This is to certify that the
dissertation entitled

Thermodynamic and Kinetic Characterization of Solute
Transfer in Reversed-Phase Liquid Chromatography

presented by

Samuel Barnett Howerton

has been accepted towards fulfillment
of the requirements for the

Chemistry - Environmental

PhD. degree In Toxicology

 

 

’UW mCWM

Major Profe'ssor’s Signature

lllmlzoos

v

 

Date

MSU is an Amnnative Action/Equal Opportunity Institution

C- .l- - 4

‘ .‘4

 

 

 

LIBRARY
Mlchlgan State
Unlversity

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDue.p65-p.15

 

THERMODYNAMIC AND KINETIC CHARACTERIZATION OF SOLUTE
TRANSFER IN REVERSED-PHASE LIQUID CHROMATOGRAPHY

By

Samuel Barnett Howerton

A Dissertation

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

DOCTOR OF PHILOSOPHY
Department of Chemistry

2003

mJ 1%
T“ M.»
, I. .

p
|
a
p

U383

Amid

LU

Nil“ E
C. u . N v
,7. w...

{s .l... ,. \ . .
.N 4 A U va N l. u

a... . c “w T? L. a-.. a u. A. a. .

.3 ~.l.. . s s i. x u .L. 1.4 u;

ABSTRACT
THERMODYNAMIC AND KINETIC CHARACTERIZATION OF SOLUTE
TRANSFER IN REVERSED-PHASE LIQUID CHROMATOGRAPHY
By

Samuel Barnett Howerton

Reversed-phase liquid chromatography is a technique by which complex
mixtures of solutes are separated from one another based upon their affinities for
a polar mobile phase and a nonpolar stationary phase. This thesis discusses the
use of functions derived from the theoretical description of retention developed
by Giddings, as well as instrumentation, the thermodynamics and kinetics of
retention are measured and quantitated in tandem.

In orderto validate the data treatment procedure, a series of simulations
was carried out to study the effect of the integration limit, the number of points
across the profile, and noise. Statistical moment analysis and the exponential
modified Gaussian model (EMG) were compared, and the data indicated that the
EMG model was more robust and resulted in smaller errors for simulated and
experimental data.

Using the conclusions drawn from the simulation studies, the
thermodynamics and kinetics of solute transfer for a series of polycyclic aromatic
hydrocarbons (PAHs) were studied as a function of ring number, annelation (i.e.,
degree of ring fusion), planarity, temperature, pressure and bonding phase

density using octadecylsilica (ODS). The data from this study indicated that an

increase in ring number results in more negative changes in molar enthalpy

 

 

(AHsm) and molar volume (AVsm). For a series of isomers, highly annelated
solutes exhibited less negative changes in molar enthalpy and molar volume than
the more linear solutes. Nonplanar solutes demonstrated changes in molar
enthalpy and molar volume that are less negative than would be expected based
upon ring number alone. These data indicated that the condensed PAHs, as well
as the nonplanar PAHs, interacted with the first few carbons near the distal
terminus, and that the more linear solutes, as well as the solutes with more rings,
interacted with the more ordered regions of the ODS closer to the proximal
terminus. In addition, the rate constants demonstrated that the rates of transfer
decreased with increasing ring number and decreasing annelation. The
enthalpic and volume barriers were found to be very large. These barriers
increased as a function of ring number, but decreased with increasing annelation.

In addition to the parent PAHs, a series of nitrogen containing polycyclic
aromatic hydrocarbons (NPAHs) were studied as a function of temperature,
pressure and mobile phase (i.e. protic or aprotic). The thermodynamics for the
NPAHs were similar to the PAHs. However, the rate constants varied
dramatically as a function of mobile phase. Comparisons between methanol and
acetonitrile demonstrated changes in the rate constants ranging from two to four
orders of magnitude. These differences are attributed to the interaction between
the nitrogen and the silica support.

In addition to the studies using octadecylsilica, the retention mechanism of
soil was also explored. Furthermore, the qualitative analysis of PAH

contaminated samples was studied using selective fluorescence quenching.

To all those who believed in me, even when I did not;

especially my mother and father.

 

ACKNOWLEDGEMENTS

And now my Charms are all o'erthrown
And what strength I have's mine own
Which is most faint; now t'is true

I must here be released by you

But release me from my bands
With the help of your good hands
Gentle breath of yours my sails
Must fill, ortelse my project fails,
Which was to please. Now I want
Spirits to enforce, art to enchant
And my ending is despair,

Unless I be relieved by prayer

Which pierces so that it assaults
Mercy itself and frees all faults
As you from your crimes would pardon'd be

Let your indulgence set me free.

-William Shakespeare, “T he Tempest”

It seems fitting that the last words I pen for this dissertation are for those
people who have made my work possible. As I sit here, contemplating where l
have been, and where I am going, I think back to the old adage that no man is an
island. I certainly am not, and the people listed below have helped me in ways
that mere words can never convey.

I extend my profound thanks to my advisor Professor Victoria McGuffin for
molding me into a better scientist and thinker. Her desire to explore the
fundamental nature of the universe will stay with me forever, driving me to always
wonder why the world is the way that it is. I would also like to thank to my
committee members Drs. Boyd, Voice, Jackson, Garrett, and Bruening. With
their help and viewpoints, l was able to see beyond the traditional boundaries of
analytical science.

Then there are my friends...those individuals, who against better judgment
actively engaged the world of Sam. First and foremost there are my peers. Dr.
Thomas Cullen, a steadfast friend who, through humor, reminds me that while
chemistry is important, it is not the only thing to live for. He is a paragon not only
of virtue, but also of dedication to excellence. Dr. John Goodpaster, who despite
my constant abuse and pestering, demonstrated the greatest work ethic l have
ever seen in a human being. He taught me not only how to be productive in a
sleep deprived state, but how to ask for help when I needed it most. Dr. Peter
Krouskop, for reminding me on a daily basis what it is that I am fighting for and
the type of man I am striving to become. Without a doubt, he is one of the most

moral men l have met, and the world would be better served if more were like

vi

hm, And
973% DE

cl fancy a

stress I
coded
:tnatei;
teen by
El deli

TI

ttlo' .

_5
(J.

‘9. I, :
”“36 IE

miles:

355.393:

l’C‘L'leg

”1P4 j

. we...
'3
FM

y,

1hr”. I

WM

”9431"; i

v.1“: 3A
lv tan

him. And finally Carl Newman...who is, by definition, indefinable to me. He has
shown patience and understanding where few others would, tolerating my flights
of fancy and my overwhelming bouts of cynicism. Without his humor this process
would have been worse.

And to my soul mates, Charlotte and Raven. There are no words that can
express my love for each of you. Your support over the years has always
provided a beacon for me, reminding me of who I was, where I came from, and
ultimately where I want to go. Through the. good and the bad, you have always
been by my side, looking out for me when no one else would, or could. I have
not doubt that I am better man for having you as my friends.

To Tiffany, my time with you has taught me many lessons that will never
be found in books; the greatest among these to have no regrets. And though my
course is different from yours, I will always remember our times together with
fondness. You have reminded me what it is to be human, though I try so
desperately not to be.

To my family, those who have shaped me into the man that I am. Without
your love and words of comfort my time in graduate school would have been
infinitely worse. All that I am, all that this document represents, is due in no small
part to your guidance throughout my life.

Finally to all the others whose names are obscured in my memory.
Though’unnamed, your aid is not unnoticed. To each of you I say thank you.
Though it is hard to imagine, my world has been impacted by each of you and l

have learned and adapted from your lessons.

vii

USTC
USN

Chapt

Cha;

Cha

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES

Chapter 1: Introduction and Background

1.1 Polycyclic Aromatic Hydrocarbons (PAHs)
1.1.1 Classification and Structure
1.1.2 Origin and Formation
1.1.3 Biological and Chemical Significance
1.2 Thermodynamic and Kinetic Theory
1.2.1 Thermodynamics
1.2.2 — Kinetics
1.3 Previous Investigations
1.3.1 Thermodynamics
1.3.2 Kinetics
1.4 Conclusions
1.5 References

Chapter 2: Experimental Methods

2.1 Introduction

2.2 Experimental Systems
2.2.1 Column Preparation
2.2.2 Chromatographic System
2.2.3 Spectroscopic System

2.3 Data Treatment and Analysis
2.3.1 Mathematical Functions

2.4 Conclusions

2.5 References

Chapter 3: Mathematical Analysis In Chromatographic Systems:
A Comparison of the Exponentially Modified Gaussian
Equation and Statistical Moments

3.1 Introduction
3.1.1 Sources of Broadening
3.1.2 Mathematical Relationships
3.2 Simulation Methods
3.2.1 Integration Limits
3.2.2 Number of Points
3.2.3 Noise
3.3 Experimental Methods
3.4 Results and Discussion

viii

xi

XV

32

32
32
32
35
36
45
45
50
51

53

53
53

55
60

65

3.4.1 Simulation Results 65

3.4.2 Experimental Results 81
3.5 Conclusions 98
3.6 References 99

Chapter 4: Retention of Polycyclic Aromatic Hydrocarbons: Effect of
Ring Number, Planarity and Stationary Phase Bonding

Density 100

4.1 Introduction 100
4.2 Experimental Methods 101
4.2.1 Solutes 101

4.2.2 Experimental System 104

4.3 Results and Discussion 104
4.3.1 Thermodynamic Behavior 104

4.3.2 Kinetic Behavior 121

4.4 Conclusions 132
4.5 References 134

Chapter 5: Retention of Polycyclic Aromatic Hydrocarbons: Effect of

Annelation 136

5.1 Introduction 136
5.2 Experimental Methods 138
5.2.1 Solutes 138

5.2.2 Experimental System 138

5.3 Results and Discussion 140
5.3.1 Thermodynamic Behavior 140

5.3.2 Kinetic Behavior 149

5.4 Conclusions 170
5.5 References 172

Chapter 6: Thermodynamic and Kinetic Characterization of
Nitrogen-Containing Polycyclic Aromatic Hydrocarbons

in Protic and Aprotic Mobile Phases 174

6.1 Introduction 174
6.2 Experimental Methods 176
6.2.1 Solutes 176

6.2.2 Experimental System 176

6.3 Results and Discussion 178
6.3.1 Peak Profiles 178

6.3.2 Methanol Mobile Phase 183

6.3.3 Acetonitrile Mobile Phase 204

6.4 Conclusions 210
6.5 References 212

7.4 CI
7.5 R'

Chapter 8: I
i
I

8.1 Ir

8.2 E

8.3 F

9.4 <
9.5 I

Chaliter 9:

9.1 I
9.2 I
9.3 l
9.4 .

I

9.5 I

Chapter 7: Retention Mechanisms in Whole Soil

7.1 Introduction
7.1.1 Composition of Soil
7.1.2 Retention Mechanisms in Soil
7.2 Experimental Methods
7.2.1 Stationary Phase
7.2.2 System Parameters
7.3 Results and Discussion
7.3.1 Batch Isotherms
7.3.2 Soil Column Characteristics
7.3.3 Retention Mechanisms
7.4 Conclusions
7.5 References

Chapter 8: Characterization of Polycyclic Aromatic Hydrocarbons
in Environmental Samples by Selective Fluorescence
Quenching

8.1 Introduction
8.1.1 Fluorescence and Fluorescence Quenching
8.2 Experimental Methods
8.2.1 Materials
8.2.2 Experimental System
8.2.3 Data Analysis
8.3 Results and Discussion
8.3.1 Standard (EPA 610)
8.3.2 Coal-Derived Fluid (SRM 1597)
8.3.3 Contaminated Soil (CRM104-100)
8.3.4 Statistical Correlation Analysis
8.4 Conclusions
8.5 References

Chapter 9: Conclusions and Future Directions

9.1 Experimental Design and Theoretical Development
9.2 Retention in Octadecylsilica

9.3 Retention in Soil

9.4 Selective Fluorescence Quenching

9.5 References

215

215
215
216
219
219
220
221
221
223
224
233
235

239

239
239
244
244
246
247
248
248
252
256
259
265
266

268

269
270
272
274
276

Table 1.11 Si

ar
Tablel.2: C
II
Table 3.1: Ir
a
Table 3.2: I
I
F
Ii
I
I
I
Table 3.3;
Table 3.4;

Table 3.5:

Table 1.1:

Table 1.2:

Table 3.1:

Table 3.2:

Table 3.3:

Table 3.4:

Table 3.5:

LIST OF TABLES

Sample values for acute toxicity of selected polycyclic
aromatic hydrocarbons

Carcinogenicity of polycyclic aromatic hydrocarbons in
mammals ~

Input parameters for simulated chromatographic peaks
according to the EMG model in Equation 2.4

Regression parameters from the EMG model (A0, A1,

A2, A3) recovered from simulated chromatographic
peaks as a function of integration interval. (A) C(t) at
limits of integration interval, expressed as % of maximum
C(t) value, (B) Error A0 (%) = (A0 - A) x 100/A0,

(C) Error AI (%) = (A1 - tG) x 100/tG, (D) Error A2 (%) =

(A2 - o) x 100/ o, (E) Error A3 (%) = (A3 - T) x 100/1

Statistical moments (M0, M1, M2) recovered from simulated
chromatographic peaks as a function of integration interval.
(A) C(t) at limits of integration interval, expressed as % of
maximum C(t) value, (B) Error Mo (%) = (M0 — A) x 100/Ao,
(C) EI’I’OI' M1 (%) = (M1 - (ta + 1)) X TOO/(ts + T),

(D) Error M2 (%) = (M2 - (o 2 + r 2)) x 100/(0 2 + r2)

Regression parameters from the EMG model (Ao,AI,A2,A3)
recovered from simulated chromatographic peaks as a

function of the number of points within the integration interval.

(A) Number of points within the integration interval at
0.10% of maximum C(t) value, (B) Error A0 (%) =

(A0 - A) x 100/A, (C) Error AI (%) = (A1 - tG) x 100/ta,
(D) Error A2 (%) = (A2 — o) x 100/ o,(E) Error A3 (%) =
(A3 - T) x 100/ T

Statistical moments (Mo,MI,M2) recovered from simulated
chromatographic peaks as a function of the number of points
within the integration interval. (A) Number of points within the
integration interval at 0.10% of maximum C(t) value,

(B) Error Mo (%) = (M0 - A) x 100/A, (C) Error MI (%) =

(M1 - (ta + 1)) x 100/(tG + 1'), (D) Error M2 (%) =
(Mg-(62+12))X100/(02+12)

xi

10

11

56

66

72

74

Table 3.

Table 3.‘

Table 3

Table 4.
Table 4,
Table 4,

Table 4,

Table 5.-
Table 5;

Table 5.:

Table 3.6: Regression parameters from the EMG model (A0, A1, A2, A3)
recovered from simulated chromatographic peaks as a
function of the noise level.(A) Standard deviation of the
random noise level, expressed as % of maximum C(t) value,
(B) Error A0 (%) = (A0 - A) x 100/A, (C) Error AI (%) =
(A1 - t5) x TOO/Te, (D) Error A2 (%) = (A2 — o) x 100/ o,
(E) Error A3 (%) = (A3 — T) x 100/1 77

Table 3.7: Statistical moments (M0, M1, M2) recovered from
simulated chromatographic peaks as a function of the
noise level. (A) Standard deviation of the random noise
level, expressed as % of maximum C(t) value,
(B) Error M0 (%) = (M0 - A) x 100/A, (C) Error MI (%) =
(M1 -— (ta + 1)) x 100/(tG + t), (D) Error M2 (%) =
(M2-(oz+1:2))x100/(oz+12) 79

Table 3.8: Statistical indicators of fit and regression parameters from
the EMG model (A0, A1, A2, A3) and statistical moments
(M0, M1, M2) recovered from experimental chromatographic
peaks in Figure 3.1. (A) AMo (%) = (M0 —Ao) x 100/Ao,
(B) AM1 (%) = (M1 - (A1 + A3» X TOO/(A1 + A3),
(C) AMz (70) = (M2 - (A22 + A32» X 100/ (A22 + A32) 92

Table 4.1: Retention factors for PAHs on monomeric and polymeric
octadecylsilica (ODS) 105

Table 4.2: Molar enthalpy and molar volume for PAHs on monomeric
and polymeric octadecylsilica 112

Table 4.3: Kinetic rate constants for PAHs on monomeric and polymeric
octadecylsilica 123

~ Table 4.4: Activation enthalpy and activation volume for selected
PAHs on polymeric octadecylsilica calculated at T=308 K

and P=1466 psi 127
Table 5.1: Retention factors for PAH isomers 141
Table 5.2: Molar enthalpy and molar volume for PAH isomers 145

Table 5.3: Rate constants for PAH isomers calculated by using the
exponentially modified Gaussian (EMG) model as a function
. of temperature 150

xii

table 5.4:

Table 5.5:
Table 5.6:

Table 5.7:

Table 5.8:

Table 6.1:
Table 82:

Table 6.3:

Ra
9X
TU!

Table“: A

Table 6.5;
Table 6.5 ;

Table 7.1:

We 8.1:

Table 8.2:

Tallle 3'3:

Table 5.4: Rate constants for PAH isomers calculated by using the
exponentially modified Gaussian (EMG) model as a

function of pressure 151
Table 5.5: Activation enthalpies for PAH isomers 155
Table 5.6: Activation volumes for PAH isomers 159

Table 5.7: Rate constants for PAH isomers calculated by using the
EMG and NLC models 165

Table 5.8: Rate constants for PAH isomers calculated by using the bi-
exponentially modified Gaussian (EZMG) model for two sites 167

Table 6.1: Retention factors for NPAHs in methanol 184
Table 6.2: Molar enthalpy and molar volume for NPAHs in methanol 188

Table 6.3: Rate constants for NPAHs in methanol as a function
of temperature and pressure 195

Table 6.4: Activation energies and activation volumes for NPAHs in

methanol 200
Table 6.5: Thermodynamic data for NPAHs in acetonitrile 205
Table 6.6 : Kinetic data for NPAHs in acetonitrile 208

Table 7.1: Concentrations and equilibrium constants for a series
of polycyclic aromatic hydrocarbons on whole and
fractionated soil 222

Table 8.1: Concentration of polycyclic aromatic hydrocarbons (PAHs)
in certified reference materials 245

Table 8.2: Correlation coefficient (r) of the product—moment method
for chromatograms obtained by using laser-induced
fluorescence detection 263

Table 8.3: Correlation coefficient (r) of the product—moment
method for chromatograms obtained by using
laser-induced fluorescence detection with quenching
by nitromethane 263

xiii

Table 8.4: Co
for

dis

Table 8.4: Correlation coefficient (r) of the product—moment method
for chromatograms obtained by using laser-induced fluorescence
detection with quenching by
diisopropylamine 263

xiv

figure1.1: I

figure 12:

figure 1.3:
figure 1.4:

figure 1.5:
figure 2.1:

figure 2.2:

Figure 2.3:

Figure 3_1.

Flgllle 32.

Figure 1.1:

Figure 1.2:
Figure 1.3:

Figure 1.4:

Figure 1.5:
Figure 2.1:

Pi gure 2.2:

Figure 2.3:

Figure 3.1:

Pi sure 3.2:

LIST OF FIGURES

Examples of A) altemant and B) nonaltemant PAH
structures with labeled atoms

Structures of altemant and nonaltemant PAHs

Fluorescence (emission) spectra of altemant and
nonaltemant PAHs

Structures of linearly fused, ortho-fused and
peri-fused PAHs

Energy coordinate diagram

Generalized schematic diagram for the post-column
detection of chromophores/fluorophores using both
ultraviolet-visible absorbance and laser-induced

fluorescence detection. I: injection valve, S: splitting tee,

R: restrictor, PMT: photomultiplier tube,
AMP: current to voltage amplifier, UV-Vis:
ultraviolet-visible absorbance spectrometer

Schematic diagram of the experimental system for
capillary liquid chromatography with on-column
laser-induced fluorescence detection. I: injection valve,
S: splitting tee, R: restrictor, FOP: fiber-optic positioner,
F: filter, PMT: photomultiplier tube, AMP: current to
voltage amplifier

Schematic diagram of the experimental system for
capillary liquid chromatography with laser-induced
fluorescence quenching detection. I: injection valve,
8: splitting tee, L: lens, F: filter; 000: Charge-coupled
device, PMT: photomultiplier tube. (Note: Collection of
fluorescence emission is orthogonal to the incident
laser beam)

Graphical representation of five simulated EMG peak
profiles described in Table 3.1.

Graphical representation of case 4 in which different
integration limits were studied.

14

37

40

43

58

61

figure 3.3: I

figure 3.4:

figure 3.5:

figure 3.6:

Furs 3.7:

FITUTE 4.1:

Flu“ 4.2:

Flgllle 4.:3

Figure 3.3:

Figure 3.4:

Figure 3.5:

Figure 3.6:

F i 9 ure 3.7:

Figure 4.1:

Figure 4.2:

Pi 9 ure 4.3:

Separation of saturated fatty acids ranging from C10

to C22 by capillary liquid chromatography with on-column
laser-induced fluorescence detection at T = 303 K

and P = 3000 psi. Experimental conditions as given in
the text. (A) Detector 1, 23.2 cm from the head of the
column, (B) Detector 2, 28.4 cm, (C) Detector 3, 51.4 cm,
(D) Detector 4, 56.9 cm.

