This item is only available as the following downloads:
CHEMODYNAMIC BEHAVIOR OF COMPLEX MIXTURES:
LIQUID-LIQUID PARTITIONING AND SORPTION OF
ORGANIC CONTAMINANTS FROM MIXED SOLVENTS
LINDA S. LEE
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
Linda S. Lee
I want to thank my committee members Professor Suresh Rao, Dean Rhue,
Joe Delfino, Kirk Hatfield and John Zachara for support and helpful guidance
leading to the successful completion of this project. I especially want to thank Dr.
Rao for his continual contribution to both my personal and professional growth. The
exceptional role Dr. Rao has played as my chairman can be best summarized by his
most recently awarded title of Graduate Research Professor.
I thank my colleagues Dr. Arthur Hornsby, Ron Jessup, Lynn Wood, Dr. Mark
Brusseau, Dr. Ken Van Reese, Dr. Sam Traina, Cheryl Bellin, Denie Augustijn, Dong
Ping Dai, Itaru Okuda, and Dr. Peter Nkedi-Kizza for their assistance, support, and
friendship. Special thanks go to Cheryl Bellin for her assistance in the acid/base
titrations, Itaru Okuda for the UNIFAC simulations, Dr. Mary Collins and Dr. Ron
Kuehl for providing numerous subsamples of Webster soil from Iowa, Vicki Neary
for her technical assistance in the completion of the laboratory experiments, and
Candace Biggerstaff for her help in the preparation and submission of my final
It has been a pleasure to be affiliated with the Soil and Water Science
Department at the University of Florida through both employment and education,
and I would like to acknowledge both staff and faculty for their continual support
throughout the past fourteen years. I also thank my family, as well as two dear
friends, Donna English and Dagne Hartman, whose long-suffering and support have
not gone unnoticed, and God for His unfailing grace, love, and guidance. I would
also like to acknowledge the unique inspiration I've received from Dr. Jim Davidson
and Dr. George Bailey.
The financial support I received from Dr. Rao as my major professor and
supervisor, as well as the Subsurface Science Program, United States Department of
Energy through a contract (DE-AC06-76RLO) to Battelle PNL; United States
Environmental Protection Agency through a cooperative agreement (CR-814512);
and the Electric Power Research Institute (contract #RP-2879-7) is gratefully
TABLE OF CONTENTS
LIST OF TABLES ..............................
LIST OF FIGURES ............................................. x
1 INTRODUCTION .........................
Partitioning from Multi-phasic Liquids .............
Sorption from Aqueous Solutions .................
Hydrophobic Organic Compounds (HOCs) ......
Hydrophobic ionogenic organic compounds (HIOCs)
Solubility in Mixed Solvents ..................
Equilibrium Sorption from Mixed Solvents ......
Hydrophobic Organic Chemicals (HOCs) ....
Hydrophobic Ionizable Organic Chemicals
. . . .. 16
. . . .. 16
2 EQUILIBRIUM PARTITIONING OF POLYAROMATIC
HYDROCARBONS FROM ORGANIC IMMISCIBLE LIQUIDS INTO
W ATER ............................................... 20
Application of Raoult's Law for Gasoline,
Motor Oil, and Diesel Fuel .......
Materials and Methods ..................
Batch Equilibration Technique ........
Chromatographic Analysis ............
Results and Discussion .........................
Coal Tar Composition ......................
Tar-Water Partitioning ...............
Analysis of Laboratory Data ..............
Analysis of Literature Data ..............
Predicting Aqueous-Phase PAH Concentrations ...
Coal Tars ...........................
Diesel Fuels .........................
Assessment of Deviations from Ideal Behavior for
for Equilibrium Conditions ...............
Sum mary ...................................
3 COSOLVENT EFFECTS ON SORPTION OF ORGANIC ACIDS BY SOILS
FROM METHANOL/WATER SOLUTIONS .....
Materials and methods ...................
Determination of Ionization Constants ....
pH of Soil Suspensions in Mixed Solvents .
Solubility Experiments ................
Miscible Displacement Experiments ......
Equilibrium Sorption Isotherms .........
pK,' Measurements ..................
Miscible Displacement Studies .........
Batch Equilibration Studies ............
Effect of Solvent Addition .............
Solute-solvent Interactions .............
Desorption Characterisics .............
Estimation of pH by pHx"P ............
4 IMPACT OF pH ON SORPTION OF BENZOIC ACID FROM
METHANOL/WATER SOLUTIONS ......................... 95
Materials and Methods ...........................
Equilibrium Sorption Isotherms .................
Results and Discussion ...........................
Effects of pH on Benzoic Acid Sorption at f,<0.5 ...
Effects of pH'ap on Benzoic Acid Sorption at f,>0.75 .
Effects of pH"PP on PCP Sorption at fc>0.75 .......
Sorption of Neutral Benzoic Acid Relative to Benzoate
Soil-Solution pH'PP ..........................
Effect of pH Treatments ......................
Sorption Domains ...........................
Sum mary .....................................
5 IMPACT OF SOLUTE STRUCTURE AND ORGANIC COSOLVENT ON
THE SORPTION OF CARBOXYLIC ACIDS BY SOILS FROM MIXED
SOLVENTS .......................................... 122
Materials and Methods
Equilibrium Sorption Isotherms .................
Determination of Octanol-Water Partition Coefficients
Results and Discussion ...........................
Sorption of Benzoic Acid in Several Solvent-
W ater Solutions .........................
Sorption of Several Substituted Carboxylic Acids
in Methanol/Water Solutions ..............
Sum mary .....................................
. . 125
6 SUMMARY AND CONCLUSIONS
Complex Mixtures ..........
Liquid -Liquid Partitioning ....
Sorption of Organic Acids .....
A SUPERCOOLED LIQUID SOLUBILITIES ................ 156
B SAMPLE pKa DETERMINATION ...................... 159
REFERENCES ............................................ 164
BIOGRAPHICAL SKETCH .................................. 183
LIST OF TABLES
2-1. Selected physico-chemical properties for the PAHs investigated ...... 28
2-2. Range of properties observed for eight coal tars (EPRI, 1993) .. .... 33
2-3. Maximum Cw values for several PAHs based on the data compiled for eight
coal tars .............................................. 45
3-1. Selected Solute Properties .................................. 73
3-2. Retardation factors for several organic acids in aqueous and methanol
solutions from Eustis Soil ................................... 84
4-1. Cation exchange capacity (CEC) in cmol(+)/kg and elemental analysis in
mg/kg of pH treated Webster soils .......................... 100
4-2. Chemical characteristics of benzoic acid and
pentachlorophenol (PCP) .................................. 101
4-3. Parameters for linear and Freundlich fits to the isotherm data for
benzoic acid sorption as a function of pH and f. ................ 103
5-1. List of various chemical and physical solvent properties ........... 124
5-2. List of various chemical and physical solute properties ............ 125
5-3. Parameters for linear and Freundlich fits to the isotherm data for
benzoic acid in several solvent/water solutions ................ 128
5-4. Parameters for linear and Freundlich fits to the isotherm data for
substituted benzoic acids in methanol/water solutions ............. 129
5-5. The logarithms of the octanol/water partition coefficients
(log Kow) for both the neutral subscriptt n) and ionized
subscriptt i) species of several substituted carboxylic acids ......... 145
LIST OF FIGURES
1-1. Comparison of measured and calculated (Raoult's law) aqueous solubilities
in binary mixtures of benzene-toluene (A) and benzene-octanol (B). Data
from: Sanemesa et al. (1987) ................................. 6
1-2. Measured and predicted sorption of flumetsulam by several soils normalized
to organic carbon content plotted as a function of pH. (Data form Fontaine
et al., 1991) ............................................. 11
1-3. Normalized sorption coefficients for several organic acids plotted as a function
of pH-pKa. [Data from Kukowski (1989) and Jafvert (1990)] ........ 13
2-1. log Kdw values plotted versus log S, for eight PAHs along with the ideal line
(solid line) calculated form Eq. (2-6) for each diesel fuel ........... 29
2-2. Comparison of measured tar-water partition coefficients (Kt) and predictions
based on Raoult's law for ID# 1(A) and ID# 2 (B) coal tars ........ 37
2-3. Comparison of measured tar-water partition coefficients (K.) and predictions
based on Raoult's law, for ID# 3(A) and ID# 4(B) coal tars ......... 38
2-4. Comparison of measured tar-water partition coefficients (K,) and predictions
based on Raoult's law, for ID# 5(A) and ID# 7(B) coal tars ........ 39
2-5 Comparison of measured tar-water partition coefficients (K,) and predictions
based on Raoult's law, for ID# 7N(A) and ID# 9(B) coal tars collected by
EPRI ......................................... ............ 40
2-6. Comparison of measured tar-water partition coefficients (K) reported in the
literature and predictions based on Raoult's law. Literature source as
indicated ............................................... 42
2-7. Comparison of laboratory-measured aqueous-phase concentrations (C,) with
those predicted on the basis of Raoult's law for eight coal tars ........ 44
2-8. Comparison of laboratory-measured aqueous-phase concentrations (C., jg/L)
with those predicted on the basis of Raoult's law for four diesel fuel. .. 48
2-9. Schematic representation of the ideal behavior (Raoult's law) and nonideality
in liquid-liquid partitioning. ................................. 51
2-10. log K, values for several aromatic hydrocarbons resulting from UNIFAC
model calculations and the average log, values experimentally determined by
Cline et al. (1991) plotted against log S, values along with the ideal line based
on Raoult's law .......................................... 52
2-11. log Kdw values for several aromatic hydrocarbons resulting from UNIFAC
model calculations plotted against log S, values along with the ideal line based
on Raoult's law ........ ........... ........................ 54
2-12. Comparison of measured and predicted tar-water partition coefficients for
several PAHs: Raoult's law (solid line) and UNIFAC model (solid
triangle). .............................................. 55
3-1. Schematic representation of cosolvency plots for solutes with a range of log
Kow values ............................ ................... 60
3-2. Example cosolvency curves that may be predicted by the use of various
parameters in Eq. (3-6). ...................................... 68
3-3. Effect of methanol content on the pK,' of benzoic acid and
pentachlorophenol. ....................................... 82
3-4. Solubility (Sb) of benzoic acid in methanol/water solutions ......... .82
3-5. Representative sorption isotherms for (A) pentachlorophenol, (B) benzoic
acid, and (C) dicamba, on Webster soil in various methanol/water
solutions. .............................................. 87
3-6. Measured and predicted sorption by Webster soil of (A) pentachlorophenol,
and (B) benzoic acid as a function of volume fraction methanol (f). . 88
4-1. Retention data for benzoic acid as a function of pHaPP at different methanol
fraction (v/v) by RPLC. .................................... 96
4-2. Representative isotherms for benzoic acid in (A) aqueous solutions; (B)
f,=0.1;and (C) f,=0.9buffered at several pH values .............. 105
4-3. Sorption of benzoic acid by Webster soil buffered at different pH values in
methanol/water solutions of fc<0.5 .................... ..... .107
4-4. Sorption of benzoic acid by Webster soil buffered at different pH values in
methanol/water solutions of f,=0.75,0.8, and 0.9. ............... 108
4-5. Sorption of PCP by Webster soil buffered at different pH values in
methanol/water solutions of fo =0.75 and 1.0 ................... 109
4-6. Sorption data obtained as a function of pH and methanol content, for neutral
benzoic acid and benzoate. ................................ 112
4-7. Isotherm data for benzoic acid on (A) A1203, AI(OH)3, and SAz-1 (pH=8);
and (B) Pahokee muck (pH=7) along with linear and Freundlich fits 117
5-1. Representative isotherms for benzoic acid in (A) acetone/water;
(B) acetonitrile/water; (C) DMSO/water; and (D) 1,4-dioxane/water
solutions. .............................................. 131
5-2. Representative isotherms for (A) anthroic acid; (B) 2-chlorobenzoic acid (C)
2,4-dichlorobenzoic acid; and (D) 2,4,6-trichlorobenzoic acid in various
methanol/water solutions. ................................. 132
5-3. (A) Benzoic acid solubility data where Sb and S, are solubilities in the binary
solution and water, respectively; and (B) benzoic sorption data with Webster
soil in binary mixtures of water and several organic cosolvents as a function
of volume fraction cosolvent (f)). ............................ 134
5-4. Trends in pH'PP of soil-suspensions in binary mixtures of water and several
organic cosolvents. ....................................... 135
5-5. Measured and predicted (Eq. 3-6) sorption of benzoic acid by Webster soil
from (A) acetone/water; (B) acetonitrile/water; (C) DMSO/water; and (D)
1,4-dioxane/water solutions as a function of volume fraction cosolvent (f338
5-6. Normalized sorption coefficients, log (Kb/K,), for the sorption of selected
substituted carboxylic acids by Webster soil as a function of volume fraction
methanol (f,). ....................................... 145
5-7. Correlation between benzoic acid sorption in neat methanol (log KMoH) and
the log Kow values for both the ionized (i) and neutral (n) forms of the
substituted carboxylic acids. ................................ 146
A-1. Schematic representation of the steps involved in the thermodynamic cycle for
producing a hypothetical supercooled liquid from a crystal solute. .... 157
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
CHEMODYNAMIC BEHAVIOR OF COMPLEX MIXTURES:
LIQUID-LIQUID PARTITIONING AND SORPTION OF
ORGANIC CONTAMINANTS FROM MIXED SOLVENTS
Linda S. Lee
Chairman: Dr. P.S.C.Rao
Major Department: Soil and Water Science
Contamination of soils and water at waste disposal sites commonly involves
various combinations of nonpolar or hydrophobic organic chemicals (HOCs) and
hydrophobic ionogenic organic chemicals (HIOCs), as well as mixtures of water and
one or more organic cosolvents (either completely or partially miscible in water).
Emphasis of this work was on understanding the chemodynamics of such complex
mixtures, specifically solubility and sorption. Experimental and theoretical analysis
presented has focused on: (1) liquid-liquid partitioning behavior of aromatic
hydrocarbons between environmentally relevant organic immiscible liquids (OILs)
and water; and (2) the solubility and sorption of HIOCs by soils from completely
miscible organic solvent/water mixtures.
Partition coefficients for several HOCs were either measured or compiled from
the literature for a wide range of OILs (e.g.,gasoline, diesel fuel, motor oil, and coal
tar). The use of the UNIFAC (UNIQUAC Functional Group Activity Coefficient)
model to estimate the likely nonidealities resulting from interactions between
components in these complex OILs is also presented. Both the UNIFAC simulations
and the observed OIL-water partition coefficients suggest that nonideality is
sufficiently small. Thus, the use of Raoult's law convention for activity coefficients
in conjunction with super-cooled liquid solubilities was considered adequate in
assessing the partitioning of HOCs between several OILs and water.
The role of solute hydrophobicity and acidity, solvent type, and pH on the
sorption of organic acids by a surface soil from mixed solvents was investigated.
