Thermodynamics and mechanisms of sorption for hydrophobic organic compounds on natural and artificial sorbent materials


Material Information

Thermodynamics and mechanisms of sorption for hydrophobic organic compounds on natural and artificial sorbent materials
Physical Description:
xv, 339 leaves : ill. ; 28 cm.
Woodburn, Kent Benson, 1956-
Publication Date:


Subjects / Keywords:
Sorbents   ( lcsh )
Liquid chromatography   ( lcsh )
Organic water pollutants   ( lcsh )
Groundwater -- Pollution   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1985.
Includes bibliographical references (leaves 325-338).
Statement of Responsibility:
by Kent Benson Woodburn.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000898075
notis - AEK6766
oclc - 15477413
System ID:

Full Text







Copyright 1985


Kent Benson Woodburn


My sincere and heartfelt thanks go to my advisor,

Dr. Joseph Delfino, whose guidance, scientific insight, and

encouragement were instrumental in the completion of this

project. I would also like to thank Dr. Suresh Rao,

Dr. Peter Nkedi-Kizza, and Dr. Arthur Hornsby for their

generous assistance and helpful comments and criticisms.

Special thanks go to Dr. Rao for his guidance and support

over the past two years and Dr. John Dorsey for technical

assistance and guidance in the liquid chromatography


I would like to thank the students and staff of the

Environmental Engineering Sciences Department and the Soil

Science Department for their kind support and assistance. I

would especially like to thank Ms. Linda Lee and Mr. Sture

Edvardsson of the Soil Science Department for their able

technical assistance in many aspects of this work. Special

thanks must go to Mr. Ron Jessup of the Soil Science Depart-

ment for assistance in statistical and mathematical opera-

tions and for his insight and guidance in the understanding

of thermodynamic processes. My sincere thanks also go to


Ms. Barbara Smerage for her expert typing and proofreading

of this dissertation.

Finally, this work would never have been possible

without the support of my wife, Janet, whose love and

friendship have tempered the pains of research during the

past three years. My thanks and loving appreciation go to

my parents and family, who have shown unfailing support

during my many years of graduate study.

This research was funded by the United States

Environmental Protection Agency, cooperative agreement

#CR-811144. This financial support is gratefully




ACKNOWLEDGMENTS.................................... iii

LIST OF TABLES ..................................... vii

LIST OF FIGURES.... ................................ x

ABSTRACT.. ..... ............ .... ...... ...... .... xiv


I INTRODUCTION ........... ...................... 1

II OBJECTIVES.............. ...................... 3

III LITERATURE REVIEW............................. 4

3.1 Introduction.. ........................... 4
3.2 Overview ........... ..................... 4
3.3 Solvophobic Model of RPLC Sorption....... 8
3.4 Thermodynamics of Sorption............... 24

IV MATERIALS AND METHODS......................... 44

4.1 Introduction............................. 44
4.2 Selection of Model Sorbents.............. 44
4.3 Selection of Natural Sorbents............ 45
4.4 Selection of Organic Solvents............ 46
4.5 Selection of Hydrophobic Compounds....... 48
4.6 Reagents...................... ............ 52
4.7 Equipment................................ 52
4.8 Experimental Techniques.................. 57

V RESULTS AND DISCUSSION........................ 67

5.1 Introduction.............................. 67
5.2 Results.......................... ........ 67
5.3 Hydrophobic Retention on RPLC Materials.. 70
5.4 Thermodynamics of Hydrophobic Sorption... 90
5.5 Enthalpy-Entropy Compensation Effects.... 149
5.6 Equilibrium Studies with RPLC Materials.. 197
5.7 Solvophobic Model of RPLC Retention...... 201

VI SUMMARY AND CONCLUSIONS....................... 219


A RPLC RETENTION DATA............................ 227




SOLUTIONS.. ....... ................ ........ ... 310

WATER SOLUTIONS ............................... 311

MODEL OF RPLC RETENTION....................... 313

THERMODYNAMIC STUDIES......................... 317

REFERENCES......................... ............ ... 325

BIOGRAPHICAL SKETCH................................. 339


Table Page

3-1 Physical meaning of compensation parameters... 40

4-1 Physical and chemical properties of Webster
soil.... ....................................... 47

4-2 List of hydrophobic compounds and their HSA
values......................................... 51

4-3 List of reagent chemicals and their respective
sources........................................ 53

4-4 Physical and chemical properties of RPLC
supports....................................... 55

5-1 Octanol/water partition coefficients of the
hydrophobic solutes.......................... 74

5-2 Regression parameters from In k' vs. log Ko
in various RPLC systems....................... 76

5-3 Regression parameters from In k' vs. HSA in
various RPLC systems......................... 82

5-4 Molecular connectivity indices of PAHs and
alkylbenzenes................................. 84

5-5 Regression parameters from In k' vs. X in
various RPLC systems......................... 88

5-6 Calculated In k values for the hydrophobic
solutes on the e8 support in several
solvent systems.................... ........... 97

5-7 Correlation of AH vs. organic solvent
content, e, for four aromatic solutes.......... 108

5-8 Correlation of AH vs. HSA in a methanol/
water eluent on three RPLC supports............ 121

5-9 Correlation of AH vs. HSA in an acetonitrile/
water eluent on three RPLC supports............ 123


5-10 AS0 vs. HSA for the PAH compounds and
alkylbenzenes in a methanol/water eluent on
three RPLC supports ........................... 130

5-11 AS0 vs. HSA for the PAH compounds and
alkylbenzenes in an acetonitrile/water eluent
on three RPLC supports ....................... 140

5-12 Compensation temperatures on three RPLC
supports in methanol/water mobile phases...... 154

5-13 Compensation temperatures on three RPLC
supports in acetonitrile/water mobile phases.. 158

5-14 TAS vs. AHo for the hydrophobic solutes on
three RPLC supports in methanol/water eluents. 172

5-15 TAS vs. AHo for the hydrophobic solutes on
three RPLC supports in acetonitrile/water
eluents .............. .......................... 180

5-16 Compensation parameters from three RPLC
supports in methanol/water eluent systems
via the three-parameter model................. 186

5-17 Regression of compensation parameters vs. HSA
for the hydrophobic solutes in a C-4,
methanol/water system......................... 189

5-18 Compensation parameters from three RPLC
supports in acetonitrile/water eluent systems
via the four-parameter model................... 191

5-19 Compensation parameters from three RPLC
supports in acetonitrile/water eluent systems
via the three-parameter model................. 192

5-20 Regression of compensation parameters vs. HSA
for the hydrophobic solutes in a C-8,
acetonitrile/water system..................... 195

5-21 Solute retention factor of pyrene as a
function of column flow rate.................. 198

5-22 Solute retention factors of pyrene and
biphenyl from batch and column RPLC systems... 200

5-23 Regression of B vs. (A + E) for the
hydrophobic solutes in C-2, C-4, C-8, and
C-18 acetonitrile/water systems.............. 209


5-24 Regression parameters from In k' vs. RPLC
chain length and critical carbon numbers
for the hydrophobic solutes in an
acetonitrile/water eluent..................... 213

5-25 Contact areas of solute-ligand interaction
for the hydrophobic solutes on four RPLC
supports...................................... 214


Figure Page

3-1 Solvophobic model of dissolving a
hydrophobic solute........................... 13

3-2 Ln k' vs. -AH for sorption of PAH
solutes............... ...................... 31

4-1 Retention volume vs. NaNO3 concentration.... 61

5-1 Ln k' vs. log K for the hydrophobic
solutes on C-8 material in 60/40 methanol/
water....................................... 72

5-2 Ln k' vs. log K for the hydrophobic
solutes on C-8 material in 50/50
acetonitrile/water......................... 73

5-3 Log Ko vs. solute HSA...................... 78
5-4 Ln k' vs. solute HSA for the hydrophobic
solutes on C-8 material in 60/40 methanol/
water......................................... 80

5-5 Ln k' vs. solute HSA for the hydrophobic
solutes on C-8 material in 60/40
acetonitrile/water......................... 81

5-6 Ln k' vs. X for the hydrophobic solutes
on C-8 material in 60/40 methanol/water..... 86

5-7 Ln k' vs. IX for the hydrophobic solutes
on C-8 material in 40/60 acetonitrile/
water...................................... 87

5-8 Ln k' of naphthalene vs. volume fraction of
organic solvent on C-8 material............. 94

5-9 Ln k' of n-butylbenzene vs. volume fraction
of organic solvent on C-8 material.......... 95

5-10 Ln k' of iodobenzene vs. volume fraction of
organic solvent on C-8 material............. 96

5-11 AHo for sorption vs. volume fraction
methanol content for four aromatic
solutes on the C-4 support................... 100

5-12 AHo for sorption vs. volume fraction
acetonitrile content for four
aromatic solutes on the C-4 support.......... 101

5-13 Surface tension vs. volume fraction methanol
content..................................... 105

5-14 Surface tension vs. volume fraction
acetonitrile content........................ 106

5-15 ASo for sorption vs. volume fraction
methanol content for four aromatic solutes
on the C-4 support.......................... 111

5-16 AS0 for sorption vs. volume fraction
acetonitrile content for four aromatic
solutes on the C-2 support................... 116

5-17 AHo vs. solute HSA for sorption of the
hydrophobic solutes on C-8 material in
60/40 methanol/water........................ 120

5-18 AHo vs. solute HSA for sorption of the
hydrophobic solutes on C-4 material in
30/70 acetonitrile/water.................... 122

5-19 AS0 vs. solute HSA for sorption of the
hydrophobic solutes on C-2 material in
35/65 methanol/water........................ 126

5-20 AS0 vs. solute HSA for sorption of the
hydrophobic solutes on C-8 material in
50/50 methanol/water........................ 128

5-21 ASo vs. solute HSA for sorption of the
hydrophobic solutes on C-8 material in
60/40 methanol/water......................... 129

5-22 AS0 vs. solute HSA for sorption of the
hydrophobic solutes on C-2 material in
25/75 acetonitrile/water.................... 133

5-23 AS0 vs. solute HSA for sorption of the
hydrophobic solutes on C-2 material in
30/70 acetonitrile/water.................... 134

5-24 AS0 vs. solute HSA for sorption of the
hydrophobic solutes on C-4 material in
40/60 acetonitrile/water.................... 136

5-25 AS0 vs. solute HSA for sorption of the
hydrophobic solutes on C-8 material in
60/40 acetonitrile/water.................... 138

5-26 AHo vs. RPLC chain length for sorption of
pyrene, chrysene, n-butylbenzene,
n-hexylbenzene in 60/40 methanol/water...... 142

5-27 AS0 vs. RPLC chain length for sorption of
anthracene, naphthalene, n-hexylbenzene,
and m-diethylbenzene in 40/60 acetonitrile/
water.................................... ......... 145

5-28 Ln k' vs. -AH/R for sorption of the
hydrophobic solutes on C-8 material in
60/40 methanol/water........................ 151

5-29 Ln k' vs. -AH /R for sorption of the
hydrophobic solutes on C-8 material in
60/40 acetonitrile/water.................... 156

5-30 Ln k' vs. -AHo/R for sorption of the
hydrophobic solutes on C-2 material in
25/75 acetonitrile/water.................... 160

5-31 Ln (Soil sorption coefficient) vs. 1/T for
three aromatic solutes on Webster soil in
30/70 methanol/water........................ 164

5-32 Ln (Soil sorption coefficient) vs. -AHo/R
for three aromatic solutes on Webster
soil in 30/70 methanol/water................ 166

5-33 TAS vs. AH for sorption of the
hydrophobic solutes on C-2 material in
35/65 methanol/water........................ 168

5-34 TAS vs. AH for sorption of the
hydrophobic solutes on C-4 material in
50/50 methanol/water........................ 170

5-35 TASo vs. AHo for sorption of the
hydrophobic solutes on C-4 material in
75/25 methanol/water........................ 171

5-36 TAS vs AH for sorption of the
hydrophobic solutes on C-2 material in
25/75 acetonitrile/water.................... 174


5-37 TASo vs. AH for sorption of the hydrophobic
solutes on C-2 material in 30/70
acetonitrile/water......................... 175

5-38 TASo vs. AHo for sorption of the
hydrophobic solutes on C-2 material in
40/60 acetonitrile/water.................... 177

5-39 TAS vs. AH for sorption of the
hydrophobic solutes on C-4 material in
50/50 acetonitrile/water.................... 178

5-40 A vs. solute HSA for sorption of the
h3drophobic solutes on C-4 material in
methanol/water eluents....................... 188

5-41 B vs. (A + E) for sorption of the
hydrophobic solutes on C-2 material in
acetonitrile/water eluents.................. 203

5-42 B vs. (A + E) for sorption of the
hydrophobic solutes on C-4 material in
acetonitrile/water eluents.................. 205

5-43 B vs. (A + E) for sorption of the
hydrophobic solutes on C-8 material in
acetonitrile/water eluents.................. 206

5-44 B vs. (A + E) for sorption of the
hydrophobic solutes on C-18 material in
acetonitrile/water eluents.................. 207

5-45 Ln k' vs. RPLC chain length for five
hydrophobic solutes in 50/50 acetonitrile/
water.......................................... 211


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



Kent Benson Woodburn

August 1985

Chairman: Professor Joseph J. Delfino
Major Department: Environmental Engineering Sciences

Reversed-phase liquid chromatography (RPLC) was used to

investigate the thermodynamics and mechanisms of sorption

for hydrophobic organic chemicals retained on C-2, C-4, C-8,

and C-18 RPLC supports in polar solvent mixtures. The

mobile phase consisted of methanol or acetonitrile in binary

combinations with water, and the test solutes were various

alkylbenzenes, polycyclic aromatic hydrocarbons (PAHs), and

the mono-substituted halobenzenes.