Logarithmic graph of the retention factor versus distance
on the column for the saturated fatty acids CI 0 (O),

012 (El), 014 (A), 016 (0), C18 (0), 020 (I),

C22 (A). Data derived from the chromatograms

in Figure 3.3

Logarithmic graph of oz versus distance on the column
for the saturated fatty acids. Data derived from the
chromatograms in Figure 3.3, symbols defined in
Figure 3.4.

Logarithmic graph of 1:2 versus distance on the column
for the saturated fatty acids. Data derived from the
chromatograms in Figure 3.3, symbols defined in
Figure 3.4. ,

Logarithmic graph of t2/52 versus carbon number for
the saturated fatty acids. Data derived from the
chromatograms in Figure 3.3.

Structure of polycyclic aromatic hydrocarbons used to
study retention in monomeric and polymeric
octadecylsilica.

Representative graph of the retention factor versus
inverse temperature used to calculate the change in
molar enthalpy according to Equation 1.5. Column:
polymeric octadecylsilica. Mobile phase: methanol,
2566 psi, 0.08 cm/s. Solutes: phenanthrene (O),
chrysene (III), picene (O), benzo[a]pyrene (A),
phenanthro[3,4-c]phenanthrene (O),
tetrabenzonaphthalene (A).

Representative graph of the retention factor versus
pressure used to calculate the change in molar volume
according to Equation 1.7. Column: polymeric
octadecylsilica. Mobile phase: methanol, 308 K,

0.08 cm/s. Symbols defined in Figure 4.1.

82

85

87

90

102

110

115

figure 4.-

figure 4.

figure 4.

l'Igure 5.

HTure 5.:

Fleure 5.:

f

EQUle 5.5

Figure 4.4:

Figure 4.5:

Figure 4.6:

Figure 5.1:

Figure 5.2:

Figure 5.3:

I==igure 5.4:

Fig ure 5.5:

Representative graph of the change in molar enthalpy
versus the Change in molar volume. Planar solutes (O),
nonplanar solutes (I).

Representative graph of the rate constants versus inverse
temperature used to calculate the activation enthalpy
according to Equation 1.20. Column: polymeric
octadecylsilica. Mobile phase: methanol,

2566 psi, 0.08 cm/s. Solutes: chrysene kms (0),

ksm (O); benzo[a]pyrene kms (O), ksm (0).

Representative graph of the rate constants versus
pressure used to calculate the change in activation
volume according to Equation 1.20. Column: polymeric
octadecylsilica. Mobile phase: methanol, 303 K,

0.08 cm/s. Symbols defined in Figure 4.5.

Structure of polycyclic aromatic hydrocarbons used to
study the effect of annelation in polymeric octadecylsilica.

Representative graph of the retention factor versus inverse
temperature used to calculate the change in molar enthalpy

according to Equation 1.5. Column: polymeric
octadecylsilica. Mobile phase: methanol, 3585 psi,

0.08 cm/s. Solutes: pyrene (O), benz[a]anthracene (I),
chrysene (A), benzo[c]phenanthrene (0).

Representative graph of the retention factor versus
pressure used to calculate the change in molar volume
according to Equation 1.7. Column: polymeric
octadecylsilica. Mobile phase: methanol, 283 K,

0.08 cm/s. Symbols defined in Figure 5.1.

Representative graph of the logarithm of the rate constant
(kms) versus inverse temperature used to calculate the
activation enthalpy according to Equation 1.20. Column:
polymeric octadecylsilica. Mobile phase: methanol,

3585 psi, 0.08 cm/s. Symbols defined in Figure 5.1.

Representative graph of the rate constant (kms) versus
pressure used to calculate the activation volume according
to Equation 1.20. Column: polymeric octadecylsilica.
Mobile phase: methanol, 283 K, 0.08 cm/s. Symbols
defined in Figure 5.1.

118

125

130

139

143

147

153

157

figure 5.6:

figure 5.7:

figure 5.8:

figure 6.1:

figure 6.2:

Flgllle 6.3:

bureau;

F'Illre 6.5:

Figure 5.6:

Figure 5.7:!

Figure 5.8:

Figure 6.1:

Fi gure 6.2:

Pi gure 6.3:

Figure 6.4:

Figure 6.5:

Representative graph from the nonlinear regression of
experimental data using the exponentially modified Gaussian
equation (EMG). The points represent collected data, while

the line results from the predicted values from the EMG
equation. Solute: chrysene, 273 K, 585 psi. 161

Representative graph from the nonlinear regression of
experimental data using the non-linear chromatography
equation (NLC). The points represent collected data,

while the line results from the predicted values from the

NLC equation. Solute: chrysene, 273 K, 585 psi. 163

Representative graph from the nonlinear regression of
experimental data using the biexponentially modified

Gaussian equation (E2MG). The points represent

collected data, while the line results from the predicted

values from the E2MG equation. Solute: chrysene,

273 K, 585 psi. 168

Structure of nitrogen containing polycyclic aromatic
hydrocarbons 177

Representative chromatograms of 4-azapyrene in

A) methanol, 303 K 2085 psi, 0.08 cm/s and B) acetonitrile,

303 K, 2175 psi, 0.08 cm/s. Column: polymeric

octadecylsilica. The inset is and expansion of the methanol
chromatogram 179

Representative chromatogram of 4-azapyrene in
acetonitrile fit with the A) exponentially modified Gaussian
equation and the B) non-linear chromatography equation. 181

Representative graph of the retention factor versus inverse
temperature used to calculate the change in molar enthalpy.
Mobile phases: methanol, 2085 psi, 0.08 cm/s ( ),
acetonitrile, 2175 psi, 0.08 cm/s (- - -). Solutes:
1-aminopyrene(<>), I-azapyrene (O), 4-azapyrene (O),
benz[a]acridine (I), dibenz[a,j]acridine (A). 186

 

Representative graph of the retention factor versus

pressure used to calculate the change in molar volume.
Experimental conditions and symbols defined in

Figure 6.4. 191

xviii

Figure 6.6:

Figure 6.7: -

Figure 7.1:

Figure 7.2:

F i sure 7.3:

Figure 8.1:

Figure 8.2:

Representative graph of the rate constant (kms) versus

inverse temperature used to calculate the activation

energy. Column: polymeric octadecylsilica. Mobile

phases: methanol, 2085 psi, 0.08 cm/s ( );

acetonitrile, 2175 psi, 0.08 cm/s (- - -). Symbols defined

in Figure 6.4. 198

 

Representative graph of the rate constant (kms) versus
pressure used to calculate the activation volume.
Column: polymeric octadecylsilica. Mobile phase:
methanol, 283 K, 0.08 cm/s. Symbols defined in

Figure 6.3. 202
Representative breakthrough curve of benzo[a]pyrene
on a soil column. Mobile phase: methanol, 0.15uL/min. 226

Graph of retention factor versus carbon number for a

series of nitroalkanes as a function of methanol/water

mobile phase. (0) 100% methanol, (I) 95% methanol,

(A) 90% methanol. 229

Graph of retention factor versus carbon number for a

series of alkylbenzenes as a function of methanol/water

mobile phase. (0) 100% methanol, (I) 95% methanol,

(A) 90% methanol. 231

Fluorescence spectra of (A) pyrene and (B) fluoranthene,

which demonstrates the spectroscopic differences

between altemant and nonaltemant polycyclic aromatic
hydrocarbons 241

Chromatogram of standard PAHs (EPA 610) without (A)
and with fluorescence quenching by nitromethane

(B) and diisopropylamine (C). Column: 1.5 m ’ 200 mm id.
fused-silica capillary, packed with 5 mm Shandon

Hypersil C18. Mobile phase: methanol, 1.0 mUmin,

24 °C, with post-column addition of (A) methanol,

1.0 mL/min, (B) 2% (v/v) nitromethane in methanol,

1.0 mUmin, (C) 50% (v/v) diisopropylamine in acetonitrile,
1.0 mL/min. Laser-induced fluorescence detection:

325 nm excitation, 350-564 nm emission. Solutes:

(1) anthracene, (2) fluoranthene, (3) pyrene,

(4) benz[a]anthracene, (5) chrysene, '

(6) benzo[b]fluoranthene, (7) benzo[klfluoranthene,

(8) benzo[a]pyrene, (9) dibenz[a,h]anthracene,

(10) indeno[1,2,3-Cd]pyrene, (1 1) benzo[ghijperylene. 250

xix

figure 8.3:

figure 8.4

Figure 8.!

“glare 8.!

llgure 8.

Figure 8. 3. Chromatogram of PAHs In a coal-derived fluid (SRM 1597)

Figure 8.4:

Figure 8.5A:

Figure 8.58:

Figure 8.5C:

without (A) and with fluorescence quenching by nitromethane

(B) and diisopropylamine (C). Solutes: (12) unknown,
possibly dibenzofluoranthene or naphthofluoranthene

isomer, (13) dibenzo[def,mno]chrysene, (14) unknown,

possibly dibenzofluoranthene or naphthofluoranthene
isomer. Other experimental conditions and solutes
as described in Figure 8.2.

Chromatogram of PAHs in a contaminated soil
(CRM104-100) without (A) and with fluorescence
quenching by nitromethane (B) and diisopropylamine
(C). Other experimental conditions and solutes as
described in Figures 8.2 and 8.3

Scatter plot demonstrating three differing degrees of
product—moment correlation. (A) Coal-derived fluid
(Figure 8.3A) versus coal-derived fluid (Figure 8.3A),
r = 1.000, P = 0.00 x 10'4930

Scatter plot demonstrating three differing degrees of
product—moment correlation. (B) Contaminated soil
(Figure 8.4A) versus coal-derived fluid (Figure 8.3A),
r = 0.877, P = 2.83 x 10"13

Scatter plot demonstrating three differing degrees
of product-moment correlation. (C) Coal-derived fluid

(Figure 8.3A) versus standard (Figure 8.2A), r = 0.120,

P = 0.0241

253

257

260

261

262

Chapter 1: Introduction and Background

Over the past fifty years, analytical science has undergone dramatic
Changes as new techniques and new instrumentation have been developed.
While many of these advances have become commonplace in both industrial and
academic laboratories, fundamental studies of these systems have often lagged
behind the applications themselves. One technique that has been widely used,

but inadequately described, is liquid chromatography.

In this dissertation, the fundamental molecular processes that govern
solute retention in synthetic and natural materials are studied. These processes
are quantitated using thermodynamic and kinetic theories that describe the entire

retention event.
This Chapter presents a description of the solutes that were used
th roughout the studies, thermodynamic and kinetic theory, as well as a review of

previous investigations of reversed-phase liquid chromatography.
1 -1 Polycyclic Aromatic Hydrocarbons (PAHs)

1 -1 .1 Classification and Structure

Polycyclic aromatic hydrocarbons are a class of solutes that belong to the
lal'ger family of chemicals known as hydrophobic organic compounds. PAHs are
Comprised of two, or more, aromatic rings and are differentiated as a function of
“'19 number, annelation, and planarity. The ring number describes the number of

fuSed rings that comprise the PAH. Altemant/nonaltemant character describes

the two subclasses of PAHs. A traditional way to differentiate the altemant PAHs
from the nonaltemant is by using a labeling scheme. In this scheme, a single
carbon is chosen as the starting point and labeled. The remaining exterior
carbons are alternatively labeled, skipping an atom between the labels (Figure
1.1). Altemant PAHs possess a structure in which no two adjacent carbons are
labeled or unlabeled. By contrast, nonaltemant PAHs contain two adjacently
labeled or unlabeled carbons. Examples of altemant PAHs are anthracene and
pyrene. Examples of nonaltemant PAHs are acenaphthylene and fluoranthene.
The structures of these PAHs can be found in Figure 1.2.

One consequence of altemant/nonaltemant character is the difference in
the spectral responses of individual PAHs. All PAHs have high-energy it-bOI'IdTl'Ig
orbitals and low energy 1r*-antibonding orbitals that allow for the absorption of
visible or ultraviolet light [1]. However, the fluorescence emission of altemant
PAHs is characterized by vibrational fine structure. By contrast, nonaltemant
PAHs have very broad spectral features in their fluorescence emission (Figure
1 -3) [1,2]. The presence or lack of fine structure is determined by the
redistribution of the electrons when the PAHs are excited. As demonstrated by
Goodpaster et al., the excited states of altemant PAHs have a high degree of
Syrnmetry relative to the ground state, wherein the excitation relocates electrons
Within the 11: system [3]. By contrast, the excitation of nonaltemant PAHs results
in a redistribution of the electrons between I: and 0' bonds within the molecule

When it is excited. These spectral responses are important for the qualitative

 

Figure 1.1: Examples of A) altemant and B) nonaltemant PAH structures with

labeled atoms

Anthracene Pyrene

Acenaphthylene Fluoranthene

Figure 1.2: Structures of altemant and nonaltemant PAHs

 

Figure 1.3: Fluorescence (emission) spectra of altemant and nonaltemant PAHs

E5 xeezm4m>§>

0mm

 

 

 

 

 

 

 

 

 

0mm 0mm owm omV 05v omv 00¢ owe omm ONm
m- n $19.“. .
“m
D.
.m
a
m

 

identification of altemant and nonaltemant molecules as demonstrated in
Chapter 8.

In addition to classification based upon altemant/nonaltemant character,
PAHs can also be differentiated based upon their annelation structure.
Annelation is a descriptor of the degree of fusion between the rings of a given
PAH. Annelation is divided into three categories: linearly fused, ortho-fused, and
pen-fused. Linearly fused PAHs are Identified as having all of the rings along a
single axis. Ortho-fused PAHs are compounds in which two rings have two, and
only two, atoms in common [4]. Peri-fused PAHs are compounds in which one
ring contains two, and only two, atoms in contact with each of two or more rings
of a contiguous series of rings [4]. Examples of linearly fused, ortho-fused, and
pert-fused PAHs are shown in Figure 1.4. Since the number of isomers
increases with ring number, the ortho- and peri-fused descriptors may describe
several isomers for any given ring number.

Similar to the altemant/nonaltemant character, the type of annelation
affects the spectroscopic behavior of the PAH. Linearly fused isomers exhibit
fluorescence emissions that are shifted to longer wavelengths than ortho-fused
isomers, which are in turn shifted to longer wavelengths than peri-fused isomers

[5,6].

 

Linearly-fused

Ortho-fused

Peri-fused

Figure 1.4: Structures of linearly fused, ortho-fused and peri-fused PAHs

1.1.2 Origin and Formation

Polycyclic aromatic hydrocarbons can be formed through natural or
anthropogenic processes. In general, any process that involves the heating of
carbon-containing compounds will form PAHs [7]. Natural processes that lead to
PAH formation include volcanic activity and forest fires, as well as subsurface
events that result in the creation of fossil fuels. Anthropogenic processes that
lead to PAH formation include the production and processing of fossil fuels, as
well as the combustion of wood and petroleum products [1].

Although PAHs can originate from a variety of sources, different structures
are formed preferentially under certain conditions. For instance, nonaltemant
PAHs tend to form at lower temperatures, where an increase in reaction time
yields an increase in the number of nonaromatic rings [8]. By contrast, altemant
PAHs, as well as PAHs with a large number of rings, require long periods of high
temperature exposure to form. PAH structure is important because it affects the
chemical stability of a PAH. Blumer reports that peri-fused PAHs are more stable

than ortho-fused isomers, which are more stable than their linear isomers (Figure

1.4) [7].

1.1.3 Biological and Chemical Significance

As a class, PAHs are important for study because of their biological and
chemical significance. While single doses of PAHs have low to moderate acute
toxicity, many PAHs have been shown to elicit mutagenic and carcinogenic

events in mammalian cells (Tables 1.1 and 1.2) [9-11]. These events include

= z 8% 8:338 am. 9.. .0 $8 5.:; a $8 .. so. .

 

 

 

 

 

 

 

 

009A .90 $305. 39.8.9323
ommA .mmcoficagc. Omzos. mcmmbco
mum .mmcoatoambc. $322 39?.
80.. .90 3:22 Ocecficmcmcn.
ooom v .90 9.50.2 0:89 £c<
man .90 3:05. 9.8.9.232
a 65:95 80.. cozngEum Lo 933. 3.095 93.90;

 

 

 

 

 

mace—822.... oszEm 9.9628 360.3 ho 3.939 238 L8 mo:.m> anmw up... 03¢...

10

= F. 28:96 5% o: I <2 .

 

90m 9m 2522 ngmoa m:m:>m_rm.ONcmm

 

 

 

 

 

 

 

 

 

 

 

 

 

a: F as 3:22 82:80 mcowbco
INT: .22 096.2 032029.50 8:95
m: m5 096.). 83203830 295355.
go to 0265. 82.3102 m:mo9£:<l
N: a{z omzos. camcozmmsc 0:835:32
.mEmo 3089385
.8622 02.32.9358 :0 3 8.8% .momwmwmwmfimwé 92cm
:oz9EEECw :0 9:3".

 

m.mEEmE :. 22:823.. oszofi 2.9620: 0o 2.2585050 "Né 03¢...

11

both initiation and propagation of carcinogenesis and mutagenesis at low doses.
For example, benzo[a]pyrene has been shown to form DNA adducts resulting in
potent carcinogenesis [12,13]. In vitro, most PAHs are converted via
intermediate epoxides into phenols, diols, and tetrols. This conversion allows the
metabolized PAH to be excreted. PAHs that are known to be toxic have been
shown to covalently bind to amino acids or DNA. It is this covalent binding that
leads to toxic responses when the damaged DNA is neither repaired, nor excised
[11].

However, not all PAH isomers have analogous toxicity. For example,

- Heidelberg demonstrated that dibenz[a,h]anthracene is a potent carcinogen,
whereas the isomer dibenz[a,c]anthracene is not [14]. Since there is no a priori
method for determining the toxicity of one isomer over another, experiments must
be conducted for each isomer in order to determine the inherent toxicity.

From an analytical viewpoint, PAHs are important because of the large
number of isomers and their associated toxicity. The separation of positional
isomers presents a Challenge that has direct benefits to the biological community.
Qualitative analysis provides information that aids not only the identification of
these toxicants, but also information about the degree of exposure in real-world
samples. At the same time, an analysis of the molecular interactions of these
solutes with chromatographic materials allows for a better understanding of
retention and transport in Chromatography. The investigation of substituted

PAHs, such as azaarenes and amino-PAHs, allows quantitative differences to be

determined and compared to the parent compounds. Substituted PAHs are

12

l M
‘Iayv

8X8.”

1.2

important because they are often more toxic than the parent PAHs. For
example, 1-aminopyrene has been shown to be fifty times more toxic than
benzo[a]pyrene [15]. By studying these compounds using Chromatography, data

that are applicable to several disciplines can be garnered.

1 -2 Thermodynamic and Kinetic Theory

Given that the study of PAHs is of benefit to a multidisciplinary audience, a
fundamental framework must be employed to demonstrate not only the utility of
the studies found within this dissertation, but also the validity of the methods.
Any description of retention in Chromatography must begin with an understanding
of the basics of thermodynamics and kinetics.

In order to characterize both the thermodynamic and kinetic aspects of
retention, traditional thermodynamic and transition state theories must be
S)lhthesized. As illustrated in Figure 1.5, the retention event can be depicted by
USing an energy coordinate diagram as the molecule moves from mobile (X...) to

stationary phase (XS). The thermodynamic changes in molar enthalpy (AHsm)

and molar volume (AVsm) are simply the difference between the final and initial
States. The kinetic aspects of the retention event are characterized by a fast
ecJuilibrium between the mobile phase and transition state (X2) with equilibrium
Constant Km, followed by a rate-limiting step between the transition state and the

stationaIy phase with rate constant km. The corresponding changes in activation

13

Pbsflflbalw—c heOlW-ERVLOAUHV \Afirwufivcu no-.. libido-Iii

92.9000 cozommm

m l rll
X I.
EWI< Em><

 

 

 

Em...m_.;.g_...§...§ H... 2......

Afiieua

14

enthalpy (Al
acomplete
constructed
mmmu

and activat‘

111 Then

The
equilibrium
systems (i.
chromat0g
mblecules

egailiann

enthalpy (AHm) and activation volume (AVm) can also be calculated, allowing for
a complete description of the retention event. A similar diagram can be
constructed to illustrate the transition from stationary to mobile phase, from which

the rate constant (kms) can be used to determine the activation enthalpy (AHts)

and activation volume (Avis).

1 -2-1 Thermodynamics

Thermodynamics is the study of systems, and their properties, in
equilibrium. While the most common experiments involving equlibria are in static
systems (i.e. materials are not entering or leaving), dynamic equilibria found in
ch romatographic systems can also be studied. Under the assumption that
molecules have reached steady state by the time that they are detected, the
equilibrium constant (K) for chromatographic separations can be expressed as

K =—ai=kI3 (1.1)
am

Where am and as are the activity of the solute in the mobile and stationary phases,
respectively. It should be noted that in liquid chromatography, true equilibrium is
”BVer attained because the system is under flow. The retention factor (k) is

defined by

k g “rt—:0) (1.2)

where tr and to are the retention times of the solute and an unretained compound,

respectively. The phase ratio ([3) is defined as the volume of the stationary

15

Ibere
molar
molar
18..
8‘] su

lTTlI :

Thus,

Cain:

phase divided by the volume of the mobile phase. The equilibrium constant is

related to the change in the molar Gibbs free energy (.063...)

