L.R. Hossner. 1996. Determination of arsenic in soil solution under flooded conditions. Soil Sci. Soc. Am. J. 60:1385-1392. Oss, C.J.V., M.K. Chaudhury, and R.J. Good. 1987. Monopolar surfaces. J. Adv. Colloid Interface Sci.. 88: 35-64. Oss, C.J. V., M.K. Chaudhury, and R.J. Good. 1988. Interfacial Lifshitz-van der Waals and polar interactions in macroscopic systems. Chem. Rev. 88:927-941. Oss, C.J. V., R.F. Giese, and P.M. Costanzo. 1990. DLVO and non-DLVO interactions inhectorite. Clays Clay Minerals. 38:151-159. Oster, J.D., and I. Shainberg. 1979. Echangeable cation hydrolysis and soil weathering as affected by exhangeable sodium. Soil Sci. Soc. Am. J. 43:70-75.

PAGE 155

144 Ouyang, Y., D. Shinde, R.S. Masell, and W. Harris. 1996. Colloid-enhanced transport of chemicals in subsurface environments: a review. Critical Reviews in Environ. Sci. Technol. 26: 189-204. Overbeek, T.Th.G. 1952. Kinetics of flocculation. In " Colloid Science" V. 1. (ed. H.R. Kruyt), Elsevier, Amsterdam. P278-301. Parker, J.C., L.W. Zelazny, S. Sampath, and W.G. Harris. 1979. Critical evaluation of the extension of zero point of charge (ZPC) theory to soil systems. Soil Sci. Soc. Am. J. 43:668-674. Parks, G. A. 1967. Aqueous surface chemistry of oxides and complex oxide minerals. In " Equilibrium concepts in natural water systems" (ed. W. Stumm,.), American Chemical Society, New York. P121-160. Parks, G.A. 1965. The isoelectric points of solid oxides, solid hydroxides, and aqueous hydroxo complex systems. Chem.Rev. 65:177-198. Parks, G.A. 1975a. Adsorption in the marine environment. In "Chemical Oceanography" (ed. Riley and Skirrow), Academic Press. P24 1-308. Parks, G.A., and P.L. De Brun. 1962. The zero point of charge of oxides. J. Phys. Chem. 66:967-973. Par-on, P., and I. Shainberg. 1970. Hydrolysis and decomposition of Namontmorillonite in distilled water. Soil Sci. 109: 241-246. Patrick Jr, W.H., and Khlid. 1974. Phosphate release and sorption by soils and: sediments: effect of aerobic and anaerobic conditions. Science. 186: 53-55. Peterson, M.L., and R. Carpenter. 1983. Biogeochemical processes affecting total arsenic and arsenic species distribution in an intermittently anoxic fjord. Mar. Chem. 12:295-321. Reerink, H., and J. Th. G. Overbeek, 1954. The rate of coagulation as a measure of the stability of silver iodide sols. Discuss. Faraday Soc. 18: 74-84. Rengasamy, P., and J. M. Oades, 1977. Interaction of monomeric and polymeric species of metal ions with clay surfaces. I. Adsorption of iron (III) species. Aust. J. Soil Res. 15,221-233. Risovic, D., and M. Martinis. 1996. Fractal dimensions of suspended particles in seawater. J. Coll. Int. Sci. 182:199-203. Ronen, D., M. Magaritz, U. Weber, A.J. Amiel, and E. Klein. 1992. Characterization of suspended particles collected in groundwater under natural gradient flow conditions. Water Resour. Res. 28: 1279-1291.

PAGE 156

145 Rose, A.W. and G.C. Bianchi-Mosquera. 1993. Adsorption of Cu, Pb, Zn, Co, Ni, and Ag on Geothite and Hematite: A control on metal mobilization from red beds into stratiform copper deposits. Econ. Geol. 88: 1226-1236. Routh, J. and M. Ikramuddin. 1996. Trace-element geochemistry of Onion Creek near Van Stone lead-zinc mine (Washington, USA)chemical analysis and geochemical modeling. Chem. Geology. 133:211-224. Roy, S.B., and D.A. Dzombk. 1996a. Colloid release and transport processes in natural and model porous media. Colloids Surfaces A. 107: 245-262. Roy, S.B., and D.A. Dzombk. 1996b. Na + -Ca 2+ exchange effects in the detachment of latex colloids deposited in glass bead porous media. Colloids Surfaces A. 119: 133-139. Ruckenstein, E., and D.C. Prieve. 1976. Adsorption and description of particles and their chromatographic separation. AICHE J. 22:276-283. Ryan, J.N., and M. Elimelech. 1996. Colloid mobilization and transport in groundwater. Colloids Surfaces A. 1 07: 1 -56. Ryan, J.N., and P.M. Gschwend. 1994a. Effects of ionic strength and flow rate on colloid release: relating kinetics to interface potential energy. J. Coll. Int. Sci. 164:21-34. Ryan, J.N., and P.M. Gschwend. 1994b. Effect of solution chemistry on clay colloid release from an iron oxide-coated aquifer sand. Environ. Sci. Technol. 28:17171726. Schindler, P.W., and W. Stumm. 1987. The surface chemistry of oxides, hydroxides, and oxide minerals. In "Aquatic Surface Chemistry Chemical Processes at the ParticleWater Interface"(ed. W. Stumm), John Wiley & Sons, New York. P83164. Seaman, J.C., Bertsch, P.M., and W.P. Miller. 1995. Chemical controls on colloid generation and transport in a sandy aquifer. Environ. Sci. Technol. 29: 1808-15. Seta, A.K., and A.D. Karathanasis. 1997. Stability and transportability of waterdispersible soil colloids. Soil Sci. Soc. Am. J. 61:604-611. Shainberg, I. 1973. Rate and mechanism of Na-montmorillonite hydrolysis in suspension. Soil Sci. Soc. Am. J. 37: 689-694. Shainberg, L, P.F. Low, and U. Kafkafi. 1974. Electrio-chemistry of soldiummontmorillonite suspension. I. Chemical stability of montmorillonite. Soil Sci. Soc. Am. J. 38: 751-756.