Predictions of a model that incorporated effects of cosolvent-enhanced solubility and
cosolvent-suppressed speciation were compared to measured data. Sorption of
neutral benzoic acid was observed to decrease with increasing methanol content,
while benzoate sorption increased. Effects of specific solvent and solute properties
were investigated by measuring (1) benzoic acid sorption from additional binary
mixtures of water and cosolvents with a wide range in solvent properties and (2)
sorption of several substituted carboxylic acids from methanol/water solutions. Of the
different solute-solvent combinations investigated, enhanced sorption by soils was only
observed with carboxylic acids in the presence of methanol or dimethylsulfoxide
(DMSO). It was postulated that enhanced sorption resulted from hydrogen-bonding
interactions combined with the formation of heterogeneous solvation shells about the
solute and the sorbent.
Environmental contamination problems at most industrial waste disposal sites
or spill sites commonly involve wastes consisting of complex mixtures of organic and
inorganic chemicals. Complex mixtures are defined here as those systems comprising
multiple organic solutes and multiple solvents. The solute mixtures of interest might
consist of various combinations of nonpolar or hydrophobic organic chemicals
(HOCs) and hydrophobic ionogenic organic chemicals (HIOCs). The solvent may
be a mixture of water and one or more organic cosolvents (either completely or
partially miscible in water). Solvent mixtures of interest may consist of water and
cosolvents in a single, homogeneous liquid phase, or multi-phases that form at least
two distinct liquid phases. The behavior of such mixtures is not well understood
because the primary chemodynamic properties have usually been characterized in
aqueous solutions which are simple in composition relative to many waste mixtures
found at or near disposal/spill sites. Several researchers have made considerable
efforts during the past decade to investigate the primary processes (e.g., solubility,
sorption, transport) governing the environmental dynamics of organic chemicals in
The release and migration of organic constituents from a waste disposal/spill
source will produce a contaminant plume, either in the vadose zone or in the
saturated zone or both. The contaminant plume composition will vary with time and
distance as the plume size increases. For discussing solubility and sorption processes
within the plume, three separate regions may be considered: a near-field region, a
transition zone, and a far-field region. The basis for such a distinction is not the
distance from the contaminant source. Rather, the criterion employed to designate
these regions is the chemistry of the contaminant mixture within the plume as
contrasted to the waste.
In the near-field region, corresponding to the source itself and its immediate
vicinity, the composition and concentrations of most waste constituents are similar
to that in the waste. There are usually two, possibly three, liquid phases in this
region. This would be the case, for example in the vadose zone, at waste disposal
sites where we may find both "dense" and "light" organic immiscible liquids (OILs)
and an aqueous phase as well as a vapor phase. In the transition zone, if it occurs
in the saturated zone, the solution phase is likely to be predominantly a single-phase,
homogeneous liquid made up of water and varying amounts of cosolvents (if they
were present in the near-field region). The concentration of one or more waste
constituents may be so high that approximations based on expected behavior in dilute
aqueous solutions are often found to be inadequate. Finally, the far-field region
corresponds to that region of the plume in which the waste constituents are present
in an aqueous solution. Most of these chemicals will be at concentrations well below
their aqueous solubility limits. During migration of the contaminant plume through
the vadose zone and the saturated zone, chromatographic separation of the waste
constituents occurs due to their different mobilities. Furthermore, dilution resulting
from hydrodynamic dispersion and attenuation resulting from abiotic/biotic
transformations could decrease contaminant concentrations. Thus, high
concentrations of multiple contaminants are less likely to be found as the distance
from the source increases. Nevertheless, it is possible that these contaminant
concentrations may be higher than the standards set by regulatory agencies.
Partitioning from Multi-phasic Liquids
An understanding of solubility (or partitioning) of HOCs from complex OILs
is essential for predicting organic contaminant release from mixtures such as fuels
(e.g., gasoline, diesel, kerosene) and industrial wastes (coal tar, creosote). The
properties of an organic mixture complex only in composition are determined by the
properties of its pure components and their concentrations in the mixture. This
implies that the chemicals of interest behave ideally in the matrix containing them.
Under these conditions, the concentration in the aqueous phase of a chemical is
proportional to the mole fraction of the chemical in the organic phase corresponding
to Raoult's law. With the stated assumptions, the concentrations of a chemical in the
aqueous phase in contact with a complex mixture can be predicted using the
following simplified expression based on Raoult's law:
C, x S (1-1)
where CW is the chemical's concentration (moles/L) in the aqueous phase in
equilibrium with the organic phase, S, is the aqueous solubility (moles/L) of the pure
liquid chemical, and xo is the mole fraction of the chemical in the organic phase. The
derivation of Eq. (1-1) was based on the pure liquid chemical as the standard state.
Many components of interest are solid in their pure form at standard state; however,
Eq. (1-1) can be extended to solid solutes by employing hypothetical super-cooled
liquid solubilities (S, ).
Raoult's law is applicable to a vast number of mixtures of organic chemicals
and its use in predicting aqueous phase concentrations in contact with a complex
organic mixture is invaluable. These mixtures may be considered complex based on
the number of chemicals that constitute the mixture. On the other hand, complexity
of a mixture can be defined by considering how the properties of the mixture deviate
from some "ideal" behavior, regardless of the number of components. The former
view corresponds to a mixture being complex in composition, whereas the latter
implies complexity in behavior. The important point is that a mixture can be complex
in composition without being complex in behavior and vice versa.
In general terms, structurally similar chemicals are likely to form "ideal"
mixtures, and solubility of such mixtures can then be estimated using Raoult's law.
A simple example of the application of Raoult's law is shown in Figure 1-1A for a
mixture of two structurally similar compounds, benzene and toluene. The pure
aqueous compound solubilities of benzene and toluene are 23.1 and 5.60 mmol/L,
respectively. Note that the pure compound solubilities are observed only in the
absence of the second component (i.e., only when xo= 0 or 1). The concentration
of either compound in the mixture is attenuated by the presence of the other. The
excellent agreement between the measured results and those predicted from Raoult's
law (lines) clearly exemplifies the role of mole-fraction on solubility.
In contrast, a mixture of benzene and n-octanol illustrates a system simple in
composition, yet nonideal in behavior. Deviations from Raoult's law assuming ideal
behavior are evident in Figure 1-1B. Such deviations, however, are not surprising
when we consider the dissimilarity in the chemical nature of these two components.
Benzene is a hydrophobic aromatic compound while octanol is an alkane with a polar
functional group (-OH). The two illustrations given in Figure 1-1 were for
compositionally simple mixtures. However, in most environmental scenarios,
mixtures with a much larger number of constituents are of interest.
Deviations from ideal behavior can arise if the activity coefficient of the solute
in the organic phase is not unity and/or if the solute activity in the aqueous phase
is significantly impacted by the presence of other components. A number of
computational schemes are available to estimate various activity coefficients such that
liquid-liquid partitioning for nonideal mixtures can be evaluated. One of the most
frequently used models for this purpose is the UNIFAC (UNIQUAC Functional-
Group Activity Coefficient) model proposed by Prausnitz et al. 1980). This model
is based on the UNIQUAC model (Abrams and Prausnitz, 1975) and the solution-of-
group concept (Wilson and Deal, 1962). In this model, a mixture of different
chemicals is treated as a mixture of functional groups constituting the components
in the mixture. The interactions between functional groups in the mixture and the
likely nonidealities, resulting from such interactions, are calculated in order to
estimate the activity coefficient of a chemical for a specified phase. Calculations
based on the UNIFAC model require the values for group interaction parameters as
well as the mole fraction of each component in the mixture. The interaction
parameters required in the UNIFAC model have been continuously reviewed and
updated since the model was first introduced (Skjold-Jorgensen et al., 1979;
Magnussen et al., 1981; Gmehling et al., 1982; Alameida-Macedo et al., 1983; Hansen
et al., 1991).
A Raoult's Law Prediction
S 20- (Ideal Behavior)
c 10 Benzene
.3 A Toluene
m CO 5 -
0 0.2 0.4 0.6 0.8 1
15 -a 3
N A (D
0.2 0.4 0.6 0.8
Benzene Mole Fraction in the Organic Phase
Comparison of measured and calculated (Raoult's law) aqueous
solubilities in binary mixtures of benzene-toluene (A) and benzene-
octanol (B). Data from Sanemesa et al. (1987).
Sorption from Aqueous Solutions
Most of the available data and theories for predicting sorption and transport
of organic chemicals may be successfully applied to predict contaminant behavior in
the far-field region (i.e., dilute aqueous solutions). The following section will
highlight the information available on equilibrium sorption of organic chemicals
relevant to this dissertation work.
Sorption is one of the dominant processes affecting the mobility of organic
contaminants in soils and groundwater. This process can be conceptualized either
as binding at a two-dimensional interface of the sorbent or as a partitioning into the
three-dimensional bulk of the sorbent. Several methods for estimating the magnitude
of sorption for organic contaminants have been developed based on the chemical and
physical properties of the sorbate, the sorbent, and the solvent.
Hydrophobic Organic Compounds (HOCs)
Equilibrium sorption of hydrophobic organic compounds (HOCs) by soils and
sediments has been successfully predicted in many cases by the "solvophobic theory"
and the use of linear free energy relationships (LFER). Excellent log-log, linear
relationships have been reported between K,, the sorption coefficient normalized to
the fraction of organic carbon (OC) of the sorbent, and Kow, the octanol-water
partition coefficient for several HOCs (c.f.,Dzombak and Luthy, 1984; Karickhoff,
1981; 1984; Kenega and Goring, 1980). Linear relationships have also been found
between log Ko and solute hydrophobic surface area (HSA) (Dzombak and Luthy,
1984; Rao et al., 1985) and solute molecular connectivity (Sabljic, 1984; 1987). The
different slopes and intercepts found in these regression equations are predominantly
determined by the characteristics of a group of compounds (i.e., class, degree of
hydrophobicity, and structure), while the sorbent properties other than OC appear
to have only minor impact in most cases (Karickhoff, 1981, 1984; Schwarzenbach and
Westall, 1985). The equations derived from LFER and experimental data obtained
for only a few sorbents provide reasonable predictions of HOC distribution in diverse
soil-water and sediment-water systems. However, the limitations of the K, concept
have been pointed out by a number of authors (e.g.,Mingelgrin and Gerstl, 1983;
Green and Karickhoff, 1991; Gerstl, 1990). The two main concerns involve the
contribution of adsorption on mineral constituents and the possibility of site-specific
interactions between functional moieties of the solute and the sorbent.
Hydrophobic ionogenic organic compounds (HIOCs)
For hydrophobic, ionogenic organic compounds (HIOCs), several factors (e.g.,
speciation, soil-solution pH, sorbent-surface pH, charge, ionic strength, ionic
composition, multiple solutes) make predicting sorption from a single parameter
difficult due to additional mechanisms that must be considered. Several mechanisms
proposed in the literature for sorption of organic solutes from aqueous solutions
include: hydrophobic interactions; London-van der Waals or dispersion forces;
hydrogen bonding; cation and water bridging; cation and anion exchange; ligand
exchange; protonation; covalent bonding or chemisorption; and interlayer adsorption
(Koskinen and Harper, 1990). Hydrophobic interactions are driven by weak solute-
solvent interactions and the preference of an organic molecule to be near an organic
surface; thus, strong inverse correlations are observed between K, and solubility of
HOCs. London-van der Waals forces result from correlations in the electron
movement between molecules that produce a small net electrostatic attraction.
Although small in magnitude (2-4 kJ/mol), these interactions are additive and have
been found to be significant for the sorption of large neutral polymeric solutes.
Hydrogen bonding interactions involve the electrostatic interaction between
protons and electronegative atoms, and can be stronger than dispersion forces (2-60
kJ/mol) (Kohl and Taylor, 1961; Stumm et al., 1980). Hydrogen-bonding interactions
may occur with both inorganic and organic surfaces, but for soils interactions with
organic matter are more important due to the abundance ofcarbonyl-type functional
groups (Sposito, 1984).
Cation bridging results if a polar organic functional group displaces a water
molecule from the primary hydration shell of an exchangeable cation (i.e.,formation
of an inner-sphere complex), whereas water bridging results when interaction occurs
without displacement of the hydrating water molecules (i.e., outer-sphere
complexation) (Farmer and Russell, 1967). The occurrence of cation bridging versus
water bridging will be a function of the heat of cation hydration, which varies with
cation size and charge (i.e., charge density). For example, water bridging would be
preferred in a Ca+2-saturated sorbent due to its large negative heat of hydration
(AH=-377 kcal/mol) compared to a saturation with K' (AH=-75 kcal/mol) (Bailey
et al., 1968).
Ion exchange involves the exchange of a cation or an anion for another ion
of similar charge at specific binding sites. Cation exchange is of much greater
importance for most soils due to the predominance of negatively charged surfaces.
Similar to cation bridging, but a much stronger interaction, is ligand exchange which
involves the formation of an inner-sphere complex with a structural cation of a soil
mineral (i.e., displacement of either water or hydroxyl molecules from iron or
aluminum oxides)(Stumm et al., 1980; Kummert and Stumm, 1980). Ligand exchange
is commonly believed to be the mechanism responsible for the adsorption of
oxyanions. Likewise, protonation involves the formation of charge-transfer complexes
with protons on mineral surfaces and organic functional groups such as amino and
carbonyl groups. Interlayer adsorption involves the sorption and entrapment of
solute molecules within clay interlayers. From infrared spectroscopic data, Farmer
and Russell (1967) infer that benzoic acid enters the interlayer space as an unionized
monomer, and then the oxygens from both the hydroxyl and carbonyl groups become
coordinated to the interlayer cation.
In many cases, it is difficult to definitively conclude what particular mechanism
is responsible for the observed sorption; however, frequently we can predict the
magnitude of sorption by incorporation a few parameters. For example, on the basis
of an analysis of a large data set for pentachlorophenol (PCP) sorption from aqueous
solutions by several sorbents over a broad pH range, Lee et al. (1990) showed that
equilibrium sorption could be predicted with a knowledge of pH, organic carbon
(OC) content of the soil, and the acid dissociation constant (pK.) for PCP. Their
model for predicting sorption coefficient is:
Koc Koc,An + Koc,i (1 4>)
n (1 + 10PH-PKa)-1
and K is the measured distribution ratio for the sorbed- and solution-phase
concentrations; Ko =(K/OC); OC is the soil organic carbon content (mass fraction);
0, is the fraction of the neutral HIOC; and the subscripts n and i refer to neutral and
ionized species, respectively.
Sorption data compiled from the literature for several other organic acids
could be, in most cases, adequately described by Eq. (1-2). Shown in Figure 1-2 for
example, is reasonable predictions by Eq. (1-2) of OC normalized sorption of the
herbicide flumetsulam compiled from Fontaine et al. (1991) for several soils.
