R?

.‘o

.u‘
_;.“f;¥i
.JA'
r}?

)

"t
v ."
4%?
‘: ‘52-»
M‘ w‘?
H 1;.“-
at

.P‘
.

.. 3-,” -|

9' f-‘t‘flf'. -\
PK‘

«33.;-

 

A;

‘ . r.
f -_ r IIIJ 1‘
1 :4 . w.

.
v"
,., L

.. v‘ ..L
Agtfff ‘ ‘1‘
05., ‘ ‘ ‘

v“ Ky ' \' - :1“

”AN.“ _ V 35...}, u“, u" '
7‘ {FUJHJ'V
vac" ”’4 " "‘V‘V'
)~_ «1' ' 'l" ‘a
n ‘ l
‘V

..;;;.

a
1
' .XF. ,

,4»: 1

‘fi
fif‘gfit‘: -
.‘ 2’, 1-9

1'
“'41:?1 ‘ . _ ‘ .
‘ 57.x" ' ..
w 2. 6:1 7 ‘7 ‘ $9.3

‘I
war ~==‘--'- “
w? 1:: I?“

;/
‘t
1.:
.f

 

”Ht”

This is to certify that the

dissertation entitled
REDUCED TRANSPORT OF ORGANIC POLLUTANTS IN SOIL
MODIFIED WITH HEXADECYLTRIMETHYLAMMONIUM

presented by

Ines Toro-Suarez

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Crop and Soil Sciences

 

Environmental Toxicology

 

Date%WéL¥/iia€~

MSU is an Affirmative Action/Equal Opportunity Inst/lunar! 0-12771

ill/ll!ill/IllllllllllllIll/llll/l/ll/llIll/Illllll

39040
F.

LIBRARY
Michigan atate
University

 

PLACE IN RETURN BOX to romovo thio checkout from your rocord.
TO AVOID FINES roturn on or botoro doto duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSUlo oAnAflirmotivoActiorVEwolO Orippom ityl notitwon
manna-m

REDUCED TRANSPORT OF ORGANIC POLLUTANTS IN SOIL
MODIFIED WITH HEXADECYLTRIMEI‘HYLAMMONIUM

By

Ines Taro-Suarez

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Crop and Soil Sciences
Institute for Environmental Toxicology

1994

ABSTRACT

REDUCED TRANSPORT OF ORGANIC POLLUTANTS IN SOIL
MODIFIED WITH HEXADECYLTRIMEI'HYLAMMONIUM

By

Ines Tom-Suarez

Groundwater use for drinking water in the U.S. justifies efforts to prevent or
remediate groundwater contamination. This laboratory is developing a remediation
technique that couples increased sorption of pollutants with microbial degradation. The
overall objective of this research was to evaluate solute sorption during transport in
Oshtemo Bt2 horizon soil exchanged with hexadecyltrimethylammonium (HDTMA). In
the first part of this study, mechanisms of interaction between this soil and
pentafluorobenzoic acid (PFBA) were evaluated.

Anion exchange of I’FBA with cationic HDTMA, formed by HDTMA C16 alkyl
moieties interacting by London dispersion forces, was demonstrated using 36Cl' and
anionic PFBA in a buffer solution (pi-I 5.4). A PFBA mass balance demonstrated ion-pair
formation between PFBA dissolved in water and desorbed HDTMA. Adsorption at the
soil/water interface, adsorption at the HDTMA/water interface and partitioning between
water and the HDTMA phase were the mechanisms considered responsible for retention
of PFBA by I-IDTMA remaining in the soil after column equilibration. Adsorption at the
HDTMA/water interface was the major contributor to retention.

In the second part of the study, sorption coefficients (K) from water for

naphthalene, benzene, and TCE were extrapolated from the predicted linear relationship

between log K values obtained in water-methanol mixture from soil column breakthrough
curves (BTCs), and the corresponding methanol fraction (Rao et al., 1985). These
values were compared to sorption coefficients from sorption isotherms. The two methods
did not coincide, and the relationship between log K obtained from batch isotherms and
the corresponding methanol fraction was not linear.

Benzene sorption coefficients from water obtained from BTCs and batch isotherms
in soil with the same organic carbon content were different. Benzene also did not follow
the linear relationship between log K and methanol fractions (0 to 0.40) predicted by
Rao’s model. These results could be explained by methanol/HDTMA interaction and the
need to redefine the phase ratio for I-IDTMA-modified soils. Evidence for
methanol/HDTMA interaction was indicated by a strong affinity of methanol for the

octane/water interface.

ACKNOWLEDGMENTS

I would like to thank my advisor, Dr. Stephen A. Boyd for financial support
during my graduate work and for give me the facilities to growth in a new area. He
arranged a cooperation with the University of Florida, where I had the privilege to learn
the secrets of soil columns personally from Dr. P.S.C. Rao. I had the opportunity to
work with Dr. Maya Gugeshashvili, with her expertise I was able to study interfacial
adsorption. I also had, for some time, the technieal assistant of Christopher Gallo.
With Surash, Maya and Chris I am personally thankful.

Each one of the members of my committee (Dr. S. A. Boyd, Dr. SJ. Anderson,
Dr. V. McGuffin, Dr. J. Tiedje and Dr. M. Zabik) has contributed in a very specific
way to my professional growth. For this. I am very thankful. I specially enjoyed the
enlighting discussions with Dr. McGuffin.

I would like to acknowledge the editorial work, at different points in the writing
process, of Linda Schimmelpfennig. Dr. Sharon Anderson, Mary Ann Bruns, Drs Judith
and Ray Frankmman and Dr. Donald Christenson. Special thanks to Mary Ann for her
consistent help.

The close cooperation between Dr. Tiedje and Dr. Boyd’s groups was always
intellectually and socially rewarding, as well as, a source of friendship and moral

support. I I would like to recognize the help in multiple occasions of Darlene Johnson,

iv

Jerry Wardwell, Linda Salemka, Dr. Dave Harris, Dr. Boyd Ellis, Dr. Delbert Mokma,

and Dr. Lee Jacobs.

TABLE OF CONTENTS .

Page

LIST OF TABLES ..................................... vm
LIST OF FIGURES .................................... x
CHAPTER 1 ........................................ 1
CHAPTER 2 ........................................ 22
2 1 Introduction ................................ 22

2.2 Materials and Methods ......................... 25

2.3 Results and Discussion ......................... 29

2.4 Summary .................................. 36
References ................................. 39

CHAPTER 3 ........................................ 40
3.1 Introduction . . . . . . . ........................ 40

3.2 Materials and Methods ......................... 45

3.3 Results and Discussion ......................... 51
References ................................. 67

CHAPTER 4 ........................................ 69
4.1 Introduction ................................ 69

4.2 Theory ................................... 72

4.3 Materials and Methods ......................... 76

4.4

vii

Table 2-1

Table 2-2
Table 2-3

Table 3—1

Table 3—2
Table 3-3
Table 3—4
Table 4-1
Table 4-2

Table 4-3

Table 44

LIST OF TABLES

Page
Properties of Oshtemo Bt horizon soil. ................ 26
Anion exchange capacity (ABC) and amount of HDTMA at
saturation coverage in an HDTMA modified soil. ......... 30
Comparison of CEC values obtained by different methods. 34
Properties of Oshtemo Bt horizon soil. ................ 46
Surface tension measurement at the air-water and octanol-water
mterface. .................................. 59
Adsorption parameters of anionic PFBA in the octane-water
system at 22°C. .............................. 63
Comparison between soil surface areas measured using N2
(BET) and estiamted using PFBA. .................. 65
Properties of Oshtemo Bt horizon soil ................ 77
General characteristics of the columns ................ 80
Retardation factor (R) and sorption coefficient (Kw) obtained
from BTC. Kw obtained from batdh sorption isotherm. ..... 87

Properties of the native soil and HDTMA modified Oshtemo soil

and of soil columns prepared with these soils.

viii

Table 4-5

Table 4-6

Summary of methanol adsorption parameters at the octane—water

SummaryoforganiccarbonpresentinthesoilafterBTC

experiments from an initial organic carbon content in the soil

ix

Figure 1-1.

Figure 1—2.
Figure 1-3.

Figure 1-4.

Figure 2-1.
Figure 2-2.

Figure 2-3.

Figure 2-4.

Figure 3-1.

Figure 3-2.

Figure 3-3.

Figure 3-4.

LIST OF FIGURES

Page
Response and schematic representation of useful parameters for
pulse (A) and step (B) input functions. ................ 4
Danckwers’ F-diagrams of fluid flow thorugh columns. ..... 7
Brenner solution for equation 1. .................... ll
Graphical representaiton of R calculation using method 4 in
Nkedi-Kizza, et al., 1987. ....................... 15
Effect of dispersion in the breakthrough curve shape. ....... 23
35cr - Anion exchange on HDTMA-modified soil. ........ 31
PFBA adsorption isotherms on HDTMA-modified Oshtemo
soil. ................................... 32

HDTMA mass balance in HDTMA-modified soil and PFBA water

system ................................... 37
Schematic diagram of miscible-displacement system. ....... 47
Breakthrough curve of PFBA in Oshtemo Bt horizon soil. . . . . 52
Breakthrough curve of PFBA/buffer in HDTMA-modified

Oshtemo Bt horizon soil ......................... 53
Change in interfacial adsorption (y) as a function

of concentration. ............................. 60

Figure 3-5.

Figure 4-1.

Figure 4-2.

Figure 4-3.

Figure 4-4.

Figure 4-5.

Figure 4-6.

Figure 4-7.

Figure 4-8.

Figure 4-9.

Figure 4-10.

Figure 4-11.

Excess surface concentration (1‘) vs PFBA concentration.

Schematic diagram of miscible-displacement system.
Log of naphthalene sorption coefficients (K) at different
fraction of methanol (fc).

Effect of HDTMA soil modification in benzene breakthrough
curve (BTC) in water.
Effect of HDTMA soil modification on TCE breakthrough
curve (BTC) in water.
Benzene sorption coefficient (K) obtained from BTCs and
sorption isotherms in different fraction of methanol (fc).

TCE sorption coefficient (K) obtained from BTCs and sorption
isotherms at different proportions of methanol (fc).
Change of interfacial tension 7 in the octane/water interface

as a function of methanol concentration.

HDTMA desorption during miscible displacement experiments. .

Benzene sorption coefficient (K) estiamted form BTCs, obtained
in seven columns and from sorption isotherms obtained in soil

from these columns.
Results of fitting benzene breakthrough curve data to a

linear and quadratic equation.

xi

62

........................... 100

..................... 102

CHAPTERI

Introduction .

It is estimated that more than 40% of the U.S. population utilizes groundwater
asasourceofdrinkingwaterandthat25% ofallfreshwaterusedintheU.S.is
groundwater. Thus, it is important to understand and predict pollutant transport from the
soil surface through the unsaturated and on saturated zones to groundwater (Bouchardet
et al. , 1989). The geologic materials through which these compounds move are very
complex porous media. This porous medium has two major components: solid particles
and free space between the particles and within their pores. Organic matter and minerals
(present in clay, silt, and sand-sized particles) in different proportions constitute the soil
solids. A dissolved organic compound entering the soil will move with the bulk flow or
be retained due to interaction with the organic matter and/or the mineral components of
the soil. It may also diffuse into the small pores of soil particles. Therefore, the main
parameters to be considered in the study of the transport of organic compounds through
natural porous media are: water flux, molecular diffusion of the compounds, organic
carbon content of soil or aquifer material, and water solubility of the compounds. The
behavior of a compound in a soil or aquifer material can not be accurately described
unless all of these parameters are considered simultaneously.

’ To study the interaction of these variables at the laboratory level, soil and
environmental scientists have normally used what they call miscible-displacement
experiments. Miscible—displacement experiments are performed by passing a solution of
the solute to be studied through a column filled with the material to be characterized

(i.e. , soil, sand, aquifer material, or sediments). The change in concentration is followed

1

2

at the outlet of the column by collecting fractions for further analysis or by continuously
feeding the outlet of the column through a concentration-sensitive detector. This method
was the original form of chromatography and is also known as frontal chromatography.
The response from frontal chromatography is transformed to obtain the effluent
breakthrough curves (BTCs). Tiselius (1940) introduced the theory of frontal
chromatography which served as a base for the further development of chromatography
theories. These theories have been used to explain the behavior of compounds of
environmental concern flowing through natural porous materials.

The overall objective of this research has been to evaluate the transport of organic
contaminants in soil columns. The soil contained in the columns has been previously
modified by substituting the native inorganic exchangeable cations with a quaternary
ammonium cation, hexadecyltrimethylammonium (HDTMA). This has been shown in
batch equilibration experiments to substantially increase the sorptive properties of the soil
for nonionic organic contaminants (NOCs) (Boyd, et al., 1988, 1991; Lee et al., 1989).
In the present study, miscible-displacement or frontal chromatography has been tools
used to study transport of organic contaminants and the processes that influence their
mobility. Thus, the purpose of this chapter is to indicate how environmental scientists
have quantitatively related the chromatographic retention and dispersion parameters of

the BTCs to physical and chemical characteristics of the system.

Components of a Miscible-displacernent Experiment

The behavior of water flowing through a soil or aquifer material is of the same
nature as that of any fluid flowing through a bed of stationary solid particles. For this
reason, it is obvious that the same mathematical equations as models applied in general
to study the movement of solutes through porous materials can be used to study the flow
of organic contaminants through soil and aquifer materials. Thus, soil scientist have

taken advantage of and improved the theories advanced by investigators in other related

3
fields, such as the theories developed for chromatography and porous media (Nielson and
Biggar, 1962). In this chapter the different aspects needed to study the behavior of
organic contaminants flowing through geological materials will be considered in a

systematic order.

Selection of the Input Function.

In any chromatographic experiment, the type of chromatogram obtained (the
response) depends on the shape of the sample input profile (the input function). Reilley
et a1. (1962) studied the chromatographic response corresponding to different input
functions. Figure 1 shows the response profiles for the two most common types of input
functions, the impulse and the step functions. In A, a concentration of the sample is
introduced instantaneously in the column (impulse function). In B, the concentration of
sample entering the column increases rapidly from zero to a finite value and remains
constant at that value (step function). The step injection mode is referenced to as frontal
chromatography. Chemical engineers and soil and environmental scientists have called
this form of sample injection miscible-displacement, due to the fact that the sample is
mixed in the porous medium at the same time that it moves through the column (Nielsen
and Biggar, 1961).

Selection of the input function depends on the objectives of the analysis and the
column diameter. During miscible-displacement experiments, the solid particles restrain,
deflect, and disperse the flow. The form of the flow pattern is determined by the
geometry of the interstitial pore space. The front of the zone advancing through the
column is spread (dispersed) due to different velocities in the flow stream. The internal
structure of the porous medium also influences the diffusion rates. Therefore, it is very
important to know as many details as possible of the flow pathway (Giddings, 1965).

Two types of dispersion can occur. Radial dispersion occurs when a sample is
injected in the impulse mode. Radial dispersion is much slower than axial dispersion at

 

 

 

 

 

 

 

 

 

 

 

 

 

R! 8'0"“

 

 

 

 

l
. [rel

 

 

 

:l

. dh—c- non-“.1
. dDuOv—I. ——-0 1r
~§
r.
I
'1
1;
I
I
I
I
I
I
I
I
I
I

 

 

 

Figure 1. Response and schematic representation of useful parameters for pulse (A)
and step (B) input functions. From Riley, et al. (1962).

5

normal flow rates. Thus, the band of solute will travel through the column contained
within a central core of the packing. Depending on the column diameter and the particle
sizediameters, thecentralcorecanbesocontainedthatinmanycasesthesolutenever
reaches the column wall. Knox (1977) quantified this phenomenon with chromatographic
columns and called it the infinite diameter effect because the column behaves as it has
a very large diameter. Axial dispersion occurs more rapidly close to the walls of the
column. Therefore, in chromatography it is considered advantageous for the sample to
avoid contact with the column walls because it decreases spreading in the signal.
However, this is not the case when part of the objectives of the research is to study
dispersion effects due to column packing, as is the case with soil and aquifer materials.
This is one of the reason why during this type of research, it is more precise to use the
step mode of injection. In this injection mode the sample volume is big enough that the
solution will contact almost all the parts of the packing material. The other reason to use
the step mode instead of the impulse mode during the characterization of the dynamic
behavior of soils, is that due to the broad range of particle size present in these materials,
wider bore columns must be used to minimize walls effect. With the impulse injection
mode, the response or peak height of a compound decreases as its retention increases .
This is a dispersion effect called the column dilution factor (Knox, 1977) and is greater
in larger columns. In the Step mode. the compound solution is introduced in the column
until the response is equal to the input concentration. In studies with soil columns, the
compound is sorbed in the column until it reaches equilibrium. The response of the
solute concentration (Co) is established first outside of the column. Thus, equilibrium
is assumed to be reached when the response of the compound eluting from the column
is equal to the Co response. Therefore. to be able to use this equilibrium criterion, it
is necessary to use the step injection mode. It is also possible to use the step method to
study the effects in sorption caused by varying solute concentration (adsorption
isotherms) (Reilley et al. ,1962).

6
Now that the reasons for selecting the step function to perform the frontal
chromatographic or miscible displacement experiments with soil columns has been
established, the construction of the ETC will be considered.

Breakthrough Curves

Danckwerts (1953) studied the problems presented by steady- flow in different
systems and showed how its behavior can be investigated and quantitatively specified.
He defined the breakthrough curve (or 'F- diagram“) as the plot of F(t) vs vt/V, where
F(t) is the fraction of concentration of the compound at the outlet of the column as a
function of time and W V is the ratio between the total volume that has flowed through
the column at time t and the volumetric water content. Lapidus and Amudson (1952)
used a similar approach to represent the change in concentration at the outlet of the
column but they used the total volume of fluid instead of the volumetric water content.
Nielsen and Biggar (1962) adapted Danckwerts’ procedure in their studies with soil
columns, adopting the notation of C/Co for F(t) and referred to the volumetric ratio as
pore volumes. This is still the most common notation used today.

Lapidus and Admundson (1952) were the first to demonstrate, with
chromatographic columns, the influence of flow velocity and diffusion in the shape of the
BTCs. Danckwerts also showed that his F- diagrams (BTCs) revealed a good deal of
information about the behavior of the fluid flowing through the column. In Figure 2,
taken from his paper, the shapes of different BTCs are presented. As he indicated, they
represent different behavior of the fluid as influenced by the column packing. Figure
2a corresponds to the ideal behavior of a fluid through a column where elements of fluids
which enter the column at the same time will move through it with constant and equal
velocity on parallel paths and will leave the column at the same time. Figure 2b is
indicating deviation from ideal behavior due to some mixing of the fluid in the column.

Figure 2c corresponds to the shape of the BTC when perfect mixing of the fluid is

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o b e . d
i V
F
e r a - r o r '. F
"W —"' '
(a) Pink.» flow; (it) l'bttott flow with (c) Complete training; (cl) Dead voter.

tantra longitudinal tttixhtm

Figure 2. Danckwers’ F—diagrams of fluid flow through columns. (Danckwerts,
P.F. , 1953).

8
occurring in the column, whereas Figure 2d represents the diagram obtained where there
is dead water present in the system (stagnant water trapped in pores).