The mechanism of retention for hydrophobic organic

compounds in the soil or groundwater environment is cur-

rently a topic of considerable interest. An enthalpy-

entropy compensation model was used to study sorption

interactions on a carbonaceous surface soil compared to the

RPLC stationary phases. For a methanol/water solution of

three PAH solutes on the surface soil material, the measured


compensation temperature (8) was 5730K. This temperature

compares favorably with 8 values obtained for PAH compounds

in methanol/water RPLC systems. This indicates that the

mechanism of retention is the same in both systems. This

finding should facilitate the study of soil sorption inter-

actions, as RPLC is an effective surrogate for studying the

sorption of hydrophobic compounds in a soil environment.

The standard sorption enthalpy and entropy changes,

AH and AS0, respectively, decreased as the water content

of the eluent and the hydrocarbonaceous surface area (HSA)

of the solute molecules were increased in methanol/water and

acetonitrile/water RPLC systems. A single linear regression

line described the relationship of AHo to solute HSA in both

eluent systems. However, plots of AS0 vs. solute HSA

produced separate linear regression lines for the alkyl-

benzenes compared with PAH and halobenzene compounds.

In methanol/water and acetonitrile/water mobile phases,

the thermodynamics and mechanisms of RPLC retention were

different for the alkylbenzene solutes compared with PAH and

halobenzene compounds. The occurrence of two distinct RPLC

retention mechanisms is a unique finding of this work.

The solvophobic theory of RPLC retention provided an

excellent model for describing the solute sorption data in

acetonitrile/water RPLC systems based on known solvent

properties. The results indicate that solute-stationary

phase interactions change considerably as one progresses

from the C-2 to C-18 stationary phase carbon chains.



The contamination of subsurface water supplies is a

topic of considerable scientific and public interest. Much

of the recent concern centers on the presence of a variety

of hydrophobic organic chemicals in these water supplies.

The sources of these chemicals include agricultural and

silvicultural practices (pesticides), accidental spills and

leaks, as well as surface and subsurface disposal of organic

wastes. Research has established that sorption and

degradation (biotic and abiotic) are the two major processes

attenuating the transport of organic chemicals in soils.

The high potential sorption capacity of soils thus allows

the sorption process to play an important role in the

potential for groundwater contamination due to these com-

pounds. A thorough understanding of sorption mechanisms and

thermodynamics would greatly improve our ability to model

and predict solute transport in the soil and groundwater


Historically, most of the sorption data for hydrophobic

organic chemicals have been collected in batch systems,

which do not easily lend themselves to thermodynamic

experimentation. Recent advances in reversed-phase liquid

chromatography (RPLC) suggest that this technique may be a

model system for the study of sorption of hydrophobic

organic chemicals on soil materials. The RPLC technique

involves the use of a chemically bonded nonpolar stationary

phase and a polar mobile phase of water mixed with a

water-miscible organic solvent, such as methanol. This form

of liquid chromatography offers a well-tested experimental

technique and a sound theoretical background for obtaining

data on solute-sorbent interactions. The method is less

cumbersome than traditional batch techniques and may be

easily modified for the collection of thermodynamic sorption


This dissertation presents a detailed discussion of the

application of RPLC as a technique for investigating the

thermodynamics and mechanisms of sorption for hydrophobic

organic chemicals on nonpolar surfaces. The sorptive

behavior of such solutes was also studied in a natural soil

environment to compare and contrast the solute-sorbent

interactions in the RPLC and soil systems.


The main objectives of this study were

(1) To examine the thermodynamics and mechanisms

of sorption for a variety of hydrophobic solutes on RPLC


(2) To examine the thermodynamics of sorption for

several hydrophobic compounds on a highly carbonaceous soil

surface; and

(3) To apply the solvophobic model of Horvath et al.

(1976) to solute retention on RPLC surfaces and to examine

solute-sorbent interactions using this retention model.


3.1 Introduction

This chapter will present a review of the pertinent

literature for each of the major subject areas covered in

this work: the solvophobic theory of hydrophobic inter-

actions; the thermodynamics of sorption processes; and

enthalpy-entropy compensation effects.

3.2 Overview

The contamination of subsurface water supplies is

currently a topic of considerable concern among scientists,

legislators, and the general public. Nearly one-half of the

population of the United States use groundwater as its

primary source of drinking water. Approximately 35 percent

of the municipal drinking water supplies come from ground-

water, and 75 percent of major U.S. cities currently depend

on groundwater as their principal water source (Pye and

Patrick, 1983; Todd, 1980). Although groundwater contamina-

tion has occurred for centuries, population demands, agri-

cultural activities, and increased industrialization have

greatly exacerbated the problem in some areas. As our

dependence on groundwater increases, its quality becomes an

increasingly important issue.

Much of the concern over groundwater quality centers

on the presence of anthropogenic organic compounds in these

systems. The route of entry for such compounds may include

agricultural practices, accidental spills, and surface and

subsurface disposal of chemical wastes. In response to this

problem, there exists a need for better understanding of the

transport of organic chemicals in the unsaturated soil

zone. Research has established the phenomenon of adsorption

onto soil materials as a principal attenuation mechanism in

the transport of organic solutes. Historically, adsorption

and desorption data have been determined by batch equi-

librium experiments using single solvent-single sorbate

systems. Solute leaching data have been obtained primarily

by the use of two methods: soil thin-layer chromatography

or STLC (Helling, 1971) and miscible displacement (MD)

through soil columns. Each technique has its own inherent

strengths and weaknesses. While STLC is less complicated to

perform than soil column experiments, the results do not

adequately reflect the dynamic conditions present in a

natural soil system. Miscible displacement techniques

reflect actual soil conditions better; however, a consider-

able amount of time and experience are required to produce

reproducible results.

High-performance liquid chromatography (HPLC) is a

chemical separation technique which exploits the differen-

tial distribution of sample components between two distinct

physical phases. One of these phases is a stationary sup-

port or sorbent, while the other is a liquid mobile phase

percolating through the column bed. The chromatographic

separation process occurs due to repeated sorption-

desorption steps during solute transport through the

stationary support. Separation of sample components is due

to differences in their distribution activities between the

two phases.

There are numerous subgroups of HPLC. In general, the

divisions are based on the nature of the stationary phase

and the separation process. One of the most popular modes

of separation, reversed-phase liquid chromatography,

involves a stationary phase which is nonpolar in nature and

a mobile phase composed of a polar liquid, such as water or

a water/methanol mixture. Thus, the more nonpolar a solute,

the longer it will be retained on the nonpolar stationary


Recent advances in reversed-phase liquid chromatography

(RPLC) suggest that this technique may be applicable as a

model for natural soil systems, given the appropriate selec-

tion of the sorbent/solvent system. This particular form of

modern liquid chromatography may best simulate the natural

sorption conditions present in soils. Adsorption and

thermodynamic data determined by RPLC may be more

representative of the conditions present in soil than batch

equilibrium techniques. It is worth noting at this time

that the term sorptionn" is often used interchangeably with

"adsorption" when speaking of solute retention on soils or

RPLC supports. The fundamental processes of solute reten-

tion on RPLC and soil surfaces are not well understood;

hence, the term sorptionn" indicates our lack of knowledge

over whether adsorptive, absorptive, or partitioning

mechanisms control retention. Mingelgrin and Gerstlz (1983)

recently reviewed this topic in considerable depth.

The evidence supporting the use of reversed-phase

columns as models of varying soil environments may be found

in the recent chromatographic literature. Veith et al.

(1979) demonstrated the relationship between a chemical's

corrected retention time (k') on an octadecylsilane RPLC

column and its octanol/water partition coefficient (K ).
The work of Karickhoff et al. (1979) and Kenega and Goring

(1980) showed the excellent correlation which exists between

K and K the carbon-normalized soil sorption coeffi-
ow oc
cient. Swann et al. (1979) and Rao and Nkedi-Kizza (1983)

reported on the correlation between measured K and octade-
cylsilane retention time for selected organic solutes.

Finally, the research of McCall et al. (1981) presented a

solute mobility classification system based on a compound's

RPLC retention time. A linear correlation was observed

between soil column leaching distance and RPLC retention


Although sorption onto soil material is critically

important in attenuating the transport of organic solutes,

little is known of the thermodynamics of the sorption

processess. Chromatographic techniques are well suited to

thermodynamic studies, and this approach may be applicable

for examining liquid-phase sorption reactions in soils and

chromatographic media. The following sections will outline

the use of RPLC as a model for solute transport in aqueous

and mixed-solvent conditions. The focus of this chapter

will be the thermodynamic basis of sorption and how the

energetic of the sorptive process may be studied and


3.3 Solvophobic Model of Hydrophobic Sorption

One of the most widely accepted models used to describe

solute retention on hydrophobic surfaces (RPLC packing,

pyrocarbon, etc.) is termed the solvophobic or hydrophobic

theory (Horvath and Melander, 1977; Horvath et al., 1976).

Among all the theories put forward to describe nonpolar

interactions in polar solvents, only the solvophobic model

treats such processes in terms of the bulk solvent prop-

erties, such as surface tension and the dielectric constant,

and solute properties, such as surface area and dipole

moment. These properties are generally available from the

literature or may be estimated from structure-activity

relationships or molecular models.

According to the solvophobic model, the hydrophobic

interaction between a solute molecule and the surface of the

nonpolar RPLC packing material is considered a reversible

association between the solute molecule, S, and the

hydrocarbonaceous ligand, L, resulting in a complex, LS

S + L LS (3-1)

Three important parameters govern the strength of associa-

tion between L and S and hence the retention of solute S:

(1) the hydrocarbonaceous surface area of the solute,


(2) the hydrocarbonaceous surface area of the ligand,


(3) the surface tension, y, of the bulk solvent.

In dissolving a hydrophobic compound in a solvent

mixture, the original solvent structure must be disrupted

and a cavity formed for the hydrophobic or nonpolar portion

of the molecule that cannot interact significantly with the

polar solvent. The energy required for cavity formation

(AG cav) is proportional to the surface tension (Y) of the
solvent and to the hydrocarbonaceous surface area of the

solute molecule (HSAs). The surface tension is used here as

an indication of intermolecular solvent forces, which

increase as a direct function of y. Additionally, the

larger the HSA value, the greater will be the AGcav
s cav
required for the molecule.

The driving force for the dispersive interactions

between the hydrophobic solute and the hydrocarbonaceous

ligand is thought to be the tendency of the polar solvent

(water, water/methanol mixture, etc.) to minimize the hydro-

phobic surface created about the solute-ligand complex

(HSAsl). Quite simply, the polar mobile phase "drives" the

hydrophobic solute toward the stationary phase rather than

any inherently strong attraction existing between S and L.

The net energy of interaction is determined largely by the

hydrophobic contact area, AA, of the LS complex (AA = HSAs +

HSA1 HSAsl), and the solvent surface tension, y. Higher

values of AA or y lead to more energy being liberated during

the association of L and S and subsequently cause the

retention of S to be stronger.