_ SlTl
_ .__AG 13
in K RT ( )

where R is the universal gas constant and T is the absolute temperature. The
molar Gibbs free energy can be further defined as a function of the changes in
molar enthalpy (AHsm) and molar entropy (ASS...)
AG srn = AHsm “T ASsm “-4)
By substitution,

— In . 1.5
RT T R B ( )

In R =
Th us, the change in molar enthalpy can be determined by graphing the natural
logarithm of the retention factor versus inverse temperature at constant pressure.
U hder the assumption that the changes in molar enthalpy and entropy are
temperature independent, the change in molar enthalpy can be calculated from
the slope of the line. The intercept contains information about the change in
melar entropy as well as the phase ratio. Because it is not known how the phase
ratio changes with temperature and pressure, the change in molar entropy
can not be reliably determined.

From the definition of the molar enthalpy,
AHsm = AEsm + P AVsm “-57

the retention factor can be expressed as a function of the pressure (P), the

change in molar internal energy (AEsm), and the change in molar volume (AVsm)

16

In k = RT lnB (1.7)

‘The change in molar volume can be determined by graphing the natural

logarithm of the retention factor versus pressure at constant temperature. Under
the assumption that the changes in molar volume, internal energy, and entropy
are pressure independent, the change in molar volume is calculated from the
slope of the line.

While the thermodynamic characterization of chromatography is important,

thermodynamics only provides a partial description of the retention event. Since
the rmodynamic theory is focused exclusively upon the initial and final states of

the system, the exact mechanism cannot be deduced without kinetic studies.

1 -2-2 Kinetics

While the thermodynamics of retention have been well established, less
Work has been conducted to explain the kinetics of retention. Although the
kir‘letics of retention have been studied (vide infra), the equations presented
IDelczw represent the first kinetic derivations from Giddings theoretical treatment of
retention.

As described above, transition state theory is used to evaluate the kinetics
of retention. The measured kinetic event contains two contributions: the sorption
eVent as well as the resistance to mass transfer in the mobile and stationary
FDTP-Flees. The rate constants include both of these contributions because

Ch romatography cannot decouple these events [16,17].

17

The rate constants can be calculated through an extrapolation of Giddings’
work [18]. The mass transfer term (Ck) for slow kinetics in a system that exhibits
a partition mechanism is given by

= 2R(1- R)

kms

OK (1 .8)

1+k (1'9)

and km is the pseudo-first-order rate constant for the solute transfer from
stationary to mobile phase. This mass transfer term can be related to the

variance (of) in the length domain via

GE =CI<UL (1.10)

where L is the column length and u is the linear velocity. By substitution, the

Va riance in the length domain is related to the rate constant by

2kL2
of = 2 (1.11)
(1+k) kms to

The variance in the length domain can be related to the variance in the time

domain (of) by

of =(1+k)2 2_ 2I<L2

 

o _ (1.12)
U2 L U2 kms to
From the definition of linear velocity,
u s L (1.13)
to

the variance in the time domain is related to the rate constant by

18

 

2 = 2kkto (1.14)
ms

When the time variance is meaSured in an experimental system, it intrinsically
contains all sources of asymmetrical broadening (i.e. instrumental, nonlinear
isotherms, and slow kinetics). If the contributions from the first two sources are

eliminated, as discussed in Chapter 3, then the asymmetrical tailing of the zone

contains only the contributions from slow kinetics (12) such that

12 = 6,2 (1.15)
The rate constants describing the transfer from stationary to mobile phase (km)

and from mobile to stationary phase (ksm) are defined by

 

 

2kt '
kms = 12° (1.16)
2k2to
ksm —kkms = T2 (1.17)

In order to calculate the derived kinetic parameters (AHm, AVm). a
thermodynamic approach toward transition state theory is applied [19,20]. The

I"‘élte constant can be described by the Arrhenius equation

 

-AE
ksm = Aim OXp[ FIT-tr“) (1.18)

Where A2... is the pre-exponential factor and AEm is the activation energy. The
activation energy. Which is the sum of the internal energy and a kinetic energy
ten“ [19,20], can be expanded using thermodynamic relationships

451," =AH¢m+RT—PAV¢m (1.19)

19

The natural logarithm of the resulting equation is useful for calculating the kinetic

parameters of interest

A +RT-PAV
In Ksm=ln Aim— Hm RT *m (1.20)

 

F rorTI the equations above, the activation enthalpy (AHtm) can be determined
from a graph of the natural logarithm of the rate constant versus inverse
temperature at constant pressure. Under the assumption that the activation
enthalpy and volume are temperature independent, the activation enthalpy can
be calculated from the slope of the line. The activation volume can be
determined by graphing the natural logarithm of the rate constant versus
pressure at constant temperature. Under the assumption that the activation
enthalpy and volume are pressure independent, the activation volume is
calculated from the slope of the line. By using this method, the thermodynamic
and kinetic parameters that characterize the partition and adsorption events in

I"e\rersed-phase liquid chromatography can be determined (Chapters 4-6).

1 -3 Previous Investigations

1 -3-1 Thermodynamics

In order to better characterize the chromatographic retention event, a
number of homologous series have been used to study the effect of repeating
stmctural units. Such series have included fatty acids, alkyl benzenes, and
pAHs. The benefit of using these series is that they offer a simple relationship
be“ween the retention factor (k) and the number of units, whether they be

methylene (CH2) or phenyl groups (Cst, x = 2,3,4). Early studies determined

20

that an increase in methylene unit number results in a logarithmic increase in the
retention factor. An early relationship between the number of methylene units (n)

and the retention factor (k) was published by Grushka et al. [21]

 

(1.21)
where or is the chromatographic selectivity
a = kn+1 (1.22)
kn

and B is the retention factor for the parent functional group of the homologous
series. Given that the solutes in a homologous series differ by a known number
of functional units, the selectivity is used to characterize the difference in the
Gibbs free energy (AAG) between solutes n and n+1. Using the thermodynamic
relationships above, the differences in the molar enthalpy (AAH), molar entropy
(MS), and molar volume (AAV) as a function of methylene number can be

examined.

One of the most pivotal studies using homologous series was published
by Berendsen and de Galan [22]. In this work, the effect of stationary phase
Chain length was studied for a number of solutes with varying methylene unit
n umber. The data from the study indicated that there is a critical chain length,
beyond which retention does not Change. According to the authors, the limiting
c“Plain length lies between 6 and 14 carbons for the alkyl-silica bonded phase
material. This critical chain length is independent of mobile phase composition,

but increases with increasing length of the solute alkyl chain, as well as

'ncreasing retention factor.

21

 

Further studies probed this idea of critical chain length in order to
elucidate the retention mechanism. Colin and Guiochon studied a number of

stationary phases for the separation of n-alkanes, n-alkyl benzenes, and n-
methyl fatty acid esters [23]. Their work concluded that carbon adsorbents were
more selective than chemically bonded phases (i.e. Ca and C13). In addition, the
authors concluded that traditional hydrophobic predictors (i.e. water solubility,
octanol-water partition coefficient) of retention failed to accurately estimate the
selectivity of positional isomers on bonded phases. Similarly, Tchapla et al.
investigated seven series of alkyl-substituted solutes (e.g. alkylbenzenes,
alkylchlorides, carboxylic acids) on different stationary phases and concluded
that the solute alkyl Chains intercalate into the alkyl-silica stationary phase [24].

l n addition to supporting the conclusions by Berendsen and de Galan, the
authors demonstrated that the critical chain length is independent of the
h Omologous series studied (i.e. it is dependent upon the number of methylene
u n its in the solute).

In addition to studying the retention, the molar enthalpy and molar entropy

h ave also been studied for homologous series. lssaq and Jaroniec calculated
th e molar enthalpy using Equation 1.5 for alkylbenzenes using binary mobile
Db ases and four stationary phases (C1, C4, C3, Cm) [25]. Using enthalpy-entropy
Q0 rnpensation theory, they conclude that the molar enthalpy and molar entropy
tiepend linearly on the number of carbon atoms (i.e. methylene units) in the alkyl
ch ain of the homologous series, not the bonded phase. Sentell and Henderson

f” l‘ther studied the effect of subambient temperatures on homologous series [26].

22

31'

r"

AM

s?

)

Their work concluded that at subambient temperatures, the structure of the
stationary phase has a large effect on the retention, and is preferential for certain
homologous series. McGuffin and Chen studied the retention of a series of even-

numbered fatty acids (010 through C22) not only as function of temperature, but
also pressure. From their experiments they quantitated the enthalpy Changes
(AAH) and volume changes (AAV) of —3.5 kcal/mol and ~14.1 mUmol,
respectively [27], for ethylene units. Thus, these studies demonstrate that a
significant amount of effort has been invested into understanding the
ch romatographic retention event.
In addition to the studies of retention as a function of methylene unit
n umber, series containing phenyl units have also been investigated. Most
notably PAHs have been used since they have repeating phenyl units in varied
Configurations. Whereas alkylated solutes can be characterized as a function of
th 9 number of methylene units and the degree of branching, PAHs have been
studied as a function of ring number, annelation and planarity [28-31]. Studies
h ave shown that an increase in the ring number leads to a logarithmic increase in
the retention factor. Similarly, linear PAHs are retained longer than ortho-fused
is'Dlners, which are more retained than peri-fused isomers. Several reviews have
ID‘s-ten published in an attempt to collate the large amount of information that is
R“ own about PAH retention [32-35]. For example, an increase in the density of
3' I'<)II Chains in the stationary phase leads to an increase In the retention of planar
co mpounds, but a decrease in the retention of nonplanar compounds

[28 ,29,32,36-38]. The current hypothesis is that an increase in density causes

23

nonplanar solutes to be excluded more efficiently. Increases in temperature

result in a decrease in the retention for both planar and nonplanar solutes
[28.39.40]. By contrast, the effect of pressure is much smaller [27,41]. As a
result, little work has been conducted using this parameter.

From these experiments, many authors have attempted to explain the
mechanism of retention (i.e. partition or adsorption) on the basis of the
thermodynamically derived value of molar enthalpy and the general shape of the
van’t Hoff plot. A van’t Hoff plot is simply a graph of the logarithm of the retention
factor versus inverse temperature (Equation 1.5). However, such experiments
have failed to stringently control pressure. As demonstrated in Chapters 4
through 6, failure to control pressure may result in erroneous values to be
calculated for the change in molar enthalpy. In addition, the change in molar

volume cannot be calculated without controlling pressure.

1 -32 Kinetics

Although the thermodynamics of retention have been extensively probed,
'ess work has been conducted into the kinetics of retention. Published work on
RE hetics has generally fallen into two classes based upon the techniques
e 'T‘Iployed: non-chromatographic and chromatographic.

Non-chromatographic experiments have utilized pressure, temperature,
aJ‘sd dipole jump methods [42-47], as well as fluorescence correlation
sli>ectroscopy [48,49] to investigate Chromatographic materials. Jump
e)<periments involve systems in which an abrupt perturbation is made to the

s)Vstem, after which the kinetic relaxation is measured. For example, pressure

24

jump methods involve two cells, which are connected to one another, but
separated by a rupture disk. One cell contains the solute and the stationary
phase, the other a compressible liquid. Once the sample cell has reached
equilibrium, the second cell is pressurized until the disk ruptures, sending a
pressure wave into the sample cell. The solutes are then studied as they
kinetically relax from this perturbation wave. While this method provided
information about the kinetics of interaction with alkyl bonded phases, the

- systems are not representative of a chromatographic column. The inherent

limitation in jump methods is that the perturbation changes the thermodynamic

and kinetic behavior of the system that is being measured. Thus, the kinetic

measurements are representative only of the jump system.
In contrast, total-internal-reflection fluorescence correlation spectroscopy

(T I FIFCS) is more suited to measuring a flowing system [48,49]. Using TIRFCS
H ansen and Harris modified a silica surface with octadecylsilica. Fluorescent
'71 olecules were flowed across the modified surface. As the molecules diffuse
toVvard the surface, and into the region that is spectroscopically probed, they add
to the low frequency noise of the system. This noise follows Poisson statistics
3“ <5 as such, allows for the number of molecules at the interface to be calculated.
Th ese data provide insight into the kinetics of the retention event. However, like
the jump methods, this system provides only limited information about a dynamic
S3'stern. The principle limitation of this method is that a highly fluorescent
first) lecule must be used. For their work, Hansen and Harris chose rhodamine 66,

a dye commonly used for lasing. This molecule is very different from both the

25

methylene and phenyl homologues and provides limited information about their
retention.

As noted, non-chromatographic techniques provide a method by which the
kinetics of retention can be probed. However, the inherent differences between
these systems and a packed chromatography bed provide limited information
about the kinetics of an actual separation. In order to circumvent these
difficulties, several chromatographic techniques have been used. These
methods have included frontal analysis and pulse methods [SO-52], with affinity
ch romatography being one of the more common applications [53,54]. Frontal
analysis involves injecting a large concentration of the solute onto the
chromatography column, and then analyzing the resultant curve. Also known as
a breakthrough curve, this curve provides information about the kinetics of
sorption. However, it does not represent the most common type of

Oh romatography, which involves injecting smaller volumes of material onto the

Go I umn.

Pulse methods better approximate a typical chromatographic separation
S i h ce they involve the injection of smaller volumes/concentrations. However, the
C: L.- rtent method to elucidate the kinetics is also questionable. Most notably,
M i.3’abe and Guiochon use a method in which plate height (H) can be related to
a)“ El dispersion, external mass transfer, intraparticle diffusivity and the
a'3'sorption rate constant. Since the plate height is a measurable quantity, and
the first three parameters can be approximated, the adsorption rate constant can

be quantitated [50,55-58]. However, this method is troublesome since there are

26

many processes that contribute to the broadening and asymmetry of the
chromatographic zone profile that do not arise from kinetic sources. As

d iscussed by Weiss [16] and Lenhoff [17], failure to accurately measure or
estimate these sources can result in spurious kinetic data. Thus, the current
ch romatographic methods are not ideal for characterizing the kinetics of retention

since these methods rely upon a number of assumptions and approximations.

1 -4 Conclusions
Although significant efforts have been invested in understanding the

molecular contributions to retention for homologous series, limitations in both

theory and experimental design have hindered definitive quantitative analysis.
Using novel theoretical and experimental designs, the thermodynamic and kinetic
contributions to retention in liquid chromatography will be explored. By
Characterizing the thermodynamics and kinetics in tandem, and by measuring the
zone profile on column versus post column, a more rigorous description of solute
transfer is developed. Ultimately, such development will lead to a better

I..- " clerstanding of retention in both synthetic and natural materials.

27

1 .5 References

I 1 1
[2]
[3]
[4]
[5]

[6]

[8]

[9]
[1 Q]
[1 1]

[12]

ML. Lee, M.V. Novotny, K.D. Bartle, Analytical Chemistry of Polycyclic
Aromatic Hydrocarbons, Academic Press, New York, NY, USA, 1981.

J. Malkin, Photophysical and Photochemical Properties of Aromatic
Compounds, CRC Press, Ann Arbor, MI, USA, 1992.

J.V. Goodpaster, J.F. Harrison, V.L. McGuffin, J. Phys. Chem. A 102
(1998) 3372-3381 .

A.D. McNaught, A. Wilkinson, IUPAC Compendium on Chemical
Terminology, Royal Society of Chemistry, Cambridge, UK, 1997.

E. Clar, Polycyclic Hydrocarbons, Academic Press, New York, NY, USA,
1 964.

E. Clar, The Aromatic Sextet, John Wiley & Sons, New York, NY, USA,
1 972.

M. Blumer, W.W. Youngblood, Science (Washington, D. C., 1883-) 188
(1975) 53-55.

M. Blumer, Sci. Am. 234 (1976) 35-45.

M. Cooke, A.J. Dennis, G.L. Fisher, Polynuclear Aromatic Hydrocarbons:
Physical and Biological Chemistry, Battelle Press, Columbus, OH, 1982.

T. Vo-Dinh, Chemical Analysis of Polycyclic Aromatic Hydrocarbons, John
Wiley & Sons, New York, NY, 1988.

RF. Hertel, G. Rosner, J. Kielhom, Selected Non-Heterocyclic Polycyclic
Aromatic Hydrocarbons, World Health Organization, Geneva, 1998.

K.M. Straub, T. Meehan, A.L. Burlingame, M. Calvin, Proc. Natl. Acad.
Sci. U. S. A. 74 (1977) 5285-5289.

28

[13]

[14]

[15]

[1 6]
[1 7]
[1 8]

[ 1 9]
[20]

[2 1]
[22]
[2 a]

[24]

[2 5]

[a 6]

[27]

M.F. Denissenko, M.-s. Tang, G.P. Pfeifer, Biomarkers Environ. Assoc.
Dis. (2002) 139-157.

C. Heidelberg, Cancer Res. 30 (1970) 1549-1569.

C.H. Ho, B.R. Clark, M.R. Guerin, B.D. Barkenbus, T.K. Rao, J.L. Epler,
Mutat. Res. 85 (1981) 335-345.

G.H. Weiss, Sep. Sci. Technol. 16 (1981) 75-80.

A.M. Lenhoff, J. Chromatogr. 384 (1987) 285-299.

J.C. Giddings, J. Chem. Phys. 31 (1959) 1462-1467.

K.J. Laidler, Chemical Kinetics, 3rd Ed., Harper & Row, New York, NY,
USA, 1987.

J.l. Steinfeld, J.S. Francisco, W.L. Hase, Chemical Kinetics and
Dynamics, 2nd Ed., Prentice Hall, Upper Saddle River, NJ, USA, 1999.

E. Grushka, H. Colin, G. Guiochon, J. Chromatogr. 248 (1982) 325-339.

G.E. Berendsen, L. de Galan, J. Chromatogr. 196 (1980) 21-37.

H. Colin, G. Guiochon, J. Chromatogr. Sci. 18 (1980) 54-63.

A. Tchapla, S. Heron, H. Colin, G. Guiochon, Anal. Chem. 60 (1988)
1443-1448.

H.J. Issaq, M. Jaroniec, J. Liq. Chromatogr. 12 (1989) 2067-2082.

K.B. Sentell, M. Ryan, A.N. Henderson, Anal. Chim. Acta 307 (1995)
203-215.

V.L. McGuffin, S.H. Chen, J. Chromatogr. A 762 (1997) 35-46.

29

[28]
[29]
[30]
[3 1 ]
[32]

[33]
[34]

[35]
[36]
[37]

[38]

[39]
[4°]
[4 ‘1 ]

L.A. Cole, J.G. Dorsey, Anal. Chem. 64 (1992) 1317-1323.

K.B. Sentell, J.G. Dorsey, J. Chromatogr. 461 (1989) 193-207.

M. Olsson, L.C. Sander, S.A. Wise, J. Chromatogr. 477 (1989) 277-290.

L.C. Sander, M. Pursch, S.A. Wise, Anal. Chem. 71 (1999) 4821-4830.

L.C. Sander, S.A. Wise, Anal. Chem. 67 (1995) 3284-3292.

J.F. Wheeler, T.L. Beck, S.J. Klatte, L.A. Cole, J.G. Dorsey, J.
Chromatogr. A 656 (1993) 317-333.

A. Tchapla, S. Heron, E. Lesellier, H. Colin, J. Chromatogr. A 656 (1993)
81-1 12.

LC. Sander, S.A. Wise, J. Chromatogr. A 656 (1993) 335-351.

K.B. Sentell, J.G. Dorsey, Anal. Chem. 61 (1989) 930-934.

S.R. Cole, J.G. Dorsey, J. Chromatogr. A 635 (1993) 177-186.

J..H Park, Y..K Lee, Y.C. Weon, L..C Tan, J. Li, L. Li, J..F Evans, PW.
Carr, J. Chromatogr. A 767 (1997) 1- 10.

LA. Cole, J.G. Dorsey, Anal. Chem. 64 (1992) 1324-1327.

M.H. Chen, C. Horvath, J. Chromatogr. A 788 (1997) 51-61.

V.L. McGuffin, C. Lee, J. Chromatogr. A 987 (2003) 3-15.

D.B. Marshall, J.W. Burns, D.E. Connolly, J. Chromatogr. 360 (1986) 13-
24.

30

[‘43]

[‘44]
[451
[461
[471
[4 81
[4 91
[501

[5 ‘1]

[52]
[53]

[54]

[55]
[56]

[S7]

[58]

SW. Waite, J.M. Harris, E.H. Ellusib, D.B. Marshall, Anal. Chem. 63
(1991) 2365-2370.

S.W. Waite, D.B. Marshall, J.M. Harris, Anal. Chem. 66 (1994) 2052-2061.
F.Y. Ren, J.M. Harris, Anal. Chem. 68 (1996) 1651 -1657.

J.M. Harris, 03. Marshall, J. Microcolumn Sep. 9 (1997) 185-191.
S.R. Shield, J.M. Harris, Anal. Chem. 74 (2002) 2248-2256.

R.L. Hansen, J.M. Harris, Anal. Chem. 70 (1998) 2565-2575.

R.L. Hansen, J.M. Harris, Anal. Chem. 70 (1998) 4247-4256.

K. Miyabe, G. Guiochon, Anal. Chem. 72 (2000) 5162-5171.

K. Miyabe, S. Sotoura, G. Guiochon, J. Chromatogr. A 919 (2001) 231-
244.

K. Miyabe, G. Guiochon, Anal. Chem. 74 (2002) 5754-5765.
F. H. Arnold, H.W. Blanch, J. Chromatogr. 255 (1986) 13-27.

P.D. Munro, D.J. Winzor, J.R. Cann, J. Chromatogr. A 659 (1994) 267-
273.

K. Miyabe, G. Guiochon, J. Chromatogr. A 857 (1999) 69-87.
K. Miyabe, G. Guiochon, J. Chromatogr. A 830 (1999) 263-274.
K. Miyabe, G. Guiochon, Anal. Chem. 72 (2000) 1475-1489.