Nielson and Biggar (1962) indicated that if any solute flowing through the column
is retained within the column by any chemical or physical process, the breakthrough
curve will move to the right. This can be concluded by simple inspection of the
compound BTC. Thus, the breakthrough curve is descriptive of the relative time taken
for the displacing fluid to flow through the medium or for the solutes in the displacing

fluid to come in chemical equilibrium with the medium.

Hydrodynamic Dispersion

The ideal behavior of a fluid flowing through a column is known as piston flow
(Figure 2a). However, in reality the fluids are Newtonian in nature. There will always
be some spreading in the column profile (Danckwerts, 1953). This longitudinal
spreading is called hydrodynamic dispersion and is due to a continuous exchange of
velocities in the flow stream. The configuration of the porous medium is very complex.
It is constituted of cavities, pores of different sizes, channels, and micropores inside the
particles, all randomly distributed in the medium. Therefore, different points in the flow
stream have different velocities; this has been called the microflow pattern (Giddings,
1963). Spreading depends on how rapidly these velocities can be exchanged.

Two mechanisms determine how fast flow exchange occurs for a compound
traveling through a column without interaction with solid material (tracer). The first one
is molecular diffusion, a process by which a molecule can move from one flow path to
another. The second mechanism has been called eddy diffusion, which occurs when the
compound follows the flow path that randomly changes velocities during its travel
through the column. In chromatography, hydrodynamic dispersion has been studied
using a theory known as the coupling theory of eddy diffusion (Giddings and Robison,
1962). In porous media, it has been well established that if the macroscopic flow is

9
accepted as one-dimensional, the longitudinal dispersion may be described by the one-
dimensional diffusion equation in which the diffusion coefficient is replaced by the
coefficient of dispersion (D).

E = DE - ”E (1)
Of 512 OZ

Where,

C(x,t) is concentration in solution (mollcm3)

t is time (s)

x is distance from the inlet (cm)

D is the dispersion coefficient (cm2/s)

U is the linear velocity (cm/s)
Also the length of the bed is L and the initial solute concentration is Co.
When this equation is solved for a non-retained compound, we have the simple case of
miscible displacement. In this case, the only effect that influences the movement of the
tracer is the flow phenomenon. Any compound flowing through the system, whether it
is retained or not, will sense this effect. Brenner (1962) gave solutions to equation (1)
in a general form and in tables, where the dimensionless exit concentration is prwented
as a function of dimensionless time (T = Ut/L) for different Peclet numbers (by analogy
with the convective heat) which defined as P = 111141) where D is the axial dispersion
coefficient. There are several techniques described in the literature to determine P from
the BTC of a tracer. The two most common will be considered here. Hashimoto et al.
(1964) determine P by fitting the BTC data of a tracer to the Brenner solutions by a least
square method. Then, the best fit is obtained by interpolating in graphs such as the one
presented in Figure 3. Rose and Passioura (1971) developed a method to determine P

in which the data are transformed in a semilogarithmic form. The data fall on a straight

10
line, and P is determined from the slope. P is related to the effective diffusion
coefficient, according to Brenners’ defintion, through the equation:
p . fl (2)
4D

This equation can be solved with just the experimental data.

So far we have indicated only the effects of dispersion on the shape of the
breakthrough curve for non-retained compounds. The shape of the BTC is also
influenced by the sorption isotherm and the mass-transfer kinetics for compounds

interacting with the media.

Sorption Isotherms

The use of frontal chromatography to measure the sorption isotherm of solutes on
solid surfaces is the oldest and, for many years, the most common application of this
technique (Parcher, 1978). When a compound has a non—linear sorption isotherm,
retention will change with concentration. For this reason, only frontal chromatography
can be used for this process; To determine the adsorption isotherm, the amount of solute
adsorbed in the solid material is determined at different concentrations (Gluckauf, 1945).
As with any BTC, the effluent concentration from the column is monitored until it
reaches a constant value equal to the inlet concentration. This indicates the saturation
of the bed by the compound under study and the end of the sorption process. This
method was recently employed by Thibaud et al. (1992) to investigate the adsorption
equilibria of several volatile organic contaminants on soil. They determined the sorption
capacity of each compound on the soil, at saturation, and compared to the adsorption
capacity corresponding to a monolayer. They found that all the compounds studied had
a BET type II adsorption isotherm and that the contaminants cover the whole surface
uniformly. They also calculated, from the linear part of the adsorption isotherm, the

BRENNER SOLUTION

 

[Nrrnrcuctivc and Conservative Salutes]

 

  
 

~-u—i==co (Lotti-in; stun.)

 

 

Figure 3. Brenner solution for equation 1 (Rose and Passioura, 1971)

12
molar heats of adsorption.

When a compound has a non-linear sorption isotherm, the boundaries of its BTC
are either diffuse or self-sharpening depending of the curvature of the isotherm. Also,
if sorption takes place but the amount retained changes with concentration, our
computational procedure can produce only an empirical description of column behavior
which may possibly be useful but certainly will have no immediate theoretical
interpretation. In consequence, a condition that must be met to be able to use the
convective-dispersive equation is that the compound under study should have a linear
sorption isotherm in the medium. That is, the ratio of the amount of solute in the mobile
phase to that in the stationary phase, the sorption coefficient (Kd), should be constant and
independent of solute concentration. Reilley, et al. (1962) concluded that if a compound
has a linear isotherm in a partition chromatographic column, the retention time (t, )
remains constant over the range of concentrations used. More importantly, the shape of
response in each step, sorption and desorption, are symmetrical. Thus, if the sorption
isotherm is linear, other factors will cause changes in the shape (symmetry) of the BTC.

Retardation Factor (R)
If'linear reversible equilibrium sorption is assumed, sorbed and aqueous solute

concentrations may be related using the partition or distribution coefficient, Kd, such
that:

s = K; (3)

where S is the concentration of solute sorbed by the soil (mg/kg) and C is the
concentration of solute in solution at equilibrium (mg/L). Hashimoto et al. (1964) solved

the diffusion equation for a column of finite length under the assumption of a linear

13
sorption isotherm. They defined a dimensionless factor that they called the retardation

factor (R) expressed as:

R=1+%xy ‘ (4)

Where p is the bulk density and 0 the volumetric water content of the column. Thus, the

difusion equation can be rewritten:

2
6C_D6C 5C (5)

__ ———U—

D! 5x2 bx

The retardation factor R is a quantitative index of a chemical compound’s
mobility, in that its value is equal to the ratio of the positions of the sorbed and
non-sorbed solute in the soil column. For a non-sorbed solute, the value of the sorption
coefficienth in the equation is equal to zero: hence, R = 1. For sorbed solutes, R is
greater than one since the value of Kd is larger than zero. A compound with a large
value of R has reduced mobility in soils.

Since there was a lack of a standard method to determine the retardation factor
from BTCs, Nkedi-Kizza et. al. (1987) decided to evaluate four different methods. The
compounds selected were diuron and atrazine, two. pesticides with non-linear sorption
isotherms. In the first method, the values of R was calculated from sorption isotherms
in mixed solvents using the equation I

pK'Co 0H)
0

R" = l + 0<N<l (6)

 

Where in indicates that the eluent is not plain water but a mixed solvent solution. The

values of Km, the sorption coefficients for diuron and atrazine were obtained from the

14
sorption isotherms using the Freundilich equation, S = K CN, for non-linear isotherms.
In the second method they calculated Rm from a series of equations deduced by Rao et
al. (1985):

R. =1 + MOO!"
6

P" = P'exp(-¢o‘f) (7)
a. , ”(HSA)
H

where, 0C is the organic carbon content of the soil, Pm the solute sorption coefficient
determined from the mix solvent and normalized by the soil organic carbon content, Pw
is the solute sorption coefficient in the soil determined from water and also normalized
by the soil organic carbon content, a is a constant with a value of 0.83 for soils, 0 is as
defined in the equation, fc is the fraction of organic cosolvent, A7 the interfacial tension
between the hydrocarbon part of the solute molecule (HSA) and the solution. In the third
method Rm was determined as the number of pore volumes at 0.5 of the input
concentration. This method assumes symmetrical BTCs and is a common method used
in chromatography to determine retention time for the impulse injection mode. In the
fourth method, Nkedi-Kizza et al. , (1985) tested a method previously suggested by
Danckwerts (1953) based on the principle of mass conservation. R is equal to the area
above the BTC (Figure 4) as given by:

rm 0
lt=fo l-Cdp (s)

where, P is pore volumes and C. is the relative concentration (C/CO). After comparing
the four methods, they suggested method 4 as very reliable way to determine R from the
solute BTC.

15

 

|n_--------— -------- -- - -

Je

 

 

 

 

 

.0 'no
8 R I ‘g ['.c.]"
‘3 .
2
O
O
5 '1
3 it... {e a.
J
U
¢
'ooo
l A , A 1 I t
0 4 I I! ‘ I. an

pone vocuue m

Figure 4. Graphical representation of R calculation using method 4 in Nkedi-Kizza,
et al. , 1987.

16

During thelastfiveyears, thevalueothasveryoften determined fromthefirst
moment of the BTC. The temporal moment analysis was developed by Aris (195 8)
traditionally used in chemical engineering and chromatography. The first moment
describes mean breakthrough time and is based in the statistical first moment, which
physically corresponds to the mean of a cumulative density curve. To determine R by
this method, the curve is digitized and the data are manipulated in the usual way as to
obtain the BTC. The C/Co and pore volumes are tabulated and entered in a computer
program. The output of the program gives the first moment. The R values are obtained
from the formula:

.1. T (9)
2

Where, M is the statistic first moment, and To the sample pulse size in pore volumes.

Non Equilibrium Parameters

As it will be review in this section, many of the parameters required to run the
bicontinuum model, used to study rate limiting process in porous media, can be obtained
from BTCs obtained for the solutes in the porous material under study.

The early theory of chromatography assumed that equilibrium between adsorbent and
adsorbate solution was immediately established (Wilson, 1940). Wilson (1940) was also
the first to recognize that the sample band (impulse) may increase because the leading
edge of the band migrates too slowly due to a low rate of desorption (non-equilibrium).

However, the most comprehensive analysis of the nonequilibrium phenomena was
carried out by Giddings in 1965. These rate-limiting or nonequilibrium processes, are
also observed during transport of solutes in structured soils and aquifer materials. These
rate-limiting process have been grouped into two general classes:transport-related and

 

 

17
sorption-related. Transport-related nonequilibrium (TNE) results from the existence
of a heterogeneous flow domain. The influence of soil structure on solute transport was
recognized more than 100 years ago (Schumacher, 1864). Non-uniform flow velocity
in the field (preferencial flow, macropore flow, channeling etc.) has been generally
accepted to result from non-unimodal pore-size distribution. To model this complex
system, the porous medium is considered to have two domains: a mobile domain where
advective flow occurs and an immobile domain in which the flow is minimal. The
mobile-immobile concept for soils was first used by Gardner and Grooks (1956). Solutes
are transported rapidly in the mobile domain and slow diffusive mass transfer also occurs
between the mobile and immobile domain. Because solutes resident in the immobile
domain move to the mobile domain by diffusion only, solutes in these systems are
considered to be in a state of nonequilibrium or TNE.
Since the early sixties TNE in soils has been studied in experimentally obtained
BTCs. Thus, in 1960 Biggar and Nielsen observed, in a soil column under unsaturated
conditions, a delayed approach to relative concentrations (C/Co) values of either 1 or 0
(tailing, according to Giddings. 1963) in the BTC. They argued that under unsaturated
conditions the large porous are eliminated for transport and the amount of water that does
not move increase. This water has been called stagned water or immobile water.
Transport-related nonequilibrium has also been observed in aggrated soils, which contain
many micropores into where solutes move only by diffusion (Biggar and Nielsen, 1962).
Transport-related nonequilibrium affects both sorbing and nonsorbing solutes; thus, since
sorption related nonequilibrium is observed only for sorbing compounds, any transport
study should start with a BTC for a nonsorbing solute (tracer).
Sorption-related nonequilibrium may result from chemical nonequilibrium or from
rate-limited diffusive mass transfer. Chemical nonequilibrium is caused by rate-limited
interactiOn between the sorbate and the sorbent. This possible source of nonequilibrium

should be important in the interaction of polar organic compounds with natural soils.

18
However, for NOCs this type of interaction is considered unimportant, since organic
matter that is the part ofthe soil interacting with them behaves as a partition medium.
Three processes are considered today as responsible for nonequilibrium due to difusive
mass transfer: film diffusion, retarded intraparticle diffusion, and intrasorbent diffusion
(Brusseau et al., 1991).

Lapidus and Amudson (1952) solved the convective/dispersive equation with the
purpose of studying longitudinal diffusion in ion exchange and other chromatographic
columns. For the first case, they implied that equilibrium was estableshed at each point
in the bed while in the second case they assumed the rate of sorption is finite and follows
a first-order kinetics, described by a first-order rate constant (k2). Under equilibrium
conditions, that as the flow velocity decreases diffusion begins to affect the system and
smear out the response (Lapidus and Amundson, 1952). Hashimoto, et al. (1964)
obtained a similar solution which included the longitudinal mixing process as well as
equilibrium and linear adsorption during one—dimensional flow. Many analytical
solutions to the convective/dispersive equation can be found in the literature for a variety
of initial and boundary conditions (Van Genuchten and Parker, 1982). However, all the
solutions are based in studing BTCs obtained from miscible displacement experiment.
Some of the most recent approaches will be considered next.

Lee et al. (1988) evaluated sorption nonequilibrium during transport of
trichloroethylene ('1' CE) and p—xylene through saturated columns of two aquifer materials.
The BTC obtained with these compounds were used to evaluate a bicontinuum sorption
model. As a secondary objective of their work, they also evaluated competitive sorption
between TCE and p-xylene using BTC and batch isotherms data. Brusseau et al. (1991)
designed experiments to determine the rate-limiting or nonequilibrium sorption of NOCs
in natural sorbents. They also analyzed their miscible-displacement results (BTCs)
using a first-order bicontinuum model. From the parameters required to run their model,
they obtained P from the BTC of a tracer (tritiated water). The value of R and thus K,

19
wasobtained fromthefirstmomentoftheBTC. ThesolutepulsesizeToalsois
established when determining the BTC. Analysing their values of (k1) and K2, obtained
from the BTCs, by a linear free energy relationship, they concluded that intraorganic
matter diffusion was the major mechanism responsible for sorption nonequilibrium in

soils.

Conclusions

There is no doubt that miscible-displacement experiments are useful tools to
characterize new sorbent materials. The transport behavior of a non-reactive tracer, in
particular the shape of the BTC and the Peclet number, gives information on the porous
medium. The sorptive properties can be determined from the value of R. Also, the
mass-transfer behavior can be determined from the non-equilibrium parameters.

From the literature reviewed in general, an obvious observation is that, for the
past forty years the same methodology has been used to study the interactions of
chemical compounds in the environment. With little modification, the diffusional
equation adapted by Lapidus and Amundson in 1952 is still used to study the flow of
compounds through porous media. There has not been much cross fertilimtion of ideas
from other disciplines, that give rise to new methodologies or provide insight into

questions that persist, about the behavior of chemicals in the environment.

20

References

Aris, R., 1958. On the dispersion of linear kinematic waves. Proc. R. Soc. Iondon.
A245:268. .

Brusseau, M.L., R.E. Jessup and RC. Rao. 1991. Nonequilibrium sorption of organic
glslerrg‘ciais:2 Elucidation of the rate-limiting processes. Environ. Sci. Technol.
:1 - 4 .

Biggar, J.W., and DR Nielsen, 1962. Miscible displacement: II. Behavior of tracers.
Soil Sci. Soc. Proc. 26:125-128.

Brenner, H. 1962. The diffusion model of longitudinal mixing in beds of finite length.
Numerical values. Chem. Eng. Sci. 17:229-243.

Bouchard, D.C.,C.G. Enfield, and MD. Piwoni. 1989. Transport processes involving
organic chemicals. in Reactions and movement of chemicals in soils. SSSA
special publication No 22.

Danckwerts, P.V. 1953. Continuous flow systems. Distribution of residence times.
Chem. Eng. Sci. 2:1-12.

De Vault, D. 1943. The theory of chromatography. J. Amer. Chem. Soc. 65:532-540.

Giddings, J .C. 1963. Evidence on the nature of eddy diffusion in gas chromatography
from inert (nonsorbing) column data. Anal. Chem. 35:1338-1341.

Giddings, J .C. 1965. Dynamics of chromatography. Part 1. Principles and theory.
Marcel Dekker, Inc., New York. 323 pp.

Giddings, LC, and RA. Robison. 1962. Failure of the eddy diffusion concept of gas
chromatography. Anal. Chem. 34:885-890.

Gluckauf, E. 1945. Adsorption isotherms from chromatographic measurements. Nature.
156:748-749.

Hashimoto, I., K.B. Deshpande, and H.C. Thomas. 1964. Peclet numbers and
getzargaéiosn factors for ion exchange columns. Ind. Eng. Chem. Fundam.
: l - 1 .

Nkedi-Kizza, P., P. Suresh, C. Rao, and A.G. Honsby. 1987. Influence of organic
cosolvents on leaching of hydrophobic organic chemicals through soils. Environ.
Sci Technol. 21:1107-1111.

Lapidus, L., and N.R. Amundson. 1952. Mathematics of adsorption in beds. J. Phys.
Chem. 56:984-988.

Lee, L.S., P.S.C. Rao, M.L. Brusseau and RA. ngada. 1988. Nonequilibrium
sorption of organic contaminants during flow through columns of aquifer
materials. Env. Tox. Chem. 7:779-793.

21

Nielsen, D..,R and J. W. Biggar. 1961. Miscible displacementin soils: 1. Experiment
information. Soil Sci Proc.

Rao, P...,SC A..G Hornsby,D.P. KilcreaseandP.Nkedi-Kizza. 1985. Sorptionand
transport of hydrophobic organic chemicals in aqueous and mixed solvent
syst3errés38h340del development and preliminary evaluation. J. Environ. Qual.
14 7

Schumacher, W. Die physik des Bodens. 1864. Berlin, Germany.

Thibaud, C.; C. Erkey, and A. Akgerman. 1992. Investigation of adsorption equilibria
of volatile organics on soil by frontal analysis chromatography. Environ. Sci.
Technol. 26: 1159-1164.

Van Genuchten, M. Th., J. C. Parker. 1984. Boundary conditions for displacement
egpgaigngggs through short laboratory soil columns. Soil Sci. Soc. Am. J.
4 -

 

CHAPTER 2
Retention of Pentafiurobenzoate in Hexadecyltrimethylamonium Modified Soil by

Anion Exchange and Ion-pairing.