In applying this theory to chromatography with nonpolar

reversed-phase supports, it is necessary to relate the

measured solute retention factor, k', to the AGo for the

binding process

k' = 1 K (3-2)

where K is the thermodynamic equilibrium binding constant

and 0 is the phase ratio or the volume ratio of the


stationary and mobile phases. The solvophobic theory may

now be used to establish a quantitative relationship between

k' and the properties of the solvent and solute.

To evaluate the equilibrium constant, K, K must be

related to the free energy change for the chromatographic

binding process. Recall that

In K = AGO/RT (3-3)

where R is the universal gas constant and T is the absolute

temperature (OK). Since the value of AGo for the binding

process is independent of the path taken, we may now write

the association reaction as follows:

S(g) + L(g) -AGG S SL(g) (3-4)
0 ,vdw,assoc

s 1 sl

S(1) + L(1) =AGassoc- SL(1) (3-5)

where g and 1 represent the gaseous and liquid phases,

respectively. The molecular associations in solution can be

conceptually broken down into two processes. One is the

interaction of S and L to yield SL in a hypothetical gas


phase without interaction of the solvent. The second, more

involved process entails the interaction of all species with

the solvent. The free energy of the binding process in

solution is merely the difference between these respective


The association of S and L in the gas phase is assumed

to occur by van der Waals forces only, and the free energy

change is denoted by AGvdw,assoc The standard free energy

change associated with bringing each component (S, L, and

SL) from the hypothetical gas phase into the solvent is

considered mathematically as the sum of two parts. The

first part corresponds to the free energy, AGcav, required to

prepare a solvent cavity of suitable size and shape for the

solute. The second term, AGint, expresses the interaction

energy between the solute and solvent. This latter expres-

sion also contains a term which accounts for the entropy of

mixing which arises upon mixing the solute and solvent. The

entropic term accounts for the distance a solvent molecule

can freely move before striking another solvent molecule.

These processes and the associated terms are shown in Figure

3-1. In summary, the total standard free energy change for

the solvation of species i is given by

AGi = Gcavi + AGint + RTln(RT/PoV)


Solvophobic Model:

SAG = AGcav

(AG > 0)

AG = AGnt + RTIn(RT/PV)

(AG < 0)

Figure 3-1. Schematic illustration of dissolving a hydro-
phobic solute into a polar solvent. The free
energy change involved in formation of a suit-
able cavity, depicted in A to B, is AGcav. The
magnitude of AGcav is determined by the cavity
surface area and the surface tension of the
solvent and is greater than zero. In B to C,
the solute is placed in the cavity and inter-
acts with the solvent. The free energy change
of interaction is AG. and the free energy
change of mixing is I ln(RT/PV)), where
R In(RT/PV) is the entropy change of mixing.
The total free energy change of interaction
is less than zero.


where V is the molar volume (cm /mole) of the solvent and P
is the standard atmospheric pressure (1 atm).

The overall standard free energy change for the

association of S and L in solution, AG assoc' is given by
d. S by


= AGvdwassoc + (A Gsl AGs AG 1)
vdw,assoc sl s I

which can be expanded to


AGvdw,assoc + (Gcav,sl + Gint,sl
- (AGcav,s + AGint,s) (Gcav,l +

AGint, ) RTln(RT/PoV)

The individual terms have been evaluated (Horvath and

Melander, 1977; Horvath et al., 1976; Sinanoglu, 1968). The

free energy of cavity formation for any species is given

approximately by

AG =

Ke.HSA. y(l Wi)N

where HSA. is the hydrocarbonaceous surface area of species

i, y is the surface tension (dyne/cm) of the bulk solvent, N

is Avogadro's number, and

W. = (1 K.i/Ke )(dlny/dlnT + 2/3A.T)
1 1 i






where Ai is the coefficient of thermal expansion for species
i. The term K is the ratio between the energy required

for formation of a suitably shaped cavity of surface area

HSAi and the energy required to expand the planar surface of

the solvent by the same amount, which is approximately

(HSAiy). In other words, because the cavity surface has a

small radius of curvature, the surface tension of the cavity

will differ from that of the bulk solvent by a proportion-
e s
ality factor, K .. The term Ks. is the corresponding
1 1
function for the entropy produced upon formation of the

solute cavity. The Ke values have been computed and

tabulated for a number of pure solvents (Halicioglu and

Sinanoglu, 1969; Horvath et al., 1976). The Ke value for

species i may be estimated (Horvath et al., 1976) by

Ke = 1 + (Ke l)(V/Vi)2/3 (3-11)

where Ke is evaluated for the pure solvent, and V and Vi are

the molar volumes (cm /mole) of the solvent and species i,

respectively. A similar relationship may be developed for

KS.. Both Ke. and Ks approach unity as the size of the
1 1 1
solute molecule increases with respect to the size of the

solvent molecules.

The second term in Eqn. (3-6) expresses the inter-

action of species i with the solvent. It is generally

assumed to be the sum of a van der Waals component, AGvdwi

and electrostatic free energy term, AG

AG. = AG + AG (3-12)
int,i = vdw,i + es,i (3-12)

The van der Waals contribution has been estimated by

Sinanoglu (1968). These calculations suggest that the van

der Waals free energy can be reliably estimated by

AGvdw, = Y + aHSA (3-13)
vdw,i 1

where Y and a are solvent-dependent parameters and HSAi is

the hydrocarbonaceous surface area of the species of

interest. Therefore, the contribution to the binding energy

from van der Waals forces may be expressed as

AGvdw,assoc = -Y aAA (3-14)

where all terms are as denoted earlier.

The electrostatic free energy change, AGes, has been

evaluated for a number of cases (Horvath et al., 1976;

Sinanoglu, 1968). In the case of simple dipoles, this

energy has been approximated using the following expression:

AGesi = -N ij.2 P/2 v. (3-15)
~ 1 X3

where pi is the static dipole moment of species i, v. is

the molecular volume of i, and F is a function of the static

dielectric constant of the solvent, E, as given by

r = 2(E 1)/(2E + 1) (3-16)

The term P depends on the polarizability of species i, ai,


P = [47 (1 -r i/vi)]-1 (3-17)

where E is the permittivity constant. The term P is essen-
tially independent of solvent composition (Horvath and

Melander, 1977).

Upon combining Eqns. (3-8), (3-9), (3-12), and (3-15),

an expression is generated for the standard free energy

change for the overall binding process

AGo = AG + [AG NI 2 P/2v
assoc vdw,assoc vdw,sl sl 2p/2s

KeslHSAsl Y(l Wsl)N [AGvdw,s

Np2 rP/2vs + Ke HSA y)l W )N]
s [s s s s

[AG N P/2v
vdw,l 1 r 1

+ KelHSA Y(l W )N] RTln(RT/PoV)



In most chromatographic systems, the solvent molecules

(water, methanol, etc.) are much smaller than the solute

molecules of interest. Hence, the following assumptions

appear to be quite reasonable


= w

= W1

= 0




= AGvdw,l

= s

= 0


Ke = Ke = 1
sl 1


An assessment of the molecular volume of the complex

(vsl) is necessary for the model, and for convenience this

volume is assumed to be a multiple of the molecular volume

of the solute (vs)

v s = v (3-24)

where X is a proportionality constant. The total hydro-

phobic surface area of the complex (SL) is expressed by





HSA = HSA + HSA AA (3-25)
sl s 1

where AA is once again the contact surface area of the

associated species. Combining the assumptions from Eqns.

(3-19)-(3-25) with Eqn. (3-18) yields

AGo = AG AG + N(A 1) s 2 P/2Xvs
assoc vdw,assoc vdw,s s s

NAAy N(Kes 1)HSAsY

RTln(RT/PoV) (3-26)

Substituting the expression for Ke from Eqn. (3-11)

produces the final expression for AG

AG0 = AG -AG + N(A 1)s PP/22v
AGassoc = vdw,assoc vdw,s + N(A ) P/

NAAy Ny(Ke )HSA (V/V)2/3

RTln(RT/PoV) (3-27)

It is now possible to relate the measured solute

retention factor, k', to the standard free energy of the

binding process,AG assoc. Combining Eqns. (3-2), (3-3), and

(3-27) produces the following expression for the solute

retention factor:

in k' = in 1/RT[AGvdw,assoc AGvdw,s

+ N(X 1) s FP/2Xvs NAAy

Ny(Kes 1)HSAs(V/V )2/3 RTln(RT/PoV)]


The above expression may be further simplified by

assuming that the solute, ligand, and complex all have

spherical shapes. Following Sinanoglu (1968) and Horvath et

al. (1976), it is possible to calculate the nonpolar surface

area of the solute by

2 2/3 2/3
HSA (cm ) = 4.836v 2/3 =4.836(V/N)23 (3-29)

where V (cm3/mole) is the solute's molar volume. Combining

Eqns. (3-9), (3-11), and (3-29) with the assumption of W = 0

(Eqn. 3-19) yields

AG = NHSA y + 4.836N /3(Ke 1)V2/3 (3-30)
cav,s s

The combination of Eqns. (3-2), (3-3), (3-8), (3-11),

(3-12), (3-15), and (3-30), with the assumption of Eqns.

(3-19) to (3-23), yields the following expression for the

retention factor, k'

In k' = lno 1/RT[AGvdw,assoc- AGvdw,s
vdw,assoc vdw,s

+ N(A 1) us rP/2 vs NAAy

4.836N /3(Ke 1)V2/3y RTln(RT/PoV) (3-31)

The expression for the solute retention factor as given

in Eqn. (3-31) may now be simplified for certain chromato-

graphic situations. For any given solute/sorbent combina-

tion, the temperature and flow rate may be considered

constant while solvent composition may be easily changed.

Under these conditions, solute and ligand properties may be

considered constant, as well as AGvdw,assoc Hence,

Eqn. (3-31) may be rewritten as

In k' = (A + E) + BT + Cy + D(Ke 1)V2/3y

+ In(RT/PoV) (3-32)

where the terms are defined as

A = lnp AGvdwassoc/RT (3-33)

B = (1 -X) 2NP/2RTV (3-34)

C = NAA/RT (3-35)

D = 4.836N /3RT = 1.67 x 107 mole2/l-atm (3-36)

E = AG vdw,s/RT
vdw, s


Rearranging terms, Eqn. (3-32) becomes

In k' D(Ke 1)V2/3y- In(RT/PoV) = (A + E) + BT+ Cy


The Ke terms for cavity formation may be determined

(Horvath et al., 1976) from the molar volume (V), the heat

of vaporization (AE ), and surface tension (y) of the
solvent at the temperature of interest, in addition to the

temperature derivative of the surface tension and the

thermal coefficient of expansion (A.) by

1/3 E
Ke = yap (3-39)
v2/3y d In y 2
d In T 3 i

These physicochemical properties are generally known for

solvents of chromatographic interest or can be reliably

estimated (Horvath et al., 1976).

There are literature data on density (Carr and Riddick,

1951; Timmermans, 1960), dielectric constant (Akerlof, 1932;

Douhert and Morenas, 1967), surface tension (Horvath et al.,

1976), and Ke (Horvath et al., 1976; Wells, 1981). Using

such data, all terms except (A + E), B, C, and In k' may be

calculated for a given solvent composition in which k' is

experimentally determined. A linear regression of the terms

on the left side of Eqn. (3-38) versus F and Y will produce

values of (A + E), B, and C as regression coefficients.

Once these terms are evaluated, estimates of k' are possible

at any solvent composition.

A number of researchers have validated the "solvo-

phobic effect" and its importance in the retention of non-

polar solutes on reversed-phase supports (Colin et al.,

1983; Horvath and Melander, 1977; Horvath et al., 1976;

Miller et al., 1982; Wells and Clark, 1982; Wells and Clark,

1984), and hydrophobic soils and sediments (Karickhoff,

1981; Karickhoff et al., 1979; McCall et al., 1980; Rao and

Nkedi-Kizza, 1983). Additionally, Horvath et al. (1977)

have successfully applied the solvophobic theory to describe

the retention of weak acids, weak bases, and ampholytes in

reversed-phase systems.