0 1 2 3 4 5 6 7 8 9 10 11 12
Measured and predicted sorption of flumetsulam by several soils
normalized to organic carbon content plotted as a function of pH.
(Data form Fontaine et al., 1991)
Flumetsulam Sorption by Soils
N-SO/ N CH,
(Data from Fontaine et al.; 1991)
I i i ; i
Data compiled from Kukowski (1989) and Jafvert (1990) for sorption of a variety
of organic acids by soils from aqueous solutions are shown in Figure 1-3. To
facilitate viewing of sorption data from different solute-sorbent combinations
simultaneously, the pH scale is referenced to the solute's pK, (i.e., pH-pK,) and
sorption is scaled to the solute's Kn and Ki values as follows: (Kobs Ki)/(Kn Ki).
Values for K, and Ki were estimated in the sorption experiments where pH-pK, was
less than or greater than one (i.e., acid was predominately neutral or ionic,
respectively). Agreement of Eq. (1-2) with the measured data suggests that the
measured bulk soil-solution pH is representative of the pH seen by the solute, and
that K. and K, are additive. Note that this does not infer a particular sorption
mechanism or that the mechanisms for the neutral and ionized species are the same.
For organic bases, sorption is affected by similar factors as for organic acids.
However, ion-exchange has been shown to be the controlling sorption mechanism for
organic bases even at pH values as much as two units greater than the solute pK.
(Zachara et al., 1987, 1990; Ainsworth et al., 1987; Bellin, 1993). Competitive
sorption between compounds has also been observed for organic cations (Zachara
et al., 1987; Felice et al., 1985) In contrast, for HOCs and neutral HIOCs
competition is minimal (Zachara et al., 1987; Karickhoff et al., 1979; Schwarzenbach
and Westall, 1981; Chiou et al., 1983; Maclntyre and deFur, 1985; Rao et al., 1986).
The predominance of ion-exchange in the sorption of organic bases suggests the use
of a sorption coefficient normalized to the cation exchange capacity of the sorbent
as a first approximation, analogous to the use of Ko for describing sorption of HOCs.
-4 -3 -2 -1 0
1 2 3 4 5 -
-4 -3 -2 -1 0
1 2 3
Normalized sorption coefficients for several organic acids plotted as a function of pH-pK,. [Data from
Kukowski (1989) and Jafvert (1990)]
A Muck Soil
__O O 2-4-D
A O 2,4,6-trichlorophenol
(Data from Kukowski, 1989) C3
_ I 1
(Data from Jafvert, 1990) oo
I_ I I 1 1 0 0%0o o '
The effects on solubility and sorption (hence, on transport) of organic
chemicals upon addition of one or more organic cosolvents to an aqueous solution
are defined here as cosolvency. This section will focus on the most significant
interactions affecting solubility and sorption of both HOCs and HIOCS. Such
interactions include solute-cosolvent, cosolvent-cosolvent, and cosolvent-water
interactions for solubility; for sorption, solvent-sorbent interactions must also be
Solubility in Mixed Solvents
The log-linear cosolvency model and the UNIFAC model are among the
theoretical approaches that have been used to examine cosolvent effects on solubility
(Fu and Luthy, 1986a; Pinal et al., 1990). The log-linear cosolvency model
(Yalkowsky and Roseman, 1981) is based on the central assumption that the
logarithm of the solute solubility in a mixed solvent is given by the weighted-average
of the logarithms of solubilities in the component solvents in the mixture; the
weighting coefficient is taken to be the volume fraction of each solvent component.
log Sm- fi log S (1-4)
where S is solubility (mg/L), f is volume fraction of the solvent, and the subscript m
denotes mixed solvent and i the i-th cosolvent. Note that averaging the logarithms
of solubilities is equivalent to averaging the free energies of solution in different
solvents in the mixture.
In many cases the UNIFAC model may be preferred over the log-linear model
because (i) it has a more sound theoretical basis, (ii) activity coefficients in mixtures
can be calculated given only pure component data, and (iii) all possible interactions
among the components in the mixture are explicitly considered. A limitation of the
UNIFAC model, however, is that although the group interaction parameters required
to estimate the solute activity coefficients are continuously reviewed and updated,
their values are not available for a number of systems of interest here. Also, there
are both experimentally-based (Banerjee, 1985; Arbuckle, 1986) and theoretically
based (Pinal, 1988) reasons that limit the applicability of UNIFAC to aqueous
A convenient measure of the impact of a cosolvent on the solubility of an
organic chemical is the cosolvency power (a), which is defined as
a log (1-5)
where the subscripts c and w refer to neat cosolvent and pure water, respectively.
HOC solubility in organic solvents is larger than that in water, thus a > 0. Larger
values of a indicate a greater solubilizing power of the solvent for a specific solute.
Rubino and Yalkowsky (1987a) and Pinal et al. (1990) have shown that a
values can be viewed as being equivalent to hypothetical partition coefficients for the
HOC between a cosolvent and water. Morris et al. (1988) have shown that a values
can be correlated to HOC octanol-water partition coefficient (K,) as follows:
o a log Ko + b (1-6)
where a and b are empirical constants unique for a given cosolvent. Other cosolvent
and solute properties may also be used to estimate a values (Rubino and Yalkowsky,
1987a,b; Morris et al., 1988).
Although both Eq. (1-5) and (1-6) provide useful first-order approximations
of the cosolvency power of a solvent for a solute, measured HOC solubility profiles
in solvent mixtures often exhibit deviations from the expected log-linear behavior
primarily due to solvent-cosolvent interactions. The observed cosolvency in a binary
mixed solvent can be more generally defined as,
log Sb log S, + ac fc (1-7)
where Sb is the solubility in the binary mixture.
Equilibrium Sorption from Mixed Solvents
Hydrophobic Organic Chemicals (HOCs)
A log-linear cosolvency model describing the decrease in sorption of HOCs
with increasing f, in a binary solvent is given by (Rao et al., 1985; Fu and Luthy,
log Kb log K, -a oc f,
where K is the equilibrium sorption coefficient (mL/g), a is an empirical constant
for describing solvent-sorbent interactions, and the subscript b stands for binary
An extensive amount of data has shown that in binary mixed solvents, HOC
solubility increases and sorption decreases in a log-linear manner as the volume
fraction of the organic cosolvent increases (Rao et al., 1985, 1986, 1989, 1990; Nkedi-
Kizza et al., 1985, 1987, 1989; Woodburn et al., 1986; Fu and Luthy, 1986a,b;
Yalkowsky 1985, 1987; Rubino and Yalkowsky, 1985, 1987a,b,c; Walters and
Guiseppi-Ellie, 1988). These experimental findings are consistent with the predictions
of both the UNIFAC model and the log-linear cosolvency model. Also, for the
sorption of HOCs, solvent-solute interactions as described by solubility are found to
predominate such that the impact of solvent-sorbent interactions has been considered
minor. However, for solutes containing specific functional groups, the impact of the
cosolvent on the sorbent may have considerable impact.
Hydrophobic Ionizable Organic Chemicals (HIOCs)
For hydrophobic ionogenic compounds (HIOCs) of environmental interest,
data on solubility, sorption, and transport in mixed solvents are limited. However,
pharmaceutical literature contains solubility data for several drugs spanning a wide
polarity range. Yalkowsky and Roseman (1981) observed that as solute polarity
increases relative to the solvent, the solubilization curves become increasingly more
parabolic in shape until an inverse relationship occurs (i.e.,decreased solubility with
cosolvent additions). Such behavior is explained on the basis of the solute-solute and
The sorption of HIOCs from mixed solvents has received little research
attention to date. For several HIOCs of environmental relevance (log Ko > 1.0),
solubility does increase with increasing f,; thus, a decrease in sorption is expected.
Fu and Luthy (1986b) observed an inverse log-linear behavior in the sorption by
three different soils of naphthol, quinoline, and dichloroaniline in methanol/water
and acetone/water solutions up to 50% by volume. Similar behavior was observed
by Zachara et al. (1986) for quinoline sorption by a natural clay isolate and
montmorillonite in the same binary mixtures. However, for the sorption of an
ionizable fluorescent dye (Rhodamine WT) from binary mixtures of methanol/water
and acetone/water, Soerens and Sabatini (1992) observed adherence to the log-linear
model only for cosolvent fractions less than 30%, while at higher fractions sorption
For hydrophobic, ionogenic organic compounds (HIOCs), several factors (e.g.,
speciation, soil-solution pH, sorbent-surface pH, charge, ionic strength, ionic
composition, multiple solutes) make predicting sorption from a single parameter
difficult due to additional mechanisms that must be considered. As discussed
previously, prediction of HIOC sorption by soils from aqueous solutions is already
complicated due to the potential for a variety of different sorption mechanisms.
Prediction of HIOC sorption from mixed solvents is further confounded by a number
of indirect effects resulting from cosolvent-induced phenomena occurring either in
the solution phase or on the sorbent. For example, for an organic acid in solvents
of low dielectric constants (e.g.,methanol, acetone, dimethylsulfoxide) an alkaline
shift in the solute pKa results in an increase in the fraction of neutral species.
Similar impacts on the ionization of sorbent functional groups and subsequent solute-
sorbent interactions must also be considered. Also, the impact of cosolvent-water
interactions that have been considered negligible in predicting the chemodynamic
behavior of HOCs may become important when assessing the behavior of HIOCs.
In addition, the different propensities of the cosolvent and water to solvate both the
solute and the sorbent will be important in understanding the sorption of HIOCs.
The existence of codisposal sites, implementation of cosolvents in remediation
schemes, and the development of alcohol-based fuels further warrants a better
understanding of the behavior of HIOCs in complex solvent mixtures.
Emphasis of this work was on understanding the solubility and sorption of
HOCs in multi-phasic mixtures, and of HIOCs in complex miscible-solvent/water
mixtures. The liquid-liquid partitioning behavior of aromatic hydrocarbons between
environmentally relevant organic immiscible liquids (OILs) and water was
investigated. The applicability of Raoult's law was assessed by measuring and
compiling partitioning data from several multi-component OILs, and the UNIFAC
model was utilized to estimate the likely nonidealities resulting from interactions
between components in these complex OILs. These results are discussed in Chapter
2. For the partitioning of HIOCs from binary miscible-cosolvent/water mixtures, the
role of solute hydrophobicity and acidity, solvent type, and pH on the sorption of
organic acids by a surface soil from mixed solvents was investigated. These studies
included (1) sorption of several organic acids from methanol/water solutions
(Chapter 3), (2) sorption of benzoic acid and PCP as a function of pH at several
fixed methanol/water compositions (Chapter 4), and (3) benzoic acid sorption from
additional binary mixtures of water and cosolvents with a wide range in solvent
properties, as well as, sorption of several substituted carboxylic acids from
methanol/water solutions (Chapter 5). The observed sorption of these HIOCs was
assessed in terms of cosolvent-enhanced solubility, cosolvent-induced speciation, as
well as specific and nonspecific solvent association mechanisms.
EQUILIBRIUM PARTITIONING OF POLYAROMATIC HYDROCARBONS
FROM ORGANIC IMMISCIBLE LIQUIDS INTO WATER
Environmental contamination problems at most industrial waste disposal sites
or spill sites commonly involve the presence of an immiscible organic phase
constituting a multi-phasic waste with multiple components. Of great concern is the
transport of organic constituents from these wastes resulting in contamination of soil
and water. Near the source of contamination where a separate organic phase is
present, solubility is the primary process controlling the release of organic chemicals
to the aqueous phase. Therefore, an understanding of the solubility (or partitioning)
of polyaromatic hydrocarbons (PAHs) from a complex liquid such as those suggested
is essential in predicting contaminant release.
Over the last few years efforts have been made to measure the partitioning
of PAHs from environmentally relevant organic liquid wastes such as gasoline, motor
oil, diesel fuel, and coal tar. Coal tars are among the most complex organic liquid
wastes and comprise a large number of hydrocarbons spanning a broad spectrum of
molecular weights. The concentrations of individual constituents in coal tars vary
significantly from one manufacturing gas plant (MGP) site to another. The
manufacturing of gas from coal and oil for residential, commercial, and industrial use
in the late 1800s and early 1900s resulted in the production of large amounts of coal
tar wastes. Eng and Menzies (1985) reported that more than 11 billion gallons of
coal tar were generated in the U.S. during the period 1816-1947, but the disposition
of several billion gallons is unknown and remains unaccounted. In many cases, the
wastes were left on-site in pits or containers, placed in near by ponds or lagoons, or
taken to off-site areas for land disposal. Such practices resulted in contamination of
soils and groundwater at most former MGP sites. Hydrophobic organic chemicals
(HOCs) have been detected at former MGP sites, and are of particular concern due
to their potential carcinogenic nature (Guerin, 1978). Several of these compounds
have already been included on the U.S. EPA list of priority pollutants.
In the past, it has often been assumed that concentrations of organic
contaminant in the aqueous phase leaving a coal tar source would be equal to their
corresponding pure-compound aqueous solubilities. This may be a reasonable
estimate if the source of interest was composed of a single contaminant (e.g.,
trichloroethylene, tetrachloroethylene). However, most complex wastes (e.g.,coal tar,
diesel, gasoline) consist of mixtures of contaminants. These mixtures may be
considered complex based on the number of chemicals that constitute the mixture.
On the other hand, complexity of a mixture can be defined by considering how the
properties of the mixture deviate from some "ideal" behavior, regardless of the
number of components. The former view corresponds to a mixture being complex
in composition, whereas the latter implies complexity in behavior. The important
point is that a mixture can be complex in composition without being complex in
behavior and vice versa.
To assess the extent of groundwater contamination and the long-term
environmental impacts from land disposal or spill sites containing multi-phasic
wastes, it is necessary to characterize the total amounts released and the release rates
of HOCs from the waste matrix. The properties of an organic mixture complex only
in composition are determined by the properties of its pure components and their
concentrations in the mixture. This implies that the chemicals of interest behave
ideally in the matrix containing them. Under these conditions Raoult's law would
suggest that the concentration in the aqueous phase of a chemical is proportional to
the mole fraction of the chemical in the organic phase.
This chapter will focus on the use of equilibrium theory to characterize the
total amounts of PAHs released from organic liquid wastes. Coal-tar/water partition
coefficients for several PAHs were measured from several coal tars spanning a wide
range in physical and chemical properties. To estimate aqueous-phase concentrations
of PAHs in equilibrium with coal tar, the utility of applying Raoult's law convention
for activity coefficients in conjunction with supercooled liquid solubilities for PAHs
that are crystalline in their pure form will be assessed. Although the majority of this
chapter is on coal tar wastes, a reassessment of diesel fuel/water and gasoline/water
partitioning data will also be presented including the use of the UNIFAC
(UNIQUAC functional group activity coefficient) model to estimate the likely
nonidealities resulting from interactions between components in these complex
The release of a chemical from an organic liquid phase can be estimated from
a liquid-liquid partition coefficient (Kd) which is defined as
Kd o (2-1)
where Co and C, are the molar concentrations (mol/L) of the chemical of interest
in the organic and aqueous phases at equilibrium, respectively. The partition
coefficients (K) for coal tar, diesel fuel, and gasoline will be designated using
subscripts tw, dw, and gw, respectively.