2.1 Introduction

The native inorganic exchangable cations on soil clays can be replaced with
quaternary ammonium cations by simple ion exchange reactions (Boyd, et a1. , 1988).
This modification increases the organic carbon content of the soils and subsoils, and
improves their sorptive properties for nonionic organic compounds (N 0Cs). Recently
the sorptive characteristics of three different soils materials (Oshtemo Bt, Mariette Bt,
and St. Clair Bt) that have been modified by the addition of
hexadecyltrimethylarnmonium bromide (HDTMA) in an amount equivalent to the CEC,
have been studied. These studies have shown that the sorptive uptake of NOCs by the
HDTMA-modified soils was greatly enhanced (lee, et al. 1989). The improvement of
the sorptive properties of the soil by chemical modification may be useful for reducing
migration of organic contaminants through soils, subsoils, and aquifer materials that have
low organic carbon content. Due to this potential application, the overall goal of this
research is to characterize the transport behavior of NOCs through an organically
modified soil, using columns packed with the chemically modified soil. For this study
we have chosen HDTMA-modified Oshtemo Bt horizon soil material.

Conservative tracers are widely used to determine the hydrodynamic properties
of a porous medium at the laboratory and field scale. A conservative tracer is a
compound that does not react chemically or biologically during transport. Typically, the
experimental evaluation of hydrodynamic dispersion consists of measuring the tracer

22

23

 

    

 

 

 

 

 

 

 

 

 

 

 

 

1 I
0.9-
0.8- ’
Piston flow
0.7-
0.6-
8 \
a 05‘ Effect of
0_4- disperssion
0.3-
0.2-
0.1r ,
o . . . /. o . . T r
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Pore Volume

Figure 1. Effect of dispersion in the breakthrough curve shape

24

concentration in eluent as a function of the number of pore volumes displaced through
the column (breakthrough curve). When a conservative tracer moves as a plug through
the column, the breakthrough curve should correspond to a vertical line (ideal piston
flow). However, breakthrough curves always deviate from ideality (Fig. 1). This is
commonly due to hydrodynamic dispersion, a process that results from mechanical
mixing and molecular diffusion (Drover, 1988; Freeze and Cherry, 1979; Giddens,
1965).

Many compounds have been used as conservative tracers in transport studies to
determine the Peclet number (P) of soils and aquifer materials. The Peclet number is a
dimensionless parameter that describes the effect of diffusion on the overall dispersion
of the column. The Peclet number is defined as P = uL/4D, where u is the fluid linear
velocity, L is the column length and D the axial dispersion coefficient of the fluid.
Pentafluorobenzoic acid (PFBA) has been widely used as a consrevative tracer since 1984
when it was evaluated with five other compounds (Bowman, 1984) to assess their
suitability as tracers in laboratory and field studies of soil water movement. PFBA and
two other fluorinated compounds were shown to move with water in a manner similar
to Br'. The fluorinated compounds have the advantage that their diffusion coefficients
are approximately 60 percent lower than Br’. This is a very important consideration in
the determination of dispersion coefficients in porous medium. Another advantage,
specifically for PFBA, is that its pKa is 1.48, which means that it will be essentially
completely anionic in most soils.

For our studies of solute transport in soil columns containing HDTMA-modified
soils, PFBA was selected for use to determine P. Although PFBA was expected to
behave as a non-retained tracer, it was strongly retained in the column. Two possible
mechanisms were identified that might account for the strong retention of PFBA in the
HDTMA-modified soil: anion exchange due to reverse charge formation (De Vivo and
Karger, 1979; Joycock and Ottewill, 1963) or ion-pairing of PFBA anion with the

25
HDTMA cation (Knox and Laird, 1976; Sood, et al., 1976), with subsequent sorption
of the neutral PFBA-HDTMA ion pair by the nonpolar alkyl tail of HDTMA.

The objective of this work was to determine the most likely mechanism(s) by
which PFBA is retained in the HDTMA-modified soil. This study is the first attempt to
understand the mechanism of interaction between quaternary alkyl ammonium modified
soils and organic acids. The elucidation of this mechanism is environmentally significant
because organic acids are often highly mobile in soils and subsoils. The retention of
organic acids, e.g. pentachlorophenol and 2,4-D, by such modified soils may be used to

reduce their transport and hence their potential for ground water contamination.

2.1 Materials and Methods
HDTMA-modified soil.

An Oshtemo Bt2 horizon (coarse-loamy, mixed, mesic Typic Hapludalfs) was
used to prepare hexadecyltrimethylammonium (HDTMA) modified soil. A 500-g of soil
material was mixed with 1000 ml water. Then an aqueous solution of HDTMA (8.2 g
in 3000 ml) was added in an amount equal to the stoichiometric cation exchange capacity
of the soil (Table 1). The suspension was stirred overnight, then transfered to centrifuge
bottles and centrifuged at 2,603xg for 20 min. The supernatant was removed using
gentle vacuum suction, and the bottles were refilled with 225 ml water. This procedure
was repeated three times to remove any excess HDTMA. After the washing process, the

soil was air dried.

l4C-HDTMA-modified soil.

The same basic procedure just described was used to prepare the 14C-HDTMA-
modified soil except that, 0.1 ml of l4C-HDTMA (specific activity 55 mCi mmol'l) was
also dissolved in the water and added to the soil together with the HDTMA required to
satisfy the CEC of the soil.

26
Table 1. Properties of Oshtemo Bt horizon soil

 

 

Particle Size (%)
Sand 89
Silt 5
Clay 6
CBC (mmol/ Kg) 45
pH 5.8
gganic Carbon Content (96) 0.1
Cation Exchange Capacity.

Method 1. The first method used for the CBC determination was reported by Lee et
a1. (1989). In summary, the CEC was taken as the summation of extractable cations:
Al3+ + Ca2+ + Mg2 + K+ + Na+ (non buffered) or H+ + Ca2+ + Mg2+ +
16‘ + Na+ (buffered). Caz+, Mg2+, K+, and Na+ were exchanged with 1N
ammonium acetate at pH 7 and then determined by atomic absorption. A13+ was

extracted with 1M KCl and determined by titration with a standarized base.

Method 2. Cation exchange capacity also was measured by ammonium saturation and
sodium displacement. The soil was saturated with 1M NH4OAc (pH 7). Entrained
solution was displaced with ethanol one time, and the entrained 14114+ concentration in
the ethanol was measured by steam distillation. Adsorbed NH4+ was displaced by
reacting the soil with an excess of NaCl. Ammonium was determined by steam
distillation. Ammonium not neutralized by the acid was back titrated with standarized
H2504 or HCl.

Method 3. Cation exchange capacity was also measured by Mg2+ saturation and 13a2+
displacement. The soil was reacted once with 0.5 M MgClz, twice with 0.1 M MgClz,
and once with 0.05 MgClz. The first three supernatant solutions were discarded, the

27
0.05 M MgC12 solution was saved, and the entrained Mg2+ concentration was
determined by atomic absorption spectroscopy. Adsorbed Mg2+ was ealculated as the
difference between the total moles of BaClz-extractable Mg2 + and the number of moles
of entrained Mg.

Anion Exchange Capacity

The anion exchange capacity of the HDTMA-modified soil, was determined by 36C1'
(ICN Biomedicals Inc., activity 0.772 mCi/ml) adsorption. Twenty ml of 0.007 N
CaC12 with 360‘ activity ranging from 6.0 x 104 to 1.2 x 104 dpm per ml were added
to 1 g of HDTMA modified Oshtemo soil. The samples were prepared in triplicate and
a blank (no soil) was prepared for each 36C1‘ concentration. The samples were shaken
for 24 hours, after which the solid and liquid phases were separated by centrifugation.
The 36C1' in 0.5 ml of supernatant was measured by liquid scintillation counting. The
amount of Cl‘ exchange was obtained by difference between the total 350' activity
present in the blanks after equilibration and the total activity present in the supernatant.
The following equation expresses to the equilibrium exchange reaction between 35Cl' and
36C1' in the modified soil where (aq) and (ex) denote aqueous and exchanger-phase

species, respectively:
350- (ex) + “Cl' (aq) .- 3"‘0' (ex) + ”Cl- (“D

= I 3"Cl'(e1t)][”(3111161);
[ ”Cl'(¢1t)][ a‘Cl'GKlH

The equilibrium constant of the isotopic reaction (Keq) is always equal to one. The
anion exchange eapacity (ABC) of the soil is equal to the [35Cl'(ex)] since 35C1’ is 99.9
96 of the Cl present. The value of ABC was calculated from the slope of a linear plot
of [35c1' (ex)] versus [35c1' (aq)], with [35c1'(aq)] = 0.007 N for this experiment:

28

35 .
["Cl’(e1)] = [—‘3‘9‘21 (“Cl-(sq):
(”Cl-(no)

Equilibrium Sorption Isotherms.

Sorption of PFBA from water and a buffer (pH 5.4) on the HDTMA-modified soil
was measured by the batch sorption isotherm method. The buffer solution was a mix of
of KHZPO4 and NaOH at concentrations of 0.5 N and 0.05 N respectively. Duplieate
l-g soil samples (for PFBA/ water) or 2-g soil sample (for PFBA/buffer) were weighed
into glass Corex centrifuge tubes, mixed with 25 ml of the PFBA solution, and closed
using teflon lined screw caps. Initial PFBA concentrations were between 50 ppm and
1,000 ug/ml. For each concentration, a no-soil blank was prepared. The samples were
shaken for 24 hrs and then centrifuged at 9,68lxg for 30 min in a Sorvall RC5-C
centrifuge at 24 °C. The aqueous-phase concentration of PFBA (Ce) was determined
using a diode-array UV/VIS spectrophotomer (Hewlett Packard, model 8452A; Palo
Alto, Ca). The amount bf PFBA sorbed in the HDTMA-modified soil (Q) was
determined by difference between initial (Co) and equilibrium (Ce) PFBA concentration
in the aqueous phase. The sorption isotherm was obtained by plotting Ce as the
independent variable and Q as the dependent variable.

The pH of the Co and Ce solutions were determined using an Orion Research

digital ion analyzer, model 701 A (Cambridge, MA.)

HDTMA Desorption.
The amount of l4C-I-IDTMA desorbed during the adsorption isotherm experiment
was determined by removing triplicate l-ml aliquots from each of the supernatant

solutions. The amount of 14C present in these samples was measured in a liquid

29
scintillation analyzer (TRI-CARB, model 1500, Canberra-Packard, Zurich, Switzerland).
The amount desorbed was ealculated based on the known activity of the 14C—HDTMA
modified soil.

2.3 Results and Discussion

In our initial experiments to characterize the hydrodynamic properties of the soil
column, we observed that PFBA was strongly retained by the HDTMA-modified
Oshtemo soil. This result was unexpected beeause PFBA is commonly used as a
conservative tracer in solute transport studies in soil columns. Also, PFBA retention was
unexpected because the pH of the 50 ppm PFBA solution was 4.8, which is slightly more
than three log units above the pKa of PFBA (1.48). At this pH, the compound is
essentially completely ionized (anionic) in solution and thus expected to be non-retained

by the soil column.

Interaction by Anion Exchange.

One possible explanation for the retention of PFBA is the presence of excess
positive charge in the HDTMA-modified soil, known as reverse charge formation.
DeVivo and Karger (1970) observed that kaolinite and montmorillonite reversed. their
charge from initially negative to positive by measuring the changes in zeta potential or
electrophoretic mobility during the reaction of the clays with dimethylarnmonium bromide
(EHDA-Br). Jaycock and Ottewill (1963) presented a model to explain the charge
reversal that they observed during the interaction of colloidal silver iodide with ionic
surfactants. DeVivo and Karger (1970) used this model to explain their results.
Essentially, a monolayer of cationic surfactant is adsorbed by electrostatic forces
neutralizing the negatively charged surface of the clay. A second layer of surfactant is
then absorbed via non-polar interactions among hydrophobic chains of the surfactant.

30
Completion of a second layer of surfactant, with ammonium groups pointing toward the
solution, results in a positively charged interface.

Based on the results above and the well known fact that CBC is not a fundamental
quantity but rather an operationally defined parameter (Drover, 1988; Rhoades, 1982),
it was postulated that the original method used (sum of exchangeble bases +Al3) may
have overestimated the HDTMA-accessible CBC resulting in the addition of an excess
amount of HDTMA during the soil modifieation. Addition of HDTMA beyond the CBC
of the soil could result in charge reversal, with the source of positive charge originating
from ionized HDTMA held by non-polar interactions with the C-16 alkyl moieties of
HDTMA adsorbed to cation exchange sites of the soil.

This hypothesis is suported by the 3601 (Fig. 2) and PFBA (Fig. 3) adsorption
data, which are summarized in Table 2. The PFBA ABC was equated with the PFBA
adsorption maxima (Arrows in Fig. 3). These results indieate that there is an excess of
positive charge in the HDTMA-modified soil, probably because the amount of HDTMA
added to the soil (45 mmolng) exceeded the number of exchange sites accessible to
HDTMA.

Table 2. Anion exchange eapacity (ABC) and amount of HDTMA at
saturation coverage in an HDTMA modified soil.

 

 

 

Experimental Source ABC (mmol/K g) Sat. Cov. (meq/ 100 g)
350' 9.0 :1: 0.24
PFBA in Buffer 7.9 :1; 0.54
PFBA in Water 16.4 :t 0.1

 

To further evaluate this possibility, the CBC of the Oshtemo soil was measured
by two other methods described in the methods section. The CBC values obtained in
each case are reported in Table 3. The variability observed among the measured CBC

Q (mmol/Kg) X10-2

Figure 2.

31

60

 

50‘

40‘

20- ' E

10“

 

 

 

T ' I ‘

0 10 20 30 40

Ce (mmol/L) X10-2

36Cl' - Anion exchange on HDTMA-modified soil

Q (mg/Kg)

Figure 3. PFBA adsorption isotherms on HDTMA-modified Oshtemo soil

32

 

 

 

 

4000
3000-1
2.... } 1
i a _
1000- E K
E
O I r I I fi I ' I
0 100 200 300 400 500 600
Ce (mg/L)

PFBAIbufler

O PFBAMater

33

values is due, in addition to the expected analytical error, to the different methods of
CBC determination. In the soil, the exchangeable cations adsorbed on negative charges
present in the aluminosilicate clay structure, at mineral edges (hydroxyl groups), and in
acid functional groups of the organic matter. Of these soil components, only the charge
present in clay minerals with isomorphic substitution is considered to be permanent. The
rest of the surface charge is variable, meaning that its magnitude depends, among other
factors, on pH, ionic strength, dielectric constant in the medium, and nature of the anion
and cation in the solution phase. Of the methods reported here, only method 3 uses
unbuffered saturating solution. This is considered an advantage because the natural soil
pH is not significantly altered (Rhoades, 1982). Also, methods 1 and 2 use NH4+ to
replace the native cations; this can result in high CBC values in soils containing
vermiculites and weathered micas, as is the case with Oshtemo Bt, because NH4+ can
replace structural K+ from clay interlayers. In method 3, the index cation is Mg2+ ,
a cation that is common in most soils and will not displace K+ fi'om micas. Therefore,
method 3 is considered to be the most precise of the analytical methods used. The way
this method corrects for . the amount of entrained saturating solution, is highly
recommended for precise determination of CBC (Rhoades, 1982). These arguments
support the conclusion that a 50 % excess [(4.5-3.0/3.0) mmol/100 g) of HDTMA was
added to the Oshtemo soil during modification. If all the excess HDTMA was retained
by tail-tail interactions, this could provide an anion exchange capacity of about 1.5
mqu 100 g, which is intermediate between the measured values reported in Table 3.

A separate independent measure of CBC (method 4) was obtained from the
difference between the organic carbon content of the modified soil and the organic carbon
of the native soil. This CBC value was corrected for the 36Cl‘ CBC. This yields a CBC
value of 2.2 meq/lOO g, which is below any of the measured CBCs. This suggests that
some HDTMA may have become bound by tail-tail interactions even before HDTMA
fully satisfied the soil CBC.

34

Table 3. Comparison of CBC values obtained by different methods.

 

 

 

Method mfg/CK g pH Sat. Cat. mm
45 7 NH4+ Yes
2 38 a: 0.2 7 NH4+ No
3 30 i 0.1 NA.“ Mg2+ Yes
4" 31 :1; 0.3 - - .-
(22 i 0.2)
aNot adjusted

bCalculated from of organic carbon contents of modified and native soil.
Value in parenthesis is corrected for the measured anion exchange capacity.

The agreement between the value of ABC calculated from the 36C1' experiment
and from the PFBA/buffer adsorption isotherm was good, supporting the conclusion that
anion exchange is responsible for the binding of PFBA by the HDTMA modified soil.
The proposed anion exchange reaction is between Br" associated with HDTMA cations
held by tail-tail interactions and the PFBA anion. The shape of the adsorption isotherm
is characteristic of ion-exchange reactions (type-I) where the added ion is strongly
preferred.

Interaction by Ion pairing. Another possible interaction will be the formation of
an ion-pair between the two ions, creating a neutral molecule in solution which would
be sorbed by nonpolar interactions with the HDTMA-coated surface. However, during
the PFBA adsorption isotherms experiments, l4C-HDTMA modified soil was used, and
the determined 14C-I-IDTMA desorbed remained essentially constant at the different
levels of PFBA. These findings argue against the formation of ion-pairs because if the
HDTMA cation is ion-pairing with PFBA its concentration in solution should decrease

with increasing concentrations of PFBA.

35

A different PFBA sorption isotherm behavior was observed in the non-buffered
system (Fig. 3). The maximum amount of PFBA adsorbed was 16.4 mmol/Kg,
approximately double the value observed with 35C1' and PFBA/buffer systems (Table 3).

Sorption of PFBA in the non-buffered systems, apparently results from the
formation of ion-pairs between PFBA and HDTMA, that has desorbed into the aqueous
phase. Knox and Laird (1976) and Haney, et al. (1976) independently introduced the
concept of using hydrophobic ions with long alkyl chains as ion-pairing reagents to
improve the chromatographic behavior of ionizable compounds. Knox and Laird (1976)
used HDTMA to ion-pair with the -SO3H group in a wide range of sulfonic acids. They
assumed that the neutral ion pair was extracted from the water-rich environment onto the
organic phase covering the porous material in the column. By analogy, HDTMA ion
pairs could be attracted to the surfaces of HDTMA-exchanged clays. The neutral ion
pairs could either be disolved in the HDTMA medium or adsorbed at the HDTMA/water
interface; either sorption process would result in the removal of PFBA from solution.
The amount of HDTMA desorbed in the blanks (soil plus water) during the PFBA/water
sorption experiments with l4C-HDTMA modified soil, was 14 mmol Kg'l. This
exceeds the amount of HDTMA held by tail-tail interactions, which was estimated as 9
mmol Kg;1 based on the 360' anion exchange capacity (Table 2), and is about one-half
of the total (initially) soil-bound HDTMA (31 mmol Kg'l; Table 1). The 35(21' ABC
capacity was about 5 mmol Kg'l less than the amount of l4C-I-IDTMA desorbed, which
means that 5 mmol kg’1 of HDTMA desorbed from cation exchange. The desorption in
pure water of 50 percent of initially bound HDTMA is consistent with other HDTMA
desorption experiments in batch or column studies where about one-half of the original
organic carbon consistently remains in soil (see chapter 4). Although the desorption
chemistry of HDTMA is not well understood, we postulate that the water- desorbable
HDTMA originates form HDTMA held by tail-tail interactions and some HDTMA held

by ion-exchange at variable charge sites or on external surfaces of clay minerals, present

36

in the soil. Desorption of the former fraction is indicated by the observation that after
exposure to pure water, the soil loses its ABC.