K. Miyabe, G. Guiochon, J. Chromatogr. A 961 (2002) 23-33.

31

Chapter 2: Experimental Methods

2.1 Introduction

In order to study the molecular level contributions to retention, a series of
i mstruments was necessary. Although each experiment was customized for a
specific measurement, a basic design was used and modified as needed. The
manufacture of the chromatographic columns, the common components, as well

as the specific modifications for each experiment are detailed below. In addition,

the data treatment procedure is also described.
2 .2 Experimental Systems

2-2.1 Column Preparation

In order for the thermodynamics and kinetics of retention to be accurately

q uantitated, the manufacture of a capillary column is necessary. This section

details the choice of materials for the stationary phases as well as the methods

employed to pack the columns.
In all instances, optically transparent, fused-silica capillary (Polymicro

I echnologies) is used for the column. The inner diameters of the columns were

200 pm for the synthetic stationary phases, and 320 pm for the natural material.

' hese sizes are chosen such that the number of particles across the column is

between 2 and 40. The number of particles across the column is important

because a larger number of particles across the column results in a minimization

of wall effects, which contribute to the asymmetrical tailing (vide infra). The

32

columns are terminated by quartz wool frits that prevent seepage of the

stationary phases from the packed column.

2.2.1.1 Stationary Phases

2.2.1 .1 .1 Synthetic Materials

Two stationary phases, monomeric and polymeric octadecylsilica (ODS),
were chosen to investigate the effects of bonding density on retention (Chapters
4-6). Both phases are prepared using an irregular silica material with 5.5 pm
particle size, 190 A pore size, and 240 m2/g surface area (lMPAQ 200, PO
Corp). Using mono- and trichlorosilanes, the silanization of these particles
p roceeds in the presence of trace water on the silica support. The presence of
water causes a covalent bond to form between the silanol groups on the surface,
and the silane functionality of the bonded phase. The reactions result in

m onmeric and polymeric stationary phases with bonding densities of 2.7 and 5.4

1.; moi/m2, respectively.

2 - 2.1 .1 .2 Natural Material

In addition to the synthetic materials, a natural material is also used as a

stationary phase. This material was chosen to represent natural soils in an
attempt to better characterize retention in the environment (Chapter 7). The
natural material used is a reference soil (CLN Soil-3 RTC Corp.). This soil
contains 5.96% organic matter by mass and is 65% sand, 25% silt, and 10%
Clay. Physically this material has a large particle size distribution that ranges

from the sub-micrometer to the millimeter scale. Since this range is too large to

33

pack efficiently, the material was sieved using a series of nylon meshes to yield a
size range from 5 to 125 pm. This fraction was tested via batch isotherms to
i nsure that the fractionation did not significantly alter the chemical behavior of the

rnaterial (Chapter 7).

2.2.1.2 Packing Method

In order to pack the stationary phases, the slurry method is used [1]. This
method involves suspension of the stationary phase in a solvent that prevents
i rnmediate precipitation. For the synthetic materials, acetone is chosen. For the
r1 atural material, methanol and ethylene glycol are used. Once suspended, the
material is transferred to a stainless steel reservoir that is connected in series
with a single piston reciprocating pump and a fused-silica capillary. The reservoir
is pressurized rapidly, thus injecting the stationary phase into the open capillary.
Depending upon the packing material, a different pressure limit is used. For the
synthetic phases, the reservoir is pressurized to 5000 psi. This high pressure
packs the column in a matter of minutes. By contrast, the natural phase is
pressurized to 1000 psi, and slowly increased to 4000 psi over the course of
Several hours in order to create a well-packed bed of particles.
Once filled to the desired length, the column is then slowly depressurized
to 1000 psi and maintained at that pressure for 24 hours. This lower pressure
allows the packed bed to compress, thus removing any voids that could be left

(3 u ring the initial packing step.

34

2.2.2 Chromatographic System
Each experiment is carried out using one of three liquid chromatography

systems. The common components of each include a single piston reciprocating
pump (Model 114M, Beckman Instruments) operating in constant pressure mode

(:15 psi) over the range of 500 to 4000 psi. The pump allows for the mobile

phase to be supplied continuously without the need for depressurization during

the refill cycle.
Samples are introduced via an injection valve (Model ECI4W1, Valco

I nstruments) with a 1 pL internal loop. Once the solutes are introduced, they are
pumped to a 50 pm i.d. fused-silica capillary (Polymicro Technologies) that
serves as the splitter. The length of the splitter is varied to change the amount of
sample that is introduced into the column so as to prevent an overload of the

stationary phase. Under normal operating conditions the volume introduced into

the column is between 12 and 50 nL.
Once the sample is split, it is carried by the mobile phase to the column.

At the end of the column, a 20 um i.d. fused-silica capillary (Polymicro
l echnologies) is attached to serve as a restrictor. The length of the restrictor is

reduced as the inlet pressure is decreased in order to maintain a constant

pressure drop along the column of (~10 psi/cm).
The injector, splitter, column, and restrictor are all housed within a

Cryogenic oven (Model 3300, Varian Associates) that maintains a constant

temperature (1 0.1 K) through the use of liquid nitrogen aspiration and resistive

heating coils.

35

 

2.2.3 Spectroscopic System

Three different types of detectors were chosen for detection. Given that
PAHs have absorption maxima in the ultraviolet portion of the electromagnetic

spectrum, the systems chosen all used excitation sources in this range. The

three systems are described in detail below.

2.2.3.1 Ultraviolet-Visible Absorbance

The default spectroscopic technique used for detection is ultraviolet-visible
absorbance. As shown in Figure 2.1, a commercially available unit (Model 2050,
Varian) is directly coupled to the capillary column such that the effluent is
d irected to a capillary flow cell. This flow cell is optically transparent, fused-silica
capillary with a 50-pm i.d. A capillary flow cell is used to minimize zone

I: roadening that arises from diffusional and mass transfer processes.

The wavelength for detection is chosen based upon the absorption
m axima of the solutes. The time constant for the detector is 0.05 seconds. The

o utput from the detector is directed to a chart recorder for further data analysis.

36

Figure 2.1: Generalized schematic diagram for the post-column detection of
chromophores/fluorophores using both ultraviolet-visible absorbance and laser-
induced fluorescence detection. I: injection valve, S: splitting tee, R: restrictor
PMT: photomultiplier tube, AMP: current to voltage amplifier, UV-Vis: ultraviolet-

visible absorbance spectrometer.

37

 

@H. m_>->:

 

 

Laser

 

 

 

 

 

 

 

 

 

 

 

 

.mEoomm

 

 

 

 

J.Wd n4" +-----------

 

 

 

 

 

:E:_oO

 

 

 

 

n=>_<

 

 

 

 

 

smSano

 

 

 

 

Pd 930E

 

 

QEzm

 

 

38

2.2.3.2 Laser-Induced Fluorescence
The second method employed for the detection of fluorophores is laser-
induced fluorescence spectroscopy. As illustrated in Figure 2.2, this
experimental system is used for the on column detection of PAHs (Chapters 4-6).
The design and validation of this system were conducted by Dr. Chen [2]. A
continuous-wave HeCd laser (Model 3074-20M, Melles Griot) is used as the
excitation source. The laser beam is carried via UV-grade optical fibers (100 um,
Polymicro Technologies) to four locations along the column where the polyimide
coating has been removed. The intensities at each location are balanced using
fiber optic positioners and a fluorophore at the inception of each experiment. At
each location, the fluorescence is collected orthogonal to the incident beam by
optical fibers (500 um, Polymicro Technologies) and is transmitted to a
photomultiplier tube (Model R760, Hamamatsu), which has a 420 nm
i nterference filter (S10-420-F, Corion). The resulting photocurrent is amplified
using a gain of 100 nA/V and a time constant of 10 ms. The signal is then
converted to the digital domain (PCl-MlO-16XE-50, National Instruments) and
stored by a user-defined program (Labview v5.1, National Instruments). The

acquisition rate of the program can be manually adjusted, and is typically set

between 1 and 4 Hz.

39

Figure 2.2: Schematic diagram of the experimental system for capillary liquid
chromatography with on-column laser-induced fluorescence detection. l:
injection valve, S: splitting tee, R: restrictor, FOP: fiber-optic positioner, F: filter,

PMT: photomultiplier tube, AMP: current to voltage amplifier.

40

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

QEzm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

n5“. mm.)
L\\ \ ixx /‘ / Ix‘lr
@ 5:200
n. Hm nMU W H<U
d .0 IO :6
w w_ T w_
I. I. I. I.
, g : .mSQEoO
_n..2< _n._>_< n.2,; n_.>.< .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mm 2:9”.

41

22.3.3 Fl.

Th
shown in
through a
from lrivi.
fluoresce
then iilte
entrance
grating,
a Chargr
was).
9.15 hr.
wmme
and the
The ac
“5‘0 r
Is Use:
”from

l!‘
'\Q
h‘vd:

2.2.3.3 Fluorescence Quenching

The third detection system employs selective florescence quenching. As
shown in Figure 2.3, a HeCd laser is used to excite the solutes as they pass
through a 75 um i.d. flow cell. This capillary flow cell minimizes the interference
from trivial processes such as primary or secondary absorption [3]. The
fluorescence emission is collected orthogonal to the incident radiation, and is
then filtered and collimated. The resulting emission is then refocused onto the
entrance slit of a Czemy-Tumer monochromator (Model 340E, 300 groove/mm
grating, Instruments SA, 10 nm/mm reciprocal linear dispersion) and detected via
a charge-coupled device.(ModeI (A)TECCD-2000x800-7, Instruments SA, 15 pm
pixels). This system provides a wavelength range of 300 nm and a resolution of
0.15 nm. Instrument control and data acquisition are achieved by using a
commercially available electronic interface (Instruments SA, Model CCD 2000)
and the associated software (Spectramax for Windows, v3.1, Instruments SA).
The acquisition rate for this system was set to 2 s, with the integration time of
~100 ms. This system differs from the preceding designs in that a second pump
is used to deliver a fluorescence quencher to the post column effluent. Using
nitromethane, altemant PAHs are selectively quenched. Diisopropylamine is
used as a post column addition to quench nonaltemant PAHs. This slight
modification allows for qualitative information to be collected about complex

samples (Chapter 8).

42

 

Figure 2.3: Schematic diagram of the experimental system for capillary liquid
chromatography with laser-induced fluorescence quenching detection. I:
injection valve, S: splitting tee, L: lens, F: fflter; CCD: charge-coupled device,
PMT: photomultiplier tube. (Note: Collection of fluorescence emission is

orthogonal to the incident laser beam)

43

 

firnnnj

 

Figure 2.3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:EEoO.

 

 

 

 

Pump 1

 

 

 

 

 

 

2
p
.. m
w P
a
IL
... ....n r IIHI I I.
BEES coco
.. .2 En.
00.0

 

 

 

 

 

 

 

.mSano

 

 

em 9:9”. .

 

 

 

44

 

2.3 D

wkd

desirei

mmmz
enad
mmow
manua
using a
&mefl
second
analysis
hncoon

flame

2.3 Data Treatment and Analysis

While the stringent control of each experimental system was critical for the
collection of the data, the analysis also provides a series of obstacles. Since the
desired data are individual peak profiles contained within a multicomponent
chromatogram, the first challenge for the data analysis is the reproducible
extraction of the indiViduaI profiles. Since the profiles must be manually
removed, a series of experiments were carried out to explore the effects of this
manual extraction (Chapter 3). Once removed, the peak profiles are iteratively fit
using a commercially available program (Peakfit v3.18, SYSTAT Software).
Since there are many functions available for the analysis of peak profiles, the
second obstacle to the analysis is choice of an appropriate function for the
analysis of the peak profiles. This section describes some of the mathematical
functions that are usedonce the data have been extracted. Using the various
parameters from the mathematical models described below, the
thermodynamics and kinetics are quantitated using the equations found in

Chapter 1.
2.3.1 Mathematical Functions

2.3.1.1 Statistical Moments

One of the most common methods to characterize peaks utilizes statistical
moments [4-6]. Statistical moments are classical functions that are used to
describe the distribution of any set of data with no assumptions about the

functional form. For the purposes ofthis dissertation, the moments related to the

45

CODCGI

by

The sec
[Ehrese
Statistic

discrete

concentration as a function of time C(t). The zeroth moment, or area, is defined
by

Mo = [omen z 20mm (2.1)
the first moment, or mean retention time, by

[100) dt 2tC(t) At
M1=——=———

2.2
Mo Mo ( )
and the higher central moments (Mn) by the general form
t- "Ctdt t- "CtAt
Mn___]< M1) () zgr M1) () (2.3)

 

 

M0 M0

The second moment (M2) represents the variance and the third moment (M3)
represents the asymmetry of the zone profile. In practical applications of the
statistical moments, the integrals in Equations 2.1 — 2.3 are usually estimated by
discrete summation.

The accurate description of asymmetrical peaks usually requires a large
number of statistical moments and a large number of data points [7]. In addition,
noise can affect the values for the moments [8]. As a result of these limitations,
alternative equations such as the exponentially modified Gaussian (EMG) [9], the
nonlinear chromatography (NLC) [10], and the Haarhoff-Van der Linde [11]
function have been developed for evaluating non-Gaussian zone profiles. The
EMG and NLC models provide variables that'can be directly related to
experimental parameters. As a result, these two functions were initially chosen to

study chromatographic zones.

46

235L2

form 0

when?
standa

compo

d ' -
Mlilor

2.3.1.2 Exponentially Modified Gaussian Equation

The EMG is a convolution of a Gaussian and an exponential function. The

form of this equation is

 

 

2 _ _
C(t)=.% expL:2 + tGT t][efi[tJ§t: — JET]+1] (2.4)

where A is the area, tG is the retention time of the Gaussian component, 0 is the
standard deviation of the Gaussian component and T is the the exponential
component.

Over the past thirty years, several authors have sought to characterize the
use of the EMG function for chromatographic analysis. While Grushka [12,13]
and Yau [14] presented some early work, Foley and Dorsey have published two
seminal papers reviewing the EMG function in the intervening years [15,16].
These reviews have demonstrated the use of the EMG function for a variety of
chromatographic applications. More recent studies have compared the EMG to
other equations for the analysis of zone profiles [17,18]. These studies have
concluded that high-order polynomials that exceed the four parameters used in
the EMG function can describe asymmetrical peaks very well. However, the
complex nature of such functions makes them impractical for common use. In
addition, the parameters for these high-order polynomials may not have any
physical significance as do those calculated using the EMG function.

Within this dissertation, the EMG is employed predominantly to calculate

the thermodynamics and kinetics of retention. The thermodynamic and kinetic

47

pararr

into E1

2.3.1.1

chrom

CURE“

fl.-
'**”9 ll

parameters can be calculated by substituting the measured values of t3 and 1
into Equations 1.2, 1.16, and 1.17, where

2.3.1 .3 Non-linear Chromatography

The second model employed to analyze the zone profiles is the nonlinear
chromatography (NLC) model. Developed initially in 1944 [19], the NLC’s most

current form was developed by Wade et al. in 1987 [10]:

 

cro- a° I-exp a3fL1EJ—x]m{—x 61] (2.6)
l [ fl [:2 iii expi- Ezii

T(u,v) = e-Vje—‘IOUEWM (2.7)

 

 

where

and lo and I1 are modified Bessel functions of the first kind. The fitting
parameters that are incorporated in the NLC model are the area (a0), position
(a), width (a2), and distortion (as). If the chromatographic data are transformed
from the time domain to the retention factor domain via Equation 1.2, then these

parameters can be used to calculate the lumped desorption rate constant

 

(2.8)

Using the established relationship between the retention factor and the rate

constants (k = ksm/km), the adsorption rate constant can also be calculated.

48

131:4

l
stereo
umIfls
expone

Gauss

2.3.1.4 Bi-exponentially Modified Gaussian Equation

. ‘ Neither the EMG nor NLC equations are capable of evaluating a multiple-
site retention model. As a result, a new equation was developed to test whether
two distinct sites were present at low temperatures (Chapter 6). The bi-
exponentially modified Gaussian (EZMG) equation is the convolution of a

Gaussian and two exponential functions, with the resulting form

2
hec‘g ex1p[:262 (t -—:t———G) -—————]
X

 

C(t)- -
f

202 (2.9)

f 2)
"t'l'tG'l': 0,2
exp 1 [1+erf[tt1 -T1tG - ]]

 

 

2120'

K 2 2)
o
[-t+tG+':c—] 02
-—exp 2 [1+erf[t12 '4th - D

 

 

 

 

\ K J 1
where h is the height, tG is the retention time of the Gaussian component, a is the
standard deviation of the Gaussian component, and I1 and 12 are the exponential
components for the first and second kinetic sites, respectively. This equation,

originally developed by Delley [20] presumes that the two kinetic events are of

equal probability.

49

2.4 Conclusions

The use of novel experimental design and theoretical treatment is
necessary in order to study retention at the molecular level. This chapter details
the experimental systems and mathematical functions that were necessary to

collect and analyze the desired data.

50

2.5 References

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

J.C. Gluckman, A. Hirose, V.L. McGuffin, M. Novotny, Chromatographia
17 (1983) 303-309.

S.-H. Chen, in Chemistry, Michigan State University, East Lansing, 1993.

S.-H. Chen, C.E. Evans, V.L. McGuffin, Anal. Chim. Acta 246 (1991) 65-
74.

E. Grushka, M.N. Myers, J.C. Giddings, Anal. Chem. 42 (1970) 21-26.

J.C. Giddings, RA. Keller, Advances in Chromatography, Marcel Dekker,
New York, NY, USA, 1966.

MS. Jeansonne, J.P. Foley, J. Chromatogr. 461 (1989) 149-163.

S.N. Chesler, SP. Cram, Anal. Chem. 43 (1971) 1922-1929.

J.V.H. Schudel, G. Guiochon, J.Chromatogr. 457 (1988) 1-12.

W.E. Barber, P.W. Carr, Anal. Chem. 53 (1981) 1939-1942.

J.L. Wade, A.F. Bergold, P.W. Carr, Anal. Chem. 59 (1987) 1286-1295.

PC. Haarhoff, H.J. Van der Linde, Anal. Chem. 38 (1966) 573-582.

E. Grushka, Anal. Chem. 44 (1972) 1733-1738.

E. Grushka, J. Phys. Chem. 76 (1972) 2586-2593.

W.W. Yau, Anal. Chem. 49 (1977) 395-398.

J.P. Foley, J.G. Dorsey, J. Chromatogr. Sci. 22 (1984) 40-46.

51

[16]

[171‘

[18]

[19]

[20]

MS. Jeansonne, J.P. Foley, J. Chromatogr. Sci. 29 (1991) 258-266.
M.L. Phillips, R.L. White, J. Chromatogr. Sci. (1997) 75-81.

J.R. Torres-Lapasio, J.J. Baeza—Baeza, M.C. Garcia-Alvarez-Coque, Anal.
Chem. 69 (1997) 3822-3831.

H.C. Thomas, J. Am. Chem. Soc. 66 (1944) 1664-1665.

H. Delley, Anal. Chem. 58 (1986) 2344-2346.

52

Chapter 3: Mathematical Analysis in Chromatographic
Systems: A Comparison of the Exponentially Modified
Gaussian Equation and Statistical Moments

3.1 Introduction

The reproducible extraction of individual zone profiles from
multicomponent chromatograms is imperative if the thermodynamic and kinetic
values are to have any meaning. This chapter describes the sources (of both
symmetrical and asymmetrical broadening that contribute to the peak shape of
individual solutes. In addition, data from simulations and experiments are
presented in order to validate the use ofathe exponentially modified Gaussian

(EMG) equation as well as the experimental methods used to collect the data.

3.1.1 Sources of Broadening

In liquid chromatography, there are two main classes of broadening:
symmetrical and asymmetrical. Symmetrical zone broadening arises from
processes that are fast on the time-scale of the separation, such as diffusion and
resistance to mass transfer in the mobile and stationary phases. These
phenomena have been well characterized in the past [1]. Asymmetrical
broadening arises from instrumental effects, nonlinear isotherms, and kinetics
that are slow on the time-scale of the separation. These processes result in zone
profiles that are skewed either to the front, or more commonly, to the rear.

Contributions to the asymmetrical broadening from instrumental effects
have a variety of sources. The first source arises from volumetric contributions,

such as exponential dilution chambers or regions of unswept fluid that occur in

53

the in
contr‘

proce

resul‘

the injector, detector, or connectors. The second source arises from electronic
contributions that occur during detection, amplification, or other signal
processing.

The other two contributions to asymmetrical zone broadening are the
result of interactions between the solute and stationary phase. Nonlinear
isotherms typically occur when the concentration of the solute exceeds the
capacity of the stationary phase, or when there are multiple sites where solute
molecules can sorb. In a dynamic system, slow kinetics occur when the time for
a sorption/desorption event is slow relative to the velocity of the mobile phase.
Slow kinetics may also occur when the diffusion coefficient of the solute in one
phase is much smaller than in the other, or when there is an interfacial resistance
to mass transfer. Kinetic information is of interest because, together with
thermodynamic information, it can be used to characterize the retention

mechanism.