For liquid-liquid partitioning, thermodynamic equilibrium is defined by the
equality of the chemical potentials in the aqueous and organic phases. This equality,
in conjunction with the choice of pure (liquid) solute as the standard state and the
Raoult's law convention for activity coefficients, results in the following expression
SYo Xw Y (2-2)
where subscripts o and w denote organic and aqueous phases, respectively; xo and
x, are the respective mole fractions of the chemical in the organic and aqueous
phase; Yo* is the activity coefficient of the chemical in the organic phase in
equilibrium with the aqueous phase; and y, is the activity coefficient of the chemical
in the aqueous phase in equilibrium with the organic phase.
From Eq. (2-2), molar concentration of a solute in the aqueous phase (Cw) can
be approximated with the following assumptions: (1) the presence of other
components in the aqueous phase is ignored, i.e., y,' is set equal to the aqueous
phase activity coefficient of the solute in equilibrium with the pure solute (y,); (2)
the solute behaves ideally in the organic phase, i.e.,Yo* is unity; (3) the aqueous mole
fraction solubility (Sx,) of the pure liquid solute is equal to 1/ y; and (4) the
solution is sufficiently dilute (i.e., moles of the solute are small relative to the total
moles of solvent; C = x V and S/ V, = S,w where S, is the aqueous solubility of the
pure liquid solute in moles/L) and V, is the molar volume of water. Application of
these four assumptions yields
Cw xo St (2-3)
Therefore, the partition coefficient (Eq. 2-1) for a solute can be approximated
For mixtures comprising a large number of constituents, each contributing a
small fraction to the total, xo/Co can be replaced by the molar volume (Vo, L/mole)
of the organic phase. The molar volume can then be approximated by the ratio of
the average molecular weight (MW,, g/mole) and density (po, g/L). The resulting
expression for Kd is:
1 (P. / MW.)
Taking logarithms of both sides of Eq. (2-5), it is evident that the inverse relationship
between log Kd and log St results in a unit negative slope and an intercept that is
dependent upon the molar volume of the organic phase (i.e.,MW,/ p):
log Kd -log S log M(2-6)
Derivation of Eq. (2-6) was based on a choice of the pure liquid solute as the
standard state. Most of the PAHs investigated in this study are solids in their pure
form; therefore, the hypothetical supercooled liquid solubilities of the solid solutes
must be employed. The supercooled liquid solubility (S) of a solute at a given
temperature can be calculated directly from the solute's measured heat of fusion
(AHf) and melting point (T) (Yalkowsky, 1980), or alternately can be estimated by
assuming a constant entropy of fusion (ASf=AH/T.) for the PAHs of interest
(Yalkowsky, 1979; Martin et al., 1979) (see Appendix A).
Application of Raoult's Law for Gasoline. Motor Oil, and Diesel Fuel
The utility of the relationship defined by Eq. (2-6) was successfully
demonstrated for several gasolines by Cline et al. (Cline et al., 1991) for several
monocyclic aromatic hydrocarbons (MAHs). Gasoline is composed of several
branched-chain paraffins, cycloparaffins, alkanes, aromatic compounds, and small
amounts of various additives. Results presented by Cline et al. (1991) revealed that
although gasoline is complex in composition, MAH partitioning into water behavior
was essentially ideal. None of these MAHs exhibit crystalline structure in their pure
form which is common to most PAHs. Chen (1993) investigated the applicability of
Raoult's law for the partitioning of MAHs as well as some PAHs from new and used
motor oil. Given the absence of experimental artifacts, nonideality was noted for the
partitioning of MAHs from the new motor oils, whereas, the one PAH investigated
(phenanthrene) partitioning was successfully predicted using Raoult's law and
supercooled liquid solubilities. However, Raoult's law appeared applicable within
a factor-of-four for the partitioning of both MAHs and several PAHs from used
Hagwall (1992) measured the partitioning of several PAHs from diesel fuel
into water and concluded that the use of supercooled liquid solubilities (S.) in
applying Raoult's law was not successful. However, Hagwall (1992) used an
inaccurate estimation of S., resulting in a wrong conclusion regarding the
applicability of Raoult's law. Using the crystal solubilities (Sw) given in Table 2-1
and assuming a constant ASf of 13.5 eu, a much better relationship was observed
between log Kd, and log S,. In Figure 2-1, the measured log Kdw values are plotted
against their log S, for the eight PAHs investigated along with the ideal line (solid
line) calculated from Eq. (2-6) for each diesel fuel using the MW, and p, given by
Hagwall (1992). For most PAHs in all four diesel fuels, the log Kdw values lie near
the ideal line suggesting that the assumption of ideal behavior may be adequate for
describing the partitioning of PAHs from diesel fuels to water. The confidence
intervals (bars) shown in Figure 2-1 were estimated using an error propagation
method (Shoemaker et al., 1980) which incorporates the errors incurred in the
analysis of both the neat fuel and aqueous phase concentrations. Arrowheads reflect
the few cases where the propagated error was larger than the average KdW value as
was the case for anthracene and fluoranthene. Note that both compounds were
present in small quantities in the neat fuel and or analytical problems were
encountered in detecting small aqueous phase concentrations. Several factors other
than nonideal behavior could result in apparent deviations such as analytical
uncertainty in Kw, as well as, errors incurred in the estimations of S, (i.e., reported
S, values and the use of a constant ASf value).
The success in applying Raoult's law for gasolines, diesel fuels, and motor oils
leads to the question of whether ideal behavior can also be assumed for coal tars.
Compared to gasolines, diesel fuels, and motor oils coal tars are even more complex
in composition, especially because over 60% of their constituents are not known.
Gasolines, diesel fuels, and coal tars collected from different sites vary greatly in
their composition, but only a small variance exists in their molecular weights (Cline
et al., 1991; Hagwall, 1992). In contrast, different coal tars exhibit a wide range in
composition, MWo and po (EPRI, 1993). The applicability of Raoult's law to
tar/water partitioning will be assessed as well as the potential for nonideal behavior.
Table 2-1. Selected physico-chemical properties for the PAHs investigated.
Point Weight" S,b
Compound (Co) (g/mole) (mg/L) log SId
Naphthalene 80.2 128.2 32 -3.05
1-methylnaphthalene -22 142.2 27" -3.72e
2-methylnaphthalene 34 142.2 26' -3.62
Acenaphthylene 82 152.2 3.93 -4.02
Acenaphthene 93 154.2 3.42 -3.98
Fluorene 116.5 166.2 1.9 -4.03
Phenanthrene 100 178.2 1.0 -4.5
Anthracene 216.3 178.2 0.07 -4.49
Fluoranthene 107 202 0.27 -5.19
Pyrene 150 202 0.16 -4.85
Chrysene 254 228.2 0.006 -5.29
Benzo(a)anthracene 156 228.2 0.0057 -6.29
Benzo(a)pyrene 179 252 0.0038 -6.28
"Verschuren (1983); b Crystal solubility at 250C (Little, 1981) unless stated otherwise;
' Miller et al. (1985); d Supercooled liquid solubility (moles/L) calculated assuming
a constant ASf for PAHs; e liquid solute at standard state.
- 3 3L
-3.5 -3 -6 -5.5
log [Sscl, moles/L]
log Kdw values plotted versus log S, for eight PAHs along with the ideal line (solid line)
Eq. 2-6 for each diesel fuel.
-5.5 -5 -4.5 -4 -3.5
-5.5 -5 -4.5 -4 -3.5 -3
Materials and Methods
For all the PAHs investigated (see Table 2-1) standards were purchased from
Aldrich Chemical Co. at > 98% purity except for acenaphthene, which was available
only at 85% purity. Methylene chloride, the solvent used for the aqueous phase
extractions, was purchased from Fisher Scientific at Fisher grade Optima.
Batch Equilibration Technique
Approximately 0.3-0.5 g of coal tar were added to a glass centrifuge tube
(nominal volume 40 mL); enough electrolyte solution (0.01 N CaC2) was added such
that no headspace remained; and tubes were closed with phenolic caps fitted with
Teflon-lined septa. Prior to sampling the coal tar for equilibration with an aqueous
phase, coal tars were rotated end-over-end at room temperature (23 + 2"C) for 12-18
hours. The coal tar/water (0.01 N CaCl) mixtures were then equilibrated for 3-7
days in the dark. Preliminary studies where samples were equilibrated for 1, 3, 5,
and 7 days showed no measurable differences in PAH concentrations after 3 days.
Following centrifugation (300 RCF for 30 minutes) of the equilibrated coal tar/water
mixtures, a portion of the aqueous phase (z25 mL) was quantitatively removed for
extraction with methylene chloride and subsequent concentration prior to analysis.
Due to the large masses of the compounds of interest present in the coal tar phase,
experimental artifacts from PAH sorption to the equilibration vessels were
considered negligible. To avoid volatilization losses and contamination of the
aqueous phase aliquot with the coal tar phase, the aqueous aliquot was removed
through the septa using a 50-mL Teflon-backed gas/liquid syringe equipped with a
3-inch needle. The equilibration vessel was vented during sampling by piercing the
septa with a second needle.
Following aqueous phase transfers, as much residual water as possible was
removed from the equilibration vessel without loss of the coal tar. The coal tar in
the equilibration vessel and the cap were rinsed with methylene chloride into a 100-
mL volumetric flask and brought to volume. Dissolved coal tar samples were filtered
(0.45 pm) prior to analysis. For the coal tar samples from which it was difficult to
remove residual water without loss (i.e., thin liquid coal tars), an aliquot of the neat
coal tar was sampled for analysis as well.
PAH concentrations in the coal tar and aqueous phases were determined using
a gas chromatograph (GC) equipped with an ion trap detector (ITD). The GC/ITD
method included an HP Ultra 2 column (95% methyl, 5% phenyl polysiloxane, 0.5
micron thickness; 30 cm x 0.32 mm ID); helium as a carrier gas at a flow rate of
approximately 1.0 Ml/min; temperature gradient program, and an ion trap detector.
The temperature gradient program consisted of a 1 minute hold at 50"C; a ramp to
1300C at 30*C/min followed by a 3 minute hold; a ramp to 1800C at 12*C/min
followed by a 1 minute hold; a ramp to 240 C at 7C/min; and a ramp to 300 C at
12 C/min followed by a 15 minute hold. The ITD was set at an electron energy of
70 eV and scanned from 45 to 450 amu at 2 scans/sec. The electron multiplier
voltage was 1650 volts and the transfer temperature from the GC was 2800C. Prior
to GC analysis, samples were usually spiked with an internal standard consisting of
naphthalene-d8 and anthracene-dg.
Results and Discussion
Coal Tar Composition
The coal tars used in this study were received from META Environmental, Inc.
Various physical and chemical properties of these coal tars had been characterized
(EPRI, 1993), including density, viscosity, water and ash content, average molecular
weight, elemental and organic analysis. The ranges observed for these properties in
terms of percentages or concentrations are summarized in Table 2-2.
The viscosity of the coal tars ranged from approximately 34 cps to 6600 cps
(40C), with the coal tar consistency varying from thin liquids (ID# 1,4, and 5) to
thick liquids (ID# 7) and from soft (ID# 3 and 9) to sticky (ID# 2) "taffy-like"
materials. Coal tar viscosity will generally increase with aging and decrease with
temperature. Some coal tars had high ash contents, suggesting the presence of other
solids. For example, coal tar ID# 7N had a high content (37%) of what appeared
to be sand and silt. The PAH concentrations for this coal tar were corrected to
represent the mass of PAH present per actual mass of coal tar. For the remaining
coal tars an occasional rock or pellet was found, which was easily removed prior to
Water content of the thin liquid coal tars was small (<1% mass basis). For
the more viscous coal tars, reported water contents were as high as 30% (mass basis);
however, high molecular weights and densities for these coal tars strongly suggests
that these high water contents were in actuality a sampling artifact. It appears that
water may have been trapped as a separate liquid phase within the taffy-like matrix
of the coal tar.
Table 2-2. Range of properties observed for eight coal tars (EPRI, 1993).
2 & 3 rings
> 3 rings
34-6,600 cps (400C)
1.06-1.43 g/mL (24C)
< 1-580 mg/kgh
< 1-150 mg/kg'
" Total Organic Carbon; b Test Methods ASTM D445 and D88; c Test Methods
ASTM D70, D369, or D1429; d Average molecular weight determined using vapor
pressure osmometry; e Exception: asphaltene-like tar 1600 g/mole; f Nitrogen
polyaromatic hydrocarbons; g Sulfur polyaromatic hydrocarbons; h Determined using
EPA Method 4500;' Determined using EPA Method 9010.
Similar compounds were found in all of the tars, but individual hydrocarbon
concentrations varied significantly from one MGP site to another. PAH
concentrations ranged from 7,000mg/kg to 220,000mg/kg, with various naphthalenes
as the dominant components. Several monocyclic aromatic hydrocarbons (e.g.,
benzene, toluene, ethylbenzene, and xylenes (BTEX), and styrene) were also present
in concentrations ranging from 13 to 25,300 mg/kg. Much smaller amounts of
nitrogen- and sulfur-containing aromatic hydrocarbons (e.g., carbazole and
dibenzothiophene) were also found.
It is important to recognize that less than 40% (on a mass basis) of the coal
tar constituents can be quantified (see Table 2-2) using common extraction and
chromatographic techniques. The unidentified tar fraction is often referred to as the
"pitch" for operational purposes. Current sophisticated analytical techniques still lack
the capability needed to identify most of the pitch constituents; however, their
general nature may be surmised based on coal composition (e.g.,Whitehurst et al.,
1980) or oil composition. A majority of the pitch constituents are aromatic
compounds with high molecular weights and low aqueous solubilities; thus, they may
not be of direct concern in terms of groundwater contamination. However, the
physical and chemical characteristics of the pitch may exert a strong influence on the
rates of release and the equilibrium partitioning of the more-soluble tar constituents
(e.g., BTEX, naphthalenes) that are of greater environmental concern. Also,
nitrogen- and sulfur-containing aromatic hydrocarbons present in coal tars may
impart nonideal behavior.