The amount of HDTMA desorbed into solution decreased as the concentration of
PFBA increased. HDTMA apparently forms ion-pairs with PFBA, and the ion-pair may
be either sorbed or adsorbed. To test the ion-pairing sorption hypothesis, we performed
a HDTMA mass balance. Assuming the amount of HDTMA desorbed by water is
constant and that the decrease in solution is due to formation of ion-pairs with PFBA
(1:1), then the amount of HDTMA determined in solution plus the amount of HDTMA-
PFBA ion- pairs (equal to the amount of PFBA sorbed) should equal the amount of
HDTMA desorbed into water. This mass balance was performed in each tube of the
PFBA sorption-from-water experiment, and the results (Fig. 4) indicate such an
equivalency up to an initial PFBA concentration of 300 ppm (pH 3.1). At higher initial
PFBA concentrations (500 to 1,000 ppm) and concurrently lower pHs (2.8 to 2.4), the
amount of HDTMA in solution combined with the amount forming ion-pairs is higher
than the amount of HDTMA originally released in pure water. This has been interpreted
as protonation on the oxide/hydroxide surfaces which depends on the ionic strength and
pH of the solutions (Rhoades, 1982; Drever, 1988). Thus, the protonated hydroxide will
desorb extra HDTMA into solution which can also ion-pair with PFBA.

2.4 Summary

Due to the reaction of Oshtemo soil with HDTMA, it was found that, some of the
native exchangeable cations are replaced by an ion exhange reaction with the quaternary
ammonium cation HDTMA. It was also found that the C16 alkyl moieties of the
exchanged HDTMA can interact with additional HDTMA, probably by London
dispersion forces. This interaction exposes the quaternary ammonium heads reversing

the normally negative charge of the soil to positive.

37

Desorb
ion-paring
I Total added

 

 

..,..”...,/1,,///7.,.?n.,..//A/v

,0.
Z

 

nrc wx 52.5.: BEE

150 250 300 500 700

50 100

0

Initial PFBA concentration (mg/L)

Figure 4. HDTMA mass balance in HDTMA-modified soil and PFBA water system

38

When PFBA, a widely use non-reactive tracer, was dissolved in a buffer solution
at a pH in which it is completely ionized, the negatively charged acid anion replaced Br"
anions present in the soil neutralizing the HDTMA positive charges. However, when
PFBA is dissolved in deionized water, the HDTMA interacting by weak dispersion forces
is desorbed from the soil and forms ion pairs with PFBA. These ion pairs interact with
the adsorbed HDTMA by dissolving in it or by adsorption at the HDTMA/water
interface.

39

References

Bowman, RS 1984. Evaluation of some new tracers for soil water studies. Soil Sci.
Soc. Am. J. 48, 987-993. '

Boyd, S.A., J-F. Lee, and M.M. Mortland. 1988. Nature. 333, 345-347.

DeVivo, D.G., and BL. Karger. 1970. Studies in the flotation of colloidal 'cules:
Effects of aggregation in the flotation process. Separation Sci. 5, 14 -16 .

Drever, 1.1. 1988. The geochemistry of natural waters. 2nd ed. Prentice Hall, Inc.
Bnglewood Cliffs, New Jersey. 437 p.

Freeze, R.A., and LA. Cherry. 1979. Groundwater. Prentice-Hall, Inc. Bnglewood
Cliffs, New Jersey. 604 p.

Giddings, J .C. 1965 . Dynamics of chromatography. Marcel Dekker, New York. 323
p.

Jaycock, M.J., and R.H. Ottewil. 1963. Inst. Mining Met. (Br). 72, 497.

Knox, J.H., and G.R. Laird. 1976. Soap chromatography - A new high performance
liquid chromatography technique for separation of ionizable materials. J.
Chromatogr. 122, 17-34.

Lee, J .F., J .R. Crum, and S.A. Boyd. 1989. Enhanced retention of organic contaminants
by soil exchanged with organic cations. Environ. Sci.Technol. 23, 1365-1372.

Rhoades, JD. 1982. Cation exchange capacity. In A.L. Page (ed.) Methods of soil
analysis, part 2. pp , ASA-SSSA, 677 S. Segoe Rd., Madison.

Sood, S.P., L.B. Sartori, D.P. Wittmar,_and W.G. Haney. 1976. High-pressure liquid
chromatogrphic determination of ascorbic acid in selected foods and multivitamin
products . Anal. Chem. 48, 796-798.

Swartzen-Allen, S.L. and B. Matijevis. 1975. Colloid and surface properties of clay

suspensions. ll. Electrophoresis and cation adsorption of montmorillonite.
J.Colloid. Interface. Sci. 50, 143-153.

Tomlinson, B.; T.M. Jefferies and C.M. Riley. 1978. Ion-pair high-performance liquid
chromatography. J. Chromatogr. 159, 315-358.

Wittmer, D.P., N.O. Nuessle, and W.G. Haney, Jr. 1975. Simultaneous analysis of
tartrazine and its intermediates by reversed phase liquid chromatography. Anal.
Chem. 47, 1422-1423.

CHAPTER 3
Interfacial Adsorption as a Major Mechanism Contributing to Pentafluorobenzoic
Acid Retention During Transport on a Hexadecyltrimethylammonium Modified Soil

3.1 Introduction

Quaternary methyl ammonium ions of the general form [(CH3)3NR] + or
[(CH3)2NRR’] + , where R and R’ correspond to large hydrocarbon groups, effectively
displace native inorganic cations from the cation exchange sites of soils (Boyd et al. ,
1988a, Lee at al., 1989). The replacement of the hydrated metal by alkylammonium
cations, increases the organic carbon content of the soil and changes the nature of the
the mineral surfaces from hydrophilic to organophilic. As a result, soils modified with
quaternary ammonium cations have been shown, under equilibrium conditions, to have
greatly enhanced capabilities for removing nonionic organic compounds (NOC) from
water (Lee et al. , 1989). Improvement of the sorptive properties of soils for NOCs due
to this chemical modification was demonstrated for both high and low clay content soils
(Lee et al., 1989) and for the major types of 2:1 clay minerals commonly found in soils
and subsoils (Jaynes and Boyd, 1991). For subsoils treated with
hexadecyltrimethyammonium ions, sorption coefficients for common ground water
contaminants were increased by over two orders of magnitude. Organic-matter-
n6rmalized sorption coefficients of HDTMA-treated subsoils versus native surface soils
showed the HDTMA-derived sorption phase was 10 to 30 times more effective on a unit-
mass basis than natural soil organic matter (Lee et a1. 1989). The HDTMA-derived
sorptive phase of modified soils and clays, like natural soil organic matter, acts as a
partition phase for NOCs (Boyd et al., 1988b; Jaynes and Boyd, 1991).

40

41

The ability to enhance the sorptive properties of soils and subsoils for NOCs
substantially reduces the transport potential of these compounds. This suggests an
approach for creating a sorbent zone hum; that could reduce down-gradient pollutant
concentrations in water (Boyd et al., 1991). The feasibility of underground injection of
organic cations to create a sorptive zone that could intercept an advancing contaminant
plume and immobilize the contaminants therein has been demonstrated (Burris and
Antworth, 1992). Due to the potential applications of this technology for in-site
remediation and contaminant plume management, the behavior of the NOCs under
dynamic conditions was evaluated in this research using columns filled with
hexadecyltrimethylammonium (HDTMA) modified soil.

The first step in a comprehensive study with soil columns is to determine the
contribution of column packing and flow phenomena (molecular diffusion and eddy
diffusion) to the total dispersion of solutes. This is known as the hydrodynamic
dispersion and is often evaluated with nonreactive tracers (see chapter 2). In this
approach, the Peclet number (1’) is estimated from the effluent breakthrough curve of a
nonreactive tracer. The apparent diffusion coefficient (D') then can be determined using
the equation 9 = uL/4D', where u is the linw velocity, and L is the column length.
KnowingD', equation (1) can be applied to obtain DI, the coefficient of hydrodynamic
dispersion:

DI - «,1: +Dt

where (11 is the dynamic dispersivity. u is the linear velocity of the fluid in the porous
medium (Freeze and Cherry. 1979).

Recently, pentaflourobenzoic acid (PFBA) has been used as a nonreactive tracer
to describe the movement of ground water under a wide set of conditions. Bowman and

Gibbens (1990) recommended PFBA as a nonreactive tracer, after it was compared with

42

several other fluorinated benzoic acids. While it is generally accepted that no perfect
nonreactive tracer exists, PFBA has many desirable characteristics (e. g. , biological and
chemical stability, low diffusion coefficients, etc.) and it has been rapidly accepted as a
tracer for soil column studies. Lee et al. (1991) compared the behavior of PFBA and
3H20 as nonreactive tracers to estimate P and concluded that the two compounds
function equally well. Brusseau et a1. (1991) also used PFBA and 3H20 as nonreactive
tracers to determine the P value of soil columns filled a wide range of organic carbon
contents (0.034 - 3.41%). They also found this compound functioned well as a
nonreactive tracer in all the soils studied . However, as reported in Chapter 2 of
this dissertation, PFBA was unexpectedly found to be strongly retained in a column filled
with hexadecyltrimethylammonium (HDTMA) modified soil.

The purpose of this research was to identify the mechanism(s) responsible for the
strong retention of PFBA in the HDTMA-modified soil column. In a previous study, we
found that PFBA dissolved in a buffer solution interacts with the HDTMA-modified
(Oshtemo Bt) soil. This occurred because excess positive charge was created during the
reaction of HDTMA with the soil. When HDTMA was added to the soil, some of the
native exchangeable cations of the soil are replaced with HDTMA by simple ion
exchange. reactions. Additionally, HDTMA is adsorbed by a non-electrostatic mechanism
when the C-16 alkyl moieties interact by London dispersion forces with the C-16 alkyl
moieties of l-IDTMA previously adsorbed by ion-exchange. This results in cationic
quaternary ammonium head groups at the modifier-water interface, and creates anion
exchange capacity in the modified soil. The positively charged head groups may then
interact with negatively charged PFBA, leading to its adsorption from solution.

Our previous results also indicated that, in non-buffered (deionized) water, PFBA
can ion-pair with HDTMA desorbed from the soil and be removed from the solution as
a neutral species by sorption to the HDTMA-derived organic phase still associated with

soil. These results were obtained in batch systems under equilibrium conditions with a

43

freshly modified soil. Studies of the desorption of HDTMA from the modified soil
performed with seven different columns (see chapter 4) revealed that under certain
conditions as much as 50% of the originally bound HDTMA desorbed from soil, leaving
a very stable and strongly adsorbed HDTMA fraction in the soil. It was also concluded
that HDTMA was desorbed during column pre-equilibration with water, which occurred
prior to introduction of PFBA into the column. Thus, the excess positive charge of
HDTMA-modified soil observed in batch buffered experiments was not present in the
HDTMA soil contained in the pre-equilibrated soil columns. Therefore, anion exchange
could not account for HDTMA retention in the columns. Additionally, since no more
HDTMA was desorbable from the pre-equilibrated columns, PFBA could not be retained
as an HDTMA-PFBA ion-pair.

In contrast to the retention of PFBA in our HDTMA-modified soil columns, Lee
et a1. (1991) and Brusseau et al. (1991) reported that PFBA behaves as a nonreactive
tracer even with natural soils containing high organic carbon contents. The difference
in reaction of PFBA in natural soils and the HDTMA-modified soil may be related to the
more nonpolar character of .the HDTMA-derived sorption phase. The very low polarity
of the HDTMA partition phase is illustrated by the fact that Kom values for NOC
sorption on modified soils and clays are nearly equal to Kow values, whereas K0m values
for NOC sorption on nature soils are typically about 10 times less than Kow (Bydetal,
1988; Lee et al., 1989; Boyd and Jayne, 1991). The low polarity of the HDTMA-
derived phase creates a defined interface with water, such as occurs between two
inmiscible liquids (e.g. hexane-water), as compared to natural soil organic matter where
polar functional groups may interact with water diffusing the interface.

Bjerrum (1926) postulated that in a medium of low dielectric constant, such as
hydrocarbons, ions of opposite electrical charge associate by columbic forces creating ion
pairs with a net electrical neutrality. Ion pairs can also be formed in aqueous bulk phase

but they have the tendency to move to phases with lower dielectric constants (Tomlinson

44

et al., 1978). Previous research (Kaiser and Valdmanis 1981; Westall et al., 1985;
Jafvert et al., 1990; Lee et al., 1990) has evaluated the distribution of chlorophenols
between octanol and water. These studies found that at high pH, where compounds such
as pentachlorophenol (PCP) are almost completely ionized, the amount of phenolate
transfer from the water to the octanol phase increases with the solution ionic strength.
These authors generally agreed that ion pairing between phenolate and the cations in
solution was one of the mechanisms responsible for the increased partitioning into
octanol. Transfer of organic ions from one phase to another, or from solution to the
interface, or diffuse layers between the two phases were also mentioned as possible
mechanisms. However, as J afvert et al. (1990) recognized, it is difficult to distinguish
between these mechanisms using data from conventional distribution experiments.

By drawing an analogy between octanol and the HDTMA-derived partition phase
of modified soils (Boyd et al.,1988 a, b; Lee et al., 1989; Jaynes and Boyd, 1991) the
mechanism postulated for phenolate partition between octanol and water could be
responsible for the strong retention of PFBA in HDTMA-modified soil. The goal of this
research was to evaluate the contribution of three potential mechanisms of retention: (1)
the interaction of PFBA with possible active sites in the Oshtemo soil; (2) the formation
of PFBA-electrolyte ion pairs followed by partitioning of the neutral species into the
liquid-like environment created by the hydrocarbon tails of HDTMA exchanged in soil
and (3) PFBA adsorption at the HDTMA/water interface with its hydrophobic part
oriented towards the HDTMA-derived phase and its carboxyl moiety oriented to water
where counter ions from the bulk solution can be transferred to the double layer to
maintain neutrality. To study the possible interaction of PFBA with the native soil,
PFBA retardation factor was obtained in column filled with Oshtemo soil. To assess the
contribution of PFBA ion-pair distribution in the solvent-er environment created by the
exchanged HDTMA, the partition coefficient of pentafluorobenzoate in octanol/water

system was determined. To evaluate the adsorption of the anionic PFBA at the

45
HDTMA/water interface, adsorption ofPFBAattheinterfacebetweenoctaneand water
was studied using the drop weight method (Adamson, 1960).

3.2 Materlab and Methods

All the solutions were prepared with reverse osmosis deionized water, polished
through a Milli-Q system (Milliport, Milford, MA). Octane and octanol were purchased
from Fluka (Ronkoma, NY) and PFBA from Aldrich Chemical Co. (Milwaukee, WI).
Hexadecyltrimethylammonium bromide was obtained from Sigma Chemical Co. (St.
Louis, MO), and 14C-HDTMA-Br, labelled on the terminal carbon of the hexadecyl
chain, from Moravek Biochemical Inc. (La Brea, CA); it had a specific activity of 55
mCi mmol’1 and a radiochemical purity of > 98% . All chemicals were used without
further purification.

HDTMA-modified soil.

An Oshtemo Bt2 horizon (coarse-loamy, mixed, mesic Typic Hapludalt) was used
to prepare the hexadecyltrimethylammonium (HDTMA) modified soil. The soil (500 g)
was mixed with 1000 ml water while an aqueous solution of HDTMA (8.2 g in 3000 ml)
was added in an amount equal to the cation exchange capacity of the soil (Table 1). The
suspension was stirred overnight, then transferred to centrifuge bottles and centrifuged
at 2,603 g for 20 min. The supernatant was removed using a gentle vacuum suction,
and the bottles were refilled with water (225 ml). This procedure was repeated three

times to remove any excess HDTMA. After the washing process, the soil was air dried.

1“c-rrnTMA-rnodified soil.
The basic procedure used to prepare the HDTMA-modified soil, described above,
was also used to prepare the l4C-HDTMA modified soil. In addition to the HDTMA

46
required to satisfy the soil CBC, 0.1 ml of l4C-I-IDTMA solution was dissolved in the
aqueous solution added to the soil suspesion containing 100 g of soil.

Table 1. Properties of Oshtemo Bt horizon soil

 

Particle Size (96)
Sand 89
Silt 5
Clay 6
CBC (mmol/Kg) 45
pH 5.8
flame Carbon Content 0.1

 

Miscible Displacement Experiments

Miscible displacement experiments were used to determine the hydrodynamic
dispersion of the column (Peclet number) containing nonmodified soil and to determine
the retention of PFBA in the HDTMA-modified soil.

A schematic diagram of the miscible displacement system is presented in Figure
l. The system consists of two glass bottles, one containing the solute (the sample) and
the second the elution solvent. Two single piston pumps (Model 302, Gilson Medical
Electronics Inc.) and (Model 510, Waters Ass.) are used in the constant pressure mode
for sample and solvent delivery. The pumps are connected to the column through a four-
way solvent selection valve (Model 5020, Reodyne). The inlet of the column is also
connected to the four way solvent selection valve. The outlet of the column is connected
to a flowthrough ultra-violet/visible variable-wavelength detector (Holochrome, Gilson).
The column isconnected to the system with low dead-volume Teflon fittings and tubing.

The various connections are kept as short as possible to minimize extra-column

47

 

589$ EoEoomEflee—afie he :5..ch cum—cocoa A «...—«E

 

 

assoc :20 25
A 05:.

t|/|\

 

 

 

anyhow—mo

 

 

(DU)

 

_ 2...; 232.5 _

e528 Ill/p I

 

 

 

 

 

_ ..ow not 58%

N mine

 

 

 

 

 

 

 

 

c258 58F

 

 

 

 

 

 

48
contribution to dispersion. Changes in concentration as sensed by the detector were
registered on a chart recorder (ABB model SE 120).

Three columns were used during these experiments. One was filled with non-
modified Osthetemo Bt horizon soil and two with the HDTMA-modified Oshtemo soil,
one with l4C-HDTMA and the other nonlabeled. The column filled with the nonlabeled
modified soil was used to obtain the BTC for PFBA/buffer solution, and the labeled one
to obtain a soil equilibrated under similar conditions as those used when the BTC was
performed.

The column used was a borosilicate glass preparative column (Candies,
Chromaflex) with an internal diameter of 2.5 cm and a length of 5 cm. Bed supports
consisted of woven Teflon (FE) diffusion mesh and a Teflon (TEE) filter membrane
(extra-fine pore-2.5 um) mesh and a 0.45-ttm nylon filter membrane. The fraction of
air-dried HDTMA-modified soil that passed a 1-mm diameter sieve was used to fill the
columns. The columns were dry-packed using the "rotate, bounce and tap method”
(Knox 1976) widely used for packing materials with particle sizes higher than 40 am.
The dry material was packed to obtain a uniform bulk density.

For pro-equilibration, the columns were connected, in standing vertical position,
to a piston pump (Model 302. Gilson Medical Electronics, Inc.). The columns were
equilibrated with pure water until steady-state water-saturation conditions were
established.