The solvophobic theory will be utilized in the follow-

ing chapters to clarify and enhance our understanding of the

hydrophobic retention of nonpolar solutes in polar solvent

systems. The model and related theory have been presented

here to aid the reader in that process. The above descrip-

tion should be considered an overview, however, and the

reader is directed to Horvath et al. (1976, 1977) for more

detailed discussions.

3.4 Thermodynamics of Sorption

3.4.1 Overview

The theory supporting the chromatographic determination

of free energy, enthalpy, and entropy evolved from the

classical equation for chromatography developed by Martin

and Synge (1941). The work of Greene and Pust (1958)

resulted in a relationship between chromatographic retention

time (k') and the heat of adsorption (AH). The later work

of Gale and Beebe (1964) examined equilibrium adsorption

models and developed a concise expression for the measure-

ment of heat of adsorption by chromatographic techniques.

The fundamental expression associated with equilibrium

sorption on chromatographic supports is Eqn. (3-2),

k' = OK. The solute retention factor is easily obtained

from k' = (tR t )/t where tR and t are the retention
R o o R 0
times for the solute under study and an unretained compound,

respectively. The retention factor may be related to

AH and AS0 by combining Eqn. (3-3) with the free
sorp sorp
energy relationship

AGo = AH TASo (3-40)

Substituting Eqn. (3-40) into Eqn. (3-3) yields

In k' = -AHo/RT + ASO/R + In (



The determination of these thermodynamic parameters for

selected organic compounds on RPLC supports and soils may

be of great usefulness in understanding and predicting

solute retention in RPLC and soil environments. For many

hydrophobic compounds in RPLC systems, the enthalpy term

dominates the entropy term in the free energy expression in

Eqn. (3-40) (Colin et al., 1978; Melander et al., 1978).

However, for many of the ubiquitous and carcinogenic poly-

cyclic aromatic hydrocarbons (PAHs), entropic processes may

control RPLC sorption (Chmielowiec and Sawatzky, 1979). It

is noteworthy that enthalpy-entropy effects have not been

extensively researched for hydrophobic solutes in natural

soil/water systems. For example, Mills and Biggar (1969)

measured the AHo for sorption of aqueous hexachlorocyclo-

hexane onto organic and inorganic surfaces, while Wauchope

et al. (1983) determined AH and AS0 for aqueous
sorp sorp
naphthalene sorbing onto a sandy loam soil.

3.4.2 Evaluation of Enthalpy and Entropy Changes

If the heat capacity change upon the binding of the

solute to the stationary phase is zero and the phase ratio

(0) is independent of temperature, then a plot of In k'

versus T- (K- ) is linear, according to Eqn. (3-41). With

such a diagram (termed a van't Hoff plot), the AH can be

obtained directly from the slope of the regression line.

Departure from linearity can occur if the heat capacities of

the bound and free forms of the solute are different.

Generally, most van't Hoff plots of RPLC data are linear and

allow easy determination of AH0 Typical values of
AH for hydrophobic solutes range from -2 to -12
kcal/mole on RPLC and pyrocarbon LC supports (Colin et al.,

1978; Melander et al., 1978).

A number of authors have successfully correlated the

enthalpy of binding in RPLC to a numerical description of

solute molecular structure. Hirata and Sumiya (1983)

reported that the AHsorp of p-nitrobenzyl esters of fatty
acids on octadecylsilane (C-18) increases almost linearly

with the number of carbon atoms in the molecule. Hornsby

and Rao (1983) employed a more complex description of solute

structure, the HSA, and found that H or increases
linearly with the HSA of the sorbate molecule. The results

of these studies agree with the solvophobic model of RPLC

retention, as described by Horvath et al. (1976).

As discussed previously, a plot of In k' vs. T-

(K ), will yield the AH sp from the slope of the regres-
sion line (Eqn. 3-41). The evaluation of the corresponding

entropy change, ASsorp, from the intercept is difficult

because the phase ratio (0) is usually not known. When

using RPLC bonded phase supports (C-4, C-8, C-18, etc.), the

"volume" or an equivalent property of the stationary phase

is not clearly defined, thus a standard method for determin-

ing 4 does not exist. Melander et al. (1980) suggested


expressing as the ratio of the surface area of the sorbent

(m2) to the column void volume (cm3). Alternatively,

Davydov et al. (1981) have used the ratio of mass of column

material (g) to column void volume (cm3). Other authors

have defined the volume of stationary phase as the fraction

of the column not occupied by the mobile phase (Chmielowiec

and Sawatzky, 1979; Jandera et al., 1982a). These latter

two assumptions appear to be particularly flawed, for they

assume the total mass or volume of solid support to consist

exclusively of the chemically bonded stationary phase. This

ignores the fact that the stationary phase occupies only a

small fraction of the total surface area of the silica gel

particle (Kikta and Grushka, 1976). Typical surface cover-

age values for C-8 and C-18 alkanes are 5 to 20% (Kikta and

Grushka, 1976; Sander and Wise, 1984).

Given the difficulties associated with experimentally

measuring a value of 0, a number of authors have instead

concentrated on determining this constant from the physical

properties of the packing material. Knox and Vasvari (1973)

estimated 4 for a C-18 column to be 0.04, while Sander and

Field (1980) calculated # to be 0.38 for their octadecyl-

silane column. In neither case did the authors detail their

method for computing the phase ratio.

The problems encountered in calculating 0 stem from

estimating the volume of stationary phase from the physical

properties of the sorbent. Only recently have researchers

begun to explore the explicit character of the n-alkane

stationary phases commonly found in RPLC (Engelhardt et al.,

1982; Gilpin and Gangoda, 1984; Sander and Wise, 1984; Wise

and May, 1983). Wise and May (1983) have proposed a simple

expression for calculating the surface concentration (Cs) of

the bonded alkylsilane

C (pmol/m2) = %C(106)/1200 NSBET (3-42)

where %C is the percent carbon (w/w) from elemental

analysis, N is the total number of carbon atoms in the

bonded silane molecule, and SBET is the specific surface

area (m2/g) of the chemically modified silica as determined

by the BET nitrogen adsorption method (Brunauer et al.,


Equation (3-42) may now be modified (Dorsey, 1984) to

yield an expression for the volume of stationary phase (VSp)

present on an RPLC column

Vp(cm3) = C (Pmol/m )SBET(MW/D)10 M (3-43)

where MW is the molecular weight (g/mole) of the bonded

alkane, D is the density (g/cm3) of the bonded alkane, and M

is the mass (g) of packing material in the RPLC column.

Combining Eqns. (3-42) and (3-43) yields


V p(cm3) = %C(MW)M/1200 N D (3-44)

Hence, with some knowledge of the physical properties

of the packing material (%C, Nc M) and bonded alkane (MW,

D), it is possible to estimate the volume of stationary

phase present on a given RPLC column. If the column void

volume, V (cm ), is also known, the phase ratio may be
expressed as

( = %C(MW)M/1200 N DV (3-45)
c o

Equation (3-45) was used in the following chapters to

estimate 0 for the RPLC columns under study. This allowed

the value of AS to be calculated from In k' versus
T-1 (K-l) regression plots (Eqn. 3-41). It should be noted,

however, that Eqn. (3-45) assumes that the molecular weight,

density, and number of alkyl carbons is the same for a

bonded alkylsilane as for the bulk alkyl liquid, i.e., C-8

is physico-chemically similar to liquid n-octane. Alkyl

ligands are rotationally and vibrationally restricted

compared to their unbonded counterparts (Gilpin and Gangoda,

1984), and polymeric stationary phases may diverge consider-

ably from simple n-alkyl groups on a silica surface (Sander

and Wise, 1984). Still, Eqn. (3-45) should represent an

improved technique for determining the phase ratio of an

RPLC column.

3.4.3 Enthalpy-Entropy Compensation Effects

In many physico-chemical interactions which are

governed by the same basic mechanism, the overall free

energy change is proportional to the change in enthalpy

(Leffler and Grunwald, 1963; Melander and Horvath, 1980).

This is indeed the case in RPLC, as In k' (a measure of AG)

is linearly related to the enthalpy of binding (Colin et

al., 1978; Knox and Vasvari, 1973; Melander et al., 1978).

From Eqn. (3-41), one would expect the slope of In k' versus

-AHo to be approximately 1/RT. In actual practice, however,

the increase in the natural logarithm of the retention

factor with the enthalpy is much less than expected, as seen

in Figure 3-2. The data for Figure 3-2 were taken from

Chmielowiec and Sawatzky (1979); similar regressions could

be developed from Appendices A and C.

In Figure 3-2, the slope (+95% confidence limits)

of the In k' versus -AHo regression lines is 0.00034

(+4.8E-5) (mole/cal) for the collected PAH compounds.

This value is significantly less than the 1/RT value at

2980K, that is, 0.00169 (mole/cal). This difference

is attributed to changes in the binding enthalpy which

are accompanied by corresponding changes in the binding

entropy (Melander et al., 1978; Melander et al., 1980).

The changes in the binding entropy are believed to be

due to structural modifications of the sorbate

molecule or changes in solvent entropy (Melander and


E 0C
0 +4
o 0

o x-

E" -
0 0 O. N
II 0 04
N on(
\ 0 =_

0 o

\ 0 3


4Ja o
\- z

\ I

\ 8 C U

\1o862) I u-1
\ 10 4a

Horvath, 1980; Sander and Field, 1980). This effect is

termed enthalpy-entropy compensation and has been observed

on RPLC supports (Jinno and Ozaki, 1984; Melander et al.,

1978; Melander et al., 1979; Melander et al., 1980; Sander

and Field, 1980) and pyrocarbon LC columns (Colin et al.,


The use of an enthalpy-entropy compensation model in

this work should allow for greater insight into the ener-

getics of solute binding. Compensation effects may be

conveniently expressed by the relationship

AH = SASo + AGo (3-46)

where AG0 denotes the change in free energy of sorption at

the temperature 8 and 8 is a proportionality constant termed

the compensation temperature (OK). Comparison of "compen-

sation temperatures" obtained from thermodynamic data can be

used to investigate whether the intrinsic mechanism of

retention for one chromatographic system is identical to

that found on another system (Melander et al., 1978).

Substituting Eqn. (3-40) into Eqn. (3-46) and rearranging


AGoT = AHo(1 T/B) + TAGo /B



where AGoT is the standard free energy change at temperature

T. Equation (3-41) may now be rewritten as

In k' = -(AHo/R)(1/T 1/B) AGO /RB + In < (3-48)

where k' is the solute retention factor at temperature T.

According to Eqn. (3-48), a plot of the natural

logarithm of the retention factor vs. the standard enthalpy

changes obtained on a constant RPLC system by various

solutes yields a straight line when compensation occurs,

i.e., when solute/sorbent binding is due to an essentially

identical mechanism for all solutes. The compensation

temperature, B, may be evaluated from the slope of the

regression line. If similar measures of B (K) are obtained

for varying solvent/sorbent systems, one may infer that the

major mechanisms of the sorptive process are identical for

those systems (Jinno and Ozaki, 1984; Melander et al., 1978;

Melander et al., 1980).

Knox and Vasvari (1973) examined the retention of

various substituted benzenes by RPLC supports and reported

enthalpy-entropy compensation effects in a 40/60 (v/v)

methanol/water eluent. Compensation effects have also been

reported by Melander et al. (1978) for buffered and ionized

aromatic acids on C-18 material in 100% aqueous and in

acetonitrile/water systems (up to 30% acetonitrile), and by

Jinno and Ozaki (1984) for alkylbenzenes on C-8 and C-18

phases in various methanol/water mixtures. Similar com-

pensation temperatures of 500 to 7000K were calculated for

each of these RPLC systems. Melander et al. (1978) hypothe-

sized that the mechanism of hydrophobic sorption is essen-

tially the same, regardless of the nature and concentration

of the organic solvent present and the chemical nature of

the sorbate molecules. Further support for this conclusion

comes from the data of Kikta and Grushka (1976). Compensa-

tion temperatures of 593 and 5120K may be computed from

their data concerning alkylphenone retention on two differ-

ent types of nonylsilica stationary phases in 50/50 (v/v)

methanol/water eluent. These 8 values are similar to those

detailed above.