The relative success in applying a model based on Raoult's law convention for
gasolines (Cline et al., 1991), diesel fuels, and motor oil prompted the investigation
of whether ideal behavior could also be assumed for coal tars. Compared to
gasolines and diesel fuels, coal tars are compositionally more complex; thus, greater
deviations from ideal behavior might be expected. The assumption of ideal behavior
for coal tar is postulated here for practical expediency, since it reduces the number
of parameters needed to estimate PAH concentrations in groundwater. Ideal
behavior is not necessarily expected for such materials, but it is hoped that the
assumption will be adequate within a specified acceptance factor; a factor-of-two has
been chosen here to be adequate for field-scale applications. Experimental
measurements of tar-water partition coefficients are difficult, and are subject to
significant errors. Thus, experimental artifacts as a possible cause must be
eliminated before attributing nonideal behavior to a given coal tar or even to one or
more constituents within a coal tar. It is with this pragmatic perspective that we will
interpret tar-water partitioning data. The investigations of tar-water partitioning
involved analysis of data collected in this study for eight tars, analysis of published
data, and theoretical analysis of solute-solute interactions that might lead to nonideal
Analysis of Laboratory Data
The tar-water partitioning data for the eight tars examined in this study are
presented in Figures 2-2 through 2-5. The logarithm of the average K, value and
the calculated standard deviations are shown along with the prediction based on Eq.
(2-6) (solid line) and the factor-of-two tolerance intervals. For most coal tars, the
data points are scattered about the ideal line within the factor-of-two bounds
suggesting that the assumption of ideal behavior suffices (again, within a factor-of-two
error) in predicting KI for the PAHs. For the one exception (ID# 1), measured
data points lie consistently above the ideal line (Figure 2-2A) indicative of an error
in the estimate of the molar volume. Specific causes for the systematic deviation
observed with coal tar ID# 1 need to be further explored.
Benzo(a)anthracene is the only PAH that consistently lies substantially below
the ideal line for most of the coal tars. Uncertainties arising from both analysis and
parameter estimation may have resulted in the observed negative deviations.
Analysis of benzo(a)anthracene in the aqueous phase approached detection limits,
thus contributing to uncertainties. A greater source of error was probably incurred
in the estimation of the supercooled liquid solubility for benzo(a)anthracene. The
S, values (given in Table 2-1) used in plotting log K, values in Figures 2-2 through
2-5 were estimated assuming a constant entropy of fusion (ASE)(Yalkowsky, 1979).
For most compounds, this method may be preferred over attempts to find reliable
measured AHf values needed for a direct calculation. However, in the case of
benzo(a)anthracene the S, values estimated using the average ASf value was about
one order of magnitude higher than that calculated using the AHf value reported by
Chio et al. (1985). Thus, the reasons for the observed deviation of
benzo(a)anthracene data points from the ideal line are indeterminate.
-5.5 -5 -4.5
-5.5 -5 -4.5 -4 -3.5 -3
log [S, moles/L]
Comparison of measured tar-water partition coefficients
predictions based on Raoult's law for ID# 1(A) and ID#
-5.5 -5 -4.5 -4 -3.5
log [S, ,moles/L]
Comparison of measured tar-water partition coefficients
predictions based on Raoult's law, for ID# 3(A) and ID#
-5.5 -5 -4.5 -4 -3.5 -3
log [S, moles/L]
Comparison of measured tar-water partition coefficients (K,) and
predictions based on Raoult's law, for ID# 5(A) and ID# 7(B) coal
-4.5 -4 -3.5
-5.5 -5 -4.5 -4 -3.5 -3
log [S, moles/L]
Comparison of measured tar-water partition coefficients
predictions based on Raoult's law, for ID# 7N(A) and ID#
tars collected by EPRI.
Analysis of Literature Data
The tar-water partition coefficients (K,) for several PAHs compiled from the
literature (Rostad, 1985; Groher, 1990; Picel, 1988) for three different coal tars, are
plotted in Figure 2-6 in a manner similar to Figures 2-2 through 2-5. For each coal
tar, the ideal line (solid line) shown was calculated from Eq. (2-6) using the best
estimates available for MW, and p,,. For the coal tar investigated by Rostad et al.
(1985) (Figure 2-6A), the ideal line was calculated using the Pet reported and a MWVt
value estimated from a weighted average of the mole fraction and molecular weight
of each known component. For the unknown fraction, an average molecular weight
of 300 g/mole was assumed. Picel et al. (1988) reported values for both pt and
MW,. Groher (1990) did not report values fore MW, and pc; therefore, data for a
coal tar, similar in composition, obtained from the same site a few years later was
used to estimate the ideal line (Figure 2-6B).
For most of the PAHs, the measured KI values are within a factor-of-two
from the ideal line, with the best agreement observed for the Picel et al. (1988) data
(Figure 2-6C). Observed deviations from the ideal line could be the result of
considerable nonideality in the tar-water system or a consequence of various
experimental artifacts including inadequate time for equilibration and poor recovery
of the PAH from the aqueous phase. The probability of such experimental artifacts
increases for the larger PAHs where a greater difficulty is often encountered in
accurately measuring the solubility of rather insoluble compounds.
Rostad et al., 198
S- dibenz(a,h)anthr cRostad et al., 1985
5.5 benzo(a)anthracene '-..
5 -... ".--.-methylnaphthalen
Ideal Line anthracen~e
MWt = 265 g/mole fluorene-..hthalen
4 p = 1.03 g/mL --
A p acenaphthene
-6.5 -6 -5.5 -5 -4.5 -4 -3.5 -3
7 -... chrysene Groher, 1990
6.5 --- fluoranthene
60 --. anthracene
0) 5~ i 'e U--. -fluorene
O 5 Ideal Line ----.....
4.5 MWct = 230 g/mole acenaphthelne--.
4 P ct = 1.064 g/mL 2-methylnaphtihaen.
-6 -5.5 -5 -4.5 -4 -3.5 -3
6.5 -.. pyrene Picel et al., 1988
6 --- -
s. phenanthrene biphenyl
I -l --... fluorene
5 Ideal Line 1.... __ ... .-- -methylnaphthalene
MWct = 150 g/mole acenaPt e h phtha
4.5 -.. aphthalene
p C = 0.99 g/mL ---.-
4 naP. ne'
3 .5 -3
log [S, moles/L]
Figure 2-6. Comparison of measured tar-water partition coefficients (K,) reported
in the literature and predictions based on Raoult's law. Literature
source as indicated.
A decrease in the measured K, values would be anticipated for deviations
resulting from sufficient nonideality as observed in Figure 2-6A for the Rostad et al.
(1985) data. The expectation of the presence of nonideality resulting in negative
deviations for PAHs is based on work by Chiou and Schmedding (1982) and Chio et
al. (1985) where the activity coefficients of several PAHs were measured in water-
saturated octanol and mixtures of benzene and cyclohexane. In both cases, the
activity coefficient of a given PAH in the organic phase (Yo') were found to be
greater than unity. Values of y., greater than unity will result in log K, values
smaller than those estimated assuming ideal behavior.
Predicting Aqueous-Phase PAH Concentrations
The log Kt, versus log S, relationship observed for several coal tars (Figures
2-2 through 2-5) suggests that the application of Raoult's law and the assumption of
ideal behavior may be adequate to predict the concentration of PAHs in groundwater
(C,) in contact with a coal-tar source. Equation (2-3) was used to estimate the
concentrations of several PAHs expected to be present in a groundwater in
equilibrium with a coal tar were estimated using Eq. (2-3) for the coal tars
investigated. The mole fraction of the PAH in the organic phase (xo) needed in Eq.
(2-3) was approximated by the product of the mass fraction (mg/g) in the coal tar
and MWt, (i.e., C, = Mi MW, St). A log-log plot comparing predicted aqueous
concentrations (converted to commonly reported units of mg/L) and those measured
during the laboratory partitioning studies is shown in Figure 2-7. The error bars
shown in Figure 2-7 for the laboratory-measured concentrations represent the
standard errors calculated from replicate averages. An arrowhead on an error bar
indicates that the lower bound approached the limit of detection. For the predicted
concentrations, the error bars shown in Figure 2-7 were estimated from the standard
errors calculated from the replicate average of Mi. Also given in Figure 2-7 is the
ideal line (i.e., 1:1 correlation) with the corresponding factor-of-two tolerance
-3 -2 -1 0 1 2
Measured [log (C, mg/L)]
Figure 2-7. Comparison of laboratory-measured aqueous-phase concentrations
(C,) with those predicted on the basis of Raoult's law for eight coal
Confidence in the C, values predicted using Eq. (2-3) is dependent on several
factors other than the premise of ideal behavior, including uncertainty about the
input parameters (e.g., Mi, 1MWt, and S). Both Mi and MW, can be determined
experimentally; therefore, errors associated with these parameters can be obtained
from replicate analysis information. Sampling and chromatographic analysis of this
heterogeneous liquid waste is prone to considerable errors; therefore, the deviation
associated with Mi is probably the greatest source of error in estimating C, values.
A majority of the data presented in Figure 2-7 lie within the factor-of-two
intervals given about the 1:1 correlation. The data that lie outside the factor-of-two
intervals result in predicted concentrations greater than those measured, with the
exception of the data points below the 1:1 correlation which correspond to
benzo(a)anthracene. Benzo(a)anthracene is present in small amounts in coal tar, thus
often approaching the limits of analytical detection. In addition, aqueous solubility
measurements for compounds with small values (i.e., <10-2 mg/L) become
increasingly less reliable. Good agreement for a majority of PAHs within a factor-of-
two suggests that the use of Eq. (2-3), based on Raoult's law, may be adequate for
estimating PAH concentrations. At the very least, aqueous concentrations estimated
using this approach should be considered more appropriate and definitive than
merely assuming crystal solubilities for aqueous-phase concentrations.
Given the variations that may exist in (1) the different coal tar deposits at a
given site, and (2) the extent of weathering at that site, it would be advantageous to
estimate maximum PAH concentrations that might be found at any site. In order to
estimate maximum CW values, the eight coal tars investigated were assumed to be
representative of coal tars that might be found at any site in the United States. The
maximum concentrations of the PAHs investigated based on the data compiled for
the eight coal tars, are given in Table 2-3 along with the ratios of C, to S,. Note
that the maximum Cw expected is the crystal aqueous solubility for anthracene,
chrysene, and benzo(a)anthracene.
Table 2-3. Maximum CW values for several PAHs based on the data compiled for
eight coal tars.
Compound SW Maximum
(mg/L) CWa CJ/S
Naphthalene 32 14b 0.44
1-methylnaphthalene 27 2 0.05
2-methylnaphthalene 26 1.4 0.05
Acenapthylene 3.93 0.5 0.13
Acenapthene 3.42 0.3 0.1
Fluorene 1.9 0.3 0.16
Phenanthrene 1.0 0.4 0.3
Anthracene 0.07 SW 1.0
Fluoranthene 0.27 0.01 0.4
Pyrene 0.16 0.1 0.5
Benzo(a)anthracene 0.0057 SW 1.0
Chrysene 0.006 SW 1.0
Benzo(a)pyrene 0.0038 0.001 0.3
Result from data compiled for seven
resulted in a prediction of 26 mg/L.
of the eight coal tars; data for one tar
In the absence of cosolvents and other solubility-enhancing adjuvants (e.g.,
dissolved organic carbon, surfactants, etc.), the maximum aqueous-phase
concentration (C,) is limited by the crystal solubility (Sw). Although the hypothetical
supercooled liquid solubility is used to obtain best estimates for C,, mixing of the
chemical with the aqueous phase is ultimately governed by interactions with the
solvent. These are expressed through the crystal solubility (S,) (Pinal, 1988). For
a PAH that has a low aqueous solubility, high melting point, and is present in high
concentration in the coal tar, the concentration predicted in the aqueous phase
assuming ideal behavior would be the crystal aqueous solubility (S,).
Reasonable agreement shown previously in the predicted and measured log
Kdw versus log S, relationships for most PAHs (Figure 2-1) also supports the use of
Raoult's law in predicting maximum PAH concentrations that may be present in the
aqueous leachate leaving a diesel-fuel contaminated area. Using Raoult's law and
assuming ideal behavior, the concentration of a constituent in the aqueous phase in
equilibrium with the organic phase is proportional to the mole fraction of that
constituent in the organic phase (see Eq. 2-3). Substituting Eq. (2-5) into Eq. (2-1)
gives the following equation for the equilibrium aqueous-phase concentrations:
C C, MW, St (2-7)
where the subscripts df and w refer to diesel fuel and water, respectively. In Figure
2-8, PAH concentrations predicted using eq 2-7 were converted to commonly
reported units (ig/L) and plotted against concentrations measured in the laboratory
partitioning studies with the four diesel fuels.
'i2- A 2 A
0 4 naphthalene
C acenaphthene fluorene
)- 1 a- phenanthrene
-1 0 1 2 3
log [Measured C, ug/L]
Comparison of laboratory-measured aqueous-phase
concentrations (C,, /g/L) with those predicted on the basis of
Raoult's law for four diesel fuels.
Also included in Figure 2-8 are the confidence intervals for both the measured and
predicted concentrations. Measured concentration errors were estimated from the
standard deviations observed in triplicate analyses of the aqueous phase; confidence
intervals with arrows reflect limits of detection. Similarly, the errors associated with
the predicted values were estimated from the standard deviations obtained from
triplicate analyses of the neat diesel fuel, i.e., the determination of Cdf. The
confidence intervals given for the predicted C, in Figure 2-8 did not include errors
incurred in estimating MWdf or pdf. Overall, the correspondence between measured
and predicted equilibrium aqueous phase concentrations shown in Figure 2-8 is to
be very good.
Assessment of Deviations from Ideal Behavior for Equilibrium Conditions
The relationship between Kd and S, assumed previously (el 2-6) was based on
the simplifying assumption of ideal behavior (i.e., yo = 1 and yw = yw). Several
factors may cause deviations from the assumed ideal behavior for diesel-water
partitioning of PAHs. For example, negative deviations from the ideal line could
result from the presence of surfactants or emulsions or sufficient nonideality, while
positive deviations can be expected if equilibrium has not been reached, and
apparent deviations (positive or negative) can result from uncertainty in parameter
For a mixture which is complex in composition and behaves in a "nonideal"
fashion, the partition coefficient (Kd) between an organic liquid and an aqueous
phase can be related to the aqueous solubility of the pure liquid (S) in the following
manner (Chiou and Schmedding, 1982):
log Kd -log S, logf MW
SP" ) (2-8)
log Yo + log
Comparison of Eqs. (2-6) and (2-8) suggests that any deviations due to nonideal
behavior will arise from the last two terms on the right hand side of Eq. (2-8).
Banerjee (1984) observed that the presence of other components in the aqueous
phase had a minimal effect on solute activity; therefore, it was assumed that y,'*/y,
= 1, thus requiring only estimates of Yo'. The UNIFAC model UNIFAC
(UNIQUAC Functional-Group Activity Coefficient) model proposed by Prausnitz et
al. (1980) for estimating activity coefficients in liquid-liquid equilibria was employed
to estimate yo values needed in Eq. (2-8). In this model, a mixture of different
chemicals is treated as a mixture of functional groups constituting the components
of the mixture. Interactions between functional groups in the mixture, and the likely
nonidealities resulting from such interactions, are calculated in order to estimate the
activity coefficient of a chemical for a specified phase. Interaction parameters
required in the UNIFAC model were obtained from the most current update
(Hansen et al., 1991).