A buffer solution containing a mixture of 0.5 N potasium phosphate monobasic
and 0.05 N NaOH (pH 5.4) filtered through a 0.45-ttm Millipore system and degassed
with helium was used to prepare a PFBA solution (50 mg L’l) and to elute the column.
After determining the inlet solution UV deflection in the recorder (Co-response) at 235
nm, the BTC of PFBA was measured by displacing the sample solution through the
column, at a linear velocity of 90 cm h'l. The sample was displaced until the detector

49
response reached Co. Then the eluent was switched to plain buffer solution. The
column was eluted until the effluent UV response was equal to the original base line.

HDTMA desorption

During the column experiment with the 14C-l-IDTMA-modified soil, eluent was
collected and its volume determined. The l4C-activity in these solutions was determined
in triplicate by LSC.

OctanollWater Partition Coefficient

Equal volumes of buffer solution (0.5 N KI-12PO4 and 0.05 N NaOH) and octanol
(Fluka, Ronkoma, N.Y.) were allowed to equilibrate for four days. PFBA was dissolved
in the octanol-saturated buffer solution to obtain a concentration of 500 pg m1“ 1 . Three
20-ml glass vials were filled with 15 ml of the PFBA/buffer solution and allowed to
equilibrate in an incubation chamber (New Brunswick Sci., Edison, N .J .) at 25 i 1°C.
Then, 5 ml of water-saturated octanol was added to each of the vials, which were sealed
with teflon-lined septa. The vials were stirred for 48 hrs. in an incubation chamber (25
j: 1°C) at approximately 200 rpm. A sample of the octanol layer was removed from
each vial using a Paster pipette, and a sample from the water layer was drawn off with
a long stainless steel needle connected to a glass syringe to avoid disturbing the octanol
layer. The PFBA concentrations in the two layers were measured using a diode array
UV/VIS spectrophotometer (Hewlett Packard, model 8452A). The K0W was calculated
as the ratio of the concentration of PFBA in octanol to the concentration of PFBA in the
buffer solution. Also, Na+ and K+ were extracted back in water from 1 ml aliquot
of the octanol phase. The Na+ and K+ concentration in the water extract as well as in
the water phase, were determined in a plasma emission spectrophotometer (Spectra Span,
model VB). ‘

50
Interfacial Adsorption of PFBA

Equal volumes of octane and 0.5 N KHZPO4 buffer (pH 5.4) were equilibrated
for 48 hrs. A series of PFBA solutions were prepared in buffer saturated with octane
solution and equilibrated for four days with an equal volume of octane saturated with
buffer. Using these solutions, the interfacial tension at the octanol/water (buffer) and
air/water (buffer) interfaces was measured by the drop method. The interfacial tension
measurement apparatus consisted of a micrometer, syringe, capillary tube and container.
A buffer solution containing PFBA was delivered to the end of one capillary tube from
a syringe, whose plunger is operated by a micrometer. The tip of the capillary must be
ground smooth so that the end is sharp, regular, free from any nicks, and perpendicular
to the tube.

The syringe was previously calibrated to determine the volume of liquid per unit
of micrometer scale. As the liquid is delivered from the capillary to the immiscible
second phase, a drop forms and eventually breaks away, and the volume of the drop is
determined by reading the micrometer.

Absorption spectra of the bulk solutions were measured by a Hewlett Packard
(Model 8452A) Diode Array spectrophotometer to determine PFBA concentrations at
equilibrium. This value was used to correct PFBA initial concentration. T h e
densities of water and organic solutions were measured using a pycnometer with a 2 ml
volume.

The interfacial tension 7 was obtained from the equation

 

Y = “d1 'd2)8f

r

51
where (11 and d2 are the densities of the immiscible liquids studied, V is the volume of
the drop, g is the gravitational acceleration, f a correction factor (from Hrakins-Brown
table), and r the radius of the tip of the capillary. The change of interfacial tension as
a function of change in solute concentration (PFBA) was used to construct the first type
adsorption isotherm, described in detail in the Results and Discussion section.

3.3 Results and D'scussion

Oshta-o Bt horizon soil PFBA interaction

To evaluate possible interactions between PFBA and the native soil, a BTC of
PFBA dissolved in 0.01 N CaC12 was obtained in a column filled with the unmodified
soil. Figure 2 shows the BTC obtained with this column, where the relative
concentration at the column outlet is plotted against the number of column pore volumes.
The calculated retardation factor for PFBA was 1.0, which indicates that the compound
is moving with the front of the solvent and is not being retained by the soil. The shape
of the BTC is sigmoidal and symmetric indicating that nonequilibrium, as would be

indicated by tailing is not present in the system.

PFBA interactions with the HDTMA-modified soil

In contrast to the observed behavior of PFBA in columns filled with the
unmodified Oshetmo Bt horizon soil, we observed retention of PFBA in the HDTMA-
modified soil. This is shown clearly in Fig. 3 which presents the BTC for PFBA (Co
50 mg/ L or 0.24 mM in 1012904 buffer at pH 5.4) in the HDTMA-modified soil
column. The calculated retardation factor was equal to 57.4 pore volumes. It is
interesting to note that it took 26.2 pore volumes for the outlet concentration to be above
the detector signal detection limit. Some tailing is observed in the complete BTC

(sorption and desorption) indicating the presence of some mass transfer nonequilibrium.

52

 

0.9-

0.8‘

0.7-

0.6-

0.5-

C/Co

0.4- ‘

0.3-

0.2-

0.1-

 

 

 

 

 

Ivent: 0.01 N CaCI2

 

 

 

0 . 1' i 5 -
Pore volumes

fi-l

Figure 2. Breakthrough Curve of PFBA in Oshtemo Bt horizon soil

 

53

 

 

0.9- I -

0.8-

 

0.7d

 

 

 

C/Co
O
‘l'
I

0.3- - '

0.2-

0.1-‘ - $

 

 

 

50 too 150 200 250
Pore volumes

Figure 3. Breakthrough Curve of PFBA/Buffer in HDTMA-modifed Oshtemo Bt
horizon soil

54

Previous studies on sorption characteristics of NOCs in HDTMA-modified clays
and soils conducted under equilibrium conditions have shown that the organic matter
normalized partition coefficients (Kom) of NOCs sorbed by HDTMA-modified subsoils
were 10to30timesgreaterthanthoseobservedonnativesurfacesoils. Intheformer
case, the organic sorption phase is composed almost entirely by HDTMA whereas in the
latter case it is natural soil organic matter. The interacting hydrocarbon moieties of
HDTMA exchanged in soil clays forms a more effective partition phase for NOCs titan
natural soil organic matter, which contains an abundance of polar oxygen functional
groups such as hydroxyls and carboxyls. In comparison, the hydrocarbon tails of
HDTMA form a phase that closely resembles a reverse-phase chromatographic material
(e.g. C18) where the long hydrocarbon tails covalently bonded to the -SiOH functional
groups of silica gel (Snyder and Kirkland, 1978) form an organic phase that is
compositionally similar to the HDTMA-derived phase. Thus, the theory and behavior
of organic acids in reverse-phase chromatography may be useful in understanding the
interaction of anionic PFBA with the HDTMA-modified soil.

Since the development of partition chromatography (Martin and Synge, 1941)
organic acids have been studied in non-polar chromatographic materials using a
simplified model, where the compounds are assumed to partition between the polar
mobile phase and the nonpolar stationary phase covering the solid support. However,
these studies have largely ignored the role of interfacial adsorption in retention. Ion-pair
formation between hydrophobic acids and quaternary alkylammonium ions was used by
Wahlund (1975) as a method to separate hydrophobic acids in reversed-phase
chromatographic columns. Partitioning of ion pairs between water and the nonpolar
phase, previously covered with l-pentanol, was the assumed mechanism.

Adsorption of the PFBA at the HDTMA/water interface is considered here as a
plausible mechanism to account for the strong retention of PFBA. Martin (1961) was

the first to suggest that solute adsorption at the gas-liquid interface can contribute to

55
retention in gas-liquid chromatography. Adsorption of polar or ionic compounds at the
interfacebetweenthenonpolarstationaryphaseandthegasphase,inreversephasegas
chromatographic systems, was later demonstrated by Bereekin (1978). For reverse phase
high pressure liquid chromatography, however, no report was found where interfacial
adsorption has been experimentally evaluated.

To account quantitatively for the strong retention of PFBA by HDTMA-modified
soil, we attempted to consider separately the contribution of PFBA (or PFBA ion pairs)
dissolution into the HDTMA phase and contribution of interfacial adsorption. To
evaluate the magnitude of these two processes, the total retention of PFBA in the
HDTMA-modified soil was compared to (a) PFBA partitioning between octanol and
water, and (b) the interfacial adsorption of PFBA in octane-water mixtures.

Model solvent selection Under the conditions in which the ETC experiment was initially
conducted, the compound was completely ionized. This was because the pH of the
system was 4.8 which is 3 units greater than the pKa of PFBA (pKa = 1.48). Thus, in
the model binary solvent systems evaluated (octanol/water and octane/water) pH was
equal to or above 4.8, so that PFBA in the aqueous phase was completely ionized.
During the interfacial adsorption experiments phosphate buffer (pH 5.4) was used to keep
PFBA completely dissociated. At pH values where acidic compounds are nearly
completely ionized (pH > pKa + 2), the formation of ion pairs has been shown to
increase with ionic strength (Kaiser and Valdmanis 1981; Westall et al. 1985; Jafvert et
al. 1990; Lee et al. 1990). Although the high ionic strength of the buffer solution used
in these experiment (a = 0.55) would favor the formation of ion-pairs, we attempted to
minimize ion-pair formation by using monovalent cations in the buffer solution (0.5 N

potassium phosphate nonobasic and 0.05 N NaOH).

Role of Partitioning in PFBA

56

The octanol-water partition coefficient (Kow) of PFBA was determined to be
0.04. To establish if ionic pentafluorobenzoate or neutral ion pair had been transferred
to octanol, the presence of cations in the buffer solution, and in the octanol fiaction, was
determined. The presence of Na+ in the octanol phase (Table 1) indicates that PFBA
was probably transferred to the octanol phase as an ion pair with Na+ from the buffer
solution. The concentration of Na+ in octanol (36.5 mg L'l) was higher than the PFBA
concentration (20.5 mg 1:1) indicating that excess Na+ had partitioned into the octanol
fiaction. Evidence for both ion pairs and free ions in water-saturated octanol was also
reported Westall et al. (1990) in a study of the distribution of several strong electrolytes
between water and octanol. It is noteworthy that K+ , present in the buffer solution at
a concentration 10 times higher than Na+, was not detected in the octanol phase.
Westall et al. (1990) also found that Na+ more readily partitioned into octanol,
consistent with the size of the hydrated ions and the Bohr model. Since PFBA is a hard
ligand, it is expected to complex more strongly with a harder metal ion (Na+) that has
lost its hydration shell in a medium with a lower dielectric constant (octanol) than in
water. Another important factor may be the size of the ion: Na+ is smaller than K+ ,
and it will be more readily complexed with a ligand having a small bite (aperture) like
PFBA. At this point, we can only speculate about the reasons why Na+ is more
favorably transfered in the octanol phase. However, this finding is interesting enough
to justify further research.

Using the experimentally-determined Kow value, the maximum amount of PFBA
retained by the soil as a result of ion-pair partitioning was calculated as an estimate of
the contribution of this mechanism (PFBA partitioning into the HDTMA organic phase)
to the overall retention of PFBA in the soil column. The distribution coefficient of
PFBA in the soil is defined as:

57

= [pmlm
0" [PFBAJW

 

where [PFBAJHDTMA is equal to the total number of pg of PFBA per unit volume of
HDTMA (ml) and [PFBA1bufl-fl. is the number of pg of PFBA per ml of water contained
in the column. If we assume that KOM for PFBA in the soil is equal to KOW (Jaynes
and Boyd 1991) and we know the HDTMA content of the soil, the total number of
mmols of PFBA retained by the soil in the column (soil loading capacity) can be obtained
as follows:

= [_|.tg PFBA/ml HDTMA]
0" [pg PFBA/ml water]

 

The volume of HDTMA present in the column is estimated from the density (p) of

hexadecane: ‘

HDTMA weigg
p hexadecane

Vol HDTMA =

 

The HDTMA content of the soil (0.3 %) is estimated from the equilibrated solid organic
carbon content (0.4%) minus the initial organic carbon content of the soil (0.1%).
Knowing the mass of organic carbon in a mole of HDTMA (228 g), then the number of
moles of HDTMA in 100 g of soil = 0.3 gC/228 gC = 0.0013 moles/ 100 g. Assuming
that PFBA will dissolve only in the hexadecane tails, the weight of the sorptive phase in
100 g of soil will be the number of moles times hexadecane molecular weight (226.45
g), or 0.0013 moles times 226.45 g/mole = 0.299 g per 100 g of soil. The weight of
hexadecane in the experimental column containing 45.42 g soil is 0.136 g. The volume
of HDTMA in the soil column then is obtained by dividing the weight of HDTMA by

58
the density of hexadecane: 0.136 g/0.773 g/ml a: 0.176 ml. The denominator of the
distribution coefficient is:

= [Total pg PFBA - X]
[pg PFBA/ml Wt] timer content (nil)

Where X equals the mass of PFBA retained by the column or the loading capacity of the
column, and total pg PFBA (in the column) is equal to the PFBA input concentration
(52.2 pg/ml) times the volumetric water content (10.36 ml). Therefore:

[(10.36 ml 1 52.2 Int!) - X]

PF minute =
[ns BA! r] 1036 ml

 

Thus, assume Kow, determined experimentally as 0.04, is equal to Kom’ then:

K _ 0.04 = [rt/0.176 ml]
0“ [(540.79 pg - xyrosomn

Solving the equation for X:

X = 0.366 pg or 1.73 x 10" mmoles

These results indicate that the HDTMA-modified soil column could retain a maximum
of 0.366 pg PFBA or 1.73 x 10’6 mmoles by partition interactions, if all of the above

assumptions are true.

Role of Interfacial Adsorption in PFBA Retention
The most common way for studying adsorption of organic compounds at liquid/air
or liquid/liquid interfaces is based on measurement of the interfacial tension. We used

59
the drop-weight method to evaluate the potential for adsorption of PFBA at the interface
between water and the HDTMA-derived organic phase (referred to hereafter as the
”hexadecane-water interface”) of the modified soil.

Model Selection. Octanol-water has been used since 1964 (Fujita et al.) as a model
system to predict certain behavior of organic pollutants in the environment, as for
example NOC partitioning into soil organic matter and the bioaccumulation is NOCs.

We used the drop-weight method to evaluate the adsorption of PFBA at the
octanol/water interface. Table 2 summarizes the PFBA adsorption data in the

equilibrated octanol/water system and at the air/water interface.

Table 2. Surface tension measurement at the air-water and octanol-water interface.

 

System Surface tension (70) Surface tension (1) Surface pressure
mN/m mN/m (70 - y) mN/m

octanol/water 8.75 8.56 0. 19

(0.05 M PFBA)

air/water 72.39 63.21 9.18

(plus 0.05 M

PFBA)

air/water 72.39 37.51 34.88

(saturat. with

octanol)

air/water 72. 39 37.5 1 34.88

(saturat. with

octanol plus

0.05 M PFBA)

 

Interfacial tension at the octanol/water interface '70 was equal to 8.75 mN/m. The
addition of PFBA (0.05 M) did not substantially change the interfacial tension (7) of the

 

.lb

3.5- I

0.5-

 

 

U
0.606 0.608 0.01 0.612 0.014

Concentration (M)

0 0.602 0.604

Figure 4. Change in interfacial adsorption (7) as a function of concentration.

 

0L

”6

61

octanol/water system (10 - 7 = 0.19). Apparently the high surface activity of octanol
obscures the surface activity of PFBA at the octanol/water interface due to the
competition between two amphiphilic compounds. The surface activity of octanol, and
the competition between octanol and PFBA, was clearly observable at the air/water
interface. The surface tension at the air/water interface (70) is equal 72.39 mN/m. If
0.05 M PFBA was dissolved in the aqueous phase instead of octanol the decrease in
surface tension (70 - 7) was only 9.18 mN/m. If octanol saturated water is used instead
of pure water to determine the surface tension (7) at the air/water interface, the value
decreases to 37.51 mN/M (10 - 'y = 34.88), due to the high surface activity of octanol.
Thus, if PFBA is added to water saturated by octanol, the surface tension at the
air/ water interface does not change significantly (Table 2). This precludes the use of
octanol-water mixtures as a model to study PFBA adsorption at the interface between
water and the HDTMA-derived organic phase. Thus, octane water was selected as the
model system to study the interfacial adsorption of PFBA at the hexadecane/water
interface of the HDTMA modified soil.

Figure 4 represents. changes in octane/water interfacial tension as a result of
increasing PFBA concentration in solution. As the graph shows, the octane/ water
interfacial tension increases as a function of PFBA concentration indicating that PFBA
is a surface-active compound. When the adsorption isotherm is plotted as I‘, where I‘
is the excess of particles of a substance adsorbed per unit of suface area of interface in
relation to the concentration of the same substance present in the bulk solution (Gibbs,
1906) versus the PFBA concentration (Fig. 5), the maximal value of Gibbs surface
excess (I‘m) is obtained. The value of I‘m corresponds to the coverage of the interface
by a monolayer of PFBA. The minimal interfacial area per adsorbed PFBA molecule
can be calculated from the equation:

.10

I' x1 0 , mole/cm2

62

 

0.8

0.74

0.6-

0.5-

0.4-

0.34

0.2-

0.1-

 

 

 

0 0.601 0.602 0.603 0.604 0.605 0.606 0.607 0.608
Concentration (M)

Figure 5. Excess surface concentration (1‘) vs PFBA concentration.

0.609

 

0.1

31

63

r - “Np"

where N A is Avogadro number and A is the interfacial. area per adsorbed molecule.
Thus, in the compact layer of the electric double layer at the octane/water interface

All 8 (r-NA)-1

where Am is the minimal area per adsorbed PFBA molecule. This value, calculated
from the I‘m obtained from figure 5, is given in Table 3.

Table 3. Adsorption parameters of anionic PFBA in the octane-water system at 22°C.

 

 

System I‘m 10'10 Am a -AG p
M/cm2 Azlmol kJ/ mole
PFBA (0.5M ' 0.716 231.84 0.455 23.73 0.63

KB P04 + 0.05 M
Naéfl) -

The PFBA adsorption parameters at the octane/ water interface can be calculated
from the general form of the adsorption isotherm of amphiphilic compounds (Markin et
al. , 1992):

 

in - (p - 061'" a 0
_ exp(-2a0) = X expl-AG IRH
p'fl-Q)’ A

64

where p is the number of adsorbed octane molecules substituted for by one molecule of
PFBA, a is the attraction constant between the particles adsorbed, X is mole fraction of
PFBA, AGO is the standard Gibbs free energy of adsorption at equilibrium, and 0 is the
surface coverage.

Using the software Mathematica (Wolfiam Research), the adsorption parameters
p, a, and AG0 were calculated and are shown in Table 3. Time data are consistent with
anionic PFBA adsorbed at the octane-water interface with the COO'-groups oriented into
water and screened from the electric double layer by inorganic cations such as H+ , K+ ,
and Na+. The phenyl rings of PFBA are present in hydrophobic octane phase
interacting due to Van der Waals forces; this is the source of the resulting attraction
constant of 0.455. The interaction energy, 23.73 kJ per mole (5.68 Kcal/mol)
corresponds to the sum of all Van der Waals forces between the adsorbed moieties.