PAGE 157

146 Shainberg, L, J.D. Rhoades, and R.J. Prather. 1981a. Effect of low electrolyte concentration on clay dispersion and hydraulic conductivity of a sodic soil. Soil Sci. Soc. Am. J. 45: 273-277. Shainberg, L, J.D. Rhoades, D.L. Suarez, and R.J. Prather. 1981b. Effect of mineral weathering on clay dispersion and hydraulic conductivity of sodic soils. Soil Sci. Soc. Am. J. 45:287-291. Shainberg, L, M.J. Singer, and P. Janitzky. 1987. Effect of aluminium and iron oxides on hydraulic conductivity of sandy loam soil. Soil Sci. Soc. Am. J. 51: 12831287. Shulepov, S.Y., S.S. Dukhin, and J. Lyklema. 1995. A dynamic approach to colloid particles interaction. J. Coll. Int. Sci. 171:340-350. Small, H. 1995. Hydrodynamic chromatography: A technique for size analysis of colloidal particles. J. Coll. Int. Sci. 48: 147-159. Song, L., P. R. Johnson, and M. Elimelech. 1994. Kinetics of colloid deposition onto heterogeneously charged surfaces in porous media. Environ. Sci. Technol. 28:1164-1171. Sposito, G. 1984. The surface chemistry of soils. Oxford University Press, Inc., New York, NY. Stevenson, F.J. 1982. Humus chemistrygenesis, composition, reactions. P309-336. John Wiley & Sons. New York. Stookey, L.L. 1970. Ferrozine-a new spectrophotometric reagent for iron. Anal. chem. 42:779-781. Stumm, W. 1984. Interpretation and measurement of redox intensity natural waters. Schweiz. Z. Hydrol. 46: 291-296. Suarez, D.L., J.D. Rhoades, R. Lavado, and CM. Grieve. 1984. Effect of pH on saturated hydraulic conductivity and soil dispersion. Soil Sci. Soc. Am. J. 48: 50-55. Sumner, M.E. 1992. The electrical double layer and clay dispersion. In "Soil Crusting Chemical and Physical Processes" (ed. M.E. Sumner and B.A. Stewart ), Lewis, Ann Arbor. PI -32. Sumner, M.E. 1993. Sodic soils: New perspectives. Aust. J. Soil Res. 31:683-750. Sverjensky, D.A. 1993. Physical surface-complexation models for sorption at the mineral-water interface. Nature. 364: 776-780.