A schematic representation of Eqs. (2-6) and (2-8) is shown in Figure 2-9 as
a plot of log KD versus log S1. Note that the expected relationship for an ideal
mixture is depicted by the solid line, with a unit slope and the intercept given as the
log Vo (see Eq. 2-8). The single data point represents a possible value for a solute
partitioning between a hypothetical nonideal mixture and water. Note that the
magnitude of deviation from the ideal line is given by the last two terms on the right
hand side of Eq. (2-8) plus an error term, E, representing experimental uncertainty.
Log SI (moles/I) 0
Pure lquid aqueous solubility
Figure 2-9. Schematic representation of the ideal behavior (Raoult's law) and
nonideality in liquid-liquid partitioning.
Application of the UNIFAC model for assessing the potential for nonideality
is presented for a gasoline, diesel fuel, and coal tar. Using the UNIFAC model,
activity coefficients (Yo') of several aromatic compounds were estimated for an
unleaded gasoline simulated to represent the relative compositions (see inset in
Figure 2-10) reported in Cline et al. (1991).
.-J 5 10
Alkanes mole fraction
1 MTBE -
2 benzene 0.07
3 toluene 0.07
4 xylene 0.07
5 ethylbenzene 0.07
6 1,2,3-trimethylbenzene -
7 n-propylbenzene -
8 3,4-ethyltoluene -
9 naphthalene 0.014
10 anthracene 5E-3
11 pyrene 1E-6
3,4-benzopyrene 1 E-7
0 Cline et al. (6)
Log S (moles/L)
Figure 2-10. log K, values for several aromatic hydrocarbons resulting from
UNIFAC model calculations and the average log, values
experimentally determined by Cline et al. (1991) plotted against log S,
values along with the ideal line based on Raoult's law.
The estimated y* values were then used to predict log K, values (shown as solid
triangles in Figure 2-10) according to Eq. (2-8). UNIFAC model calculations for the
monocyclic aromatic compounds represented in Figure 2-10 (compounds 2-5) confirm
the experimental observations of Cline et al. (1991) that gasoline-water partition
coefficients of several liquid hydrocarbons can be approximated by assuming ideal
behavior. However, for compounds with increasingly more aromaticity and are solids
in their standard state (PAH compounds 9-11 in Figure 2-10), the UNIFAC model
predicted some negative deviation from ideal behavior. Partition coefficients for
these compounds were not measured by Cline et al. (1991) as they are present only
in small quantities in gasoline. Compared to gasolines, diesel fuels contain a larger
fraction of low-solubility PAHs. Therefore, it was of interest to see if the UNIFAC
model estimations of Yo* for these PAHs resulted in deviations from ideality.
The composition of the diesel fuel assumed in the UNIFAC model
calculations is shown in Figure 2-11. The concentrations of the eight PAHs chosen
were comparable to those found in the diesel fuels used in this investigation; the
concentrations of monocyclic aromatic hydrocarbons used were based on analyses
reported by Thomas and Delfino (1991); and the mole fraction of water was selected
based on the maximum ASTM limiting requirement for diesel fuel (Kirk-Othmer,
1980). To simulate the alkane fraction of the diesel fuel, a representative compound
for each alkane (n-, iso-, and cyclo-alkane) was selected (see Figure 2-11) in
proportion to those reported by Mackay et al. (1985). The UNIFAC model
calculations for the yo* values of the PAHs ranged between 0.99 for toluene to 1.16
for fluoranthene. The close proximity of the calculated log Kdw values (solid triangles
in Figure 2-6) to the ideal line based on Raoult's law for the simulated diesel fuel
suggest that deviations from ideal behavior for PAHs smaller than fluoranthene may
be negligible. These calculations suggest that deviations from the ideal line for the
larger PAHs noted in Figure 2-1 cannot be attributed to solute-solute interactions,
lending support to analytical sources of error for the observed deviations.
Independently assessing the potential for nonideal behavior emphasizes the need to
account for experimental and analytical sources of errors when judging whether the
deviation noted from the ideal line is indeed the result of nonideal behavior.
Figure 2-11. log Kd, values for several aromatic hydrocarbons resulting from
UNIFAC model calculations plotted against log S, values along with
the ideal line based on Raoult's law.
Based on the success for gasoline and diesel fuel, an attempt was made to use
the UNIFAC model to assess the likelihood of nonideality for coal tar ID#4. Since
less than 40% of the composition of this coal tar was unknown (as usually is the
case), it was represented by a single compound indicated in Figure 2-12. The
UNIFAC model simulations suggested that nonidealities are indeed small, and that
Raoult's law approximation was justified (Figure 2-3). A note of caution is in order,
however, the UNIFAC model results depend heavily on the presumed composition
of the pitch (62% mole fraction in our example with coal tar ID#4), and on the
presence of polar constituents in coal tar (none were present in significant
quantities in this example).
-5 -4.5 -4
log [S, moles/L]
Figure 2-12. Comparison of measured and predicted tar-water partition coefficients
for several PAHs: Raoult's law (solid line) and UNIFAC model (solid
Release of aromatic hydrocarbons from an immiscible organic liquid waste is
governed primarily by solubility phenomena. In assessing the likelihood of soil and
water contamination from complex organic wastes (e.g.,gasoline, diesel fuel, and coal
tar), it is incorrect to assume that PAH concentrations in groundwater would be
equal to the corresponding aqueous solubilities of the pure compounds. Such an
assumption usually leads to considerable over-predictions of the PAH concentrations
likely to be found in groundwater.
According to the model based on Raoult's law, the concentration of an
organic constituent in the aqueous phase in equilibrium with an "ideal" organic
mixture is proportional to the mole fraction of that constituent in the organic phase.
An experimental evaluation of a model based on ideal behavior was presented for
the partitioning of aromatic hydrocarbons from diesel fuel and coal tar into water,
and the results compared to data reported earlier for gasoline/water and motor
oil/water partitioning. The diesel fuel/water and tar/water partitioning of several
PAHs, all solids in their standard state, was well described within a factor of four for
diesel fuels, and within a factor of two for coal tars by employing supercooled liquid
solubilities and assuming ideal behavior. Good agreement between the observed
partitioning of several PAHs and UNIFAC model calculations for a simulated
gasoline, diesel fuel, and coal tar further suggests that the extent of deviations from
ideal behavior may be relatively small.
Agreement between the model predictions based on Raoult's law and
measured liquid-liquid partitioning data for several aromatic hydrocarbons is not to
be taken as evidence that such compositionally-complex organic liquid wastes are
indeed ideal mixtures. Rather, the assumption of ideal behavior might suffice for
practical considerations in providing first-order estimates for maximum PAH
concentrations likely to be found in groundwater leaving an area contaminated with
residual OILs. Several site-specific hydrogeologic factors might lead to significant
mass transfer constraints for solute partitioning. Such factors include: random
spatial variability in aquifer hydraulic properties, the patterns of residual fuel
entrapment, and the source of fuel contamination (e.g.,surface spill versus subsurface
leaks). Under nonequilibrium mass transfer conditions, the concentrations of organic
constituents detected in groundwater are likely to be smaller than those estimated
using the equilibrium approach presented here. In contrast, larger concentrations
might be observed in the presence of surfactants, emulsifiers, or cosolvents.
COSOLVENT EFFECTS ON SORPTION OF ORGANIC ACIDS
BY SOILS FROM METHANOL/WATER SOLUTIONS
The codisposal of contaminants, as well as the potential use of alternative
fuels and mixing of contaminant plumes from different sources, will result in
environmental contamination problems consisting of a complex mixture of chemicals
including both polar and nonpolar organic in miscible and immiscible solvent
mixtures. Solubility, sorption, and transport of hydrophobic organic compounds
(HOCs) are well characterized in aqueous solutions and various complex mixtures.
Solubility of HOCs increases with increasing volume fraction cosolvent of an organic
cosolvent (Yalkowsky and Roseman, 1981; Yalkowsky, 1985; 1987; Rubino and
Yalkowsky, 1987a; 1987b; Fu and Luthy, 1986; Pinal et al., 1990; 1991). Sorption of
HOCs is inversely related to solubility and as a result, an increase in solubility from
the addition of a cosolvent leads to a proportional decrease in sorption (Rao et al.,
1985; 1990; Nkedi-Kizza et al., 1985; 1987; Rao and Lee, 1988; Woodburn et al.,
1986; Fu and Luthy, 1986).
For hydrophobic ionizable compounds (HIOCs) of environmental interest,
data on solubility, sorption, and transport in mixed solvents are limited. Some
research investigating the impact of multiple solutes on HIOC sorption (i.e.,
competitive sorption) by soils from aqueous solutions has been documented (Felice
et al., 1985; Zachara et al., 1987; Rao and Lee, 1987); however, little attention has
been given to the behavior of HIOCs in solvent mixtures.
Pharmaceutical literature contains solubility data for several drugs spanning
a wide polarity range. As shown in Figure 3-1, Yalkowsky and Roseman (1981)
observed that as solute polarity increases relative to the solvent, cosolvency curves
become increasingly more parabolic in shape until an inverse relationship occurs (i.e.,
decreased solubility with cosolvent additions). Such behavior is explained on the
basis of the solute-solute and solute-cosolvent interactions. Therefore, for
compounds that exhibit a decrease in solubility with addition of a cosolvent (log Ko,
< 1), sorption may increase with increasing cosolvent composition.
For the sorption of naphthol, quinoline, and dichloroaniline by three different
soils from methanol/water and acetone/water solutions up to 50% by volume, Fu
and Luthy (1986b) observed log-linear behavior inversely proportional to
corresponding solubility data (Fu and Luthy, 1986a) as observed with HOCs. Similar
behavior was observed by Zachara et al. (1986) for quinoline sorption by a natural
clay isolate and montmorillonite in binary mixtures of methanol or acetone and water
regardless if the protonated or neutral species predominated in solution. For these
HIOCs it appears that the cosolvent effect on sorption is dominated by solvation
forces (i.e., solubility) similar to that observed with HOCs even though sorption
mechanisms for HIOCs and HOCs are different (electrostatic and ion exchange
versus hydrophobic partitioning).
5.0 2.0 1.5 0.0 -1.0
Volume Fraction Cosolvent, f
Figure 3-1. Schematic representation of cosolvency plots for solutes with a range
of log Kow values.
For an acidic fluorescent dye (Rhodamine WT) in binary mixtures of
methanol/water and acetone/water at cosolvent fractions above 30%, sorption was
observed to increase even though at lower cosolvent fractions (< 30%) sorption
appeared to follow an inverse log-linear relationship (Soerens and Sabatini, 1992).
Previous use of Rhodamine WT as a surface and groundwater tracer prompted an
investigation on the potential use of this dye as a tracer in alternative fuel research
(i.e.,alcohol-based fuels). In soil thin-layer chromatography (TLC) studies (Hassett
et al., 1981), the herbicide dicamba (3,6-dichloro-2-methoxybenzoic acid) moved with
the solvent front in both aqueous and 50/50 (v/v) ethanol/water solutions, but was
strongly retained by soil with neat ethanol as the mobile phase.
For analysis of various carboxylic acids and substituted phenols by reversed-
phase chromatography using an acidified mobile phase (i.e., when neutral species
dominate), retention is inversely proportional to cosolvent content as observed for
HOCs. Lewis and Wilson (1984) observed similar behavior for several carboxylic
acids in non-acidified methanol/water solutions (0 to 80%) using reversed-phase TLC
plates impregnated with an ion-pairing agent tetra-n-butylammonium bromide (TBA).
However, in the absence of an acidified mobile phase or an ion-pairing reagent,
cosolvent effects were minor. These data suggest that the retention behavior of
organic acids by a hydrophobic surface is similar to HOCs only when the charge is
compensated (e.g., neutral by protonation or paired with a counterion).
In considering the impact of cosolvents on sorption, the following interactions
must be considered: solute-solute; solvent-solvent; solvent-solute, solvent-sorbent,
and solute-solvent-sorbent interactions. The first three pairs of interactions can be
assessed from solubility studies. For HOCs, solute-solute interactions are ignored
due to the absence of both specific functional groups and high solute concentrations.
For most miscible solvents (e.g., alcohols), irregular behavior due to specific
complexation of solvent and water molecules doesn't appear to be of concern. For
sorption of HOCs, solvent-solute interactions, as described by solubility, are found
to predominate such that the impact of solvent-sorbent interactions have been
considered minor. However, for solutes containing specific functional groups, the
impact of the cosolvent on the sorbent may have considerable impact.
For HIOCs, the impact of adding a cosolvent to aqueous solutions on the
conditional ionization constant of a HIOC must be considered. Likewise, similar
impacts on the ionization of sorbent functional groups and subsequent solute-sorbent
interactions must also be considered. Also, the impact of solvent-water interactions
that were considered negligible in predicting HOC behavior may be of importance
in understanding the chemodynamic behavior of HIOCs, as well as the different
propensities of the cosolvent and water to hydrate both the solute and the sorbent.
Cosolvent-induced interactions involving the sorbent surface include:
speciation of organic matter functional groups, clay surface acidity, and ion-
association with the surface. Both acidic and basic groups tend to become neutral
with increasing cosolvent content as a result of shifts in the pKa' (Perrin et al., 1981),
leading to a net increase in hydrophobicity of soil organic matter. This phenomenon
may explain why decreases in HOC sorption with increasing f, are smaller in
magnitude than would be predicted from solubility profiles in mixed solvents (Rao
et al., 1990; Nkedi-Kizza et al., 1985; 1987; Rao and Lee, 1988). Parallel to changes
in pK,', Kan and Tomson (1990) observed a decrease in naphthalene sorption by
Lincoln fine sand from aqueous solutions by increasing pH (pK, fixed, but pH
varied). However, the increase in sorption resulting from such changes on surface
hydrophobicity are likely to be more than compensated by cosolvency effects.
The presence of cosolvents may also alter the surface acidity of the clay
fraction. Loeppert et al. (1977, 1979) found that the amount of base required to
titrate the pH-dependent sites of kaolinite varied in the following manner: methanol
< water < DMSO < acetonitrile. The fact that larger titers in DMSO and
acetonitrile were required was attributed to (1) pH-dependent sites for which a
quantitative endpoint was not obtained in aqueous media due to the acidic properties
of water, and (2) increased surface acidity in organic solvents. Loeppert et al. (1986)
also observed an increase acidity of montmorillonite in acetonitrile and
dimethylformamide. No apparent changes were observed in surface acidity with neat
methanol (Loeppert et al., 1979). Similar conclusions were made by Hesleitner et
al. (1991), who noted that addition of methanol (f <5 0.5) caused no apparent change
in the surface charge density of a hematite surface (iron oxide) or in the point of
zero charge which coincided with the isoelectric point.