Estimation of Anionic PFBA Retention Due to Interfacial Adsorption

To integrate the results obtained in the octane/water system with the results
observed in the BTC, it has been assumed that, at the moment when PFBA started to
elute from the column, a compact layer of PFBA has been formed at HDTMA/water
interface .of the soil. Therefore, the minimum area A1 covered by a molecule of PFBA
at the compact layer in the model octane-water system was used to calculate the surface
area of the soil covered by the anionic PFBA, as shown below.

The number of molecules retained by the soil before any PFBA elution occurs
was calculated as follows:

V [PFBA] = mass of PFBA adsorbed at the base line

65
where V (271.53 ml) is equal to the number of base line pore volumes before PFBA
elution commenced multiplied by the volume of one pore volume. The input
concentration of PFBA [PFBA] = 52.2 pg/ml. Therefore, 251.53 ml x 52.20 pg/ml
= 14173.75 pg, or 0.067 mmols. Multiplying by Avogadros’s number gives the
corresponding number of molecules:
6.02 x 10'23 molecules/mol x 0.067 mmol/1000 = 40.243 x 1010
Now, the number of molecules is multiplied by the area covered per molecule (Table 3)
at the compact. layer: so, 4.024 x 1019 molecules. 231.84
A2/molecule = 9,330.1 x 1021 A2 To convert this value to square meter we divide by
1020:
9,330.1 x 1021 A211020 = 93.301 rn2

Thus, the maximum surface area coverage by adsorbed PFBA per gram of modified soil,
obtained after dividing by the column soil weight (45.42 g), yields a value of 2.05 m2/g.

In Table 3 the surface area of the soil calculated with the model value is
compared to the value obtained using the BET method.

Table 3. Comparison between soil surface areas measured using N2 (BET) and estimated
using PFBA

 

 

 

 

Method of Determination Surface Area (mzlg)
BET 2.36
BTC base line and area cover/molecule 2.04

 

Interestingly, a very close agreement exists between the surface area calculated
from the octane/water model and the N2 BET surface area of the modified soil. This
would seem to be a reasonable result supporting the concept of interfacial adsorption of
PFBA in the modified soil, realizing that the majority of surface area in this soil is due
to clay mineral surfaces which are presumably covered with exchanged HDTMA.

66
Conder et al. (1969) established that the different contributions to gas-
chromatographic retention (partition, adsorption at solid/liquid and liquid/liquid
interfaces) are additive, and we have adapted this idea and applied to our system. The
retardation factor (R) calculated from the BTC first moment and expressed column pore
volumes for anionic PFBA in the HDTMA-modified soil column will be:

Rtotal = (R)Ad.Soil/water + (Rldisol. + (”Ads Liquid/liquid

Where the total retention Rtotal obtained from the BTC is equal to 57.4 pore volumes,
the retention due to interaction of PFBA with the native soil (R) Ad. Soil/water is 1 pore
volume, the number of pore volumes due to dissolution (mdisol is equal to 3.4, and the
number of pore volumes due to interfacial adsorption in the HDTMA/water interface
(R) Ads lquidlliquid is equal to 26.2 pore volumes. The sum of all terms in the right side
is equal to 30.6, which is 26.8 pore volumes smaller than the Rtotal° The difference is
approximately equal to the number of pore volumes due to interfacial adsorption. This
could conceivably be accounted for by the formation of a second PFBA layer at the
interface. At any rate, these results suggest that among the different contributors to
retention of PFBA in the HDTMA modified soil, adsorption of PFBA at the hexadecane-
water inbrface of the HDTMA-soil is the primary mechanism accounting for the strong
retention of PFBA.

67
References
Adamson, A.W. 1960. Interscience Publishers, Inc. New York, N.Y.
Berezkin, V.G. 1978. J. Chromatogr. 159. 359.
Boguslasvssgii, L. 1., A.N. Frumkin, and M. I. Gugeshashvili. 1976. Electrokhimiya. 12,

Boyd, S.A., J-F. Lee, and M.M. Mortland. 1988a. Nature. 333, 345-347.
Boyd S.A., M.M. Mortland, and GT. Chiou. 1988b. Soil Sci. Soc. Am. J. 52,652.

Boyd, S.A.,W.F. Jaynes, and BS. Ross. 1991. pp. 181-200. In In R.A. Baker (ed),
vol. 1, Lewis Publishers, Chelsea, MI.

Bowman, R. S. 1984. Soil Sci. Soc. Am. J. 48, 987.
Bowman, R. S. 1992. Ground Water, 30, 8.

Brusseaué4M. g“, R. E. Jessup, and P. S. C. Rao. 1991. Environ. Sci. Technol. 26,
1 -14 .

Burris, D. R. and C. P. Antworth, 1992. J. Contain. Hydrol. 10, 325-337.
Conder, J. R., D. C. Locke, and J. H. Parnell. 1969. J. Phys. Chem. 73, 700-708.
Damaskin, B. B. 1964. Doklady Acad. Nauk SSSR, 156, 128.

Damaskin, B. B., O. Petrii, and V. Batrokov. 1971. Plenum press, New York-
I.ondon. .

Fujita, T., J. Iwasa, and C. Hansch. 1964. J. Am. Che. Soc. 86, 5175.

Getmanskii, I. K., and L. I. Bavika. Assay Methods of Aqueous Surfactant Solutions
(in Russian), 1965. Part 1, NIITEKhM, Mocow.

Gibbs, J .W. 1906. Dover Reprint, 1931. Longmans and Green & Co., New York, N.Y.

Gugeshashvili, M. 1., M. A. Manvelyan, and L. I. Boguslavskii. 1974.
Electr'okhimiya, 10, 819.

Jaynes, W.F. and S.A. Boyd. 1991. Soil Sci. Soc. Am. J. 55, 1991.
L6e,J.F., J.R. Crum, and S.A. Boyd. 1989. Environ. Sci.Technol. 23,1365-1372.

Lee, L. S., P.S.C. Rao, P. Nkedi-Kim, and JJ. Delfino. 1990. Envir. Sci. Tecnol.
24, 654.

Lee, L. S., P. S. C. Rao, and M. L. Brusseau. 1991. Envir. Sci. Tecnol. 2, 722.
Lyklema, J. 1991. Vol. I. Adacemic Press Inc. San Diego, Ca.

68

Markin, V. S., M. I. Gugeshasvili, A. G. Volkov, G. Munger, and R. Leblanc, 1992.
J. Electrochem. Soc., 139, 455.

Markin, V. S. and A. G. Volkov. 1989. Progr. Surface Sci., 30, 233.
Martin, A. J. P. and R. L. Synge. 1941. Biochem. J. 35, 1358.
Martin R.L. 1961. Anal. Chem. 33, 347.

Me Bride, M.B. Surface chemistry of soil minerals. In Minerals in soil environments.
1989. SSSA Book Series, no. 1. Madison, WI.

Mbellenberg, K., C. Leuenberger, and R. P. Schwarzenbach. 1984. Environ. Sci.
Technol. 18, 625.

Reilley, (29.8N.2, g3. P. Hildebrand, and J. W. Ashley, Jr. 1962. Anal. Chem. 34,
11 -1 l .

Rittich, B. and M. Pirochtova. 1990. J. Chromatogr. 523, 227.

Tomlingora, B., T. M. Jefferies, and C. M. Riley. 1978. J. Chromatogr. 159, 315-
5 .

Wahlund, K-G. 1975. J. Chromatogr. 115, 411-422.

CHAPTER 4
Directly Measured vs Predicted Sorption Coefficients During Transport for Nonionic
Organic Compounds in a Hexadecyltrimethylammonium Modified S011.

4.1 Introduction

Sorption and degradation (biological or chemical) of nonionic organic compounds
(NOCs) cause decreased transport through soils, thereby reducing ground water
contamination. Sorption of N OCs by soil and sediments in aqueous systems is controlled
mainly by the organic carbon content of the sorbing material and by the water solubility
of the NOC. With these facts in mind, this laboratory has been working to develop an
integrated approach to remediate contaminated soils and aquifer materials. This approach
consists of reducing transport of common ground water contaminants and then
subsequently degrading them microbiologically (Nye et a1. , 1993). To reduce transport,
the sorptive capability of sandy soils has been increased by exchanging the inorganic
native cations of the soils with quaternary ammonium compounds (cationic surfactants)
of the form (CH3)3NR where R is a large alkyl hydrocarbon. This method has been
proven to increase the sorption of N OCs on surfactant treated B-horizon soils by over
two orders of magnitude (Boyd et al. , 1988). Sorption studies with these modified soils
indicate that the surfactant-derived sorptive phases are 10 to 30 times more effective than
native soil organic matter as partition media for NOCs (Lee et al. , 1989).

The two tools most commonly used to study sorption behavior of organic
compounds in soils and sediments are batch sorption isotherms and breakthrough curves
(BTC). With these methods, the thermodynamic distribution constant, or the sorption

coefficient, K, of the compounds are determined. The sorption coefficient in the batch

69

70
method corresponds to the slope of the sorption isotherm constructed by plotting the
concentration of the sorbed compound in soil versus its concentration in water contacting
that soil at equilibrium. From the BTCs the retardation factor (R) of the compound, a
parameter indicating the mobility of a solute eluted through a soil column, is obtained.
K is determined from the R value by using the equation

R=l+r£ (1)

where p is the bulk density and 0 is the volumetric water content of the soil. The phase
ratio, p/0, is the ratio between the two phases in the system, in this case the solid phase
and the liquid phase, in which the compound flowing through the system will be
distributed. The basic difference between the two methods is that one value is obtained
under equilibrium conditions (batch method) and the other under dynamic conditions
(BTCs). However, since K is an equilibrium parameter the two values should be the
same. For natural soils, in most cases, these values have been shown to be the same
(Nkedi-Kizza et al., 1987; Wood et al., 1990). The results obtained with laboratory soil
columns are considered to provide a better understanding of the transport behavior of
NOCs through soil or aquifer materials.

Before this work, only batch sorption isotherms had been used to characterize the
impact of chemical modification, with cationic surfactants, on the sorption capacity of
soils. The first objective of this part of the study was to compare K values obtained
during transport, i.e., from BTCs, with K values obtained from sorption isotherms. To
obtain the ETC, a constant concentration of the compound must be fed into the column
until equilibrium is obtained (step function injection mode). Based on the sorption
coefficients obtained with the batch method, NOCs are expected to be highly retained in
the soils and thus, their R values are expected to be high. This suggests an operational

limitation to obtain the BTC directly from water, for volatile organic compounds,

71

because the elution volumes and hence elution times required are large. Nkedi-Kim et
al. (1985) suggested for solutes with low water solubility and large sorption coefficients
that sorption coefficients from water can be estimated fiom BTCs obtained from mixtures
of water and organic solvents using a model developed by Rao et al. (1985). This
model predicts a log-linear relationship between K and the fraction of organic solvent in
water (fa. This relationship was developed by combining the Karickhoff (1984) equation
for sorption of hydrophobic organic compounds (HOC) from water on soils and
sediments with a solubility model developed by Amidon et al. (1974) and Yalkosky et
al. (1975). A summary of the theory is presented in the next section.

The theory was used by Nkedi-Kim et al. (1985) to predict the sorption of
diuron, atrazine and anthracene from water in five different soils. They used different
proportions of methanol-water and acetone-water. They found that, as expected, a log-
linear relationship exits between the K values of the compounds and the corresponding
fractions (fc) of methanol-water. However, when different proportions of acetone-water
were used as solvent mixtures. a quadratic relationship appeared to fit the data better.

The overall goal of this part of the study was to evaluate the retention of common
ground water contaminants (benzene. naphthalene. and TCE) in soil columns containing
quaternary ammonium cation modified soils. We compared the K values determined in
columns filled with hexadecyltrimethylammonium (HDTMA) modified soil with the K
values obtained in the same soil from batch equilibrium sorption isotherms. Because
NOCs are strongly retained in the organically modified soils (Boyd et al., 1988; Lee of
al. , 1989), the approach was to use the log-linear relationship between K and fc of water-
methanol mixtures (Rao et al. , 1985) to estimate K values in pure water by extrapolation.
Because the K values, for the compounds above, obtained by extrapolation using the
model did not coincide with the K values calculated from the sorption isotherms and also
because it was observed that the organic carbon content of the soil was reduced
significantly during the BTC experiments. The values of K from effluent BTCs were
obtained directly from pure water and compared to the extrapolated Ks. Also, K values

72
were obtained by batch sorption isotherms using soil removed from the column at the end
of the BTC experiments.

4.2 Theory

The model of Rao et al. ( 1985) was developed to describe quantitatively the
sorption and transport of hydrophobic organic chemicals (HOC) dissolved in water and
mixtures of water-organic solvents in sorbent materials such as soils and sediments.
These sorbents are known to contain both mineral and organic constituents. However,
based on Karickhoff (1984) findings, it was assumed that the surfaces in direct contact
with the interstitial solution were predominantly hydrocarbonaceous in nature. The
sorbate molecules were considered to be composed of hydrocarbon and polar moieties,
with hydrocarbonaceous portions predominating. The theory considered sorbate-solvent
and solvent-solvent interactions to be most important in describing HOC sorption on soils
and sediments from single or mixed solvents. Sorption (sorbate-sorbent interaction) is
assumed to be controlled by sorbate solubility in the solvent from which sorption occurs.
Thus, the major theoretical assumption was that sorbate-solvent hydrophobic forces are
responsible for sorption. Changes in sorbent physico-chemical properties due to solvent-
sorbent interactions, as well as competition between solvent and sorbate for the sorbent
surface, were neglected in the theory (Rao et al. 1985).

Karickhoff (1984) showed that for soil and sediments, the logarithm of the HOC

sorption coefficient normalized by the soil organic carbon content (PW) is equal to:

AS
In P a in X [—1] (T. T) 0

where a and 6 are constants, X‘" is the mole fraction solubility, ASf is entropy of
fusion, Tm is the melting point (0K), T is temperature (0K) and R the gas constant
(KJ/mol 0K). Karickhoff (1984) found the values of a and B to be 0.83 and 2.142,
respectively, in soil and sediments.

73

Amidon et al. (1974) and Yalkowsky et al. (1975) modified the basic concept of
solubility of Hildebrand et al. (1970), which was used to describe solvent-solute
interactions in regular solutions, to study solubility in polar solvents. In an ideal solution
all intermolecular forces are equal and there is no change in heat or volume during
mixing. In a non-ideal solution, the deviation fiom ideal is known to arise as a result
of the unequal intermolecular forces between solvent-solvent and solvent-solute
molecules, and it is described by the activity coefficient (ac). Thus, the mole fiactional
solute solubility (X2) is expressed as:

dog X. do: xi“ + los (ac) 6’
The ac is considered to reflect the work required to remove the solute from its own
environment (W22), plus the work required to create a cavity in the solvent large enough
to contain the solute molecule (W 1 1), plus the work gained after the molecule is inserted
into the cavity (W 12). Mathematically the log of ac can be expressed as:

(W22 + W11 ' 2W12) "2"2 (4)
2.303”

 

”108 (a) =

where V2 is partial molal volume, and 411 is the solvent volume fraction. For regular
solutions, where the size and polarity of the solvent do not differ greatly, Hildebrand et
al. (1970) approximated the term W12 by the geometric mean of W11 and W22. Thus,
the equation for diluted solutions (4’1) becomes:

V2
2.303RT

 

log (ac) = - (wt? - W3): (5)

where the square roots of the work terms are known as solubility parameters 61 and 32
of the solute and the solvent, respectively. When Yalkowsky et al. (1975) considered
that for solutions in aqueous or polar solvents, the geometric mean approximation is not
valid, they proposed a two dimensional analog of the equation of Hildebrand et al.
( 1970). Thus, instead of using the geometric mean approximation to obtain W12, they

74

used the concept that the work required to remove a solute molecule from the bulk phase
is equal to the work of cohesion of the solute. This work is equal to the surface area of
the solute (A) times the surface tension of the solute (11). The work required to create
a cavity of area A in the solvent is considered to be equal to the work of cohesion of the
solvent, i.e. , A72, where 72 is the solvent surface tension. Also, the work involved in
the insertion of the solute molecule into the solvent corresponds to the work of adhesion,
which is equal to A712, where 112 is the interfacial tension between solute and solvent.
Thus, they arrived at the equation:

'12‘2 (6)
2.303RT

 

4% (ac) = -

Furthermore, considering that both interfacial tension and molecular surface area vary
with structure, they adopted the approach of Langmuir (1911) of separating the
contributions to interfacial tension of the nonpolar and polar portions of the molecule.

Thus, equation 6 becomes:

YlPAP +71?“ (7)
2.303”

 

- 108 (a) =

where 71b is the microscopic aliphatic hydrocarbon-solvent interfacial tension, and 71p
is an analogous two-dimensional term dependent on the interaction between the solvent
and the polar portion of the solute. Rao et al. (1975) adapted this basic concept of
solubility. They expressed Yalkowsky mole fraction solubility for any crystalline solute

in any pure solvent as:

m,- g _ [tr/HSA) + (e'mn _ (ll—St)“ - 7) (s)
kT RT '

where yj and ej correspond to ‘Ylh and 71p, respectively; HSA and PSA correspond
to Ah and AP, respectively, according to Yalkosky et al. (1975) nomenclature; and k is
the Boltzmann constant (kJ/K). They also used an extension of the cavity model
developed by Yalkowsky et al. (1976) to predict HOC solubility in binary solvent

75
mixtures which is expressed as a linear combination of terms representing the pair-wise

interactions of each solvent with each topographic component of the solute molecule:

In X" = 11: X" +f‘[(Ay‘HSA) + (A e‘PSA)]IkT (9)

where A 7 = (7w - 7° ); Ac 0 = (tsw - c); and the superscripts w, m and c
correspond to water, mixed-solvent, and organic miscible solvent (cosolvent),

respectively.
Rao et al. ( 1985) extended the Karickhoff (1984) sorption equation to mixed

solvents:

AS
1n P alnX [ )(T. 7). B

By substituting lnXm as defined by Yalkosky et al. (1976) in equation 9, they obtained

the following equation:

mp~=-m

 

. , 11
x" + filmy HSA)” (Ae‘PSA)] _ (35.1](1'. _ 1) _ p ( )

Next, Rao et al. (1985) defined the term

0‘ = [(Ay‘HSA)k; (AE‘PSAH (11b)

 

Combining Eq.(2) with Eq.(lla) and (11b) gives

lnP" = lnP“'- ao‘ffc (12)

The above equation expresses the log-linear relationship between the normalized sorption
coefficient and the fraction of organic cosolvent. Rao et al. (1985) extended the concept
of retardation factor to considered the effect of cosolvents on the transport of HOCs in

soils:

76

p K" (13)

 

R.=1+

where Rm is the solute retardation factor in a mixed solvent system, KIn is the non-
normalized sorption coefficient and 0m is the liquid phase content (cm3 / cm3)

They generalize Eq (11) for the non-normalized sorption coefficient and substituted in
Eq (12) rearranged giving:

log (R. - 1) .. rogue, — 1) - «07¢ (14)

This equation predicts an exponential decrease of (Rm - 1) with increasing f0 and was
the equation used in this research to obtain K in water by extrapolation.