It is interesting to note that in chromatographic

systems employing polar stationary phases and nonpolar

eluents (normal-phase chromatography), the calculated

compensation temperature, 1400K, was markedly lower than

those obtained in RPLC (Knox and Vasvari, 1973). The lower

8 value suggests that the retention mechanism in normal-

phase chromatography is different from that operating in


3.4.4 Enthalpy-Entropy Compensation with Changing Solvent

The work of Melander et al. (1978, 1979, 1982), Colin

et al. (1983), and Martire and Boehm (1983) suggests the

presence of a rigorous mathematical relationship between the

effect of temperature and solvent composition on solute

retention in RPLC. The following presentation will outline

a thermodynamic model for this relationship and is taken

primarily from research performed by C. Horvath and

W. Melander of Yale University.

Enthalpy-entropy compensation effects have been

observed in RPLC by numerous authors (Boumahraz et al.,

1983; Colin et al., 1978; Knox and Vasvari, 1973; Melander

et al., 1978). This behavior was reviewed in Section 3.4.3,

and it suggests that the change in the enthalpy of solute

binding to the stationary phase as solvent composition

changes is proportional to a change in the corresponding

entropy; the proportionality constant is termed the compen-

sation temperature. This relationship may be expressed as

the derivative of Eqn. (3-46) with respect to solvent


dAH0()/d6 = B dAS(e)/de (3-49)

where dAH(8) and dAS(8) are the incremental changes in

enthalpy and entropy at solvent composition 8, upon change

in composition, de (as measured by volume fraction of

organic modifier), and 8 is the compensation temperature


Integrating Eqn. (3-49) from a reference solvent

composition, 8 = 0, to a final composition, 8, yields


AH(6) AHo(0) = 8AS(e) BASo(0) (3-50)

The dependence of the change of entropy of binding on

solvent composition may be eliminated by combining Eqns.

(3-41) and (3-50), with the result that

SAHo(6) AHo() AHo(0) ASo(0)
in k' + + + in

Equation (3-51) relates the natural logarithm of the

retention factor, In k', at a particular solvent composition

and temperature to the change in the standard enthalpy of

binding at that composition and the change in the standard

binding enthalpy and entropy at the reference composition,

taken to be 100% water (6 = 0). It is now useful to relate

AH (6) to the enthalpy at the reference state, AHo(0), where

AH (0) is the standard enthalpy change for solute binding in

100% water. Enthalpy as a function of solvent composition

The dependence of the change in standard binding

enthalpy upon solvent composition may be expressed in

several ways. The simplest relationship is given by

AH(8) = AH c(0)f(9)



where AH c(0) is the standard enthalpy change in 100% water

which exhibits complete enthalpy-entropy compensation, and

f(0) is the solvent compensation function which is unity for

100% aqueous solutions.

When some portion of the standard binding enthalpy

change does not undergo compensation, the relationship may

be written as

AH(0) = AH n(0) + AHoc(0)f(0) (3-53)

where AH (0) is the noncompensated portion of the total

standard enthalpy change (in 100% water). This relationship

is commonly observed in RPLC systems exhibiting enthalpy-

entropy compensation effects (Melander and Horvath, 1984;

Melander et al., 1978).

Equations (3-51) and (3-52) may be combined to yield an

expression for solutes undergoing complete enthalpy-entropy


-A H (0)f(0) [AH (0)(f(0) 1)] ASO(0)
in k' c + + + In


Likewise, a relationship may be developed for the partially

compensated binding enthalpy change by combining Eqns.

(3-51) and (3-53)

-AH (0)f(O) AH (0) AH (0)(f(e) 1)
c n c
In k' = + +

+ + In
R (3-54b)

A linear relationship between the natural logarithm of

the retention factor (In k') and the solvent composition (8)

is frequently observed with water-miscible organic eluents

(Abbott et al., 1976; Horvath et al., 1976). Since 0 is the

volume fraction of organic solvent, the value of f(8) is

unity with a 100% aqueous eluent. The mathematical expres-

sion for this simple case is

f(9) = 1 + ae (3-55)

where a is a constant and will have a negative value. For

example, a 30/70 (v/v) acetonitrile/water eluent (8 = 0.30),

where a may be -2.0, has a f(6) value of 0.40.

Karger et al. (1976), on the other hand, noted a marked

deviation from linearity for In k' vs. 6 plots of straight-

chain alcohols in acetonitrile/water mobile phases.

Schoenmaker et al. (1978) studied the retention of aromatic

solutes in methanol/water, ethanol/water, and n-propanol/

water mobile phases and suggested a quadratic relationship

for the dependence of In k' on solvent composition


f(6) = 1 + ae + 'e2 (3-56)

where a and Y are constants.

If the simple linear model for f(O) is considered, the

following relationship results from combining Eqns. (3-54a)

and (3-54b) with (3-55)

In k' = A 1(l B/T) + A2/T + A3 (3-57)

Similarly, using Eqn. (3-56) for the quadratic f(8)


In k' = A 1(1 8/T) + A2/T + A3 + A42(1 6/T)


The mathematical expressions for Al, A2, A3, and A4 for

fully compensated (Eqn. 3-54a) and partially compensated

(Eqn. 3-54b) standard binding enthalpy changes are given in

Table 3-1. Model evaluation Solvent compensation function, f(8)

To properly discriminate between the linear and

quadratic forms of the solvent compensation function, it is

generally convenient to examine the dependence of In k' on

solvent composition, 6. A linear or quadratic model is

Table 3-1.

Mathematical description of parameters A A2,
A and A4 in Eqns. (3-57) and (3-58).

Full enthalpy compen- Partial enthalpy compen-
tion according to station according to
Parameter Eqn. (3-54a) Eqn. (3-54b)

A1 aAH c(0)/RS aAH c(0)/Re

A2 -AHoc(0)/R -(AH c(0) + AHon(0))/R
A3 AS (0)/R + In ( AS (0)/R + In

A4 YAH c(0)/RB TAH0 (0)/RB

The parameters AH c(0) and AH n(0) are the compensating and
noncompensating portions of the standard binding enthalpy
change in a 100% aqueous mobile phase, and AS (0) is the
standard binding entropy change in 100% water. R, a, (, a,
and are the gas constant, compensation temperature (OK),
column phase ratio, and the coefficient for the first- and
second-order solvent dependence of the natural logarithm of
the solute retention factor, In k', respectively.

selected by statistical fit of experimental data to

Eqn. (3-57) or (3-58), respectively. Full versus partial enthalpy compensation

An examination of the coefficients in Table 3-1 reveals

that Al, A3, and A4 are identical for the two enthalpy

compensation models, Eqns. (3-52) and (3-53). The physical

interpretation of A2 is different for each model, however,

and this fact permits selection of the proper compensation

model based on the chromatographic data. If full enthalpy-

entropy compensation occurs (Eqn. 3-52), then from Table


A2 = -SA /a (3-59)

However, if enthalpy compensation is incomplete, then from

Table 3-1

A2 = -AHon(0)/R BA1/a (3-60)

Upon comparing Eqns. (3-59) and (3-60), it is clear

that the behavior of the A2/A1 ratio for a given chromato-

graphic system may be used to select the appropriate

enthalpy compensation function. The ratio of A2 to A1 will

remain constant only if full enthalpy-entropy compensation

is exhibited by the test solutes. A homologous series of

solutes is preferred for such an investigation.

The research of Melander et al. (1979, 1982) explored

the application of the above thermodynamic model to the RPLC

retention of n-alkylbenzenes in binary mobile phases of

methanol, acetonitrile, dioxane, tetrahydrofuran, and

isopropanol in water. They found that Eqn. (3-57) best fit

the retention data. Additionally, the A2/A1 ratio was found

to generally increase with chain length of the

n-alkylbenzene solute, indicating incomplete enthalpy-

entropy compensation effects (Eqn. 3-53). Therefore,

Melander et al. (1979, 1982) suggested that Eqn. (3-57) best

described the dependence of n-alkylbenzene retention on

solvent composition and temperature and that the expressions

for A1, A2, and A3 given in Table 3-1 are for partial

enthalpy compensation.

Further research by Melander et al. (1985) examined the

application of Eqns. (3-57) and (3-58) for describing the

octadecylsilane retention of 54 polar and nonpolar sorbates

in acetonitrile/water solvent mixtures. They found that

the average relative errors in predicting solute retention

were 7.8% and 6.0% for the four-parameter (Eqn. 3-58) and

three-parameter (Eqn. 3-57) equations, respectively. In

view of the small decrease in error, but considerable

increase in complexity and in number of data points associ-

ated with use of the four-parameter equation, Melander et

al. (1985) recommended that the three-parameter equation,

Eqn. (3-57), be applied in general use.

The results obtained with the enthalpy-entropy compen-

sation model compare favorably with attempts by Colin et

al. (1983), Grant et al. (1979), and Martire and Boehm

(1983) to model RPLC retention dependence on solvent

composition and temperature. The model has a further

advantage in that it may allow greater insight into the

energetic and mechanisms of the sorptive process. The

dependence of the regression parameters A1, A2, and A3 on

the carbon number in n-alkylbenzenes (Melander et al., 1982)

suggests a close relationship between the model parameters

and solute molecular structure. Additionally, Melander and

Horvath (1984) have used this model to examine three

proposed mechanisms of RPLC retention. It is the broad

range of applications that makes this thermodynamic approach

a useful technique for contrasting and comparing sorption

processes on reversed-phase LC supports and natural soil

surfaces. In the following chapters, thermodynamic and

retention data will be interpreted with the aid of the

enthalpy-entropy compensation model and the solvophobic

theory of hydrophobic sorption.


4.1 Introduction

This chapter will discuss the materials and experi-

mental methods employed in all the experiments performed

during this study. The guidelines for selecting model

sorbents, natural sorbents, organic solvents, and hydro-

phobic solutes will initially be presented, followed by a

description of reagents, equipment, and the experimental

design for model and natural sorbent studies.

4.2 Selection of Model Sorbents

The choice of RPLC packing material, which was to serve

as a surrogate sorbent for a soil, was a critical component

of the model chromatographic system. The factors that

influence the sorption and selective attenuation of solutes

by a sorbent include the surface loading of the bonded alkyl

phase (Sadek and Carr, 1984), the n-alkyl chain length of

the bonded phase (Berendsen and DeGalan, 1980), and the

extent of cross-linkage or polymerization of the stationary

phase (Scott and Simpson, 1980). The wide range of RPLC

packing materials currently available made it possible to

select almost any combination of sorbent properties.


It is worth noting that the n-alkyl stationary phases

usually found in RPLC systems are covalently bonded to the

parent silica surface. The resulting n-alkyl bonded film

is retarded both in rotational and vibrational degrees of

freedom when compared to organic matter adsorbed onto soil

mineral surfaces. The sorptive capacity of RPLC stationary

phases is much greater than that of soil materials

(including organic soils and peat), due to their higher

organic loading and bonded attachment to the silica gel

support. Four RPLC stationary phases were selected for

investigation because of interest in the effect of RPLC

chain length on sorptive interactions of hydrophobic

solutes. For this reason, the following 10 pm (outside

particle diameter), reversed-phase, polymeric stationary

phases on silica gel (Analytichem International Inc.) were

chosen for study:

(1) C-2 (ethyl), 5.57% carbon

(2) C-4 (n-butyl), 7.92% carbon

(3) C-8 (n-octyl), 12.05% carbon

(4) C-18 (n-octadecyl), 14.73% carbon.

4.3 Selection of Natural Sorbents

The principal natural sorbent used in this study was a

Webster sandy clay loam surface soil, collected in Iowa at

0-30 cm depth from a profile classified as a Typic

Haplaquoll. The physical and chemical properties of this

soil are listed in Table 4-1. The Webster soil has been

used extensively in earlier studies of sorption and leach-

ing of hydrophobic pesticides (Nkedi-Kizza et al., 1983;

Nkedi-Kizza et al., 1985). This soil was chosen for study

since it represents a highly carbonaceous, hydrophobic

surface soil. The high organic carbon content (3.9%)

allowed an evaluation to be made of hydrophobic sorption on

a natural soil surface. The Webster soil was air-dried and

passed through a 2 mm sieve to remove stones and root frag-

ments prior to use.