Cosolvent-enhanced formation of ion-pairs with positive charges on the
sorbent surface may cause an increase in sorption of organic acids with addition of
an organic cosolvent even if solubility increases. As previously mentioned, Hesleitner
et al. (1991) observed no changes in the total surface charge of hematite in the f,
range investigated (f, < 0.5), but noted a pronounced decrease in electrokinetic
potentials with increasing methanol fractions (i.e., effective surface charge was
lowered). They attributed the decrease in electrokinetic permittivity to an
enhancement of counterion association with the surface charged groups. This
counterion association could include both the formation of outer-sphere complexes
by bridging of the carbonyl to the solvent (water and/or cosolvent) coordinated on
the exchange cation (Farmer and Russell, 1967) and inner-sphere complexation by
hydrogen bonding of the carbonyl group with protonated hydroxyls on the surface
(Kohl and Taylor, 1961; Stumm et al., 1980). These mechanisms have been included
among those proposed in the literature for sorption of organic acids in aqueous
solutions (Farmer and Russell, 1967; Kohl and Taylor, 1961; Stumm et al., 1980;
Davis, 1982; Kummert and Stumm, 1980); however, the impact of cosolvents on such
interactions has yet to be investigated.
In this chapter, the overall impact of methanol additions on (1) the
enhancement of solute-solvent interactions as described by solubility; and (2)
speciation changes due to cosolvent induced changes in the solute's pK,' will be
assessed for the sorption of several organic acids by soils. Subsequent chapters will
assess (1) speciation changes due to changing pH at several fixed methanol/water
compositions for benzoic acid and PCP sorption; (2) the overall impact of several
solvents with a wide range in solvent properties on the sorption of benzoic acid; and
(3) the relationship between solute properties, such as acidity and hydrophobicity, on
the shapes of the sorption curves observed in methanol/water solutions.
The following log-linear model successfully describes (Yalkowsky and
Roseman, 1981; Fu and Luthy,1986; Pinal et al., 1990; Rao et al., 1985; Nkedi-Kizza
et al., 1985, 1987; Rao and Lee, 1988; Woodburn et al., 1986) solubility and sorption
of HOCs in miscible solvent-water systems,
log Sb log S, + of, (3-1)
log(S,/Sw ) a (3-2)
log Kb log K, aof, (3-3)
where S is solubility (mg/L), K is sorption coefficient (mL/g) with subscripts b, c,
and w referring to binary mixtures, pure cosolvent, and water, respectively; f is
volume fraction cosolvent; a describes the cosolvency power of a solvent for a solute;
and a accounts for solvent-sorbent interactions.
Sorption of HIOCs is dependent on the formation of neutral and ionized
species, as determined by pH and the solute acid dissociation constant (pK,). For
many organic acids, the neutral species is sorbed more than its dissociated (anionic)
species, and the differences in the sorption coefficient values can be rather large.
Thus, the measured sorption coefficient for HIOCs is a strong function of pH and
conditional dissociation constants (pKa') of the solute in the solvent system of
interest. Lee et al. (1990) showed that the pH-dependence of pentachlorophenol
(PCP) sorption from aqueous solutions can be described by,
K Kw,nF + Kwi(1-#) (3-4)
c" (1 + 1O3-P')-1
and subscripts n and i refer to neutral and ionized species, respectively. Similar
findings for the sorption of several other organic acids by various sorbents have been
reported in the literature (Jafvert, 1990; Kukowski, 1989; Fontaine et al., 1991).
If solubility of a solute increases with addition of a cosolvent to an aqueous
solution (see Figure 3-1), a decrease in sorption is expected. Also, the addition of
a solvent with a low dielectric constant will result in an alkaline shift in the pK,' of
an organic acid (Perrin et al., 1981), leading to an increase in the fraction of neutral
species. In the absence of specific adsorption reactions, the neutral species will be
sorbed to a greater extent. Therefore, the addition of a cosolvent brings about two
opposing effects. To incorporate both speciation and cosolvent effects, Eq. (3-1) and
Eqs. (3-3, 3-4, and 3-5) were combined,
Kb KwA, n n + K,,(1-n) (3-6)
3, 10-l-"'" ; Pi 10-a,'a (3-7)
The cosolvency power for the neutral species (as) will increase relative to
hydrophobicity. The cosolvency power for the ionized species (ai) will be a function
of the relative hydrophobicity of the anion and the potential for ion-pairing. For
example, Lee et al. (1990) observed a log-linear decrease in sorption of PCP by
Webster soil in methanol/water (0.01 N CaC12) solutions (f,= 0 to 40%) for both the
neutral species (pH < 3) and ionized species (pH < 9) with resulting values for aiai and
ana, were 2.56 and 3.88, respectively. The decrease in sorption observed with
increasing f, for pentachlorophenolate was attributed to the relatively large
hydrophobicity of the anion and the formation of neutral ion-pairs.
Figure 3-2 illustrates the types of cosolvency curves for the sorption of organic
acids that might be predicted using Eq. (3-6). Using parameter set #3 results in the
presence of primarily the neutral species of the HIOC (pH-pK,' <-1) thus yielding
cosolvency curves similar to that observed for HOCs (Eq. 3-3). In the absence of
specific interactions, a reduction in solubility with increasing cosolvent content might
be expected for a solute existing as an anion in solution, thus potentially increasing
sorption (parameter set #6). Similar results are predicted using Eq. (3-6) for a
solute with relatively small hydrophobicity (a=l ) and assuming no impact of
cosolvency on the anionic species (ai=0) (parameter set #2 and #5). Note how the
magnitude of the increased sorption predicted by Eq. (3-6) is a function of the inter-
relationship between initial soil-solution pH (i.e.,pH-pK,,) and the a values. The
a values used in sets #1 and #3 are larger than those used for sets #2 and #5
changing the impact of pH variations. For sets #1 and #3, enhanced linearity and
an upward shift is observed with decreasing pH; whereas, for sets #2 and #5, the
shape of the sorption curve changes from a convex to a concave shape as pH
decreased. Therefore, the overall magnitude and direction of the sorption observed
will vary as a function of cosolvency power (a), soil-solution pH, and cosolvent
induced shifts in the observed pK,'.
K, = 0.1
-1.4 --- 1
-1.6 .................. 5
Figure 3-2. Example
0.2 0.4 0.6 0.8 1
Volume Fraction Methanol
cosolvency curves that may be predicted by the use of various parameters in Eq. (3-6).
The success of Eq. (3-6) in describing sorption of organic acids is predicated
on the ability to measure (or define) the ionization constant (pK,') and pH in the
solutions of interest. Defining pK,' and pH is fairly straightforward for aqueous
systems; however, various complications must be considered for mixed solvent
systems. The pH of an aqueous solution is thermodynamically defined as the
negative logarithm of the hydrogen ion activity (aH+)
pH -log aR. -log H[H'] (3-8)
where YH+ and [H+] are the hydrogen activity coefficient and concentration,
respectively. Experimentally, an electrometric method is usually employed (e.g.,pH
meter) where the determination of pH is based on the measurement of the
electromotive forces (e.m.f.)of standard aqueous buffer solutions. Therefore, the pH
of an unknown solution (pHx) can be determined by
pH, pH, + x E (3-9)
(RT In 10/F)
or at T=298,
pH H pH, + E- E (3-10)
where Ex and E, are the e.m.f. values of the solutions, R and F are the gas and
Faraday's constants, respectively, and T is absolute temperature. Not shown in Eq.
(3-9) are the potentials that arise from the liquid junction and the standard potential
of the glass electrode. The difference in these potentials between the standard and
unknown solutions are assumed to be the same when the solution matrix is similar,
thus cancelling out in the (E,-E) term.
Likewise, in mixed solvents (denoted by *), pH is thermodynamically defined
pH* -log y.t[H] (3-11)
If standard mixed solvent buffers are employed, pH can be operationally defined as
pH* pH + E -E (3-12)
It is usually expedient to employ readily available standard aqueous buffers in which
case Eq. (3-12) must be modified to estimate the pH of a solution in mixed solvents,
pH pH Es (3-13)
where pHaPP is the measured pH of a mixed solvent solution relative to a standard
aqueous buffer solution. The differences in the liquid junction potential and the
standard potential of the glass electrode between mixed solvents and aqueous
solutions cannot be assumed to be the same and must be considered. However,
Gelsema et al. (1967) have shown that differences in the standard potentials of the
glass electrodes between mixed solvents and aqueous solutions are negligible. The
operational definition of pH for a mixed solvent solution (pHx*) referenced to an
aqueous standard can then be written as
pH E, E, E-E (3-14)
pH. pH, + -1
Therefore, differences in the apparent pH and the actual pH arise from the
difference in the liquid junction potentials:
8 pHxa pH* jE'- (3-15)
Van Uitert and Haas (1953) achieved a practical standardization of pH
measurements in dioxane-water solutions by measuring the pH of a series of HCI
solutions of known concentration in the mixed solvents. The difference between the
measured pH and the known hydrogen concentration was assumed to be a
reasonable estimate for 6. This approach yields estimates for 6 that encompass all
differences (i.e.,liquid junction potential, standard potential, activity, solvent medium
effects) observed between a measurement in aqueous versus solvent/water solutions
irrespective of the source. The magnitude of 6 increases with increasing amounts of
an organic solvent. For methanol/water solutions at fc<0.8, 6 values are relatively
small, but 6 values may become greater than 2 as the neat organic solvent is
approached (Van Uitert and Haas, 1953; De Ligny and Rehbach, 1960). Although
there are obvious shortcomings to this simplified approach, it appears adequate in
many cases for estimating pH in several mixed solvent solutions.
Materials and Methods
The primary sorbents used in this study were Eustis fine sand (Psammentic
Paleudult) from Florida containing 96.4%, 1.8%, 1.8%,and 0.39% sand, silt, clay, and
organic carbon (OC), respectively; and Webster silty clay loam (Typic Haplaquoll)
from Iowa (5 miles north and 3 miles east of Ames) containing 30.7%,42.8%, 27%,
3.0% sand, silt, clay (predominately montmorillonite), and OC, respectively. Specific
surface measurements by N2-BET of approximately 4 m2/g was obtained for a similar
Webster soil subsample used in previous studies (Rao et al., 1988). Both the Eustis
and Webster soils were collected from the surface horizon (0-30 cm). The soil OC
contents were determined using the Walkley-Black method (Nelson and Sommers,
1982). The soil-solution pH in 0.01 N CaC12 was 5.0 and 6.9 for Eustis and Webster
soils, respectively. Soils were air-dried and passed through a 2 mm sieve prior to use.
The organic acids used in this study are listed in Table 3-1 along with selected
physical and chemical properties. All crystalline compounds had a chemical purity
of >98%. All solvents were purchased from J.T. Baker (high purity, HPLC grade)
and used without further preparation. For sorption experiments with
pentachlorophenol (PCP), 14C uniformly ring-labeled compound was purchased from
Sigma Chemical Co. with a specific activity of 12 mCi/mmol and a reported
radiochemical purity of >98%.
Table 3-1. Selected Solute Properties
Melting Molecular pK, Solubility' Log
Solute Point' (oC) Weight' Aqueous Methanol2 (mg/L) Kow
Pentachlorophenol 190 266.3 4.744 8.6 14 5.01
2,4-Dichlorophenol 42 163.0 7.854 11.9 4,500 3.23
Picric Acid (2,4,6-trinitrophenol) 121 229.1 0.419' 4.1 14,000 2.03
Gentisic Acid 205 154.1 2.97' 7.6 21,5002 NA9
2,4,5-trichlorophenoxy acetic acid 156 255.5 2.85 7.4 278 NA
2,4-dichlorophenoxy acetic acid 138 221.0 2.64' 7.6 890 NA
Benzoic Acid 122 122.1 4.201 9.0 2,900 1.87
Pentafluorobenzoic acid 101 212.1 1.496 5.8 NA NA
Dicamba (3,6-dichloro-o-anisic acid) 115 221.9 1.947 6.9 7,900 2.468
'From Dean (1985); 2 This study; 3 From Verschueren (1983); 4 From Callahan et al.(1979); S Koskinen and O'Connor
(1979); 6 From Walters (1982); From Kearney and Kaufman (1976); a EPA Environmental Fate One-Line Data Base,
Version 3.04; 9 Not available.
Determination of Ionization Constants
The conditional ionization constants (pKa') for benzoic acid, gentisic acid, 2,4-
dichlorophenoxyacetic acid, PCP, 2,4-dichlorophenol, and dicamba were determined
in methanol/water solutions by measuring pH as a function of NaOH additions
(Albert and Sergeant, 1984). Solvent mixtures were prepared with 0 to 100%
methanol and degassed prior to use. For all solutes except for PCP and dicamba,
0.01 M solutions were titrated with 0.1 M NaOH. For PCP and dicamba, 0.001 M
solutions were titrated with 0.01 M NaOH. A Metrohm 686 Titroprocessor,
employing a combined pH glass electrode (6.0202.100)and a resistance thermometer
(6.1103.000), continuously measured pH and temperature, respectively. The pH
meter was calibrated using aqueous buffers. The temperature of the solutions was
24 0.5 C. Titrations were performed in duplicate in 50 mL beakers placed on a
stirring plate to mix the solutions. The ionization constants determined in this study
are mixed ionization constants (Albert and Sergeant, 1987) rather than true
thermodynamic ionization constants. A brief discussion on the difference between
the various ionization constants are given in Appendix B along with sample sets of
titration data from this study and corresponding calculations.
In calculating pK.' values, adjustments for the impact of methanol on pH
measurements were made using a method similar to that employed by Van Uitert
and Haas (1953) as described above for the measurement of pH in mixed solvents.
This method consisted of measuring the pH of 0.001 M hydrochloric acid in the
mixed solvent. The difference (6) between the pH measured in water and in the
mixed solvent was added to the pKa estimated from the titration curve (i.e.,
pK,'=pK. + 6). From 0 to 70% methanol, 6 values were negligible. At higher
methanol fractions, 6 values were approximately 0.1,0.4, and 2.3 for f,values of 0.8,
0.9 and 1.0, respectively. Similar results were obtained by De Ligny and Rehbach
(1960) for methanol/water solutions by comparing pH measured in aqueous standard
buffers (KC1 saturated solutions) and standard buffers prepared in the appropriate
mixed solvent using the method proposed by the National Bureau of Standards
(Bates et al., 1963). Therefore, the corrections needed to adjust the pK, determined
in mixed solvents relative to the use of aqueous standard buffers are only significant
pH of Soil Suspensions in Mixed Solvents
When considering the measurement of pH in mixed solvent soil-suspensions,
problems in addition to those previously discussed for pH measurements in mixed
solvents arise. It has long been recognized that the ambiguity of measuring the pH
in aqueous soil solutions, and even more so in soil suspensions, is due to the inability
to accurately determine liquid junction potential differences between standard buffer
solutions (Ej,) and soil-solutions (Ej,x). Even with this ambiguity, the error in the
measured pH resulting from differences in the liquid junction potentials (EFj, E,,
is usually assumed to be within 0.2 pH units for an aqueous soil-solution or dilute
soil-suspension given a background electrolyte concentration of approximately 0.01
N (Sposito, 1989).