PAGE 158

Sverjensky, D.A. 1994. Zero-point-of -charge prediction from cristal chemistry and solvation theory. Geochim. Cosm. Acta. 58: 3123-3129. Swanton, S.W. 1995. Modeling colloid transport in groundwater: the prediction of colloid stability and retention behavior. Adv. Coll. Int. Sci. 54: 129-208. Tanner, C.B., and M.L. Jackson. 1947. Nomographs of sedimentation times for soil particles under gravity or centrifugal acceleration. Soil Sci. Soc. Proc. 12: 60-65. Tejedor-Tejedor, ML, and M.A. Anderson. 1986. "In situ" attenuated total reflection Fourier transform infrared studies of the goethite (cc-FeOOH)-Aqueous solution interface. Langmuir. 2:203-210. Temminghoff, E.J.M., E.A.T.M.V.D.Z. Sjoerd, and A.M.D.H. Frans. 1997. Copper mobility in a copper-contaminated sandy soil as affected by pH and dissolved organic matter. Environ. Sci. Technol. 31:1109-1115. Tessier, A., F. Rapin and R. Carignan. 1985. Trace metals in oxic lake sediments: possible adsorption onto iron oxyhydroxides. Geochim. Cosm.. Acta. 49: 183194. Theis, T.L. and Singer, P.C. 1974. Complexation of iron(II) by organic matter and its effect on iron(II) oxygenation. Environ. Sci. Technol. 8: 569-572. Thellier, C, and G. Sposito. 1989. Influence of electrolyte concentration and exchangeable cations on the flocculation of silver hill illite. Soil Sci. Soc. Am. J. 53: 711-715. Tien, C. 1989. Granular filtration of aersols and hydrosol. Butterworth, Stoneham, MA. Tipping, E., And D. Cooke. 1982. The effects of adsorbed humic substances on the surface charge of goethite (oc-FeOOH) in fresh-waters. Geochim. Cosm. Acta. 46: 75-80. Turner, D.R., M. Whitfield, and A. G. Dickson. 1981. The equilibrium speciation of dissolved components in freshwater and seawater at 25 C and 1 atm pressure. Geochim. Cosm. Acta. 45: 855-881. Vaughan D., And B.G. Ord. 1984. Soil organic matterA perspective on its nature, extraction, turnover and role in soil fertility. In "Soil Organic Matter and Biological Activity" (ed. Vaughan D. And R.E. Malcolm), W. Junk Publishers, Boston. P. 4-28. Verwey, E.J.W., and J.Th. G. Overbeek. 1948. Theory of the stability of lyophobic colloids. Elsevier. Amsterdam. Wan, J., and J.L. Wilson. 1994a. Colloid transport in unsaturated porous media. Water Res. Res. 30: 857-864.

PAGE 159

148 Wan, J., and J.L. Wilson. 1994b. Visualization of the role of the gas-water interface on the fate and transport of colloids in porous media. Water Res. Res. 30: 1 1-23. Williams, D.F., and J.C. Berg. 1991 . The aggregation of colloidal particles at the airwater interface. J. Coll. Int. Sci. 152: 218-229. Wu, W., R.F. Giese, Jr., and C.J. Oss. 1994a. Linkage between ^-potential and electron donacity of charged polar surfaces. 1 . Implications for the mechanism of flocculation of particle suspension with plurivalent counterions. Colloids Surfaces A. 89:241-252. Wu, W., R.F. Giese, Jr., and C.J. Oss. 1994b. Linkage between ^-potential and electron donacity of charged polar surfaces. 2. Repeptization of flocculation caused by plurivalent counterions by means of complexing agents. Colloids Surfaces A. 89: 253-262. Yao, K.M., M.T. Habibian, and C.R. O Melia. 1971. Water and waste water filtration: concepts and applications. Environ. Sci. Technol. 5:1193-1208. Yoon, R.H., T. Salman, and G. Donnay. 1979. Predicting points of zero charge of oxides and hydroxides. J. Coll. Int. Sci. 70: 483-493. ZacharaJ.M., C.E. Cowan, and C.T. Resch. 1991. Sorption of divalent metals on calcite. Geochim. Cosmochim. Acta. 55:1549-1562.

PAGE 160

BIOGRAPHICAL SKETCH Yan Dong was born on September 16, 1962, in Heilongjiang Province, People's Repulic of China. In P.R. China, he obtained his B.S. from the Institute of Light Industry of Qiqiharr in 1985. He majored in synthetic-fibers engineering with a thesis titled "Synthesis of the nonionic surfactant (polyoxyethylenated glyceryl esters of Cio-h alkyl acids) and its application as a textile softener." Then he continued his graduate education in Harbin Institute of Technology, and received an M.S. degree in 1988, majoring in polymer materials with a thesis titled "Modification of wheat pulp paper with poly(vinyl formal) fibers both in laboratory and industry scales." In the following seven years, he worked at Department of Applied Chemistry, the Northeast Forestry University, China, on several projects with widely different subjects, such as wood-polymer composites, selective adsorption of acrylates monomers onto the interfaces, nonaqueous dispersion resin for road paints, wooden materials treatment using rare earth, base-hydrolyzed starch-graft-polyacrylonitrile water absorbent, and mushroom preservation. Meanwhile, he enjoyed teaching courses such as general chemistry, organic chemistry, polymer chemistry, and fine chemicals and its fundamental chemistry. 149

PAGE 161

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Lena Q Ma, Chair Assistant Professor of Soil and Water Science I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ^ R. Dean Rhue Professor of Soil and Water Science I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Willie G. Harris Professor of Soil and Water Science I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Peter Nkedi-Kizza Professor of Soil and Water Science I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosot Timothy Q. Xownsend Assistant Professor of Environmental Science and Engineering

PAGE 162

This dissertation was submitted to the Graduate Faculty of the College of Agriculture and to the Graduate School and was acceptejj as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1999 £ Dean, College of Agriculture Dean, Graduate School

PAGE 163

This dissertation was submitted to the Graduate Faculty of the College of Agriculture and to the Graduate School and was acceptejj as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1999 £ Dean, College of Agriculture Dean, Graduate School