3.1.2 Mathematical Relationships

While the EMG function (Equation 2.4) has proven to be robust at
describing asymmetrical peaks, many authors have chosen to use statistical
moment analysis (Equations 2.1 — 2.3) since it requires fewer assumptions.
Presented herein is a detailed comparison of these two methods. In order to
facilitate this comparison, a relationship between the moments and the EMG
parameters must be defined [2,3]:

M0 =A (3.1)

54

M1 = tG + I (3.2)
M2 = 0’2 + 12 (3.3)
M3 = 2 1:3 (3.4)
It should be noted that the exponential component of the variance (1:2) has been
previously calculated by subtracting the Gaussian variance (02) from the second
moment via Equation 3.3 [2,3]. This approach implicitly assumes that o and ‘t
are independent of one another. However, the use of Equation 3.3 in this
manner is problematic since there has been no proof to support the validity of
this summation. This chapter directly tests this assumption of independence by
using both simulated and experimental data. The simulation studies compare the
results from both the statistical moment and EMG methods for a series of
generated zone profiles. The experimental data characterize the differences

between the two mathematical methods for a series of fatty acids separated by

reversed-phase capillary liquid chromatography.

3.2 Simulation Methods

Using a conceptual framework by Chomin Lee, five different zone profiles
were created by using commercially available software (Peakfit Version 3.18,
SYSTAT Software) to be representative of the experimental data (vide infra).
The profiles were generated by using the EMG equation with constant values for

t6 and o. The exponential contribution ’t was varied, which results in a gradually

increasing value for the ’t/O' ratio from 0.5 to 10.0 (Table 3.1, Figure 3.1). The

55

Table 3.1: Input parameters for simulated chromatographic peaks according to

the EMG model in Equation 2.4.

56

 

 

 

ed? 06 md o.m deo m
o.m 9N md o.m mmdom v
0N oé md oh Ned—N m
oé md md o.m name N
md mNd md oh 3..me _.
to p b 3 m9< mmmo

 

 

_..m min...

57

Figure 3.1: Graphical representation of five simulated EMG peak profiles

described in Table 3.1.

58

 

 

 

 

 

m 9.60

v memo

m 9.30

m memo

F ammo

 

 

 

 

5 9:9...

59

increa
symrn
simule
integra
mm
by nor
the El

that is

3.2.1

nonlim
”lathe;
descn't
P-Oinl o
the Pro
The lirr
[FEQUre
Visual it
he: the
filmy.
of 0.0%,»:

increasing value of this ratio is indicative of profiles that vary from the
symmetrical to the highly asymmetrical. The profiles were subjected to three
simulation experiments, as described below, to examine the effect of the
integration limits, number of points, and random noise. The profiles were then
imported into Peakfit and evaluated by statistical moment calculation as well as
by nonlinear regression using the EMG equation. The extracted parameters from
the EMG equation (A0, A1, A2, A3) are the best fit for the true values (A, t5, 0, I),

that is they yield the smallest standard error of the regression.

3.2.1 Integration Limits

The first simulation evaluated the effect of the integration limits on the
nonlinear regression parameters and the resulting accuracy of the two
mathematical methods. The process involved generating the profiles, as
described above, and then truncating at different points along the baseline. The
point of truncation corresponded to a fixed percentage of the maximum value for
the profile and represented the boundaries of integration for Equations 2.1-2.3.
The limits that were chosen varied from 0.0 to 1.0 % of the total peak height
(Figure 3.2) and are representative of integration limits that might be selected by
visual inspection of experimental data. For example, a limit of 0.0% indicates
that the integrals in Equations 2.1-2.3 were evaluated from negative to positive
infinity. Since the maximum peak height was equal to 100, an integration interval
of 0.0% contained all points with a value greater than zero. A larger integration
limit indicates that more of the zone profile was truncated before it was analyzed

(9.9., a 1.0% truncation contains all points with a value greater than or equal to

60

Figure 3.2: Graphical representation of case 4 in which different integration

limits were studied.

61

 

 

 

 

 

o\ooo.o
ox... ooé
o\o o to
o\o Pod

 

 

 

 

 

 

 

we 2:9“.

62

as mig
simula
asymr
profile
10 em
The t1

PiOfile

3.2.3

”else

1). The resultant zone profiles were analyzed using the methodology described

above.

3.2.2 Number of Points

The second simulation evaluated the effect of the number of data points,
as might occur at different sampling rates. Case 4 was selected for this
simulation, since it is representative of profiles with moderate levels of
asymmetrical broadening. An integration limit of 0.10% was used to truncate the
profile. In order to vary the number of points, a user-defined program was used
to extract every nth point, which were then exported to a separate file for analysis.
The total number of points was varied from 10 to 398. The resultant zone

profiles were analyzed using the methodology described above.

3.2.3 Noise

The final simulation evaluated the effect of random noise. Five different
noise levels were chosen with standard deviations ranging from 0.1 to 10% of the
maximum C(t) value. The corresponding signal-to-noise ratio varied from 1000
to 10. The random noise at each level was generated several different times
using the random number generator in Microsoft Excel. Each of these generated
noises was added separately to each profile, resulting in several different peaks
at each ‘t/O' ratio. These replicate profiles were necessary to represent
statistically the effect of random noise [4,5]. The resultant zone profiles were

analyzed using the methodology described above.

63

3.3 Experimental Methods

A 10'3 M solution of saturated fatty acids ranging from C10 to 022 (Sigma)
in anhydrous acetone was combined (1:1) with dry sodium sulfate, potassium
bicarbonate, and dibenzo-18-crown-6 (Sigma). An excess of 4-bromomethyI-7-
methoxycoumarin (Sigma) was added to the mixture and the solution was then
allowed to react in the dark for 2 hours at 50 °C [6]. This reaction results in the
addition of a fluorescent tag to the fatty acids and allows for detection via laser-
induced fluorescence. The resultant solution was evaporated under a nitrogen
stream and redissolved in high purity methanol (Baxter Healthcare, Burdick &
Jackson Division) to yield a final concentration of 10" M. The unreacted 4-
bromomethyl-7-methoxycoumarin serves as a void marker since it is unretained
in this system. The fatty acids were then separated by capillary liquid
chromatography [7] using the system (Figure 2.2) described in Chapter 2.2.3.2
by Dr. Chen. The zone profiles were extracted and analyzed by the statistical
moment (Section 2.3.1.1) and EMG (Section 2.3.1.2) methods. The regression
of the zone profiles to the EMG equation was excellent, with typical values for the
square of the correlation coefficient (3) of 0.99 or greater. Other equations of
kinetic origin, the Haarhoff-Van der Linde equation [8], and the non-linear
chromatography equation [9] were also used. However, these functions were

found to poorly fit the experimental data and evince nonrandom residuals.

3.4

3.4.1

from t
Table
of inte
of rec<
Statisti
st tisti

conditi

3.4.1.1

3.4 Results and Discussion

3.4.1 Simulation Results

Zone profiles were generated with independent and varying contributions
from the symmetrical (c) and asymmetrical (1) contributions, as summarized in
Table 3.1. The purpose of the simulation experiments is to determine the effects
of integration limits, number of points, and signal-to-noise ratio on the accuracy
of recovery of the parameters of the EMG equation (A0, A1, A2, A3) and the
statistical moments (M0, M1, M2). In principle, both the EMG function and the
statistical moments should accurately represent the simulated profiles if these

conditions are properly chosen.

3.4.1.1 Integration Limits

The first simulation, detailed in Tables 3.2 and 3.3, varied the integration
interval for the regression analysis from 0.0 to 1.0% of the maximum peak height.
An integration interval of 0.00% results in no error between the EMG or statistical
moments and the initial parameters, as would be expected. However, when the
larger integration limits are evaluated, a deviation between the two methods
emerges. The data in Table 3.2 demonstrate that there is no error in the
regression parameters derived from the EMG equation for all five cases. This
lack of error is attributed to the continuous nature of the EMG equation and its
consequent ability to extrapolate beyond the provided integration limits. While

the statistical moment method also returned no error at the 0.00% integration

65

Table 3.2: Regression parameters from the EMG model (A0, A1, A2, A3)
recovered from simulated chromatographic peaks as a function of integration
interval. (A) C(t) at limits of integration interval, expressed as % of maximum C(t)
value, (B) Error Ao (%) = (A0 — A) x 100/A0, (C) Error A1 (%) = (A1 — ta) x 100/tG,

(D) Error A2 (%) = (A2 - o) x 100/ o, (E) Error A3 (%) = (A3 - 1:) x 100/1:

66

 

 

 

 

67

 

 

 

88 83 88 888 88 83 88 :38 3
88 83 88 83 88 83 88 :38 F8
88 83 88 83 88 83 88 :38 88
88 83 88 888 88 83 88 :38 88 m
88 83 88 83 88 83 88 888 3
88 83 88 83 88 83 88 8.88 F8
88 83 88 83 88 83 88 888 88
88 83 88 83 88 83 88 888 88 v
88 83 88 83 88 83 88 ~33 3
88 83 88 83 88 83 88 N38 3
88 83 88 83 88 83 88 ~33 88
88 83 88 83 88 83 88 ~33 88 m
88 83 88 83 88 83 88 $82 83
88 83 88 83 88 83 88 882 3
88 83 88 83 88 83 88 $82 88
88 83 88 83 88 83 88 882 88 N
88 838 88 83 88 83 88 :32 3
88 83 88 888 88 83 88 3.3 F8
88 83 88 888 88 83 88 :32 88
88 83 88 83 88 83. 88 3.3 88 F
m g E a g _ E o g E m g (as .325
n< hotw n< ~< hotw ~< —< hotm r< o< hotw o< 5:359:— mme
29meth 02m
3 033

ll
its.

‘1
Jt.
)

 

Table 3.3: Statistical moments (M0, M1, M2) recovered from simulated
chromatographic peaks as a function of integration interval. (A) C(t) at limits of

integration interval, expressed as % of maximum C(t) value, (B) Error Mo (%) =
(M0 - A) x 100/Ao, (C) Error M1 (%) = (M1 — (ta + 1)) x 100/(tG + t), (D) Error M2

(%) = (M2 — (o 2 + x 2)) x 100/(0 2 + 12)

68

 

 

 

0.07 K00~ 00. Pl .000 00.0: 00.50 0. ..

0:1 03.3 0N0: 2.0.0 00.0r 00.-0 v.0.

V0.0: $0.0m 00.0r 000.0 3.0: «0.30 3.0

00.0 00~.0~ 00.0 000.09 00.0 3.0.0.0 0.0 0

v.2: 000.0 . 0?. _.r 2;: 050: 00.000 0.9

05.0: 00N.0 5.0: Ev.» 00.0.. 00.000 3.0

00.0.. 030 00.0: 00.1. 3.0.. 5.000 30.0

00.0 000.0 00.0 000.5 00.0 0.0.000 0.0 v

0.3.. V00... 30: 30.0 3.0.. 3.0 K 0.?

0v.~r 0 E; 50.0: 000.0 00.0: 0N0 5 F0

0Y0: 03.? ~00- 000.0 _.0.0.. 3.0 E 8.0

00.0 00m... 00.0 000.0 00.0 3.08 0.0 0

00K: Etc 0.0: 03.0 00.0: 00.00“ 00.?

0v. Pr 000.0 «0.0: 03.0 00.0: 0600? v.0

0~.0r 03.0 00.0 000.0 5.0: «0.00? _.0.0

00.0 000.0 00.0 000.0 00.0 00.00.. 0.0 N

~31 03.0 00.0: 23.0 3.0r 00.09 0.?

3.0: 2.0.0 900: 00~.0 3.0.. 00.»? F0

9.0: ~50 00.0 03.0 3.0: of»? 5.0

00.0 ~50 00.0 00.0.0 00.0 0“.an 0.0 F

83.0 ms _ 03 E mg as. (3.0 sass 88
N.2 .otw N5. :2 .otm :2 8.2 Lotm 0008020.

mEoEofi

 

 

 

md 030...

69

llm't, am
errors be
in Table
cases. l
the max
43.88%
error is
momen
incapal
undere
the mc
dlfiere]
he ett
vanan
Chang
lO~1g
lfibrea

limits.

limit, as would be expected, the larger integration limits resulted in noticeable
errors between the input parameters and the recovered parameters. As shown
in Table 3.3, the error in the zeroth and first moments is relatively small for all five
cases. While there is an increase in the error as the integration limits increase,
the maximum error between Mo and the true area (A) is only

-0.88%, and between M1 and the true retention time (ta + t) is only -1.99%. This
error is attributed to the discrete nature of the summation for the statiStical
moments in Equations 2.1 - 2.3. Consequently, the moment method is
incapable of extrapolating beyond the provided data, which results in an
underestimation of the true value. Note that the error generally increases with
the moment number (n) for all cases. These larger errors arise because the
difference (t - M1) is raised to the nth power in Equation 2.3, which exacerbates
the effect of the integration limits. Accordingly, the error between M2 and the true
variance (02 + 12) becomes more pronounced as the integration limits are
changed. The difference ranges from —0.1 6 to —4.32% for case 1 and from —0.84
to -19.3% for case 5. Thus, an increase in the asymmetry of the zone profile
increases the error of the statistical moments that arises from the integration
limits.

The conclusion that can be drawn from this simulation is that if statistical
moments are to be used for experimental data, the integration limits must be
chosen carefully (i.e. <0.10%) in order to insure that the retrieved values are
meaningful. This observation has been noted previously [10]. In contrast, the

EMG method can be used for larger integration limits without error.

70

3.4.1.2

numbe
Once a
the irri‘
ol the
Becat
error l
where
Point:
numt

for G.

3.4.1.2 Number of Points

The second simulation, detailed in Tables 3.4 and 3.5, varied the point
number used for the regression analyses from 10 to 398 points for case 4 only.
Once again, there is no difference between the EMG regression parameters and
the initial parameters (Table 3.4). This is another result of the continuous nature
of the EMG function, which can interpolate between the provided data points.
Because of its discrete nature, however, the statistical moment method results in
error (Table 3.5). The overall error in the zeroth and first moments is small,
whereas the error in the second moment becomes greater as the number of
points is decreased. In general, the error is relatively minor as long as the
number of points is greater than 50. This conclusion has been noted previously

for Gaussian as well as more asymmetrical zone profiles [11].

71

Table 3.4: Regression parameters from the EMG model (Ao,A1,A2,A3)
recovered from simulated chromatographic peaks as a function of the number of
points within the integration interval. (A) Number of points within the integration
interval at 0.10% of maximum C(t) value, (B) Error A0 (%) = (A0 — A) x 100/A,

(C) Error A1 (%) = (A1 - tG) x 100/tG, (D) Error A2 (%) = (A2 - o) x 100/ c,

(E) Error A3 (%) = (A3 — r) x 100/ ‘t

72

 

 

 

 

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 3

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 .00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 2 P

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00F

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00—

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 00F

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 .00

00.0 000.0 00.0 000.0 00.0 000.0 00.0 00.000 000 0

mach E 83.0 E 03.0 E mg 8 (.8582 88
04 .600 0< N< Lotw ~< P.80 .000 .< 8< BEm .500

920E900 0.2m.

 

 

 

V0. 030...

73

Table 3.5: Statistical moments (Mo,M1,M2) recovered from simulated
chromatographic peaks as a function of the number of points within the
integration interval. (A) Number of points within the integration interval at 0.10%
of maximum C(t) value, (B) Error Mo (%) = (M0 - A) x 100/A, (C) Error M1 (%) =

(M1 — (ta + 1)) x 100/(tG + 1), (D) Error M2 (%) = (M2 — (o 2 + 1 2)) x 100/(0 2 + 1 2)

74

 

 

 

09.0: 000.0 30.0: ~00N 00.0 00.000 0?

000.0: 000.0 000.0: 053.5 .0 w .0: 00.000 om

N006: n 5.0 F .00: V0.05 0 _. .0: 00.000 0v

304.01 03.0 0.0.0: v0.00 0...0: 00.000 00

009v: 00N.0 00 *0: 00vK 00.0: 00.000 00

0.. _...VI 000.0 00 F0: 00¢.N 00.0: 00.000 00

000.0: 30.0 .0 w .0: 00vK 00.0: 00.000 00

050.0: 00.0.0 #0 _..0I 00in 00.0: 00.000 P F F

000.0: 000.0 N0 P .0: 00vN 00.0: 00.000 00 F

.000: 000 N0 F .0: 00.05 00.0: 00.000 00 F

3.50: N000 b0 00: h0vfi 00.0: 0 P .000 09

000.0: 00.0 «h 00: 50.0.5 00.0: _. 0.000 #00

000.0: 000.0 00 *0: 00‘s 50.0: N F .000 000 v

on 0 oo 0 30. 00:52
éflmnpwm ANNE. 3%mmbwm Ms.3 85..“ 8:0 8.). «L 00.00 030
mEoEofi

 

 

 

0.0 0.00...

75

3.4.1.:

noise
posit:
repre:
devia

the p

statis
a fin:
zero.
More
than

disc:

for t
\eve
'vev.
Thu

VGA:

3.4.1.3 Noise

The final simulation, detailed in Table 3.6 and 3.7, varied the random
noise from 0.1 to 10% of the total peak height. Since noise can take on both
positive and negative values, a single profile would not be statistically
representative of the effect on the regression parameters. Thus, the standard
deviation for multiple profiles is reported in addition to the error as calculated in
the previous studies.

As the number of profiles increases, the error in both the EMG and
statistical moment methods is theoretically expected to approach zero. Because
a finite number of profiles was used, typically five to ten, the error is small but not
zero. In general, the error increases slightly as the noise level increases.
Moreover, the error in the EMG regression parameters (Table 3.6) is slightly less
than that in the statistical moments (Table 3.7)‘because of the continuous and
discrete natures, respectively.

The standard deviations are more meaningful than the errors since they
demonstrate the variability with increasing amounts of noise. The relative
standard deviation for the EMG parameters is very small, typically ranging from
0.01 to 1.5% for A0 to A2. The greatest relative standard deviation is observed
for the exponential component A3, which is as high as 5.9% for the highest noise
levels. An increase in asymmetry causes an increase in the relative standard
deviation for A0 to A2, but a decrease in the relative standard deviation for A3.

This trend is observed because it is more difficult to accurately determine small

values for the exponential component when noise is present.

76

Table 3.6: Regression parameters from the EMG model (A0. A1, A2, A3)

recovered from simulated chromatographic peaks as a function of the noise level.

(A) Standard deviation of the random noise level, expressed as "/0 of maximum
C(t) value, (B) Error A0 (%) = (A0 — A) x 100/A, (C) Error A1 (%) = (A, — t6) x

100/te, (D) Error A2 (%) = (A2 - o) x 100/ o, (E) Error A3 (%) = (A3 - T) x 100/1

77

Table 3.6

 

 

 

 

 

 

 

 

EMG Parameters

 

 

 

 

 

 

 

000.0 050.0 H 000.0 500.0: 05.0 H 00V0 5V00: F50 H 000V 0V0 00.0 H 0...000 0.0F
00F.0 000.0 H 05.0 000.0: 500.0 H 000.0 500.0 000.0 H 000.0 :0: 00.F H F5000 0.0
000.0: 000.0 H 000.V 000.0 50.0 H 000.0 000.0: 50.0 H 000.0 V0.0: 00.0 H 0F.000 0.F
000.0 50.0 H 000.0 050.0 50.0 H 000.0 000.0 000.0 H 000.0 5.0: 3.0 H 00.000 0.0
000.0 50.0 H 000.0 000.0 000.0 H 000.0 50.0 000.0 H 000.0 00.0 00.0 H 0V.000 F0 0
000.0: 000.0 H 00V.0 00V0 500.0 H 000.0 000.0 000.0 H 50.0 00.0 00F H 00.000 0.3
05.0 500.0 H 000.0 0V00 V000 H 50.0 000.0: V000 H 000.0 9.0 05.0 H 05.000 0.0
05.0 V000 H 000.0 000.0 50.0 H 0000 05.0 000.0 H 000.0 00.0 0F.0 H 00.000 0...
55.0: 50.0 H 00V.0 0000: 000.0 H 000.0 000.0 000.0 H 000.0 5.0 50.0 H 0V.000 0.0
000.0 000.0 H 000.0 000.0 000.0 H 000.0 50.0 000.0 H 000.0 00.0 00.0 H F0000 F0 V
000.0: 05.0 H 000.F 05.0 000.0 H 000.0 500.0 000.0 H 000.0 00.0 00.0 H 00.90 0.0..
000.0 F50 H V00.F 000.0: 0000 H 50V0 05.0: V000 H 000V 0F.0 00.0 H V5.0F0 0.0
000.0: 50.0 H 000... 05.0: 000.0 H 000.0 50.0: 000.0 H 000.0 00.0: 0F0 H 00.05 0.F
0000: 50.0 H 000.0 500.0 000.0 H 000.0 000.0 000.0 H 000.0 00.0: 00.0 H 50.90 0.0
50.0: 000.0 H 000.F 000.0 000.0 H 000.0 50.0 000.0 H 000.0 5.0: 5.0 H 0005 F0 0
0V0.0: 05.0 H 50V0 000.0 000.0 H 000.0 50.0 000.0 H 000.0 00.0: 00F H 00.00 F 0.9
05F.0 000.0 H 50.0 0V00: 0000 H 00V0 000.0: V000 H 000.0 50.0: 00.0 H 05.00.. 0.0
000.0: 50.0 H 000.0 000.0: 000.0 H 000.0 000.0 000.0 H 000.0 5.0: 00.0 H F0.00F 0...
500.0: 000.0 H 000.0 500.0 000.0 H 000.0 000.0 000.0 H 000.0 5.0: V0.0 H 00.00F 0.0
V000: 000.0 H 000.0 000.0 000.0 H 000.0 000.0 000.0 H 000.0 00.0 5.0 H 00.00 F F0 0
FFO.V: V5.0 H 000.0 050.0 000.0 H 000.0 05F.0 F50 H 000.0 3.0: 00.0 H 00.52 0.9
00V.F 500.0 H V000 000.0 000.0 H 000.0 000.0: 000.0 H 500V 00.0 00.0 H V0.50F 0.0
000.0: 50.0 H 0000 000.0 000.0 H 000.0 V000 50.0 H 000.0 5.0 5.0 H 0V.50F 0.F
V000 50.0 H 000.0 05.0: 000.0 H 000.0 V000: 50.0 H 000.0 5.0 00.0 H 0V.50F 0.0
000.0: 000.0 H 000.0 000.0: 000.0 H 000.0 50.0 000.0 H 000.0 00.0 5.0 H ..V.50F ”.00 F
g a. e... E 3 E E . .\.
3% 8:0 3. «(08:0 N.o. chotm Pa. oxwhotw c< “My ammo

9992995. wiwi

 

 

 

0.0 0Ba...