4.3 Materials and Methods

All solutions were prepared with reverse osmosis deionized water, purified further
through a Milli-Q system (Milliport, Milford, MA). Benzene and TCE, both 99 %
purity, were purchased from Aldrich Chem. Co. (Milwaukee, WI); naphthalene and
hexadecyltrimethylammonium bromide, both 99 % purity, from Sigma Chem. Co. (St.
Louis, MO); methanol, high purity, from Burdick & Jackson (Muskegon, MI); and
octane from Fluka (Ronkoma, NY). l4C-HDTMA-Br, labelled on the terminal carbon
of the hexadecyl chain, was obtained from Moravek Biochemicals, Inc. (La Brea, CA);
it had a specific activity of 55 mCi mmol'1 and a radiochemical purity of > 98 %. All

chemicals were used without further purification.

HDTMA-modified soil.

An Oshtemo Bt2 horizon (coarse-loamy, mixed, mesic Typic Hapludalt) was used
to prepare the hexadecyltrimethylammonium (HDTMA) modified soil. The soil (500 g)
was nlixed with 1000 ml water, after which an aqueous solution of HDTMA (8.2 g in
3000 ml) was added in an amount equal to the cation exchange capacity of the soil

77
(Table 1). The suspension was stirred overnight, then transferred to centrifuge bottles
and centrifuged at 2,603 g for 40 min. The supernatant was removed using gentle
vacuum suction, and the bottles were refilled with water (225 ml). This procedure was
repeated three times to remove any excess HDTMA. After the washing process, the soil
was air dried.

Table 1. Properties of Oshtemo Bt horizon soil

 

 

Particle Size (%)
Sand 89
Silt 5
Clay 6
CBC (mmol/Kg) 45
pH 5.8
giant Carbon Content (%) 0.1

 

14C-HDT‘MA-modified soil.

The basic procedure used to prepare the nonradioactive HDTMA-modified soil
was also used to prepare the "enema modified soil. In addition to the HDTMA
required to satisfy the soil sec. 0.8 ml of “(Z-HDTMA solution (3 mCi in 31 ml
ethanol) was dissolved in the aqueous solution added to the soil suspension containing
800 g of soil. The resulting specific soil activity and organic carbon content were
determined with 50 mg of finely ground soil samples. Triplicate samples of the soil were
combusted at 1020°C in a Carbon Nitrogen -Mass spectrometer (CN-MS) analyzer
(Roboprep, Europa Scientific, U.K.). The C02 released from the samples was split, a
small portion was used in the mass spectrometer for carbon analysis, and the rest was
collected. in liquid scintillation vials and analyzed on a Packard 1500 TriCarb Liquid
Scintillation Analyzer (Packard Instrument Co., Downer’s Grove, IL). The obtained

disintegration per minute (dpm) values were corrected for background levels.

78

Miscible Displacement Experiments

Miscible displacement experiments were used to determine the hydrodynamic
dispersion (Peclet number) and the retardation factor (R) of the column containing
nonmodified soil and all the columns containing the HDTMA-modified soil used during
this study.

A schematic diagram of the miscible displacement system is presented in Figure
l. The system consisted of two glass bottles, one containing the solute (the sample) and
the second containing elution solvent. Two single piston pumps (Model 302, Gilson
Medical Electronics Inc.) with water were used in the constant pressure mode for sample
and solvent delivery. The pumps were connected to the column through a four-way
solvent selection valve (Model 5020, Reodyne). The inlet of the column was also
connected to the four-way solvent selection valve, and the outlet of the column connected
to a flowthrough (Holochrome, Gilson) ultra-violet/visible variable-wavelength detector.
The column was connected to the system with low dead-volume Teflon fittings and
Teflon tubes. The various connections were kept as short as possible to minimize extra-
column contributions to dispersion. Changes in concentration as sensed by the detector
were registered on a chart recorder (ABB model SE 120).

The column for the miscible displacement experiments performed with the native
Oshtemo Bt horizon soil was a borosilicate glass preparative column (Beckman, Altex
Division) with an intemai diameter of 2.5 cm. The column length was adjusted to 5 cm,
using a moveable plunger. Bed supports consisted of woven Teflon (TFE) diffusion
mesh and a Teflon (TFE) filter membrane (extra-fine pore-2.5 pm). For the miscible
displacement experiments performed with the HDTMA-modified soil, fit-bed columns
were used. They were also made from glass borosilicate (Candies, Chromoflex), with
an intemai diameter of 2.5 cm., and a fit length of 5 cm. Bed supports consisted of
woven FE teflon diffusion mesh and a 2.5-pm Teflon filter membrane together with a
0.45-pm nylon filter membrane. The fraction of air-dried, non-modified or HDTMA-
modified soil that passed a 1-mm diameter sieve was used to fill the columns. The

79

 

 

528$. :20 new

 

 

A an:

d

 

ESQ? 3083395058? .8 =5me cums—2.0m A Eaufi

 

i meeommo

 

 

m
m
<

 

 

2:5 9.22.5

 

 

e58 111/?

 

 

 

 

 

..ow men. ..momHN
m mize

 

 

 

 

 

 

 

 

 

 

 

.5228 .89....

 

80
columns were dry-packed using the "rotate, bounce and tap method" (Knox 1976) widely
used for packing materials with particle sizes higher than 40 pm. The dry material was
packed to obtain a uniform bulk density.

For pro—equilibration, the columns were connected, in standing vertical position,
to a piston pump (Model 302, Gilson Medical Electronics, Inc.). The non-modified
column was equilibrated with a 0.01 N CaC12 solution until steady-state water saturation
conditions were established. The modified columns were equilibrated in the same way

but with pure water. Columns specification are summarized in Table 2.

Table 2. General characteristics of the columns

 

Length (cm) 5

Diameter (cm) 2.5
Input flow rate (ml/min) 3.0
Linear velocity (cm/ hr) 91. 8

 

To obtain the BTCs in the native Oshtemo Bt horizon soil, a 0.01 N solution of
CaC12 filtered through a 0.45 pm Millipore system and degassed with helium was used
to prepare solute samples. After determining the sample solution UV deflection (Co)
outside of the column at the specific selected wavelength for each solute, the BTC was
measured by displacing the sample solution through the column at a linear velocity of
approximately 90 cm h'l . The sample was displaced until the detector response reached
Co. Then the eluent was switched to 0.01 N CaClz. The column was eluted until the
effluent UV response was equal to the original base line. The same basic procedure was
used to obtain the BTC through the HDTMA-modified soil column, with the only
difference that the solutions were prepared in pure water or different proportions of

methanol water mixtures.

81

Benzene BTCs with presmrized system

Thesamebasicapparatus (Fig. 1) andthesarnebasicprocedurewereusedtorun
miscible displacement experiments through an HDTMA-modified soil column with water
and different proportions of methanol and water. However, the solvent-sample delivery
system consisted of two glass lined bottles connected through a valve to a He tank which
degassed the solutions and maintained the head space slightly pressurized to control
evaporation of solvent and indirectly of solute. For the last part of this research, seven
columns were gradually equilibrated with increasing proportion of methanol (0 to 40 %)
in water and used to obtain benzene BTCs. At the end of each experiment each column
was unpacked and the soil air dried and used to determine benzene sorption isotherms,
described below.

Organic carbon mass balance

During the experiments with the HDTMA-modified soil, eluent from each step
of the experiment was collected, its volume determined, and the solutions analyzed for
l4C-activity in triplicate by. LSC. These data were added together and compared to the
specific activity of the soil at the beginning of the experiments and at the end of the
experiments. After the experiment, the soils from each of the seven columns were air
dried and the soil specific activity and organic carbon content were determined in them.
For these analyses, the same procedures described above to obtain the l4C-I-IDT'MA
modified soil organic carbon content and specific activity were followed.

Sorption Isotherms

Batch sorption isotherms from water were measured for solutes (benzene,
naphthalene and TCE) in the native Oshtemo Bt soil. For the HDTMA-modified
Oshtemo Bt soil, batch sorption isotherms were measured for all solutes in increasing
proportions of methanol-water (fc). In addition, benzene sorption isotherms were
measured on air-dry HDTMA-treated soil that was removed from the columns at the end

82
of the BTC experiments. Different soil:solution ratios (1:1 to 1:5) were used for the
experiment depending upon fc . Larger soilzsolution ratios were used at higher fc in
order to improve the precision of the measurement. Duplicate samples were prepared .
in 5 ml glass amber tubes with increasing concentrations of14c-labe1ed solute, up to 75
% the water solubility of the solute. The tubes were closed with teflon-lined screw-tops
and shaken for 24 h. Following this equilibration period, the solid and liquid phases
were separated by centrifugation (1,993 x g) and 1 ml aliquots of the supernatant were
analyzed by liquid scintillation counting (LSC). The concentration of solute sorbed in
the soil (Q) was determined from the difference between initial and final solution-phase

concentrations.

Methanol Interfacial Adsorption

Equal volumes of octane and water were equilibrated for 48 hours. A series of
methanol solutions were prepared in octane-saturated water and equilibrated for four days
with an equal volume of water-saturated octane. Using these solutions, the interfacial
tension at the octane/water. and air/water interfaces was measured by the drop weight
method.

The interfacial tension measurement apparatus consisted of a micrometer, syringe,
capillary tube, and container. A water solution containing methanol was delivered to the
end of one capillary tube with a syringe, with the plunger operated by a micrometer.
The tip of the capillary had to be ground smooth so that the end was sharp, regular, fiee
from any nicks, and perpendicular to the tube.

- A The syringe was previously calibrated to determine the volume of liquid per unit
of micrometer scale. As the liquid was delivered from the capillary to the immiscible
second phase, a drop formed, and on the break away, the volume of the drop was
determined by reading the micrometer. The methanol concentration after equilibrium
was determined by gas chromatography (G.C) analysis of 0.5 ml taken from the head
space of 10 ml sealed vials containing 1 ml aliquots of the different solutions. The G.C.

83
wasequippedaflameionizationdetectoratatemperatureof 150°C, andaDB—624
column with 3 pm film thickness, 30 m x 0.543 mm at a temperature of 600C. The
injection port temperature was 100°C.

The densities of water and organic solutions were measured using a picnometer

with a 2 ml volume.

4.4 Results and Discussion
Estimation of Kw

To overcome the experimental limitations expected when dealing with strongly
retained volatile organic compounds, the linear relationship between log Km and fc,
predicted by the model of Rao et al. ( 1985), discussed in the theory section was used to
estimate the sorption coefficients of several solutes from water (Kw). Thus, to obtain
Kw, the logarithms of the solutes‘ sorption coefficients obtained from BTCs and batch
equilibrium isotherms in mixtures of methanol-water were linearly extrapolated until fo
= 0.

Five NOCs were used in preliminary tests, benzene, naphthalene, toluene, p-
xylene and TCE. Among them, naphthalene had the lowest water solubility (30 ppm)
and was therefore most strongly retained in the column. requiring higher proportions of
methanol to obtain the BTCs. This allowed use of a wider range of methanol
concentrations which provided a more rigorous test of the proposed log-linear
relationship between Km and fc. Log Km values for naphthalene as a function of fc,
obtained by the two methods as well as the log Kw obtained by the batch method, are
presented in Figure 2. The logarithm of Km values estimated from the BTCs decreased
linearly as a function of fc. for the range of fc value studied (0.5 to 0.65). A linear
equation fitting these data was used to obtain the Kw value of naphthalene, which was
ca. 1280. However, the naphthalene Kw value measured directly using the sorption
isotherm method (100.9) is approximately one order of magnitude smaller than the value

calculated by direct extrapolation assuming a log-linear relationship between Km and fc

 

 

 

 

 

 

 

 

 

 

 

2A Naphthalene
5
1

 

0.9 T
0 . ‘

 

 

 

f
I

 

A
'0.5 I I I I I ‘ l
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Methanol fraction

 

I column A batch

 

 

 

Figure 2. Log of naphthalene sorption coefficients (K) at different fraction of
methanol(fc).

85
over the entire range of methanol concentrations. The Kw values obtained by the two
methods correspond physically to the equilibrium distribution constant of the compound
between the HDTMA-modified soil and the interstitial water and therefore, their values
should be the same.

An important observation is that the values of Km obtained by the two methods
are very close in value in the range of methanol concentrations 0.5 to 0.65 used to obtain
the BTCs, although they do not coincide exactly. In the lower range of methanol
concentrations, where Km values were only obtained from sorption isotherm, the log of
Km values were not linearly dependent on fc° Negative deviations from linearity were
observed below fc = 0.5. Similar results were obtained for benzene and TCE as shown
in figures 5 and 6. Deviation from linearity has been observed before with natural soils
when acetone has been used as cosolvent (Nkedi-Kizza et al., 1985), with other binary
solutions (Morris et al. , 1988), and with reversed phase chromatographic materials
(Dorsey and Dill, 1989). To verify the real behavior of the system, BTCs were obtained
from water over a complete range of fc values,including pure water (f(3 = 0).

BTCs were obtained for two solutes with the relatively high water solubilities
(SW), benzene (SW = 1,790 mg/L) and trichloroethylene (TCE) (SW = 1,100 mg/L).
The higher water solubilities of benzene and TCE as compared to naphthalene manifests
lower K values and enables the determination of BTC in pure water as well as water-
methanol mixtures. We were also able to control solute evaporation, a common problem
found with volatile compounds, using a new chromatographic commercial pump where
the solvent reservoir is pressurized, which minimizes volatilization of the solutes during
elution times associated with low fc values (including pure water). In Figures 4 and 5,
the BTCs, from water, for benzene and TCE obtained with the HDTMA-modified
Oshtemo soil are compared with the BTCs, also from water, obtained for the same
compounds in a column filled with the native Oshtemo soil. A shift to the right of the
natural Oshtemo soil BTCs is observed in the BTCs of both compounds in HDTMA-
modified soil. This shift in the BTCs corresponds to an increase in the R factor for the

86

over the entire range of methanol concentrations. The Kw values obtained by the two
methods correspond physically to the equilibrium distribution constant of the compound
between the HDTMA-modified soil and the interstitial water and therefore, their values
should be the same. 1

An important observation is that the values of Km obtained by the two methods
are very close in value in the range of methanol concentrations 0.5 to 0.65 used to obtain
the BTCs, although they do not coincide exactly. In the lower range of methanol
concentrations, where Km values were only obtained from sorption isotherm, the log of
Km values were not linearly dependent on fc. Negative deviations from linearity were
observed below fc = 0.5. Similar results were obtained for benzene and TCE as shown
in figures 5 and 6. Deviation from linearity has been observed before with natural soils
when acetone has been used as cosolvent (Nkedi-Kizza et al., 1985), with other binary
solutions (Morris et al. , 1988), and with reversed phase chromatographic materials
(Dorsey and Dill, 1989). To verify the real behavior of the system, BTCs were obtained
from water over a complete range of fc values,including pure water (fc = 0).

BTCs were obtained- for two solutes with the relatively high water solubilities
(SW), benzene (SW = 1,790 mg/L) and trichloroethylene (T CE) (Sw = 1,100 mglL).
The higher water solubilities of benzene and TCE as compared to naphthalene manifests
lower K values and enables the determination of BTC in pure water as well as water-
methanol mixtures. We were also able to control solute evaporation, a common problem
found with volatile compounds, using a new chromatographic commercial pump where
the- solvent reservoir is pressurized, which minimizes volatilization of the solutes during
elution times associated with low fc values (including pure water). In Figures 4 and 5,
the BTCs, from water, for benzene and TCE obtained with the HDTMA-modified
Oshtemo soil are compared with the BTCs, also from water, obtained for the same
compounds in a column filled with the native Oshtemo soil. A shift to the right of the
natural Oshtemo soil BTCs is observed in the BTCs of both compounds in HDTMA-

modified soil. This shift in the BTCs corresponds to an increase in the R factor for the

87
nvocompoundsandmmiscaseindicatesdlatbmzeneandTCEaremoresuongly
retained in the HDTMA-modified soil than in the native soil. The R and K values
obtained form these BTC experiments are summarized in Table 3. The R factors
increased by a factor of about 5 in the HDTMA-modified soil and the K values increased
about 15 to 30 times.

Some physical and chemical properties of the native and HDTMA-modified
Oshtemo soil are given in Table 4. The only apparent changes in the HDTMA-modified
soil are a significant increase in the organic carbon content of the soil and a slight
decrease (2x) in surface area of the soil. Thus, the increased retention is most likely due
to the increase in organic carbon content in the soil produced by the exchange of
inorganic cations by HDTMA which manifests higher sorption coefficients and hence
larger R values.

Table 3. Retardation factor (R) and sorption coefficient (Kw) obtained from
BTC. KW obtained from batch sorption isotherm.

 

 

 

 

 

 

:arameter 1 Benzene TCE
Oshtemo HDTMA-Osh Oshtemo HDTMA-
‘ Osh
R 1. 15 6.2 1.36 6.6
Kw(BTC) 0.04 1.3 0.09 1.4
Kw(batch) 0.05 10.005 3.0:t0.09 0.08:1:0.00 2.910.05

5

 

BTCs and sorption isotherm were experiments used to obtain K’s for benzene and
TCE in the HDTMA-treated soil for a range of methanol-water mixtures (fc from 0 to
0.4). Figures 6 and 7 summarize the log of Km values calculated by the two different
methods over the fc range. The Kw values obtained fiom the sorption isotherms and the
BTCs are different and the relationship between the log of Km and f6 is not a straight

88

 

 

 

 

 

1? r AAA A A
0.9“ - A A
0.8“ - A
0.7- ‘
0.6- ‘
O A -
o _ I
B 0.5 A
0.4-
0.3-
0.2- _ A
0.1-
OJ I I I I I I
0 5_ 10 15 20 25 30
Pore volumes
II non-modified soil A HDTMA-modified soil

 

 

 

Figure 4. Effect of HDTMA soil modification in benzene breakthrough curve (BTC) in
water

89

 

C/Co
O
‘l‘

I
b

 

 

 

00' s 10 125 2'0 2'5 8'0 85

Pore volumes

 

I non-modified soil A HDTMA-modified soil

 

 

 

Figure 5. Effect of HDTMA soil modification on TCE breakthrough curve (BTC) in
water

90

 

0.5:
0.4-

 

Benzene

 

 

 

0.1

 

Log K
C)
i

-0.1- _

-0.2-
I
-0.3- A

-0.4-
-O.5 l I I JI

0 0.05 0.1 0'15 02 0'25 0.8 0.85 0.4
Methanol fraction (to)

 

 

 

II BTCs A sorption isotherms

 

 

 

Figure 6. Benzene sorption coefficient (K) obtained from BTCs and sorption isotherms
in different fraction of methanol (fc).

91

 

0.5;
0.4-
0.3‘
0.2“
0.1 1

 

TCE

 

 

 

 

Log K
C)

-0.1 -
-0.2‘
-0.3~
-O.4‘

 

 

. Jr

 

-0.5

0.05 01 0'15 012 0.125 03 0.35 0.4
Methanol fraction (fc)

 

I BTC A sorption isotherm

 

 

 

Figure 7. TCE sorption coefficient (K) obtained from BTCs and sorption isotherms at
different proportions of methanol (fc)

92
line as predicted by the model of Rao’s et al., (1985). It was observed that the organic
carbon content in a sample of soil obtained from the column at the end of the miscible
displacement experiments was substantially different from the organic carbon content of
the freshly modified soil. Difference in the organic ear’oon content of the soil used to
obtain the BTC and the sorption isotherms with certainly give different Kw values.
Experiments were later designed to study this possibility.