4.4 Selection of Organic Solvents

The interactions occurring between the solute and the

solvent can greatly affect solute retention and transport in

a soil system or chromatographic column. The four major

interactions are dispersion forces, dipole interactions,

hydrogen bonding, and dielectric interactions (Snyder and

Kirkland, 1979). The organic solvents chosen for intensive

study were methanol (CH3OH) and acetonitrile (CH3CN), which

represent two distinct classes of organic solvents. These

solvents were used in binary combinations with water in

order to provide a wide range of solute-solvent interac-


Table 4-1. Physical and chemical properties of Webster

SMajor clay
Soil Mechanical analysis pH OCa CEC minerals

sand silt clay 1:1 % meq/100g

Webster 55 20 25 7.3 3.9 21.8 smectites

*Taken with permission from Nkedi-Kizza et al. (1985).
Percent organic carbon.

bcation exchange capacity.

4.5 Selection of Hydrophobic Compounds

The hydrophobic organic solutes chosen for investiga-

tion were compounds of general environmental concern.

Chemicals that were distinctly different in their molecular

conformations were studied to examine the effect of solute

structure on retention thermodynamics. One of the classes

of compounds studied was the polycyclic aromatic hydro-

carbons (PAHs), which have been the focus of numerous

studies concerning their fate and sorption in the aqueous

environment (Dzombak and Luthy, 1984; Karickhoff, 1981;

Means et al., 1980; Waters and Luthy, 1984). Members of

this class of ubiquitous environmental pollutants have been

found widely distributed in air (Handa et al., 1984; Harkov

et al., 1984), rainwater (Pankow et al., 1984), freshwater

(Hase and Hites, 1976), wastewater (Adams and Giam, 1984),

and sediments and soils (Boehm and Farrington, 1984; Prahl

et al., 1984). The environmental sources of PAHs include

anthropogenic inputs such as petroleum spills and energy

production (Hase and Hites, 1976) and natural inputs such as

the combustion of organic material (Kamens et al., 1985;

Mast et al., 1984). A number of the PAHs are potent car-

cinogens (Neff, 1979). Reversed-phase liquid chromatography

is one of the techniques generally used in the analysis of

environmental samples containing PAHs. Due to their exten-

sive aromaticity and lack of substituent groups, the PAHs

are generally rigid, planar molecules with little or no

internal degrees of movement. Recent research has concen-

trated on developing a mathematical relationship between the

molecular structure of PAHs and their RPLC retention factors

(Hanai and Hubert, 1984; Hasan and Jurs, 1983; Jinno and

Kawasaki, 1983a; Wise et al., 1981).

Another class of compounds studied was the substituted

benzenes. In particular, the alkylbenzenes were selected

for study since members of this series of compounds have

been the subject of considerable thermodynamic and RPLC

retention research (Jinno and Kawasaki, 1983b; Melander and

Horvath, 1984; Melander et al., 1982) and have been discov-

ered in landfill leachate plumes (Reinhard et al., 1984).

The alkylbenzenes are also of interest by way of structural

contrast when compared to the rigid, planar PAHs. The

alkyl-chain portion of a molecule in the solid state has

been observed to exist in a fully extended form (Mizushima,

1954). The extended form is also a stable conformation in

the liquid state, but it may not predominate because its

statistical weight is small compared with the sum of other

possible conformers (Testa, 1979). When placed in aqueous

solution, the C1-C4 n-alkanes are suggested to exist pre-

dominately in an extended conformation (Nemethy and

Scheraga, 1962), while the C5 and large aliphatic hydro-

carbons are proposed to exist in folded or coiled conforma-

tions (Edward, 1970; Nemethy and Scheraga, 1962; Herman,

1972). It is these coiled and folded conformations which


present the smallest amount of hydrophobic surface area for

contact with water molecules.

The space inside a coiled or folded alkyl-chain may

consist of a nonsolvated, empty interior volume stabilized

by intramolecular interactions (Edward, 1970; Nemethy and

Scheraga, 1962), or the hydrocarbon chains may be separated

by one or more layers of solvent molecules (Herman, 1972).

An RPLC retention study encompassing a number of straight

chained and branched alkylbenzenes, as well as the rigid

PAHs, may therefore provide useful information for determin-

ing the importance of solute conformation upon the

mechanisms and energetic of hydrophobic sorption.

Nitrobenzene and the monohalobenzenes were chosen for

study because of their structural simplicity and concern

over their environmental fate (Chiou, 1985). A listing of

the solute compounds used in experiments reported in this

dissertation and their respective hydrocarbonaceous surface

area (HSA) values is shown in Table 4-2. Except for nitro-

benzene, it was assumed that the total surface area (TSA)

was equivalent to HSA. The HSA value for nitrobenzene was

calculated using a modification of the surface area model

proposed by Herman (1972). The actual HSA value for nitro-

benzene was calculated by Dr. G. Belfort of Rensselaer

Polytechnic Institute (Belfort, 1982).

Table. 4-2. List of hydrophobic compounds used in
chromatographic studies and their respective
hydrocarbonaceous surface area (HSA) values.

Compound HSA (R2) Compound HSA (R2)

A. Polycyclic aromaticsa


B. Alkylbenzenesa


C. Halobenzenesa




D. Substituted Benzenes




aHSA values taken from Herman (1972),Valvani et al. (1976),
Yalkowsky and Valvani (1979), and Yalkowsky et al. (1979).

HSA value computed using a modified Herman (1972) model by
Dr. G. Belfort of Rensselaer Polytechnic Institute.

4.6 Reagents

A listing of the reagent chemicals used in the

discussed experiments and their respective industrial

sources is shown in Table 4-3. The solvents (water, aceto-

nitrile, and methanol) used in the RPLC and soil experiments

were HPLC-grade and were obtained from Fisher Scientific


4.7 Equipment

The isocratic elution of the hydrophobic solutes

through packed columns of the RPLC sorbents was performed

using a modular liquid chromatography system consisting of

two Gilson Model 302 metering pumps, a Gilson 1.5 mL

analytical mixer, and two Gilson Model 802 manometric

modules interfaced with an Apple lie microcomputer system.

The absorbance of the column effluent was monitored at 254

nm using a Waters 450 variable wavelength UV detector, with

the chromatograms recorded with a Hewlett-Packard 3390A

reporting integrator. Batch soil studies employed a Gilson

Model 121 filter fluorometer as a detector for the solution

phase solute, with excitation and emission filters of

305-395 and 430-470 nm, respectively. In the RPLC retention

studies, injections of the hydrophobic solutes onto the RPLC

supports were made using a Rheodyne 7161 switching valve

with a 20 pL injection loop. For quantitative analysis of

the solution phase in soil batch studies, however, a 200 pL

Table 4-3. List of reagent chemicals and their respective


Polycyclic aromatic



n-Butylbenzene and

Remaining alkylbenzenes
and substituted benzenes


Aldrich Chemical Co.--98+% pure

Fisher Scientific Co.--Reagent

Mallinckrodt Inc.--Nanograde,
99+% pure

Aldrich Chemical Co.--99+% pure

Eastman Kodak Inc.--Reagent grade


injection loop was required. All RPLC sorbate mixtures were

5-500 mg/L in 100% methanol. The flow rate for RPLC thermo-

dynamic retention experiments was set at 1.0 mL/min; the

actual flow rate was measured using a 10 mL graduated

cylinder and a stopwatch and found to be within 5% of this

value. All RPLC experiments were performed at least in


All column packing were purchased from Analytichem

International Inc. (Harbor City, CA). The precolumn packing

material was 40 pm Sepralyte unbonded silica gel. As

discussed in Section 4.2, the reversed-phase stationary

phases consisted of porous, irregularly shaped, 10 um

diameter silica gel particles chemically bonded with the

following trichloroalkylsilanes: C-2, C-4, C-8, and C-18.

A listing of these stationary phases and their physico-

chemical properties is shown in Table 4-4. These stationary

phases were slurry-packed into 5 cm x 4.6 mm (i.d.) x 1/4"

(o.d.) stainless steel HPLC columns, equipped with 2 um

frits at each end. The unbonded silica gel was dry-packed

into a similar column for use as a precolumn; however, the

outlet end of this column contained a 0.5 Pm frit to prevent

fines from clogging the injection valve and downstream


Both the precolumn and analytical HPLC column were

thermostated by circulating water jackets. The precolumn

was filled with unbonded silica gel and was placed before

Table 4-4. Physical and chemical properties of reversed-
phase, 10 pm diameter particle size, stationary
phases used in chromatographic studies.

Alkyl Density
group (200K)
molecular of parent
Stationary Percent Alkyl wt. n-alkaje
phase carbon ligand (g/mole) (g/cm )

C-2 5.57 -CH2CH3 29 0.5235

C-4 7.92 -(CH2)3CH3 57 0.5788

C-8 12.05 -(CH2)7CH3 113 0.7025

C-18 14.73 -(CH2)17CH3 253 0.7768

the injection valve to assist in bringing the mobile phase

to the required temperature while also saturating the

solvent mixture (water/methanol, etc.) with dissolved

silica. The former effect is desired to avoid differential

temperature bands in the analytical column while the latter

minimized the loss of silica support from the downstream

analytical column. The circulating water bath was a

Brinkman Model RC-20T, capable of operating over a tempera-

ture range of -15 to 1000C, with an accuracy of + 0.20C.

The temperature of the water bath was monitored with a

Bailey Model BAT-8 digital thermocouple thermometer. This

instrument has a stated range of 0 to 1000C and a maximum

sensor error of + 0.10C at 1000C. The RPLC retention

studies were routinely done at temperatures of 298, 308,

318, and 3280K for a single isocratic solvent mixture. In

some experiments, the temperature 288K was substituted for

3280K to avoid possible heat capacity effects in the hydro-

phobic sorption reaction.

A Shandon HPLC packing pump (courtesy of Dr. J. Dorsey)

was used to slurry-pack the RPLC sorbent materials into the

HPLC columns. Approximately 2 mL of the selected sorbent

material was slurried in 20 mL of chloroform (Reagent

grade, Fisher Scientific Co.). This slurry was shaken and

placed in an ultrasonic bath for 20 minutes to insure

adequate distribution of the 10 m material. The sorbent-

chloroform slurry was then poured into the packing reservoir

and packed at 6000 psi with a succession of increasingly

viscous solvent mixtures: 50/50 (v/v) chloroform/methanol,

100% methanol, and 50/50 (v/v) methanol/water. The first

two solvent mixtures were run for 1-2 minutes, with the

methanol/water mixture then packing the column for 10-15

minutes. Following the packing procedure, the column was

removed and installed on the Gilson HPLC system.

A constant temperature room (courtesy of Dr. D. Silvia)

was used to perform all soil thermodynamic sorption experi-

ments. The room was manufactured by the Electric Hotpak

Co. (Philadelphia, PA) and had an operating range of 0 to

400C. The observed temperature variability was +1.00C in

studies from 5 to 350C.

4.8 Experimental Techniques

4.8.1 General RPLC Experiments

A typical RPLC experiment involved the isocratic

elution of all compounds from a given RPLC sorbent at a

known solvent composition. For any given methanol/water or

acetonitrile/water eluent, retention studies were performed

at a minimum of four temperatures, e.g., 298 to 328K in

increments of 100K. The solvent composition was then

changed, usually by steps of 8 = 0.10, and the temperature

studies repeated. Four to five methanol/water or

acetonitrile/water solvent systems were studied in this

manner for each RPLC material. Solute retention factor (k')


data were collected as functions of solvent composition (6)

and temperature (T) on C-2, C-4, and C-8 stationary phases

in methanol/water and acetonitrile/water mobile phases. To

explore further the importance of stationary phase chain

length in hydrophobic sorption reactions, isothermal reten-

tion data were collected on a C-18 sorbent in various

acetonitrile/water solvent systems at 2980K. The collected

In k' data for all sorbent/solvent combinations may be

found in Appendix A.

All RPLC experiments were performed on the Gilson HPLC

unit using a 1.0 mL/minute flow rate. Measured flow rates

agreed with this value within a 95% confidence limit.

Isocratic solvent systems prepared using 250 mL burets gave

identical solute retention as eluents prepared by the Apple

IIe/analytical mixer system; i.e., k' values agreed within

1% of mean values. All retention studies were run at least

in triplicate, and triplicate analyses typically agreed to

within +1-4% of the mean k' value.

4.8.2 Chromatographic Data Analysis

The coefficients for Eqns. (3-57) and (3-58) were ob-

tained from multiple linear regression analyses of retention

factor-composition-temperature data for each solute using

the GLM package in SAS (Statistical Analysis System, Inc.,

Box 8000, Cary, NC) on an IBM 3081D/3033N/4341 computer at

the Northeast Regional Data Center (Gainesville, FL). A

similar GLM analysis of retention factor-dielectric

constant-surface tension data resulted in the regression

coefficients for Eqn. (3-38).