78

Table 3.7: Statistical moments (M0, M1, M2) recovered from simulated chromatographic
peaks as a function of the noise level. (A) Standard deviation of the random noise level,
expressed as °/o of maximum C(t) value, (B) Error M0 (%) = (M0 — A) x 100/A, (C) Error
M1 (%) = (M1 — (ta + 1)) x 1OO/(tG + T), (D) Error M2 (%) = (M2 - (o 2 + 1.‘ 2)) x 100/(o 2 +

12)

79

Table 3.7

Moments

 

 

 

 

 

 

:':I

«r
I
or

:log.

mew“.-
(Cf: Er"!
m"

 

00F.0: 55F.0 H 0V0.00 V000: 00F.0 H 50.0 0000 F00 H 00.000 00F

000V: 5500 H 09.0 9.00: 0V0.0 H 0500 050.0: V5.F H 00.000 0.0

000V: 0V0.0 H 09.0 0000: 0000 H 000.0 03.0: 00.0 H F0000 0.F

000.»: 0000 H VFO.V0 000.0: 000.0 H F500 5000: 00.0 H F0000 00

0V0.V: 0000 H 000.V0 0000: 50.0 H F500 V000: 000 H 00.00 F0 0
0000: 0000 H 00V0 0VFO: VV00 H 00V.5 0F F0 0 F0 H F0000 0.0F

05F. F: 05. 0 H VOV0 000.0: 0000 H 00V.5 000.0 00.F H F0000 0.0

53.0: V000 H 050.0 00F.0: 050 H 00V.5 05.0: 00.0 H V0000 0.F

000.0: 0V0.0 H 000.0 00 F0: 000.0 H 00V.5 V000: 0 F0 H 0F000 0.0

000.0: V000 H 000.0 5000: 000.0 H 00V.5 0.00: 00.0 H 00.000 F0 V
000V: 0V0.0 H 000.F 050.0: 0000 H 000.0 00F0 F0.F H V50F0 0.0F

00F. F: 0000 H 000.F 000.0: 0000 H 000.0 0000 05.0 H 00.90 0.0

90.0: 050 H 0 F0.F 000.0: 000.0 H 000.0 0000: 0 F0 H 00.30 0.F

000.0: 0000 H 0F0.F F500: 500 H 000.0 0000: 000 H VO0F0 0.0

50V.0: 500 H 000.F 000.0: 000.0 H 000.0 0500: 5.0 H 5005 F0 0
000. F: 0000 H FOV0 00F.0: 0000 H VOV.0 000.0: 5V.F H 0000F 00F

00F.0: 050 H 00.0 000.0: 500.0 H 000.0 200: 00.0 H 0VO0F 0.0

F00. F: 000.0 H VOV0 0000: 50.0 H 00V.0 9.0.0: 000 H 05.00F 0.F

V00. F: 500 H 00V0 0000: 50.0 H 00V.0 0.00: 00.0 H 0500F 00

F00. F: 50.0 H VOV0 000.0: 000.0 H 00V.0 0000: 5.0 H 55.00F F0 0
000.0: 0000 H 0 F00 F000: 0000 H 0V0.0 00F0: 000 H V0.50F 00F

FOV0: 0000 H F F00 0 F00: 0000 H 0V0.0 FV00: VVO H 5V.50F 0.0

000.0: 000.0 H F F00 0000: 50.0 H 000.0 000.0: 00.0 H 50.59 0. F

V000: 000.0 H F F00 0000: 000.0 H 000.0 05.0: V0.0 H 00.50F 00

000.0: 000.0 H F F00 0000: 000.0 H 000.0 0000: 5.0 H 50.50F F0 F

o m co 3 300 30:05...
NFQmmwbw AMEV .Swmmwam :2 2% 8:0 0.2 < $62 300
£00.55.
hm 2an

80

:mge
conu
M28:
forfl
EAAG
no$e

most

unde
kneg
anat:
Conn

«flafi

The standard deviation for the statistical moment method is somewhat
larger than that for the EMG method. The relative standard deviation for M0 is
comparable to that for A0. However, the relative standard deviations for M1 and
M2 are significantly larger, typically ranging from 0.06 to 13.2%. The deviations
for the moments increase with the asymmetry and the noise level, just as for the
EMG parameters. The results summarized in Table 3.5 indicate that random
noise does affect the statistical moments, with the second moment being the
most susceptible to such fluctuations.

When all of the tables are analyzed, a single conclusion emerges: only
under conditions in which the solute zone has a small amount of tailing, small
integration limits, a large number of points, and low noise can statistical moment
analysis be used without significant contributions from user defined errors. In
contrast, for each of the simulations, the EMG function has been shown to be
relatively insensitive to the variables and, as a result, should be better suited for

the analysis of experimental data in which the amount of tailing is significant.

3.4.2 Experimental Results

The simulation studies provide insight into the differences between the two
mathematical methods and the accuracy of the recovered parameters. The
results of these studies are important because they can be used to determine the
reliability of reported chromatographic figures of merit such as the plate number
and plate height. In addition, the accuracy of thermodynamic and kinetic data
can be established on the basis of the calculated errors and the limitations of

each method.

81

Figure 3.3: Separation of saturated fatty acids ranging from C10 to 022 by
capillary liquid chromatography with on-column laser-induced fluorescence
detection at T = 303 K and P = 3000 psi. Experimental conditions as given in the
text. (A) Detector 1, 23.2 cm from the head of the column, (B) Detector 2, 28.4

cm, (C) Detector 3, 51.4 cm,(D) Detector 4, 56.9 cm.

82

Amy m2;

 

 

 

 

 

 

 

 

 

 

0000 F 0000 0000 000x. 0000 0000 000V 0000 0000 000 F
Z - :24 Z 2
D
Z J g 4 iii
0
j g j
m
I/L j‘ 4 fijjl
<

a.» 2:9“.

 

 

 

 

 

 

 

 

 

 

 

83

Figure

dislanc
column
volume
small a
contrlb
to the s
mmar

be dire

the sol
the :et
d'lslan:
flierrnc
litmus
3.5, g?
i5 9%.
linear],
reachs
‘32 Wit:

30mm:

A series of chromatograms of the saturated fatty acids is included in
Figure 3.3. Each chromatogram is representative of the separation at specified
distances of 23.2, 28.4, 51.4, and 56.9 cm along the column. The use of on-
column detection minimizes extracolumn contributions to the zone profile. Any
volumetric contributions that arise from the injector, detector, or connectors are
small and constant at each of the detectors. In addition, the electronic
contributions are also constant since each photomultiplier tube is directly coupled
to the same amplifier and data acquisition system. The result of this design is
that any changes in the solute zone profile that occur between the detectors can
be directly attributed to mass transfer and retention events [12].

To ensure that the behavior of the solutes is consistent, the properties of
the solute zones are graphed as a function of distance. As shown in Figure 3.4,
the retention factor for all of the fatty acids remains constant with increasing
distance. Having established that the retention factor and, hence, the
thermodynamic behavior is constant, the kinetic behavior is then investigated
through the symmetrical and asymmetrical contributions to broadening. In Figure
3.5, o2 is plotted versus distance on a logarithmic scale. The slope of these lines
is equal to unity (1.000 :t 0.002), which indicates that variance is increasing
linearly with distance and that the symmetrical broadening processes have
reached steady state for all solutes. For C10 and C12, there is little difference in

02 with increasing distance. Thus, the symmetrical broadening processes

contribute only a minimal amount to the overall value of o2 for these solutes.

84

Figure 3.4: Logarithmic graph of the retention factor versus distance on the
column for the saturated fatty acids C10 (0), C12 (C1), C14 (A), C16 (0), C18 (0),

C20 (I), 022 (A). Data derived from the chromatograms in Figure 3.3.

85

Figure 3.4

 

100
r

09.

050 8250.0

0F

 

 

 

 

 

 

 

 

Ollo lolullo
nlln Ullllu
II ltd a.
«lei o 10
9+ 7 6
1+ Hill.
I’ «T i

 

F.0

Iij—rj Ti

"2

 

.ooF

 

V.” 93...

HOlOVd NOLLNSLBH

86

Figure 3.5: Logarithmic graph of 02 versus distance on the column for the

saturated fatty acids. Data derived from the chromatograms in Figure 3.3,

symbols defined in Figure 3.4.

87

Figure 3.5

 

1 0000 /

00F

A53 m02<._.w_o

 

 

 

 

 

 

0.0 95mm

2
F
.,., 2
DZ
, 00F 0.1,.
m89
. L“, 88.

88

Howe

lorC‘

grapt
{0995
bmao
intin
beoor
possH
vanun
the fin
aSyrn:
lhEn :.
Contnl
(“Stan

However, a systematic increase in 02 is observed with increasing carbon number
for C14 to 022.

In Figure 3.6, 1:2 is plotted versus distance on a logarithmic scale. This
graph also exhibits a linear increase with distance with a slope equal to unity
(0.999 t 0.002). This value of the slope indicates that the asymmetrical
broadening processes have achieved steady state for all solutes. This increase
in 1 indicates that there is either a kinetic effect or nonlinear isotherm that
becomes more pronounced with increased distance along the column. The
possibility of a nonlinear isotherm can be rejected, since a series of solutions with
varying concentration was injected to insure that the final concentration was on
the linear portion of the isotherm. Similarly, instrumental contributions to the
asymmetry can be discounted. Were T to arise solely from instrumental effects,
then the value would be constant with distance because the extracolumn
contributions are the same at each detector. Thus, any increase in 1: with
distance must be the effect of slow kinetics. In addition, a systematic increase in
12 is observed with increasing carbon number for all solutes.

In order to demonstrate these trends more effectively, the data for the fatty
acids are tabulated in Table 3.8 as a function of carbon number for both the EMG
and statistical moment methods. The regression parameters from the EMG
equation (A0, A1, A2, A3) show the expected increase with carbon number, as
mentioned above. The statistical moments (M0, M1, M2) show similar trends.

The most interesting data, however, are contained in the final three columns that

describe the differences between the moment and EMG methods. The first of

89

 

Figure 3.6: Logarithmic graph of 12 versus distance on the column for the
saturated fatty acids. Data derived from the chromatograms in Figure 3.3,

symbols defined in Figure 3.4.

90

 

Figyre 3.6

 

 

 

[OE-+06;

 

 

00..

A83 woz<hw5

0F

 

 

r

[II-Irv Vf IIIIII I I

ItiIII I I

U [UUUUUV I j

I'UI'" U U

 

'r'UUU I I

 

0.0 930E

00+w0.F

F0+m0.F

00+mo. F

00+mo. F

V0+mo.F

00+m0. F

00+w0. F

(as) 21

91

Table 3.8: Statistical indicators of fit and regression parameters from the EMG
model (A0, A1, A2, A3) and statistical moments (M0, M1, M2) recovered from
experimental chromatographic peaks in Figure 3.1. (A) AMo (%) = (M0 -Ao) x
100/Ao, (8) AM (%) = (M1 - (A1 + A,» x 100/(A1 + A3), (C) AM2 (%) = (M2 - (A22
+ A32)) x 100/(A22 + A32)

92

 

Yon:

05. F:

0.3:

00V.0VF

000. FF

000.00F.F

0.500

00.00

000. FF

000.000. F

800000

 

 

 

 

 

 

 

 

 

 

 

 

 

08¢ NN
. . i . l . . . . . . . . F820
0 Fe: 8 F 0 NF 80 FV NFm F 80 N8 o N8 FN ON 80 F 8o 08 F N85 ON
o. FF: 8. F: 0.07 80.0 FmNV 8080 on: FF. FF SFV SQVmF Fwwwomv 0F
. i . . l . . . . . . . SFNFFV
o FF 8 o: m 0F 0V0 F 8” N 2: FOV 8 NF on F an N 8V VmV 08.0 0F
. .. . . u . . . . . . . 6880
V 2 F0 on N 2 V FNN mNV F 2: 8N Fa mN NV 0 8V F 2: NmN F86 VF
. . : . . . . . . . . 6&8
VFV: ONo V00: :8 83 838 F NF mFV 008 SF mFF 80o NF
. i . . u . . . , . . . . F889
m 8 3 o: mN N 8 8 F 8» 2: V2 3 V VF V N 8F 80 02 Sad 2
000 mg <00 E E as. E E E °< E .355
£2 .52 £2 «.2 :2 9V ~< PV. em 828

 

 

0.0 630...

EMG

93

the
am
ext
bet
ran
lim:
less
larg
bet:
the
is 0

70°.

bee

that

these columns describes the error that is inherent in the area (Mo, A0). For C10
and 012, the errors are relatively small. However, for the larger molecules that
exhibit greater symmetrical and asymmetrical broadening, the difference
between the moment and the corresponding EMG parameter is much larger,
ranging from —12.8 to —18.2%. The errors associated with the mean retention
time (M1, A1 + A3) are much smaller than those associated with the area, typically
less than -1.7%. However, the error in the variance (M2, A22 + A32) is much
larger. Although the simulation results above suggest that the difference
between the moment analysis and the EMG parameters may be as high as 20%,
the experimental data demonstrate even larger errors. The error for C10 and C12
is on the order of 20 and 48%, respectively, whereas there is a consistent error of
70% for C14 to C22. It should be noted that that any error arising from the sources
discussed previously (integration limits, number of points, and noise) has already
been minimized. Thus, the large difference between the two methods indicates
that the second moment is not equal to the sum of 0’2 and 12, as given by
Equation 3.3. As mentioned previously, this equation presumes that the
contributions from 0’2 and 12 are independent, which may not be true for this
experimental system. Rather, the data suggest that the retention processes
leading to symmetrical broadening are coupled with those leading to
asymmetrical broadening. This suggestion is reasonable, since each solute will
have a wide range of possible residence times within the stationary phase rather
than a fixed value [13]. An individual sojourn in the stationary phase with a short

residence time will contribute to the symmetrical broadening, whereas those with

94

 

hc
C:
oc
for
a5)
the

dat.

longer residence times will contribute to the asymmetrical broadening. As shown
in Figure 3.7, the ratio of the asymmetrical and symmetrical contributions (oz/oz)
increases systematically with carbon number for solutes up to C13. In other
words, the asymmetrical contribution from slow kinetics becomes more important
with increasing carbon number. This observation is intuitively reasonable
because the diffusion coefficient in the stationary phase will decrease with
increasing carbon number, leading to longer residence times. It is noteworthy,
however, that the 12ch2 ratio remains relatively constant for solutes larger than
C13. Since the C20 and C22 solutes cannot insert completely into the
octadecylsilica stationary phase, their kinetic contribution is no greater than that
for C18. It is apparent from these results that the symmetrical (o) and
asymmetrical (1) contributions are not independent in this system. Consequently,
the method of calculating T from Equation 3.3 [2,3,10] is incorrect and kinetic

data derived in this manner should be considered suspect.

95

Figure 3.7: Logarithmic graph of 12/o2 versus carbon number for the saturated

fatty acids. Data derived from the chromatograms in Figure 3.3.

96

00

 

mmmEDz 20m¢<0
00 0 F 0F . VF 0F 0..

F F _ P r

T!

 

 

 

 

5.0 050E

I OF

00F

003

.01
2,2

97

3.5 Conclusions

Although the use of statistical moments for the analysis of
chromatographic data is commonplace, the use of moments in the derivation of
secondary variables is problematic. As demonstrated by the simulation data, the
use of moments to calculate parameters from the EMG model can lead to errors.
As a result, the use of an appropriate model is critical for characterizing the
molecular level processes that contribute to retention. In addition, the constant
values for the retention factors as a function of distance, as well as the linear

increases in 0' and T as a function of distance demonstrate that the experimental

design and data collection do not lead to significant error. Thus, this chapter
demonstrates that the method by which the data are collected and analyzed is

valid for the analysis of the molecular contributions to retention.

98

3.6

[1]

l2]

l3]

l4]

l5]

l5]

3.6 References

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

J.C. Giddings, Dynamics of Chromatography, Marcel Decker, New York,
NY, USA, 1965.

JP. Foley, J.G. Dorsey, J. Chromatogr. Sci. 22 (1983) 40-46.

M.S. Jeansonne, J.P. Foley, J. Chromatogr. 461 (1989) 149-163.
J.V.H. Schudel, G. Guiochon, J. Chromatogr. 457 (1988) 142.8

V8. Di Marco, C.G. Bombi, J. Chromatogr. A 931 (2001) 1-30.

V.L. McGuffin, R.N. Zare, Appl. Spectrosc. 39 (1985) 847-853.

V.L. McGuffin, S.-H. Chen, J. Chromatogr. A 762 (1997) 3546.

RC. Haarhoff, H.J. Van der Linde, Anal. Chem. 38 (1966) 573-582.
J.L. Wade, A.F. Bergold, P.W. Carr, Anal. Chem. 59 (1987) 1286-1295.
D.J. Anderson, R.R. Walters, J. Chromatogr. Sci. 22 (1984) 353-359.
S.N. Chesler, S.P. Cram, Anal. Chem. 43 (1971) 1922-1933.

C.E. Evans, V.L. McGuffin, Anal. Chem. 60 (1988) 573-577.

V..L McGuffin, P. E. Krouskop, D. L. Hopkins, inJ. F. Parcher, T..L Chester
(Editors), ACS Symp. Ser., Washington, DC, 1999, p. 37.

99

he

he

he
an
ac
en
da

ihc

Chapter 4: Retention of Polycyclic Aromatic
Hydrocarbons: Effect of Ring Number, Planarity and
Stationary Phase Bonding Density

4.1 Introduction

One method traditionally used to investigate PAHs has been reversed-
phase liquid chromatography (RPLC). By using RPLC, the retention of PAHs
has been studied as a function of many parameters. Previous investigations
have sought to explore the retention event as a function of solute structure [1 -4],
stationary phase chain length and bonding density [1 ,2,5-8], mobile phase
composition [1,8], temperature [1.9.10], and pressure[11,12]. The investigations
into solute structure demonstrated that ring number, annelation, and planarity
have a large effect on the overall retention [2-5]. From these experiments, many
’ authors have attempted to explain the mechanism of retention (i.e. partition or
adsorption) on the basis. of the thermodynamically derived value of molar
enthalpy and the general shape of the van’t Hoff plot. However, thermodynamic
data alone fail to consider the kinetic aspects of retention. Without kinetic data, a
thorough description of the retention mechanism cannot be developed.

Using the mathematical analyses in Chapter 1 and the chromatographic
system described in Chapter 2, the retention behavior of a series of six polycyclic
aromatic hydrocarbons (PAHs) was studied and characterized using
thermodynamic and kinetic theory. These solutes were studied as a function of
ring number and planarity. The thermodynamics and kinetics are measured

simultaneously, permitting a correlation of these processes that has not

100

me

Cor

Sac

previously been possible. These studies were completed as a function of

temperature, pressure, and stationary phase bonding density.
4.2 Experimental Methods

4.2.1 Solutes

As depicted in Figure 4.1, six polycyclic aromatic hydrocarbons have been
chosen to investigate the effect of ring number and planarity on the
thermodynamics and kinetics of retention. Phenanthrene (Ph) and chrysene
(Chr) (Sigma-Aldrich), and picene (Pic) (National Institute of Standards and
Technology) were obtained as solids and dissolved in high-purity methanol
(Burdick and Jackson, Baxter Healthcare) to yield standard solutions at a
concentration of 10“ M. These three solutes were chosen as a homologous
series to study the effects of ring addition on retention. In addition a sample
(SRM 869, National Institute of Standards and Technology) containing
benzo[a]pyrene (BaP), tetrabenzonaphthalene (T BN) and phenanthro[3,4-
c]phenanthrene (PhPh) was used as received. These three solutes were chosen
to investigate the effect of planarity on retention. A nonretained marker, 4-
methylhydroxy-7-methoxycoumarin, was added to each solution at a
concentration of 10" M. To ensure solubility, the solutions were equilibrated at

each temperature prior to analysis.

101

Figure 4.1: Structure of polycyclic aromatic hydrocarbons used to study

retention in monomeric and polymeric octadecylsilica.

102

Figure 4.1

Phenanthrene Chrysene

*

Benzo[a]pyrene

o o ‘0
00% “”0

Tetrabe nzonaphthalene Phenanthro[3,4-c]phenanthrene

103

 

TE

allc

COn

Wei.