Table 4. Properties of the native soil and HDTMA modified Oshtemo soil and of soil
columns prepared with these soils.

       

Soil Properties Natural Soil HDTMA-modified

 

Soil
Org. Carb. Cont. 0.1 i 0.01 0.8" :t 0.02
(95)

Surface Area 5.4 1; 0.1 2.4 :1; 0.08

(111 lg)
Column Properties

Bulk d 'ty p 1.61 1.60
(slot???)i

Poro' 03 - 0.4 0.4
(cm cm )

 

Organic carbon content prior to elution with water or water-methanol
mixtures.

The observed deviation from linearity in the relationship between log Km and fc
may be related to fundamental differences in the chemical composition of the sorptive
phases in native and HDTMA-modified soils. Natural organic matter has a much more
polar nature than the HDTMA-derived sorptive phase in the modified soil, which is
comprised of the C16 alkyl hydrocarbon matter. Polar functional groups (-COOH, ~OH,
-NH2) present in the soil organic matter in either the ionized or nonionized form can
interact strongly with water. This interaction decreases the interfacial resistance between

the organic matter and aqueous phase and in effect eliminates a well-defined interface.

93

These types of interactions, however, do not likely exist between the non-polar
hydrocarbon tailsofHDTMAandthepolarwater, givingrisetoamoredefinedinterface
between theHDTMA hydrocarbon tailsandtheintersticial water. Thenatureofthis
interface is similar to that between two immiscible liquids (e.g., hexane and water) and
that present between water and reverse-phase chromatographic material where
hydrocarbons are covalently bonded to the Si-OH functional groups of silica.
Amphiphilic molecules such as alcohols have high affinity for these interfaces where they
orient according to the polarity of their functional groups (Davis and Rideal, 1961). This
creates the possibility of interaction between methanol with the HDTMA-modified soil.
Examples of analogous solvent-sorbent interactions have been found in the
chromatographic literature (Wahlund, 1979; Scott and Kucera, 1977; Schoenmaher,
1983). Since solvent-sorbent interactions were explicitly ignored in the development of
the theory of Rao et al. (1985), interaction of methanol with the organic sorptive phases
of the modified soil (sorbent) may explain the deviation of our data from the log-linear
relationship predicted by their model. The next section present results of experiments
in which we have further explored the possibility of solvent/sorbent interaction.

Possible interaction or methanol with the HDTMA on the soil

The replacement of inorganic adsorbed cations by HDTMA, which has a long
hydrocarbon (C16) chain, separate soil and water by a nonpolar phase. The presence of
a sorptive phase comprised of nonpolar hydrocarbon chains that do not interact with
water creates a well defined interface between sorbed HDTMA and water. Solute
molecules transported to this interface by convective water flow in a column filled with
the HDTMA-modified soil will need to cross the interfacial barrier to contact the
nonpolar hydrocarbon phase. For HOCs there are two cooperative forces that help them
overcome the interfacial barrier: the Van der Waals attractive forces between the solute
molecules and the hydrocarbon tail of the HDTMA, and the repulsive hydrophobic force

created when a HOC tries to enter in the water and disrupt its hydrogen bond network.

94

In contrast to H003, amphiphilic molecules have a strong tendency to accumulate at the
interface, where they can orient themselves with their non-polar part interacting with the
hydrocarbon tail and the polar part oriented to the water. This orientation at the
interfacedecreases theinterfacialbarrierbetween thewaterandthehydrocarbon phase.
Methanol is an amphiphilic molecule and alcohols are well known as surface active
compounds (surfactants) (Davies and Rideal, 1961). The interfacial adsorption of
methanol could affect the sorption of N OCs by HDTMA-modified soils in methanol-
water systems. For this reason, the interfacial adsorption of methanol was studied using
the immiscible solvent system octane-water as a model for the HDTMA-water interface
in the HDTMA-treated soil.

Table 5. Summary of methanol adsorption parameters at the octane-water interface.

 

 

CMC Am - A G
System (%) Azlmolecul Kj / mole
e
Methanol in 35.8 19.4 5.33
octane-water

 

The results indicate that methanol, as a surface active compound, decreases the
interfacial tension between water and octane as a function of concentration (Fig. 8).
The decrease in the interfacial tension is nearly linear and follows Henry’s adsorption
isotherm. The adsorption parameters of methanol at the octane-water interface obtained
from these mesurements are summarized in Table 5. The formation of a single compact
layer is observed at a methanol concentration of 35.8 % . The surface area covered
(cross sectional) by each molecule at this point is equal to 19.4 A2. The low energy of
interaction (5.33 kJ/ mole) indicates that the particles are interacting by London dispersion

95

forces. The concentration of methanol at compact layer is considered to be the critical
micelle concentration (CMC). Above the CMC molecules of methanol start to interact
with each other in the aqueous phase forming micelles. The observed interfacial
adsorption of methanol in an octane-water system could by analogy occur at the
HDTMA-water interface in the modified soils. This could affect the sorption of NOCs
and in part be responsable for the relationship observed experimentally between the K
values for benzene and TCE, and fe One possible mechanism is that the dissolution of
methanol into the HDTMA phase, alone or in combination with the adsorption of
methanol at the interface could increase the overall polarity of the sorptive phase. This
would in turn cause lower sorption coefficients.

If we carefully look to all the results obtained for naphthalene, TCE, and benzene
during the experiments (Fig. 2, Fig. 5, Fig. 6), two slopes can be discerned rather than
the predicted log-linear relationship based on the work by Rao et al. (1985). In the
HDTMA-modified soil-water system, the intercept between the two lines of different
slopes occurs at fc values between 0.2 and 0.3 (20 to 30 per cent). At fc values above
this, the K values are lower than those produced by an extension of the line relating K
and fc between 0 and 0.20 methanol. One possible explanation is that at these high
methanol. concentrations (> 22 i) the dissolution of methanol into the HDTMA phase
could increase the overall polarity of the sorptive phase and hence manifest lower
sorption coefficients for NOCs.

The observation of an apparent CMC l'or methanol at a concentration of 35.8 96
is also interesting. This suggests a physical change in the methanol-water system that
might increase the solubility of the system for NOCs. For example, Kile and Chiou
(1989) showed that the solubility enhancement of DDT due to the presence of surfactants
is a two-stage process. Below the CMC, the monomeric surfactant slightly increases the
solubility of the compound in water; however, above the CMC, a sharp increase in
solubility enhancement was observed. With our system the solubility below methanol
CMC may be different than above it. Although the CMC of 35.8 % is higher than the

96

 

A Y‘ (mN/m)

 

 

 

I

6
Concentration (M)

1
1O 12

O
O
”-4
‘q
04

Figure 8. Change of interfacial tension 7 in the octane/water interface as a function of
methanol concentration

97
intercepts of the experimental lines, it does suggest the possibility of two distinct regions
of solubility. At higher fc values relative solvency may increase and account for the

change in slope observed experimentally in the plots of log K vs. fc.

Organic carbon mass balance

The results of the mass balance performed with the 14c determined in solvent
collected during the BTC and the organic carbon determined in the columns soil at the
end of the BTC are summmarized in Table 6. These results indicate that approximately
half of the HDTMA initially present in soil remains adsorbed at the end of each
experiment, regardless of methanol concentration. A plot of the amount of HDTMA
released from the column at each step of the process (Figure 9) shows that after the
equilibration process the amount of carbon eluted from the column is almost zero. These
results allow us to conclude that after the equilibration period, a very stable material with
an organic carbon content of about 0.4 percent is obtained which corresponds to about
half the original HDTMA content. The results also allow us to be sure that the organic
carbon content of the soil is the same during the miscible displacement experiments and

the corresponding sorption isotherms.

98

Table 6. Summary of organic carbon present in the soil after BTC
experiments from an initial organic carbon content in the soil

 

 

 

of 0.8 %.
95 left 96 left org.carb.cont.
Column from mass from (%)
balance org.carb.cont after BTCs
1 63.9 46.2 0.37 :I: 0.001
2 59.9 45.0 0.36 :I: 0.003
3 57.8 46.2 0.37 1; 0.003
4 (") 52.5 0.42 :l: 0.14
5 59.1 45.0 0.36 1; 0.01
6 58.6 45.0 0.36 :1: 0.003
7 53.8 43.8 0.35 i 0.10
(*) loss data

Benzene K values from BTCs in pressurized system and sorption isotherms on
equilibrated soil

The log K values for benzene obtained from BTCs in individual columns with
methanol’fractions (fc) ranging from 0 to 0.4 and from sorption isotherms on equilibrated
soil removed from the columns after the BTC experiments are presented in Figure 10.
As it was demostrated in the previous section, the organic carbon content of the soil was
the same during the two determinations of K. However, it is observed in Figure 10 that
the values of K obtained by the two methods is different. The gap between the values
of K obtained by the two methods (see Fig. 6) has decreased using soils with the same

organic but there is still an observable difference.

When the equation R = 1 + K p/0 is used to determine K values from the BTC,
the conventional approach is to use the bulk density (p), which considers the whole mass

of soil as one of the phases with where the compound is distributed. It is assumed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

re ‘

16-
8

14-
$3 . 1 Water
7,} ,2, 2 5%Meth.
< 310%Meth.
E ,0- 415%Meth.
% 520%Meth.
“5 B- 1 630%Meth.
o 740%Meth.
5’ 1
*3 6‘
a: 2
o A
‘- 4‘ ‘
g ‘A E, 4 E 6

A ' 7
2..
A
AA A A
c , r 1.4%
o 1000 2000 3000 4000 5000 6000 7000

Desorption Volume (ml)

Figure 9. HDTMA desorption during miscible displacement experiments.

100

 

 

 

 

1
#1
4i
0 -
El Sorp.lso.
x
.5
o BTC
-1 .-
‘2 ' I ' I ' I ' I '
0.0 0.1 0.2 0.3 0.4 0.5

Methanol traetlcn

Figure 10. Benzene sorption coefficient (K) estimated from BTCs, obtained in seven
columns, and from sorption isotherms obtained in soil from these columns.

101

implicitly that the compound will be uniformly distributed throughout the soil. In the
HDTMA-modified soil, the sorption of NOCs will occur in the nonpolar phase created
by the HDTMA hydrocarbon tails present on the Oshtemo soil surface. This suggests
that the whole soil density will not be the best descriptor of one of the phases where
NOCs will be distributed, which means that nature of the phase ratio has changed.
However, the K values were obtained from the BTCs were obtained using the traditional
phase ratio applied to soil columns. We believe that this change in the nature of the
phase ratio is responsible for the difference observed between the two values. However,
since we have demostrated the possibility HDTMA/ methanol interaction, calculation of
this phase ratio will be a matter of further research.

As observed previously, there is a change in slope in the log of Km versus fc at
about 22% methanol (0.2), probably explained by the solvent/sorbent interaction. For
reverse-phase chromatographic materials, which we believe resemble the HDTMA-
modified soil more than a natural soil does, a quadratic relationship between In K and
the fraction of organic solvent has been demostrated (Dill, 1987, Dorsey and Dill, 1989;
Schoenmakers, et al., 1982). Dill ( 1987) developed a theory, based on lattice statistical
thermodynamic theories, to account for retention and selectivity of solutes in reverse-
phase liquid chromatography. ln Dill's model the equilibrium constant (K) is given as
a simple quadratic function of the permtage of organic solvent. Ying et al., (1989)
applied the linearized form of Dill‘s equation to a data base of 346 data sets and found
that 80% of the data sets had R2 of 0.9 or higher and almost 50% with R2 of 0.99 or
higher. We fitted the benzene BTC data to linear and quadratic equations; the results are
presented in Figure 11. A R2 of 0.993 for the quadratic fit vs a R2 of 0.982 for the
linear indicates that the HDTMA-modified soil behaves more like a reverse-phase
chromatographic material (e.g. C13) than a natural soil.

In summary, exchange of native cations in the soil by HDTMA increases retention
of NOCs during transport. Sorption coefficients from BTCs run with methanol fractions

ranging from 0 to 0.4 were used to demostrate that linear extrapolation of the relationship

102

y . 0345 - 247x - 3.29m R"2 - 0.993

1
. y-o.42-3.79x m2. 0.982

 

 

 

8
....

9.

s:

5

'2 ' I V I '— I ' I ' I
0.0 0.1 0.2 0.3 0.4 0.5
Methanol traetlon
Figure 11. Results of fitting benzene breakthrough curve data to a linear and quadratic

equation

103
between log Km and fc can not be used to predict sorption coefficients of NOCs from
water in the HDTMA-modified soil because the relationship deviates from linearity below
30 % methanol. Interaction between methanol and HDTMA is probably responsible for
this behavior. It was also found that benzene sorption coefficients determined directly
from water using BTCs and sorption isotherms are different. We believe that these
values are different because I(W is calculated from the BTC using the phase ratio (p/O)
applied to natural soils. In the modified soil, the nature of the phase ratio appears to be
different, although it is difficult if not impossible to calculate with the present level of
knowledge. The relationship between log K vs fc fits a quadratic better than a linear
equation, reinforcing our belief that the modified soil resembles more a reverse-phase

chromatographic material than a natural soil.

104
References

Amidon, G.L., S.H. Yalkosky, and S. Leung. 1974. Solubility of nonelectrolytes in
1133?; solvents ll: solubility of aliphatic alcohols in water. J. Pharm. Sci. 63,

Axilrod, R.M. and E. Teller. 1943. Interaction of the van der Waals type between three
atoms. J. Chem. Phys. 11, 299.

Boyd S.A., J-F. Lee, and M.M. Mortland. 1988. Attenuating organic compound
mobility by soil modification. Nature. 333, 345.

Davis, J.T:7and E.K. Rideal. Interfacial phenomena. Academic Press, New York, 1961
p 4 4.

Dill, K.A. 1987 . The mechanism of solute retention in reversed-phase chromatoraphy.
J. Phys. Chem. 1980-1988

Dorsey J .G. and K.A. Dill. 1989. The molecular mechanism of retention in reversed-
phase liquid chromatography. Chem. Rev. 89, 331.

Halicioglu, T. and O. Sinanoglu. 1969. Solvent effects on cis-trans azobenzene
isomerization: A detailed application of a theo of solvent effects on molecular
associations. Ann. New York Academic Sci. 1 8, 308.

Hildebrand, J .H., J .M. Prausnits, and R.L. Scott. Regular and related solutions. Van
Nostrand, New York, 1970.

Horvath, C. , W. Melander and I. Molnar. 1976. Solvophobic interactions in liquid
chromatography with nonpolar stationary phases. J. Chrom. 125, 129.

Karickhogf, SW. 1984. Pollutant sorption in aquatic systems. J. Hydraulic Eng. 110,
7 7.

Kestrler, N.R. and O. Sinanoglu. 1962. Effective intermolecular pair potentials in
nonpolar media. J. Chem. Phys. 38, 1730.

Kile, D.E. and CT. Chiou. 1989. Water solubility enhancements of DDT and
trichlorobenzene by some surfactants above and below the critical micelle
concentration. Environ. Sci. Technol. 23, 832.

Knox, J .H. 1976. Practical high performance liquid chromatography. Ed. Simpson,
Heyden & Son Ltd. , London.

Lee, 1.1=., J.R. Crum, and S.A. Boyd. 1989. Environ. Sci. Technol. 23, 1365.

McCormick, R.M. and B.L. Karger. 1980. Distribution phenomena of mobile-phase
components and determination of dead volume in reversed-phase liquid
chromatography. Anal. Chem. 52, 2249.

Morris, K.R.; R. Abramowitz, R. Final, P. Davis, and S.H. Yalkowsky. 1988.
Solubility of aromatic pollutants in mixed solvents. Chemosphere. 17, 285.

105

Nkedi-Kim, P., P.S.C. Rao, and A.G. Homsby. 1985. Influence of organic
cosohlll‘reiltslgn évigigption of hydrophobic organic chemicals by soils. Environ. Sci.
Tee 0 . , .

Nkedi-Kim, P., P.S.C. Rao, and A.G. Homsby. 1987. Influence of organic
cosolvents on leachin87of hydrophobic organic chemicals through soils. Environ.
Sci. Technol. 21, 11 . -

Nogare, SD. and J. Chiu. 1962. A study of the performance of packed gas
chromatography columns. Anal. Chem. 34, 890.

Nye, J.V. Biocompatibility of quaternary ammonium cations used to enhance the
imlgobilization of nonionic organic contaminants in soils. M.Sc., Thesis. 1993.
M. .U.

Rao, P.S.C., A.G. Hornsby, D.P. Kilcrease, and P. Nkedi-Kim. 1985. Sorption and
transport of hydrophobic organic chemicals in aqueous and mixed solvent
sys6tems: Model development and preliminary evaluation. J. Environ. Qual. 14,
37 .

Schoenmakers, P.J., H.A.H. Billiet and L. de Galan. 1983. Description of solute
detention over the full range of mobile phase com sitions in reversed-phase
liquid chromatography. J. Chromatography, 282, l .

Scott, R.P.W. and P. Kucera. 1977. Examination of five commercially available
chromatographic reversed phases (including the nature of the solute-solvent-
stationary phase interactions associated with them. J. Chromatography, 142, 213.

Slaats, E.H., W. Markovski, J. Fekete and H. Poppe. 1981. Distribution equilibria of
solvent components in reversed-phase liquid chromatographic columns and
relationship with the mobile phase volume. J. Chromatography, 207, 299.

Wahlund, K-G. 1975 . Reversed-phase ion-pair partition chromatography of carboxilates
and sulphonates. J. Chromatogr. 115, 411-422.

Wekhove-Goewie, C.E., U.A.Th. Brikman and R.W. Frei. 1981. Trace enrichment of
polar compounds on chemically bonded and carbonaceous sorbents and application
to chlorophenols. Anal. Chem. 53, 2072.

Wood, A.L., D.C. Bouchard, M.L. Brusseau, and P.S.C. Rao. 1990. Cosolvent effects
on sorption and mobility of organic contaminants in soils. Chemosphere. 21 ,
575.

Yalkowsky, S.H., G.L. Amidon, G. Zografi, and G.L. Flynn. 1975. Solubility of
nonelectrolytes in polar solvents lll: alkyl p-amonobenzcates in polar and mixed
solvents. J. Pharm. Sci. 64, 48.

 

Yaynes, W.F. and S.A. Boyd. 1991. Clay mineral type and organic sorption by
hexadecyltrimathylammonium-exchanged clays. Soil Sci. Soc. Am. J. 55, 1991.

Ying, P.T.,J.G. Dorsey, and K.A. Dill. 1989. Retention mechanisms of reversed-phase
liquid chromatography: Determination of solute-solvent interaction free
energies. Anal. Chem. 61, 2540-2546.

 

MICHIGAN $7an UNIV. LIBRnRIEs
illWIN“liIUIIHWIiIWII”lllllllllllllllllllmlll
31293010330409