4.8.3 Determination of Column Void Volume, V

In the previous discussions, the solute retention

factor (k') was described by k' = (tR to)/t where tR and

to are the retention times of the compound under study and

an unretained solute, respectively. It is evident that

proper evaluation of retention factors requires a knowledge

of the chromatographic void volume (t or V ) of the column,

i.e., the mobile phase volume that is experienced by the

solute in the course of the chromatographic process.

Ideally, the column void volume is represented by the

elution volume of an inert or ionic solute that explores the

available mobile phase volume but does not interact with the

stationary phase.

Recently, the problem of determining column void volume

has received considerable attention in the RPLC literature.

The investigative approach to this problem has taken several

fronts. Some researchers used the linearity of solute

retention for homologous series to determine V (Berendsen

et al., 1980; Krstulovic et al., 1982; Tchapla et al.,

1984). McCormick and Karger (1980) have made a convincing

case for the use of D20, while others used various organic


and ionic solutes for V determination (Jinno et al., 1983;

Wells and Clark, 1981).

Melander et al. (1983) examined both the theory and

empirical determination of the void volume. They concluded

that the nitrate ion (as in NaNO3 or KNO3) was an adequate

V indicator, if the mobile phase contained between 25 and

75% methanol or between 25 and 95% acetonitrile and if the

ionic strength of the eluent was sufficiently high to avoid

the effect of Donnan salt-exclusion (Donnan, 1924). This

strategy is in agreement with the research of Wells and

Clark (1981), who found that the column void volume could be

best estimated by an injection of approximately 3 x 106

mole or more of NaNO3 in an unbuffered methanol/water

system. The large amount of NaNO3 was recommended to avoid

Donnan salt-exclusion effects that may occur in an

unbuffered mobile phase.

The NaNO3 salt was chosen for use as the void volume

indicator in the RPLC chromatographic studies. Since unbuf-

fered solutions of methanol/water and acetonitrile/water

were used, it was necessary to study the effect of NaNO3

concentration upon measured void volume. The results are

shown in Figure 4-1. The effect of increasing salt concen-

tration upon measured retention volume was attributed to

Donnan ion-exclusion. A concentration of 25 g/L was

selected for use in V determinations for methanol/water and

acetonitrile/water systems. If an injection volume of






in ra


4) r
r. f

\ w awOJ l

\ o z
<< 00
\ ^rf "1 t\
\ (0 >

(lus) 0~lo uo^a~
's, *H

-s^ -P

20 x 10-6 L is assumed, this corresponds to an injection of

approximately 6 x 106 mole of NaNO3. This exceeds the

3 moles suggested by Wells and Clark (1981) to avoid

exclusion effects in unbuffered methanol/water eluents.

The Donnan ion-exclusion effect has also been docu-

mented by Buytenhuys and van der Maeden (1978), who investi-

gated the use of sodium heparin as a void volume indicator

for silica gel columns. The effect they noted was quite

similar to that described by Wells and Clark (1981) and

similar to that shown in Figure 4-1.

4.8.4 Equilibrium Experiments with RPLC Materials

It is an implicit assumption in RPLC thermodynamic

studies that a dynamic equilibrium is established between

the sorbed and solution phases of the solute of interest.

To investigate this assumption for the RPLC systems used in

my studies, two experiments were performed:

(1) The solute retention factor, k', was examined as a

function of column flow rate. On the C-8 sorbent, in a

60/40 (v/v) acetonitrile/water mobile phase, the k' of

pyrene at 2980K was recorded at flow rates of 0.10 to 1.0

mL/min. If solution-phase kinetics were a limiting factor,

one would expect some differentiation of k' as flow rate is

changed over a significant range.

(2) Individual batch equilibrium sorption isotherms of

biphenyl and pyrene were performed on RPLC material

dissolved in a 50/50 or 60/40 (v/v) acetonitrile/water

solvent system. The linear equilibrium sorption coeffi-

cients for these systems were then calculated and compared

to the column k' values for an identical system. When

corrected for column mass and dead volume, the equilibrium

sorption coefficient should be equivalent to k' measured on

the column. The sorbent/solvent mixtures were shaken for 24

hours prior to centrifugation and analysis and are therefore

assumed to represent equilibrium systems.

Approximately 0.1 to 0.4 g of the RPLC material was

weighed into eight 25 mL glass centrifuge tubes with Teflon-

backed rubber stoppers. Five milliliters each of four

solute standards were then added to the tubes for duplicate

analysis. The tubes were shaken at ambient temperature

(22 + 30C) for 24 hours and centrifuged for 30 minutes at

1000 rpm. Quantitative HPLC analysis of the solution phase

employed a 15 cm, 10 pm, C-8 Zorbax column (DuPont column,

via Fisher Scientific Co.) with a 60/40 (v/v) acetonitrile/

water mobile phase at 2980K and 1.5 mL/min flow rate.

Column effluent was monitored by the Waters 450 variable

wavelength UV detector set at 254 nm. The solution phase

concentration (C ) of the solute of interest was determined

from a four or five point standard curve; the standards were

prepared in similar tubes and solutions and had been taken

through the entire procedure. The sorbed solute concentra-

tion (S ) was calcualted from


S = V/m (Co C ) (4-1)

where S is the sorbed solute concentration (pg/g), C is

the solution phase solute concentration (pg/mL), V is the

volume (mL) of standard added initially to the tube, m is

the mass (g) of sorbent present, and C0 is the standard

solute concentration (ug/mL) added to the tube.

The batch equilibrium isotherm data were interpreted

with the aid of the Freundlich equation, which expresses the

equilibrium relationship between the solute in solution (C)

and the sorbed solute (Se) as


S = K CN (4-2)
e e

In S = In K + N In C (4-3)
e e

where K is the thermodynamic equilibrium sorption coeffi-

cient, and N is a constant. The linear version of the

Freundlich equation, where N = 1.0, was used by a number of

investigators to describe the sorption of organic pesticides

onto soil (Davidson and McDougal, 1973; Kay and Elrick,

1967). Other researchers have indicated that the linear

Freundlich equation is inappropriate under certain soil

conditions (Rao and Davidson, 1979). Since linear sorptive

behavior is generally assumed to occur in RPLC (Snyder and

Kirkland, 1979), the linear form of the Freundlich equation

was used to interpret the results of these RPLC batch

equilibrium studies.

4.8.5 Soil Thermodynamic Sorption Studies

To properly study the thermodynamics of the sorption of

hydrophobic solutes onto soils, the sorption of three hydro-

phobic solutes in a natural soil/solvent mixture was studied

at several temperatures using the batch equilibrium sorption

technique outlined in Section 4.8.4. The solutes chosen

were biphenyl, anthracene, and pyrene, which were dissolved

in a 30/70 methanol/water solvent mixture. Four standards

of each solute were prepared, and 5 mL of each were

individually applied to 1 g samples of Webster soil in glass

centrifuge tubes with Teflon-lined caps. The samples were

then shaken for 24 hours and then centrifuged for 1 hour at

1000 rpm in the constant temperature room. Four tempera-

tures were used: 5, 15, 25, and 350C, with an observed room

temperature variation of +10C. Aliquots (50 pL) of the

solution phase were removed from the tubes in the constant

temperature room, and then were analyzed using the same HPLC

conditions outlined in Section 4.8.4. However, a 65/35

acetonitrile/water (v/v) mobile phase was used for most

analyses. In addition, for the solutes anthracene and

pyrene, the Gilson Model 121 filter fluorometer (excitation

and emission filters of 305-395 and 430-470 nm,

respectively) was required for analyzing the low solution


phase concentrations of these solutes. The collected soil

solution data were then analyzed using the nonlinear form of

the Freundlich equation (Eqn. 4-2).


5.1 Introduction

This chapter will review the results of all reversed-

phase liquid chromatography (RPLC) and soil experiments that

were performed and presents an extensive discussion of those

results. The initial section outlines the experimental data

and indicates their placement in the appendices. The

discussion and interpretation of the data comprise the

remainder of the chapter. The chapter will present an

analysis of the thermodynamics of hydrophobic sorption on

both RPLC materials and the Webster soil, the application of

the enthalpy-entropy compensation model for RPLC retention,

and use of the solvophobic model for examining the

mechanisms of hydrophobic sorption.

5.2 Results

This section outlines the experimental and statistical

results developed during this study and identifies the

location of those results in Appendices A to I.

Sorption data in the form of In k' as a function of

temperature and solvent content were obtained for all

experimental solutes retained on C-2, C-4, and C-8 sorbents

in methanol/water and acetonitrile/water solvent systems.

Similar data were generated at a single temperature (2980K)

for the solutes on a C-18 sorbent in an acetonitrile/water

mobile phase. The collected retention data may be found in

Appendix A. The HSA value for each solute is supplied for


The In k' data collected for each solute-sorbent

system of interest have been linearly regressed against the

inverse of absolute temperature, as presented in Eqn.

(3-41). The regression coefficients for these individual

van't Hoff plots are tabulated in Appendix B, where the

correlation coefficient, slope, and intercept are listed, in

addition to the 95% confidence limits for the latter two

regression parameters.

The van't Hoff regression data (Appendix B) were mathe-

matically modified to represent the mean AHo and AS0 values

for solute sorption in the RPLC systems under study. The

AH p data are presented in Appendix C, while the AS0
sorp sorp
values, corrected for the column phase ratio, are listed in

Appendix D.

The results of the RPLC retention experiments in

methanol/water and acetonitrile/water eluents were analyzed

using the enthalpy-entropy compensation model outlined in

Chapter III, Section 3.4.3. Both the three-parameter

(Eqn. 3-57) and four-parameter (Eqn. 3-58) models were

studied, and the respective regression coefficients and

correlation coefficients are listed in Appendix E. The

three-parameter model was applied to methanol/water and

acetonitrile/water systems, while the nonlinear response of

In k' to acetonitrile content ( ACN suggested the addi-

tional application of the four-parameter model to all

acetonitrile/water systems for comparison purposes. To

examine the importance of the compensation temperature upon

the calculated thermodynamic parameters (AH (0), a, and Y),

the 8 value was varied for the C-8 stationary phase material

in the methanol/water and acetonitrile/water solvent

systems. The results of these calculations also appear in

Appendix E.

The data for the physicochemical properties of

methanol/water and acetonitrile/water solvent systems,

respectively, appear in Appendices F and G. These prop-

erties include density, refractive index, dielectric

constant, molar volume, surface tension, and Ke; all data

are at 250C.

The physicochemical data tabulated in Appendices F and

G may be used in conjunction with the In k' data of Appendix

A to examine the applicability of the solvophobic model

(Horvath et al., 1976) to the RPLC systems under study.

The solvophobic model (Eqn. 3-38) has been applied to the

C-2, C-4, C-8, and C-18 stationary phases in an

acetonitrile/water solvent system at 250C (2980K). The

calculated regression parameters and correlation coeffi-

cients are shown in Appendix H.

The experimental data and statistical analyses of the

Webster soil thermodynamic solute sorption studies are

listed in Appendix I. Data for each of the three solutes

involved in the experiments (biphenyl, anthracene, and

pyrene) are detailed for each system temperature.

5.3 Hydrophobic Retention on RPLC Materials

An initial objective of the RPLC studies was to examine

differences in sorptive behavior between structurally

distinct classes of chemicals, i.e., the PAHs and the alkyl-

benzenes. To properly evaluate differences in hydrophobic

retention, the compounds of interest must be described and

differentiated on the basis of chemical properties or

molecular structure. Parameters that are commonly used in

such studies on RPLC or soil materials include topological

descriptors such as the molecular connectivity index

(Bojarski and Ekiert, 1982; Hanai and Hubert, 1984; Sabljic,

1984; Wells et al., 1982), and measures of hydrophobicity

such as aqueous solubility (Chiou et al., 1979; Chiou et

al., 1983) and the octanol/water partition coefficient

(Chiou et al., 1983; Hammers et al., 1982; Harnisch et al.,

1983; Karickhoff, 1981; Koopmans and Rekker, 1984; Rao and

Nkedi-Kizza, 1981; Veith et al., 1979; Wells and Clark,

1984). Geometric descriptors such as the length/breadth


ratio (Wise et al., 1981) or the van der Waals volume (Jinno

and Kawasaki, 1983b; Hanai and Hubert, 1984), along with

chemical properties such as molecular polarizability (Jinno

and Kawasaki, 1984b; Lamparcz et al., 1983) and the solu-

bility parameter (Hafkenscheid and Tomlinson, 1983) are also

commonly used to describe solute sorption in RPLC systems.