4.2.2 Experimental System

The system used to study the effect of ring number and planarity is
depicted in Figure 2.2, and the data were collected by Dr. Chen. The mobile
phase was pure methanol. Polymeric (5.4 umol/mz) and monomeric (2.7
umol/mz) octadecylsilica were used as the stationary phases to study the effect
of bonding density. The temperatures for these experiments were 293, 298, 303,
308 and 313 K. The inlet pressures for these experiments were 830, 1000, 1930
and 3000 psi. These values correspond to average pressures of 400, 570, 1466,
and 2566 psi, respectively, which were calculated assuming a linear pressure
drop along the column. The inlet pressures were achieved by trimming the
restrictor and splitter proportionally as described in Chapter 2. The data were
analyzed using the exponentially modified Gaussian (EMG), which is described

in section 2.3.1.
4.3 Results and Discussion

4.3.1 Thermodynamic Behavior

4.3.1.1 Retention Factor

The retention factor (k) is the simplest of the measurable parameters that
allow for the comparison of the relative strength of interaction. A larger retention
factor indicates that the PAH has greater affinity for the stationary phase. This
comparison is useful to evaluate the effect of different PAH structural features as

well as different stationary phase bonding densities.

104

 

-. :c 2. £5ch :czcht ..F .V Sock

200:0; .8 u «02 a

v. 000 n... 6 09030.00 0.298 50:0me 0

._0a 0000.."0 F0 020.023 0680.. 50:22“. 0

.00 0000ua F0 020.328 0.683 000:0me V

 

 

 

 

 

 

 

 

 

 

 

 

 

0F.F F0.F F0.F N0.F 0N.F 0 F. F 0F.F F0.F 20F
NF.F 0F.F 00.0 00.0 0F.0 NF.F F00 00.0 £20
000 F00 V0.0 00.0 00.N 00.0 00.0 V0.0 000
$2 $2 8.0 00.0 3.0 . Rd 00.0 20
.32
00.N F0.N 0V.o 90 2F 00.N FV.0 0V.o .00
00.0 00.0 0N.0 FN.o VV.0 00.0 0N0 0N0 0n.
00 000N 00 8V 00 000N an 000 x 20 v. 00N x 00 x 00N 9200
0000 0000 ..000 0.000
otoEZoa otoEocoE otmEzom otoEocoE

 

 

 

 

 

 

.3000 mo___m_>omanoo 2008200 0:0 25:65:. :0 01$. .2 0.209 200:0me ”F.V 200...

105

su
re:
rel
no
be:
ph:
W
has
elul

pre:

Using Equation 1.2, the retention factor for each solute on the monomeric
octadecylsilica was calculated. The results of theses calculations are
summarized in Table 4.1. For the homologous series of planar PAHs, the
retention factor increases with ring number (i.e. Pic > Chr > Ph). However, the
retention factor decreases with increased ring condensation (i.e. Pic > BaP). As
noted for the five-ring analogs, the retention factor for the more condensed
benzo[a]pyrene is smaller than that for picene. The nonplanar PAH, '
phenanthro[3,4,-c]phenanthrene, has a smaller retention factor than would be
expected based upon the number of rings. By contrast, tetrabenzonaphthalene
has a retention factor that is consistent with an increased number of rings. The
elution order follows the trend of PhPh < BaP < TBN at all temperatures and
pressures. Of the three solutes, PhPh is the most nonplanar and cannot
penetrate far into the bonded phase, resulting in small retention factors. The
planar BaP, and the slightly nonplanar TBN, penetrate further into the stationary
phase and thus have larger retention factors. The planar solute is less retained
than TBN on the monomeric octadecylsilica due to the decreased interactions
with the stationary phase that result from the decreased bonding density. The
nonplanar TBN’s larger retention factor likely results from a more optimal
interaction distance with the stationary phase molecules, which leads to
increased van der Waals interactions. Sander and Wise have previously
observed this elution order and attributed it to the three-dimensional structure of

the monomeric stationary phase [7,13].

106

all
firs
ev;
lar'
alk
aris
of 1
co:
witl
fact

0Ct:

larg
beg
(~71
COn-
tetra
Whe
Dilly
The 1

Dory

By comparison, the retention factors for all solutes are larger on the
polymeric phase than on the monomeric phase (Table 4.1). This increase can be
attributed to two differences between the monomeric and polymeric phases. The
first difference is simply a change in the phase ratio (Equation 1.1). As
evidenced by the higher bonding density, 5.4 jimon2 for the polymeric phase, a
larger volume of stationary phase (Vs) allows for more interactions between the
alkyl chains and the solutes. A second difference is the molecular structure
arising from the trifunctional silane, which allows higher density and greater order
of the alkyl chains than the monofunctional silane. This structural difference,
coupled with the increased carbon load, allows solutes to interact more strongly
with the polymeric phase. The increased interactions lead to higher retention
factors, which are in agreement with previous work using high-density
octadecylsilica [7,13-1 5].

The retention factor for phenanthrene on the polymeric phase is 76%
larger than that on the monomeric phase. This increase in retention factor
becomes systematically larger with ring number for-chrysene (180%) and picene.
(~750%), whereas the more condensed benzo[a]pyrene changes by 260%. In
contrast, the nonplanar phenanthro[3,4-c]phenanthrene and
tetrabenzonaphthalene demonstrate smaller changes in the retention factors
when compared to the planar solutes. In addition, the elution order on the
polymeric phase has also changed from that on the monomeric phase, following
the trend of PhPh<TBN<BaP at all temperatures and pressures. For the

polymeric ODS, the nonplanar solutes are more excluded than the planar BaP.

107

tee

bo

pic
ter
Ph
lei
co
Ph

37

an
ret
00
NC
We
Ph.
ieti

Cor

This results in a decreased retention factor. Sander and Wise have observed
this retention behavior previously and have used these solutes as an empirical
test for the characterization of stationary phases [7,13].

The effect of temperature on the retention factor is very pronounced on
both phases. For all solutes, an increase in temperature results in a decrease in
the retention factor. For the homologous series, phenanthrene, chrysene and
picene exhibit changes of —10.7%, —16.3%, and -28%, respectively, over the
temperature range of 293 to 313 K at 2605 psi on the monomeric phase.
Phenanthro[3,4-c]phenanthrene exhibits a change of —13%, whereas
tetrabenzonaphthalene demonstrates a change of -19% under the same
conditions. On the polymeric phase, the changes in retention factor are larger:
phenanthrene —32.3 %, chrysene —56.7 %, phenanthro[3,4-c]phenanthrene -
37.5 %, and tetrabenzonaphthalene —26.2%.

In contrast, the effect of pressure is relatively small on both the monomeric
and polymeric phases. An increase in pressure results in an increase in
retention factor for the planar solutes, but a decrease for the nonplanar solutes
on both phases. For the homologous series, phenanthrene, chrysene, and
picene exhibit changes of 3.7%, ~0.0%, and 3.1%, respectively, over the
pressure range of 395 to 2605 psi at 293 K on the monomeric phase.
Phenanthro[3,4-c]phenanthrene exhibits a change of ~0.0%, whereas
tetrabenzonaphthalene demonstrates a change of —0.7% under the same

conditions. On the polymeric phase, the changes in retention factor are slightly

108

 

de

8F
lin
in:
PC
in:
EF

a:

pt

ire

larger for most solutes: phenanthrene ~0.0%, chrysene 5.6%, phenanthro[3,4-

c]phenanthrene -2.6%, and tetrabenzonaphthalene —6.4%.

4.3.1 .2 Molar Enthalpy

By applying the thermodynamic theory detailed above, the change in
molar enthalpy can be calculated. Using Equation 1.5 and the methods
described in Sections 1.3.1 and 2.3.1.2, the change in molar enthalpy was
calculated. A representative graph used in the calculation of the change in molar
enthalpy is shown in Figure 4.2. For all solutes at all pressures, the data are
linear (R2 = 0.994 - 0.999) and the slope of the line is positive. A linear graph
indicates that the change in molar enthalpy is constant with temperature. A
positive slope is demonstrative of a negative change in molar enthalpy, which
indicates that the transition from the mobile to stationary phase is an
energetically favorable exothermic process. Representative data for the PAHs
are summarized in Table 4.2.

Overall, the change in molar enthalpy is relatively small on the monomeric
phase, ranging from -0.8 to -2.9 kcanol. The homologous series illustrates a
trend of decreasing molar enthalpy, where retention becomes more exothermic
as ring number increases. A comparison of the five-ring analogs, picene and
benzo[a]pyrene, indicates that the more condensed solute has a smaller change
in molar enthalpy. This trend is conserved at all pressures in the range from 400
to 2600 psi. Previous work by Sentell and Henderson demonstrates the same

trend for the four-ring analogs, pyrene and naphthacene [16].

109

Figure 4.2: Representative graph of the retention factor versus inverse
temperature used to calculate the change in molar enthalpy according to
Equation 1.5. Column: polymeric octadecylsilica. Mobile phase: methanol, 2566
psi, 0.08 crn/s. Solutes: phenanthrene ( <> ), chrysene ( Cl ), picene ( O ),
benzo[a]pyrene ( A ), phenanthro[3,4-c]phenanthrene ( O ),

tetrabenzonaphthalene ( A ).

110

 

FL. m0:0 000250.20: F

 

 

 

 

 

 

 

 

 

 

 

 

 

59

j to

end, 2566
l.

O

0V0 80 00.0 00.0 0N0 80 000
p L F b - —..o
0 lo
+ 10.1
T lie
lloil
lll...‘
cl 0 . F
4‘ lei
. 2
SF
N.V 050:

HOlOVd NOILNBJEH

111

.JIIIL
J- itToFLC 0:05300 DEM 0.:0EOEOE CO 0I<Q FOF 0:030) 0505 020

AQFNCFFQ L030: uN.V 0200.5

 

.2. 000N 2 8V". 0:0 v. 000uF .0 00.0.8.8 .52.. 2.5.2. 0.22 o
.8 000N 0. 000". 0:0 v. 000uF .0 00.0.8.8 05>... 9.5.2, .222 o
v. 90 0. 00NuF 2.0 .8 8V". .0 00.0.8.8 05:3 305:0 .03.). a

v. 80 s 00NuF 0:0 .8 000nm .0 00.0.8.8 05:3 3.05:0 .065. V

 

 

 

 

 

 

 

 

 

 

 

F H 0.0 F H 0.F N0 H N0: 2. H 0.F: 20F
FHFF FH0F F.0H0.V: F.0H0F: 0%..
FHF.0F: FH F.N: F.0H F0: F.0HN.N: .00
F H V.0N: F H F0: N0 H 0.0T F... H 0N: o...
FHN0: FH0.F: F.0H0.F: 2.0.7 .00
NHFF: FHmFi F.0H0.0: F.0H0.o: F...
0.000 0.000 0 000 V 000
0_.0E>.0n. 000E985. 0..0E>.on. 000E822 05.00
0855. e.>< 025.09.. 5.2

 

 

 

 

.00....03000000 0..0E>_on 0cm 0_.0Eo..oE co 01$. .2 0E:.o> .0.0E 0cm 3.05:0 .065. "0.F. 030...

112

phe

let:
the

oh.

oh
pt
th
3‘:

b:

rE

1h

The changes in molar enthalpy for the nonplanar PAHs on the monomeric
phase are smaller than those for the planar PAHs. The change in molar enthalpy
is -1.0 kcal/mol for phenanthro[3,4-c]phenanthrene and -1.5 kcal/mol for
tetrabenzonaphthalene. These small values indicate that the nonplanar nature of
these molecules inhibits their ability to interact optimally with the stationary
phase. The changes in molar enthalpy for the PAHs on the polymeric phase
follow the same trends as seen on the monomeric phase. However, as indicated
in Table 4.2, the trends are more pronounced. For the planar solutes, the
change in molar enthalpy ranges from -3.6 to -15.8 kcal/mol on the polymeric
phase. The change in molar enthalpy for phenanthrene is 350% greater than
that on the monomeric phase. This increase in molar enthalpy becomes
systematically larger with ring number for chrysene (375%) and picene (444%),
but becomes smaller for benzo[a]pyrene (314%). This is likely the result of
increased interactions between the solute and the alkyl chains that comprise the
stationary phase, as purported by Sander and Wise [17,18].

The data for the nonplanar solutes also demonstrate the large differences
in molar enthalpy between the polymeric and monomeric phases. For
phenanthro[3,4-c]phenanthrene and tetrabenzonaphthalene, the changes are
350% and 113%, respectively. On the polymeric phase, the absolute values of
the molar enthalpy exhibit the trend of PhPh>TBN. However, the retention
factors show the trend of TBN>PhPh. As demonstrated in Equation 1.5,
retention is a function of both enthalpy and entropy. It is the entropic contribution

that causes the retention factor to be larger for the TBN even though the

113

enthalpic term is smaller. Specifically, the difference in retention factors
suggests that phenanthro[3,4-c]phenanthrene has greater entropic contributions
that compensate for the enthalpic contributions, resulting in a shorter retention

time.

4.3.1.3 Molar Volume

Similar to the molar enthalpy, the molar volume also provides insight into
the effects that arise from structure. Using Equation 1.7 and the methods
described in Sections 1.3.1 and 2.3.1.2, the change in molar volume was
calculated. A representative graph used in the calculation of the change in molar
volume is depicted in Figure 4.2. For the planar solutes at all temperatures, the
data are linear (R2=0.949 - 0.999) and the slope of the line is positive. A linear
graph indicates that the molar volume is constant with pressure. A positive slope
results in a negative molar volume, which indicates that the molecule occupies
less space in the stationary phase than in the mobile phase. However, the
nonplanar solutes exhibit linear behavior with a negative slope. A negative slope
indicates that the transition from the mobile to stationary phase results in a larger
volume for these nonplanar molecules. Representative data are contained in
Table 4.2.

Overall, the change in molar volume is relatively small on the monomeric
phase, ranging from —1.8 to —3.1 mUmol for the planar solutes. The changes in
molar volume for phenanthrene and chrysene are statistically indistinguishable.
For chrysene and picene, however, an increase in ring number results in a more

negative change in molar volume. This trend arises because the smaller PAHs

114

Figure 4.3: Representative graph of the retention factor versus pressure used to
calculate the change in molar volume according to Equation 1.7. Column:
polymeric octadecylsilica. Mobile phase: methanol, 308 K, 0.08 cm/s. Symbols

defined in Figure 4.1.

115

0000

003 mewwwmn.

 

 

 

 

 

 

 

 

 

oomN OOON comF oooF com o

h 1- D I h Fco
O 9 Tlo
. . I

1F
@l 1m.) I
«r . . 0F
00..

 

 

0V 2.0.“.

HOiOVd NOLLNBJSH

116

penetrate only a short distance into the stationary phase [19], where the alkyl
chains exhibit a high degree of disorder [15,20,21]. Since the larger PAHs can
penetrate more deeply, they are able to probe regions that are more ordered and
more dense. This closer arrangement of the PAHs and alkyl chains leads to a
smaller molar volume than in the disordered regions of the stationary phase and
in the randomly oriented mobile phase. A similar rationale can be used to
explain the effect of annelation structure. The more condensed solute,
benzo[a]pyrene, penetrates to a smaller depth than does picene. As a result, the
overall change in molar volume is less negative for benzo[a]pyrene than for
picene. In contrast, the change in molar volume is positive and statistically
indistinguishable for the nonplanar PAHs, tetrabenzonaphthalene and
phenanthro[3,4-c]phenanthrene. The positive molar volume implies that these
molecules do not penetrate deeply and are not effectively solvated by the alkyl
chains. The increase in molar volume for TBN results from the inability of the
polymeric phase to encapsulate the solute as efficiently as the monomeric phase.
These trends are conserved at all temperatures in the range from 293 to 313 K.
The changes in molar volume on the polymeric phase are, in general,
more pronounced than those on the monomeric phase. For the planar solutes,
the change in molar volume ranges from —1.7i 1 to —28.4 i 1 mL/mol on the
polymeric phase. The change in molar volume for phenanthrene is statistically
indistinguishable on the monomeric and polymeric phases. Chrysene and picene

exhibit changes of 355% and 816%, respectively, whereas the more condensed

117

Figure 4.4: Representative graph of the change in molar enthalpy versus the

change in molar volume. Planar solutes ( O ), nonplanar solutes ( I ).

118

00:509.. >n.-_<I...Zm mSOs.
0- F1 0- 0. 0 F- N F- V F-

F F F F F F

 

 

1 ON!

.2.

0F-

 

 

V.V 050.”.

0F

(:ownw) awmorx awow

119

benzo[a]pyrene changes by 380%. As with the monomeric phase, all of the
planar PAHs exhibit negative changes in molar volume, whereas the nonplanar
molecules exhibit positive changes. Phenanthro[3,4-c]phenanthrene undergoes
changes that are statistically indistinguishable, whereas tetrabenzonaphthalene
undergoes a much larger change of 325% on the monomeric and polymeric
phases.

Finally, it is interesting to note the correlation between the changes in
molar enthalpy and molar volume. As illustrated in Figure 4.4, for the planar
PAHs, a linear correlation (R2 = 0.98) is observed between these thermodynamic
parameters on the polymeric phase. The nonplanar PAHs lie nearly along the
same line, such that the overall correlation is still quite high (R2 = 0.96). This
trend occurs because the molar enthalpy and molar volume can be compensated
through the internal energy (Equation 1.6), similar to the compensation of molar
enthalpy and molar entropy through the free energy (Equation 1.4) [22,23]. The
physical reason for this correlation can be attributed to the varying extents of
solute penetration and the varying extents of disorder in the stationary phase. At
the proximal end that is bound to the silica, the alkyl chains are highly ordered in
the all-trans configuration. As distance increases, some flexibility is possible and
gauche configurations may be formed. At the distal end, these gauche bonds
are most prevalent and result in a highly disordered region [20,21]. As the PAH
penetrates into the more ordered regions, there arises simultaneously an

increase in the interaction with the alkyl chains and a decrease in the available

120

space. Accordingly, the changes in molar enthalpy (AHsm) and molar volume

(AVsm) are observed to be correlated.
4.3.2 Kinetic Behavior

4.3.2.1 Rate Constant

Although the thermodynamic data demonstrate the steady-state aspects,
they do not fully explain the mechanism of retention. Using Equations 1.16, 1.17
and 1.20 and the methods described in Sections 1.3.2 and 2.3.1.2, the pseudo-
first-order rate constants, activation enthalpies, and activation volumes were
calculated. These values help to quantify the kinetic aspects of solute transfer
between the mobile and stationary phases.

Representative values of the rate constants are summarized in Table 4.3.
Several solutes, notably phenanthrene, chrysene, phenanthro[3,4-
c]phenanthrene, and tetrabenzonaphthalene, have rate constants that exceed
the resolution capabilities of this method. The upper limit of reliability was
determined through a series of simulations and experiments. For the data
contained herein, the #0 ratio must be greater than 0.4 to yield reliable rate
constants. For smaller ‘t/O’ ratios, the zones are nearly Gaussian in shape and
the exponential component cannot be accurately determined. The rlo ratio of 0.4

corresponds to a rate constant of 400 s".
For the monomeric phase, the transfer of phenanthrene and chrysene is
sufficiently fast that no reliable conclusions can be drawn. However, the rate

Constants decrease with increasing ring number, such that picene and

121

benzo[a]pyrene can be evaluated. The more condensed structure of
benzo[a]pyrene results in larger rate constants for the mobile to stationary phase
(km), and the stationary to mobile phase (kms), when compared to picene. Both
molecules exhibit a retention mechanism in which the rate-limiting step is the
transfer from mobile to stationary phase (i.e., kms> ksm). In addition, an increase
in temperature yields an increase in the rate constants for all molecules. For
example, the rate constant (km) for picene ranges from 26 to 78 s'1 for
temperatures from 288 to 313 K. In contrast, an increase in pressure yields only
slight changes. The rate constant (ksm) for picene ranges from 38 to 34 s’1 for
pressures from 605 to 2605 psi at 303 K.

As the rate constants for all planar compounds with ring numbers equal to
five can be determined, those for the six-ring compounds were also anticipated to
be sufficiently small. However, phenanthro[3,4-c]phenanthrene and
tetrabenzonaphthalene are fast and cannot be quantitated on either the
monomeric or polymeric phases. This suggests that the kinetic behavior, similar
to the thermodynamic behavior, is heavily influenced by the nonplanar structure
of these solutes.

On the polymeric phase, the rate constant for phenanthrene is very large
but those for chrysene, picene, and benzo[a]pyrene are small enough to be
quantified. As shown in Table 4.3, the rate constants for the transfer from
stationary to mobile phase are smaller than those observed on the monomeric
phase. In contrast, the rate constants for the transfer from mobile to stationary

phase are generally larger than those observed on the monomeric phase. Thus,

122

 

 

 

 

 

 

 

 

 

 

 

0on ooVN ooVN ooVN 20F

ooVN 8VN 8VN ooVN 0%:
F0N F0 F0 0F :00
0: 0 00 0V 0..
FNN N0F ooVN 8VN .00
00VN 00VN ooVN 00Vm 0:

0.0.2.... 0.0.2.... .7002... .0. a... 0.200

m.000 2.0029...

 

<000 000E055.

 

 

 

.8 00°F": 0:0 v. 000uF .0 00.0.8.8 0.5.008 0.0m 0

.0... 000an 0.0

v. 000"... .0 00.0.0200 05... 50.05.. 000000 00:20.0 0. 0.30:. 0:0 .12... 0.50:. 0. 00:20.0 0.... .0. 3:00:00 0.0m. <

.00...0.>0000.0o 000E200 0:0 0:080:06 :o 01$. .0. 0:00:00 0.0. 000:5. .0.V 0.00...