In these studies, pH of the supernatant and/or the resuspended soil sample
were measured using a Coming Model 130 pH meter and a Fisher Scientific or
Orion combination micro-electrode (AgCl saturated 3 M KC1 filling solution)
following equilibration and analysis of the sample. For suspensions of Webster silty
clay loam in methanol/water solutions with a background electrolyte of 0.01N CaCl2,
changes in the measured pH (pH"xPP) of less than 0.5 pH units were observed going
from aqueous to methanol solutions. Recall that a change (6) of over 2 pH units
were previously noted for solutions going from aqueous to methanol solutions. This
prompts questions regarding (1) the interactions between the liquid junction
potentials arising from the solvent and soil medium; and (2) the effect of methanol
on the activity of hydronium ions on the soil surface. Given the difficulty of
answering such questions at this time, pH/,"P will be used in combination with pK,'
to estimate solute speciation.
Experimental techniques described by Pinal et al. (1990) were employed to
measure benzoic acid solubility in methanol/water solutions that were either acidified
with 0.01 M HC1 or made basic with 0.3 N NaOH. These data were compared to
solubilities obtained by Yalkowsky (1985) without additions of an acid or a base.
Solute concentrations were analyzed using reversed-phase liquid chromatography
(RPLC) techniques. The RPLC system consisted of a ternary solvent pump (LDC
Milton Roy Model CM4000, Eldex Model 9600, or Gilson Model 302), a Waters
Radial Compression Column with a C-18 cartridge, a UV detector (Gilson Model
115 or Waters Model 490), and a Waters Intelligent Sample Processor (Model 710B
or 715). The composition of the mobile phase (acetonitrile/methanol/water; pH2
w/HCl) and the UV wavelength were optimized for analysis of each solute. When
necessary, samples were diluted to within an optimal concentration range of the
specific analytical method.
Miscible Displacement Experiments
Miscible displacement techniques described by Brusseau et al. (1990) and Lee
et al. (1991), were used to estimate retardation factors with water and neat methanol
as the eluent for the solutes listed in Table 3-1. The column was packed with air-
dried Eustis soil and 0.01 N CaCl2 solution was pumped through the column until
steady-state, water-saturated conditions were established. All solutions were filtered
(0.45 pm) and degassed with helium before use. The physical properties of the
Eustis soil column were as follows: 5 cm length, 0.4 mL/cm3 volumetric water
content (0), 1.69g/cm3 bulk density (p), and a column pore volume of 9.64 mL. A
pore-water velocity (v) of about 90 cm/hr was used for all experiments.
Solute concentrations in the influent solutions were approximately 100 pg/mL,
except for PCP which was 3 4g/mL. Solute concentrations in the column effluent
were monitored continuously as described by Brusseau et al. (1990) using a flow-
through, variable-wavelength UV detector (Gilson Holochrome, Waters 450 or LDC
UV) connected to a linear chart recorder (Fisher Recordall 5000). Retardation
factors (R) were obtained by calculating the area above the measured breakthrough
curves (Nkedi-Kizza et al., 1987). Periodic measurements were made of the column
effluent pH with a combination glass electrode using a Coming 130 pH meter or a
Brinkman 686 unit.
Equilibrium Sorption Isotherms
Equilibrium sorption isotherms were measured using the batch-equilibration
method (Rao et al., 1990). The vials used for this study were 5 mL (1 dram) screw
cap borosilicate glass autosampler vials with teflon-lined septa inserts. Amber vials
were employed to minimize photolysis. Soil mass to solution volume ratios ranged
from 1:2 to 2:3 to achieve sorption of 50% ( 20%) of the chemical added. All
solutions used had a 0.01 N CaC1, matrix unless noted otherwise. Initial solution
concentrations added to the soils ranged from 5 to 45 jig/mL for all solutes with the
exception of PCP. A concentration range of 0.25 to 3 gg/mL was used in the PCP
equilibration studies. All sorption isotherms were measured at room temperature
(T=22-25" C). Following equilibration, the solution and solid phases were separated
by centrifuging the soil samples at approximately 300 RCF (relative centrifugal force)
using a Sorvall RT6000 centrifuge.
Each isotherm consisted of sorption measured in duplicate at four or more
concentrations and at least 30% of the isotherms were replicated. Also for each
isotherm, blanks containing the solvent with and without soil were run to check for
coelution of any peaks from the soil. Samples were usually equilibrated by rotating
for 16-24 hours. For f. <0.2,degradation was noted after 4 hours in control studies
where the soil-solution matrix was removed from the soil and spiked with the
appropriate solute concentrations. Therefore, equilibration of samples in solutions
of f,<0.2was reduced to a maximum of 2 hours with no differences observed in
sorption coefficients measured after 1 and 2 hours.
Nonradiolabeled samples were analyzed by RPLC techniques as described
previously for the solubility studies. The use of autosampler vials in conjunction with
the Waters Intelligent Sample Processor (WISP 715) enabled direct analysis of the
samples by RPLC techniques without further sample transfer. The WISP 715 has the
capability of varying sampling depths within a vial allowing sampling of the
supernatant without removal of the soil. The higher mass to volume ratios (2:3),
however, necessitated transfer of the supernatant to a new vial. For 4C-labeled
solutes, 0.5 mL aliquots of the supernatant were taken from each sample and mixed
with 20 mL of Scinti-Verse II for analysis. Solute concentrations were then assayed
using liquid scintillation counting (LSC) methods employing a Searle Delta 300 liquid
Sorption coefficients, K (mL/g), were estimated by fitting the sorption data
to a linear isotherm: S. = K Ce, where Se and C. are sorbed (mg/g) and solution
(Ag/mL) concentrations, respectively, at equilibrium. The solution concentrations
were directly determined, whereas S. values were determined by difference: S. =
(Ci C)(V/M), where Ci is the initial solution concentration (Ajg/mL) of the solute;
V is the solution volume (mL); and M is the soil mass (g).
The pK,' values measured as a function of volume fraction (fQ) methanol
increased linearly up to approximately f,=0.6,and then increased markedly at higher
cosolvent contents. Representative data for benzoic acid and pentachlorophenol
(PCP) are presented in Figure 3-3. Similar results were observed for the other
compounds. For organic bases, a decrease in pK,' (an acid shift) occurs upon
addition of a cosolvent (shift towards neutral species); however, the overall shift from
aqueous to neat solvents is usually much less than a single pH unit. Also shown in
Figure 3-3 are pK,' values for benzoic acid determined conductometrically by Pal et
al. (1983) up to 80% methanol, and the pK,' value reported by Bacarella et al. (1955)
in neat methanol using a different type of potentiometric method with an electrode
system void of a liquid junction. Good agreement between our data and the
published data suggests that the procedure used in this study was adequate. The
lower pKa' value obtained in this study for benzoic acid in neat methanol is most
likely due to the use of hydrated methanol (0.05%); residual water was removed from
the methanol used in the cited studies. Also the constants determined in this study
are mixed ionization constants, whereas thermodynamic ionization constants were
reported by Bacarella et al. (1955).
Solubility data reported by Yalkowsky (1985) for benzoic acid in
methanol/water solutions are shown in Figure 3-4. Solubility increased with
increasing volume fraction methanol. For the solubility of an organic acid in an
unbuffered solution, the pH at saturation will be less than the solute pKa. For
example, the pH of an aqueous solution saturated with benzoic acid is approximately
2.8 (Bates, 1973). Thus, the neutral species dominates over the solubility profile,
with over 90% existing in the neutral form at f_>0.3. Also shown in Figure 3-4 are
the solubilities of benzoic acid in acidified methanol/water solutions (0.01 M HC1)
measured in this study. At saturation, the acidified samples remained near a pH of
2. Minimal differences were observed between the solubility of benzoic acid in
acidified and nonacidified methanol/water solutions. Solubility curves were not
measured for other solutes in this study, but benzoic acid is believed to be
representative of the general behavior of carboxylic acids in methanol/water
solutions. For example, the solubility reported for dicamba in ethanol is over a 100
times greater than its aqueous solubility (Humberg et al., 1989). The observed
solubility of benzoic acid in methanol/water solutions is similar to the curve shown
in Figure 3-1 for a log Ko of 2.
To further investigate the effect of speciation on solubility, benzoic acid
solubility in solutions containing approximately 0.3 M NaOH was measured for 0 to
40% volume fraction methanol. In the presence of a base, the solubility of benzoic
acid was greater than that observed in the unbuffered or acidified solutions. At the
solubility limits, the saturated solution pH was 5.0. Given the pK,' and measured pH
(pH"P), speciation of benzoic acid in saturated solutions was estimated to range from
approximately 90% to 60% ionized going from aqueous solutions to f,=0.4. The
increase observed in solubility with increasing f, parallels the increase in the neutral
species suggesting that cosolvent effects on benzoate solubility are negligible in the
range investigated (i.e., aari = 0).
---e--* Benzoic Acid (This study)
Benzoic Acid (Pal et al., 1983)
A Benzoic Acid (Bacarella et al., 1955)
) 0.2 0.4 0.6 0.8
Volume Fraction Methanol, fo
Effect of methanol
content on the pKa' of benzoic acid and
Benzoic Acid (0.01 M HCI)
A Benzoic Acid (Yalkowsky, 1985)
O Benzoic Acid (0.3 M NaOH)
0.4 0.6 0.8
Fraction Cosolvent, fo
Solubility (Sb) of benzoic acid in methanol/water solutions.
Miscible Displacement Studies
The retardation factors (R) estimated from the miscible displacement studies
with Eustis soil column are shown in Table 3-2. The column effluent pH ranged
between approximately 4 and 4.8 for the different solute/solvent combinations.
Comparisons of influent pulse sizes with zero-th moments showed greater than 97%
mass recovery for all solute pulses. Loss of soil organic matter from the soil column
during elution with methanol was considered negligible as shown previously by Lee
et al. (1991). For the substituted phenols, R values determined in methanol were
smaller than those water (f,=0). For both chlorophenols in neat methanol,
retardation factors decreased to one corresponding to no sorption (R= 1). A decrease
in R with increasing f is expected from the log-linear cosolvency model in the
absence of any specific interactions. The opposite trend, however, was observed for
all of the substituted benzoic acids. Benzoic acid and dicamba were chosen for
further investigation in several methanol/water solutions using batch techniques with
PCP included as a control. For the batch studies, Webster soil with a higher organic
carbon content was used to better differentiate sorption in the various
methanol/water solutions. Although the greatest increase in retardation was
observed with gentisic acid, preliminary batch isotherm data exhibited extreme
nonlinearity (data not shown). Since problems associated with isotherm nonlinearity
may confound assessment of the proposed model, further investigation of gentisic
acid was not pursued.
Table 3-2. Retardation factors for several organic acids in aqueous and methanol
solutions from Eustis Soil.
Solute Aqueous Methanol
Pentachlorophenol 4.7 1.0
2,4-Dichlorophenol 3.6 1.0
Picric Acid 1.9 1.4
Substituted Benzoic Acids
Gentisic Acid 1.9 3.1
(2,5-dihydroxy acetic acid)
2,4,5 Trichlorophenoxy Acetic Acid 1.7 2.1
2,4-Dichlorophenoxy Acetic Acid 1.4 2.7
Benzoic Acid 1.2 2.2
Pentafluorobenzoic Acid 1.0 1.6
Dicamba 1.0 2.0
Batch Equilibration Studies
Sorption of benzoic acid, dicamba, and PCP by Webster soil was measured
from several methanol/water solutions. Representative isotherms are shown in
Figure 3-5. Sorption isotherms were linear for PCP and dicamba in both aqueous
and mixed-solvent systems over the concentration range investigated. Sorption
isotherms for benzoic acid were slightly nonlinear, but a linear approximation of the
sorption coefficients (K) adequately described the data. The correlation coefficients
(r2) ranged between 0.95 and 1.0.
Effect of Solvent Addition
As noted previously, addition of an organic cosolvent to an aqueous solution
results in an increase in the pKa' for organic acids. Changes in speciation become
significant at fe >0.5 as marked changes occur in the pK,' values. In neat methanol,
the measured soil-solution pH for Webster soil ranged between 6.2 and 6.5;
therefore, essentially all the benzoic acid and PCP existed in the neutral form, while
20% to 30% of dicamba remained ionized.
The sorption coefficients estimated from batch equilibration studies of PCP
and benzoic acid are plotted in Figure 3-6 as a function of volume fraction methanol
(fo). Sorption of PCP in methanol/water systems was well described by the log-linear
model with speciation given by Eq. (3-6) (Figure 3-6A) except in neat methanol. Of
the required model parameters, bulk pH and pK, 'were measured; pn was estimated
using Eq. (3-5); and K,,K,, ,,a,, and a.on were taken from Lee et al. (1990) where
sorption of PCP was measured as a function of f, while pH was maintained such that
PCP was either completely ionized or completely neutral. The values for KE,i and
Kw,n were adjusted for differences in the OC content of the Webster soil used in the
two studies (i.e., K=Ko OC).
Benzoic acid sorption decreased with the addition of methanol up to fc0.2,
but then increased with f, thereafter (Figure 3-6B). Eq. (3-6) was applied to the
benzoic acid data using four reasonable parameter sets to investigate if this behavior
was mostly due to changes in speciation with methanol additions. For all cases, the
sorption coefficient for benzoate (K,,i) was measured at pH = 6.9; K,, was
estimated by measuring the K, at pH = 3.0 and applying Eq. (3-4); and ao was
estimated by regressing benzoic acid solubility data in methanol/water solutions (data
in f,=0 to 0.8; Yalkowsky, 1985). Two values for ai were used. In one case, ai was
set equal to zero as suggested by the solubility data (Figure 3-4), and in the second
case, ai was set equal to 0.65 as estimated from the initial portion of the log K.
versus f, curve (i.e., fe 0.2) where benzoic acid remained 2 99% ionized. For two
parameter sets, solvent-sorbent interactions were ignored (a,=ai=l) while in the
remaining two parameter sets an average a value of 0.5 observed by Fu and Luthy
(1986b) for several solute, soil, and solvent combinations was used as an initial
estimate of solvent-sorbent interactions. In all cases, Eq. (3-6) failed to adequately
predict the magnitude of sorption observed for benzoic acid at higher methanol
contents (Figure 3-6B). Similar sorption data were observed for dicamba (data not
shown). Model parameters were estimated for the dicamba sorption data in a
manner analogous to the calculations for benzoic acid with similar results.
xml version 1.0 standalone yes