Jinno and Kawasaki (1984a) reported that although the sorp-

tion of PAHs and alkylbenzenes is dominated by hydrophobic

interactions, the size and shape of these molecules are of

considerable importance in determining overall RPLC


The natural logarithm of the solute retention factor

(In k') was linearly regressed against the logarithm of the

octanol/water partition coefficient (log K ) for sorption

on C-8 material in 60/40 methanol/water and 50/50

acetonitrile/water. These plots are shown in Figures 5-1

and 5-2, respectively. The log Kow is generally considered

a measure of the solute's hydrophobic nature (Leo et al.,

1971). The log Kow values for the solutes used in the

regression are listed in Table 5-1. In both the methanol/

water and acetonitrile/water mobile phase systems, there is

a difference in overall sorptive behavior between the PAH

compounds and alkylbenzenes, based on a measure of their

respective hydrophobicities. The regression parameters for

both classes of compounds, taken from the data in Figures









C) 0
-4 CO




0 C


0 I

> 0


(>Io86Z) ,MI u7







0 O,





*L 0
4- -,1




(>1~86) r.u-




(N.8613) U-1~


Table 5-1. Octanol/water partition coefficients (K ) of the
hydrophobic solutes used in the RPLC retention

Compound Log K Reference



Bruggeman et al., 1982
Bruggeman et al., 1982
Bruggeman et al., 1982
Bruggeman et al., 1982
Bruggeman et al., 1982
Bruggeman et al., 1982
Bruggeman et al., 1982
Leo et al., 1971
Nys and Rekker, 1973
Nys and Rekker, 1973
Wasik et al., 1981
Bruggeman et al., 1982
Bruggeman et al., 1982
Wasik et al., 1981
Wasik et al., 1981
Wasik et al., 1981
Wasik et al., 1981
Wasik et al., 1981
Wasik et al., 1981

5-1 and 5-2, are listed in Table 5-2 for the two RPLC


Also shown in Table 5-2 are data for similar In k'

vs. log Kow correlations on the C-2 sorbent in various

methanol/water and acetonitrile/water mobile phases. Over-

all, there appears to be little to no difference in sorptive

behavior as the alkyl chain of the stationary phase is

increased in length. These findings differ considerably

from those of Jinno and Kawasaki (1983b), who found that the

size and shape of a solute molecule are the dominant factors

controlling retention on a C-2 stationary phase, with hydro-

phobic interactions becoming more important as stationary

phase chain length is increased. Their conclusions were not

supported for the RPLC sorbents used in this study.

The hydrocarbonaceous surface area (HSA) is a useful

descriptor of the solute area available for hydrophobic

interactions (Horvath and Melander, 1977; Horvath et al.,

1976; Nkedi-Kizza et al., 1985; Rao et al., 1985; Woodburn

et al., 1985). It has been used extensively by Horvath et

al. (1976) in their solvophobic model of hydrophobic reten-

tion and by Rao et al. (1985) and Woodburn et al. (1985) as a

tool in modeling solute sorption in soil systems containing

organic solvent/water mobile phases. The excellent linear

correlation between the HSA and the classic hydrophobicity

parameter, log Kow, is shown in Figure 5-3, where a single

regression line was found to fit the PAHs, alkylbenzenes,

Table 5-2. Regression parameters from linear correlation of
In k' (2980K) vs. log Kow in various RPLC systems.

Regression parameters (+95% conf.
Column Solvent system limit) from In k' (2980K) vs. log K

60/40 MeH/H20a

60/40 MeOH/H20

50/50 ACN/H20b


(1) PAHs, Benzene, Halobenzenes, and

n = 12, R = 0.9967
Slope = 0.52 + 0.03
Intercept = -1.83 + 0.12

(2) Alkylbenzenes

n = 7, R = 0.9995
Slope = 1.85 + 0.03
Intercept = -2.68 + 0.11

(1) PAHs, Benzene, Halobenzenes, and

n = 12, R = 0.9989
Slope = 0.81 + 0.03
Intercept = -1.09 + 0.11

(2) Alkylbenzenes

n = 7, R = 0.9997
Slope = 1.11 + 0.03
Intercept = -1.80 + 0.11

(1) PAHs, Benzene, Halobenzenes, and

n = 12, R = 0.9959
Slope = 0.34 + 0.02
Intercept = -0.52 + 0.09

(2) Alkylbenzenes

n = 7, R = 0.9991
Slope = 0.57 + 0.03
Intercept = -1.10 + 0.10



Table 5-2. Continued

Column Solvent system

50/50 ACN/H20

Regression parameters (+95% conf.
limit) from In k' (2980K) vs. log K

(1) PAHs, Benzene, Halobenzenes, and

n = 12, R = 0.9981
Slope = 0.54 + 0.02
Intercept = -0.14 + 0.12

(2) Alkylbenzenes

n = 7, R = 0.9994
Slope = 0.82 + 0.03
Intercept = -0.87 + 0.12

aMethanol/water solvent system

Acetonitrile/water solvent system


MO > 601



and substituted benzenes. As a result, it is not surprising

that In k' vs. HSA plots are similar to those developed for

In k' vs. log Ko. Plots of In k' vs. HSA are shown in

Figures 5-4 and 5-5 for the C-8 stationary phase with 60/40

methanol/water and 60/40 acetonitrile/water eluents,

respectively. The solute nitrobenzene has been omitted from

any linear regressions involving HSA because it exhibits

unusually strong retention relative to its small HSA value

(86 A2). This behavior is believed related to the inter-

action of the polar nitro group with the available silanol

groups on the RPLC surface. Tanaka et al. (1978) reported

similar behavior for nitrobenzene and attributed it to

preferential polar interactions with the silanol groups

still present on the RPLC surface.

The difference in retention behavior for the PAH and

alkylbenzene solutes is again quite distinct in the In k'

vs. HSA plots for C-8 sorption in the two solvent systems

(Figures 5-4 and 5-5). The pertinent regression parameters

for the linear correlation of ln k' vs. HSA in several

sorbent/solvent systems are listed in Table 5-3. As in the

In k' vs. log Kow relationships, stationary phase chain
length does not appear to affect the quality of the correla-

tion of retention to solute structure.

The molecular connectivity index, or chi factor, is an

easily computable topological parameter devised by Randic

(1975) and extended by Kier and Hall (1976) for describing

C(o86Z) l u-1








0 CN


4 c


m C

C (0





\ \ ru co

o NJ o 0
CL o
\- \

-) I


\ 0 c

< = \ v3 (

\ \:r >

\ \ )

~\ \ -*S -4 1

\A8 Z M U-1

\ \ ^ 4

\-3 \r (4

Table 5-3. Regression parameters from linear correlation of
ln k' (2980K) vs. HSA in various RPLC systems

Regression parameters (+95% conf.
Column Solvent system limit) from In k' (2980K) vs. HSA

C-2 60/40 MeOH/H2Oa (1) PAHs, Benzene, and Halobenzenes
n = 12, R = 0.9941
Slope = 0.0141 + 0.0010
Intercept = -2.21 + 0.19

60/40 MeOH/H20

50/50 ACN/H20b

50/50 ACN/H20

(2) Alkylbenzenes
n = 9, R = 0.9955
Slope = 0.0245 + 0.0021
Intercept = -3.58 + 0.34

(1) PAHs, Benzene, and Halobenzenes
n = 12, R = 0.9975
Slope = 0.0224 + 0.0018
Intercept = -1.74 + 0.31

(2) Alkylbenzenes
n = 9, R = 0.9975
Slope = 0.0313 + 0.0018
Intercept = -2.86 + 0.29

(1) PAHs, Benzene, and Halobenzenes
n = 12, R = 0.9944
Slope = 0.0094 + 0.0007
Intercept = -0.77 + 0.12

(2) Alkylbenzenes
n = 9, R = 0.9945
Slope = 0.0160 + 0.0014
Intercept = -1.61 + 0.22

(1) PAHs, Benzene, and Halobenzenes
n = 12, R = 0.9940
Slope = 0.0149 + 0.0012
Intercept = -0.59 + 0.20

(2) Alkylbenzenes
n = 9, R = 0.9948
Slope = 0.0236 + 0.0022
Intercept = -1.73 + 0.35

aMethanol/water solvent system

bAcetonitrile/water solvent system




molecular shape and the degree of molecular branching. A

large number of studies have demonstrated that many

physicochemical and biological properties depend upon the

topology of a molecule, which may be related to the con-

nectivity index. These properties include water solubility

and boiling point (Hall et al., 1975), the octanol/water

partition coefficient (Murray et al., 1975), the activity of

general anesthetics (DiPaolo et al., 1977), solute retention

in RPLC and GC systems (Bojarski and Ekiert, 1982; Hanai and

Hubert, 1984; Jinno and Kawasaki, 1983a, 1983b, 1984a;

Kaliszan and Lamparczyk, 1978; Wells et al., 1982) and

solute sorption on soil material (Sabljic, 1984).

The first-order connectivity index, IX, may be computed

from the expression

1X = E (6i6)1/2 (5-1)

where the sum is over all bonds in the molecule. Atoms i

and j are directly bonded; 6 is a number assigned to each

atom reflecting the number of nonhydrogen atoms bonded to

it. The IX values for the PAH compounds and alkylbenzenes

are listed in Table 5-4, which includes a sample IX calcu-

lation for ethylbenzene.

Chromatographic data (In k') from Appendix A were used

to develop linear correlations between In k' and 1X for the

PAHs and alkylbenzenes in methanol/water and acetonitrile/


Table 5-4. First-order molecular connectivity indices of X of
the PAHs and alkylbenzenes. A sample calculation
for the X value of ethylbenzene is shown.

Compound Calculated X value

Biphenyl 4.071
Naphthalene 3.405
Phenanthrene 4.815
Anthracene 4.809
Pyrene 5.559
Chrysene 6.226
Fluoranthene 5.565
Benzene 2.000
Toluene 2.411
Ethylbenzene 2.971
n-Propylbenzene 3.471
n-Butylbenzene 3.971
n-Hexylbenzene 4.971
o-Xylene 2.827
p-Xylene 2.821
m-Diethylbenzene 3.943
1,2,4-Trimethylbenzene 3.238

Sample X calculation for ethylbenzene:

3 c dC 3
c/ \e
3C C-C ---C
b \ /f4 g2 h 1
3 a 3

1 1 + 1 + 1 1 1 1
(3x3) (3x3) (3x3) (3x3) (3x4) (3x4)
a b c d e f

1 1
+ +
(4x2) (2xl)
g h

X = 2.971

water eluents. The resulting correlations are shown in

Figures 5-6 and 5-7 for sorption on C-8 material in 60/40

methanol/water and 40/60 acetonitrile/water solvent systems,

respectively. The distinction between PAH and alkylbenzene

retention as a function of molecular shape is evident from

the figures and from the In k' vs. lX regression data,

presented in Table 5-5. As the data in Table 5-5 demon-

strate, stationary phase chain length does not significantly

affect the correlation of retention behavior to molecular

topology. This conclusion differs from that of Jinno and

Kawasaki (1983b), who reported that the correlation of

solute retention to molecular shape ('X) improved consider-

ably as the stationary phase chain length was decreased.

The explanation for the different conclusions reached

in these two retention studies may lie in the steric dis-

tinction between the "brush-type" stationary phases used by

Jinno and Kawasaki (1983b), and the polymericc type" RPLC

phases employed in my experiments. The brush-type reversed-

phases are prepared by reacting dried silica gel with a

monohalosilane, producing one bonded hydrocarbon chain per

available silanol group. The polymeric or bulk-type phase,

on the other hand, is prepared from silica reacting with a

trichlorosilane in the presence of water, forming a cross-

linked polymeric structure upon reaction (Scott and Simpson,

1980). The findings of Jinno and Kawasaki (1983b) may be in

agreement with the finding that shorter (C-2) brush-type