123

the data indicate barrier for entry into the higher density phase requires more
time and more energy (vide infra), and that the barrier for exit is noticeably lower.
For chrysene and picene, the rate constants decrease sharply as the ring
number increases. For picene and benzo[a]pyrene, the more condensed
structure allows faster transitions.

Overall, the data indicate that the rate constants (kms and ksm) decrease as
the retention factor increases. These results imply that the solutes become more
retained as ring number and bonding density increase, with each retention event

becoming longer in duration.

4.3.2.2 Activation Enthalpy

Given that many of the molecules exhibit very fast transitions, the
additional kinetic parameters cannot be determined on the monomeric phase.
Although the rate constants for picene and benzo[a]pyrene are quantifiable at low
temperature, they too become unreliable at higher temperatures. As a result, the
activation enthalpies and activation volumes can only be calculated on the
polymeric phase with any statistical certainty.

Using Equation 1.20 and the methods described in Sections 1.3.2 and
2.3.1.2, the activation enthalpies were calculated. Figure 4.5 illustrates a typical
graph of the natural logarithm of the rate constant versus inverse temperature
used in the calculation of the activation enthalpy. Table 4.4 contains those
activation enthalpies that could be reliably calculated from the kinetic rate
constants. As demonstrated for chrysene and picene, an increase in ring

number results in an increase in the activation enthalpy. Similarly, a comparison

124

Figure 4.5: Representative graph of the rate constants versus inverse
temperature used to calculate the activation enthalpy according to Equation 1.20.
Column: polymeric octadecylsilica. Mobile phase: methanol, 2566 psi, 0.08 cm/s.

Solutes: chrysene kms( 0 ), ksm ( O ); benzo[a]pyrene kms ( O ), ksm( O ).

125

9‘...”

00.0

no. m.00.. 0:35.00sz .
00.0 . 00.0 mm...” 0N0

b n h P

mfim

 

 

’
I

 

 

_.
. e w
I.
a
O
O
N
B
V
N
l.
u cow Sn.»
coop

 

0... 05y.

126

Table 4.4: Activation enthalpy and activation volume for selected PAHs on

polymeric octadecylsilica calculated at T=308 K and P=1466 psi.

 

 

 

 

AH” AHtm AV” Avtm
Solute
(kcal/mol)‘ (kcal/mol) (mL/mol)B (mUmol) .
cm 2711 19:1 73:13 72:17
Pic 1 71:23 54:23 139119 107013
BaP 26:3 17:3 26:23 15:27

 

 

 

 

 

 

 

A Activation enthalpies for the stationary phase to transition state (AHts), and
mobile phase to transition state (AHm) were calculated at T=308 K and P=1466
psi

B Activation volumes for the stationary phase to transition state (AV.), and
mobile phase to transition state (Av...) were calculated at T=308 K and P = 400

to 2566 psi

127

of benzo[a]pyrene and picene indicates that the more condensed solute has a
smaller activation enthalpy. Thus, the structure of the molecule affects the
energy needed to transfer between mobile and stationary phases. This general
conclusion is consistent with the results of Miyabe, Sotoura, and Guiochon [24]
who have estimated the activation energy for surface diffusion of the
alkylbenzenes on octadecylsilica to be in the range of 3.5 to 5.2 kcal/mol. These
values are significantly smaller than those reported herein because of the smaller
molecular size, the lower bonding density of the octadecylsilica, as well as the
different computational method used by Miyabe and coworkers. Nevertheless,
they confirm the importance of molecular structure in the kinetic processes. As
the activation enthalpies for the stationary to transition state (AHts), and the
mobile phase to transition state (AHm) in Table 4.4 are compared to one another
and to the change in molar enthalpy (AHsm) in Table 4.2, a trend emerges:
AH¢3>AH¢m>>AHsm. Thus, this trend reveals the inherent danger of drawing
conclusions about mechanism and the relative energies of transition from the
thermodynamic parameter alone. It should be noted that the difference between
the activation enthalpies is equivalent to the change in molar enthalpy, indicating
an internal consistency in the method of calculation.

In addition to the slope of the line, the intercept of the line contains
information about collisional frequency of the system. For a bimolecular system,
the pre-exponential factor (A), also known as the frequency factor, has values on
the order of 1013 [25]. For the data contained in this chapter, as well as the

following chapters, the pre-exponential factor ranges from 1016 to 102‘, implying a

128

larger number of collisions per unit time. The large differences between the
typical values and those shown here could result from the large concentrations of
the solutes, or a large contribution from the activation entropy (ASS, ASm), which
is a component of the pre-exponential factor. However, it should be noted that
the kinetic theory is derived for gas phase systems. Thus, the assumptions
inherent not only to the kinetic theory of gases, but also the differences between
the liquid and gas states should be recognized as likely contributions to the

differences in the values for the pre-exponential factor.

4.3.2.3 Activation Volume

Using Equation 1.20 and the methods described in Sections 1.3.2 and
2.3.1.2, the activation volumes were calculated. A representative graph of the
natural logarithm of the rate constant versus pressure is depicted in Figure 4.6
and the resulting values for the activation volume are presented in Table 4.4. For
chrysene and picene, an increase in ring number correlates with an increase in
the activation volume. The activation volume for benzo[a]pyrene is smaller than
that for picene, indicating once again that annelation structure has a significant
effect on the kinetics. Similar to the activation enthalpy, when the activation
volumes for the stationary to transition state (Avis), and the mobile phase to
transition state (AVm) in Table 4.4 are compared to one another and to the
change in molar volume (Av...) in Table 4.2, another trend emerges:
AV15>AVm>>AVsm. This trend indicates that the barrier for the transition event is
very large, although the overall change in molar volume may be small. Akin to

the activation enthalpies and the molar enthalpy, the difference between the.

129

Figure 4.6: Representative graph of the rate constants versus pressure used to
calculate the change in activation volume according to Equation 1.20. Column:
polymeric octadecylsilica. Mobile phase: methanol, 303 K, 0.08 cm/s. Symbols

defined in Figure 4.5.

130

coon

can. mmamwwmn.
800 080 80. 89 80

P

 

 

 

 

 

 

.053.

I
O
‘—

. cow

....

(,Js) .LNVLSNOO 31W

131

 

activation volumes and the molar volume are similar, demonstrating and internal
consistency in the method of calculation.

Similar to the thermodynamic parameters, the activation enthalpy and
activation volume are positively correlated for the planar PAHs. However, there

are insufficient data to determine whether or not this correlation is linear

4.4 Conclusions

To illustrate the full implications of the data presented above, it is helpful
to describe the entire retention event for a single solute. For example, as picene
transfers from the mobile to stationary phase, it passes through a high-energy
transition state. This transition state may be envisioned as a cavity created
within the stationary phase that incorporates the solute as well as some
associated solvent molecules. The creation of this transition state requires an
enthalpy change of 54 kcanoI and a volume change of 107 mUmol. Given
these large changes, it is likely that this transfer does not occur in a single step,
but in a series of discrete and sequential steps. As the system approaches
equilibrium, the solvent molecules are released and the alkyl chains rearrange to
optimally solvate the solute. As a result of this rearrangement, the net energy of
the system decreases by —15.8 kcal/mol and the volume decreases by -28.4
mL/mol. This transfer occurs with a rate constant of 118 s". As picene transfers
from the stationary to mobile phase, it passes through another high-energy
transition state. This transition requires an enthalpy change of 71 kcal/mol and a
volume change of 139 mL/mol. This large change suggests, once again, that the

transfer is characterized by a series of discrete steps. This transfer occurs with a

132

rate constant of 9 s", which is 13-fold slower than the transfer from mobile to
stationary phase.

This holistic view of the transition event illustrates that both the
thermodynamics and kinetics are necessary to fully describe the retention event.
Thus, previous investigations that have used the change in molar enthalpy to
describe the mechanism of retention are inherently flawed and highly suspect.
By using the necessary theoretical foundation and instrumental design, this
chapter demonstrates the first comprehensive study of retention for a series of
known toxicants/pollutants.

Throughout this chapter, and those that follow, the main driving force for
retention is described as the solvation of the nonpolar solute into the nonpolar
stationary phase. As noted by Ranatunga and Carr[23] the enthalpy of retention
(i.e. transfer from the mobile phase to stationary phase) is favorable and large.
The authors demonstrate that the contribution from the mobile phase is .
enthalpically unfavorable, but is overcome by the large enthalpy imparted by the
solvation of the solute into the lipophilic stationary phase. In addition, the entropy
is shown to contribute very little to the retention event. Thus, the contributions
from the mobile phase are not incorporated into the description of the retention

event.

133

4.5 References

[1 ]
[2]
[3]
[4]
[5]
[5]
[7]
[8]

[9]

[10]
[11]
[12]
[13]

[14]

[15]

[16]

[17]

[18]

LA. Cole, J.G. Dorsey, Anal. Chem. 64 (1992) 1317-1323.

K.B.~ Sentell, J.G. Dorsey, J. Chromatogr. 461 (1989) 193-207.

M. Olsson, L.C. Sander, S.A. Wise, J. Chromatogr. 477 (1989) 277-290.
L.C. Sander, M. Pursch, S.A. Wise, Anal. Chem. 71 (1999) 4821-4830.
K.B. Sentell, J.G. Dorsey, Anal. Chem. 61 (1989) 930-934.

L.A. Cole, J.G. Dorsey, J. Chromatogr. 635 (1993) 177-186.

L.C. Sander, S.A. Wise, Anal. Chem. 67 (1995) 3284-3292.

J.H. Park, Y.K. Lee, Y.C. Weon, LC. Tan, J. Li, L. Li, J.F. Evans, P.W.
Carr, J. Chromatogr. A 767 (1997) 1-10.

L.A. Cole, J.G. Dorsey, Anal. Chem. 64 (1992) 1324-1327.
M.H. Chen, C. Horvéth, J. Chromatogr. A 788 (1997) 51-61.
V.L. McGuffin, S.-H. Chen, J. Chromatogr. A 762 (1997) 35-46.
V.L. McGuffin, C. Lee, J. Chromatogr. A 987 (2003) 3-15.

L.C. Sander, S.A. Wise, J. Chromatogr. A 656 (1993)335-351.

A. Tchapla, S. Heron, E. Lesellier, H. Colin, J. Chromatogr. A 656 (1993)
81-1 12.

J.F. Wheeler, T.L. Beck, S.J. Klatte, L.A. Cole, J.G. Dorsey, J.
Chromatogr. A 656 (1993) 317-333.

K.B. Sentell, A.N. Henderson, Anal. Chim. Acta 246 (1991) 139-149.
L.C. Sander, S.A. Wise, CRC Crit. Rev. Anal. Chem. 18 (1987) 299-415.

L.C. Sander, S.A. Wise, Anal. Chem. 61 (1989) 1749-1754.

134

[19]

[20]

[21]

[22]

[23]

[24]

[25]

GE. Berendsen, L. de Galan, J. Chromatogr. 196 (1980) 21 -37.

MW. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5576-5584.

M.W. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5585-5592.

W. Melander, D.E. Campbell, C. Horvath, J. Chromatogr. 158 (1978) 215-
225.

R.P.J. Ranatunga, P.W. Carr, Anal. Chem. 72 (2000) 5679-5692.

K. Miyabe, S. Sotoura, G. Guiochon, J. Chromatogr. A 919 (2001) 231-
244.

SW. Benson, The Foundations of Chemical Kinetics, Robert E. Krieger
Publishing, Malabar, FL, USA, 1982.

135

Chapter 5: Retention of Polycyclic Aromatic
Hydrocarbons: Effect of Annelation

5.1 Introduction

As demonstrated in Chapter 4, the effect of ring number and planarity are
very important to the thermodynamics and kinetics of retention in liquid
chromatography. While a large number of studies have been carried out to study
the aforementioned parameters, fewer investigations have sought to characterize
the effect of annelation [1-5]. The effect of annelation is important since it can
affect the biological impact of one isomer over another.

Four-ring polycyclic aromatic hydrocarbon (PAH) isomers have been used
to study the effect of annelation structure with regard to stationary-phase bonding
density [3,4], mobile-phase composition [2], and temperature [4,5] in reversed-
phase systems. Sentell and Dorsey reported that selectivity increases with
increasing bonding density [3]. Chmielowiec and Sawatzky reported
thermodynamic values for three- and four-ring homologues, which indicate that
an increase in Iength-to-breadth ratio results in more negative changes in molar
enthalpy and molar entropy [2]. The more recent investigations by Sentell and
Henderson [4] and Sentell et al. [5] have used a homologous series of four-ring
PAHs to study the effect of sub-ambient temperature on retention and selectivity.
Their data indicate that the selectivity between these solutes increases at lower
temperature. This increased selectivity is attributed to the larger surface area for
interaction that results from the higher order of the alkyl chains near the proximal

terminus, as purported by Stalcup et al. [6]. In addition to four-ring PAHs, six-ring

136

PAHs have also been used as probes of reversed-phase systems. Most notably,
phenanthro[3,4-c]phenanthrene and tetrabenzonaphthalene have been used to
examine the effect of planarity on retention [1 ,6-10]. However, without the
comparison to other six-ring compounds, these nonplanar PAHs provide little
direct information about the effect of annelation structure.

To date, there have been no systematic studies that consider both the
thermodynamic and kinetic behavior as a function of annelation structure. In the
present study, a series of four-ring PAHs with planar and nonplanar structures
are separated on a polymeric octadecylsilica stationary phase with a methanol
mobile phase. Using a temperature range from 273 to 303 K and an average
pressure range from 585 to 3585 psi, the thermodynamic and kinetic behavior
are characterized by previously established methodology [11-13] detailed in
Chapters 1,2 and 4. The retention factors are calculated, together with the
concomitant changes in molar enthalpy and molar volume, to characterize the
thermodynamic behavior. The rate constants are calculated, together with the
concomitant changes in activation enthalpy and activation volume, to
characterize the kinetic behavior. These data provide a holistic view of the effect
of annelation structure on the retention mechanism in reversed-phase liquid

chromatography.

137

5.2 Experimental Methods

5.2.1 Solutes

As depicted in Figure 5.1, four polycyclic aromatic hydrocarbons have
been chosen to investigate the effect of annelation on the thermodynamics and
kinetics of retention. Pyrene, benz[a]anthracene, chrysene, and
benzo[c]phenanthrene (Sigma-Aldrich) were obtained as solids and dissolved in
high-purity methanol (Burdick and Jackson, Baxter Healthcare) to yield standard
solutions at a concentration of 10“ M. A nonretained marker, 4-methylhydroxy-7-
methoxycoumarin, was added to each solution at a concentration of 10“ M. To
ensure solubility, the solutions were equilibrated at each temperature prior to

analysis.

5.2.2 Experimental System

The system used to study the effect of annelation is depicted in Figure 2.2.
The mobile phase was pure methanol, and the stationary phase polymeric
octadcecylsilica (5.4 pmol/mz). The temperatures for these experiments were
273, 283, 288, 293 and 303 K. The inlet pressures for these experiments were
1000, 1750, 2500, 3250 and 4000 psi. These values correspond to average
pressures of 585, 1335, 2085, 2835, and 3585 psi, respectively, which were
calculated assuming a linear pressure drop along the column. The data were
analyzed using the exponentially modified Gaussian (EMG), non-linear
chromatography (NLC) and bi-exponentially modified Gaussian (E2MG)

functions, which are described in section 2.3.1.

138

Pyre ne Benz[a]anthracene

 

Benzo[c]phenanthrene

Figure 5.1: Structure of polycyclic aromatic hydrocarbons used to study the

effect of annelation in polymeric octadecylsilica.

139

SC

fa

5.3 Results and Discussion

5.3.1 Thermodynamic Behavior

5.3.1.1 Retention Factor

Using Equations 1.2 and 2.5, the retention factor was calculated for each
solute at each temperature and pressure. Representative values of the retention
factor are summarized in Table 5.1. It is apparent that the retention factor for all
PAHs decreases with increasing temperature and increases with increasing
pressure. Retention also decreases as a function of the length-to-breadth ratio
(i.e., pyrene < benz[a]anthracene < chrysene), with the exception of
benzo[c]phenanthrene. As noted previously, benzo[c]phenanthrene is slightly
nonplanar due to steric hindrance of the hydrogen atoms in the bay region.
When compared to data reported in Chapter 4 for nonplanar six-ring PAHs [12],
the retention factor for benzo[c]phenanthrene is larger than those for
phenanthro[3,4-c]phenanthrene and tetrabenzonaphthalene. This suggests that
the smaller degree of nonplanarity of benzo[c]phenanthrene allows for greater
interaction with the stationary phase. Moreover, an increase in pressure causes
a slight increase in retention for benzo[c]phenanthrene, but causes a decrease in
retention for phenanthro[3,4-c]phenanthrene and tetrabenzonaphthalene.

Hence, compression of the stationary. phase causes greater interaction of the
alkyl chains with benzo[c]phenanthrene, similar to the planar solutes, but causes

expulsion the other nonplanar solutes.

140

Table 5.1: Retention factors for PAH isomers.

 

 

 

 

 

 

k B k C
S°'”teA 273 K 303 K 535 psi 3535 psi
Pyr 1.89 0.84 1.31 1.36
BaA 5.33 1.45 2.99 3.25
Chr 3.12 1.33 4.20 4.54
BcP 1.59 0.30 1.19 1.21

 

 

 

 

 

 

 

A Pyr = pyrene, BaA= benz[a]anthracene, Chr = chrysene,
BcP = benzo[c]phenanthrene
B Retention factor (k) calculated at 3585 psi

C Retention factor (k) calculated at 283 K

141

5.3.1 .2 Molar Enthalpy

A representative graph of the logarithm of the retention factor versus the
inverse temperature is shown in Figure 5.1. The change in molar enthalpy is
calculated from the slope of this graph, according to Equation 1.5. The graph for
each PAH is linear (R2 = 0.991 — 0.999) and the slope is positive. A positive
slope is indicative of a negative change in molar enthalpy and suggests that the
transfer from mobile to stationary phase is enthalpically favorable. As shown in
Table 5.2, the most condensed of the planar solutes, pyrene, has the smallest
change whereas the least condensed, chrysene, has the greatest change in
molar enthalpy. These differences in molar enthalpy can be attributed to the
depth that each PAH penetrates into the stationary phase. The proximal regions,
where the alkyl group is bound to the silica surface, are highly ordered with all
trans carbon-carbon bonds. As the distance from the surface increases, there
are more gauche bonds and greater disorder [14,15]. The more condensed
PAHs, such as pyrene, probe only the distal regions, whereas less condensed
PAHs, such as chrysene, penetrate more deeply into the ordered regions of the
stationary phase. The change in molar enthalpy becomes more negative the
farther the PAH penetrates into the stationary phase. It is noteworthy that
benzo[c]phenanthrene exhibits changes in molar enthalpy that are smaller than
the other PAHs, even though it is not the most condensed. It is hypothesized
that the nonplanar character of benzo[c]phenanthrene does not allow it to
penetrate as deeply into the high-density polymeric stationary phase. This effect

has been noted previously with other nonplanar PAHs [4,5,7-9]. The molar

142

Figure 5.2: Representative graph of the retention factor versus inverse
temperature used to calculate the change in molar enthalpy according to

Equation 1.5. Column: polymeric octadecylsilica. Mobile phase: methanol, 3585
psi, 0.08 cm/s. Solutes: pyrene ( O ), benz[a]anthracene ( I ),

chrysene ( A ), benzo[c]phenanthrene ( O ).

143

on...”

mod

_

om...”

b

.7: 3-2 0.. 0:350:20»:

mmd

omd

-

mvd

ofm

-

mad

om.m

-

mud

 

 

 

...o

 

r1 l I

 

0.0 050.:

or

HOLOVd NOLLNBlElU

144

Table 5.2: Molar enthalpy and molar volume for PAH isomers

 

 

 

 

 

 

 

 

Solute AHsm (kcal/mol)A AVsm (mL/mol)B
Pyr —4.4 :l: 0.2 —1.9 i 1
BaA -7.1 i 0.2 -9.6 i 1
Chr -8.2 i 0.2 -11.7 :t 1
BOP -3.8 :t 0.2 -1.3 :1: 1

 

A Molar enthalpy (AHsm) Calculated at 3585 psi

3 Molar volume (AVsm) calculated at 283 K

145

 

enthalpies reported herein are consistent with those for other PAHs in Chapter 4

[12].

5.3.1.3 Molar Volume

A representative graph of the logarithm of the retention factor versus
pressure is shown in Figure 5.2. The change in molar volume is calculated from
the slope of this graph, according to Equation 1.7. Again, the graph for each
PAH is linear (R2 = 0.984 - 0.998) and the slope is positive. A positive slope is
indicative of a negative change in molar volume and suggests that the PAH
occupies less volume in the stationary phase than in the mobile phase. As
shown in Table 5.2, the most condensed of the planar solutes, pyrene, has the
smallest change whereas the least condensed, chrysene, has the greatest
change in molar volume. These differences in molar volume, like the changes in
molar enthalpy, can be attributed to the depth that each PAH penetrates into the
stationary phase. The more condensed PAHs, such as pyrene, probe only the
distal regions, whereas less condensed PAHs, such as chrysene, penetrate more
deeply into the ordered regions of the stationary phase. The change in molar
volume becomes more negative the farther the PAH penetrates into the
stationary phase. Again, it is noteworthy that the nonplanar solute,
benzo[c]phenanthrene, has a smaller change in molar volume than the planar
solutes. This is consistent with a smaller depth of penetration, as discussed
above. However, it is not as extreme as for the six-ring nonplanar PAHs,

phenanthro[3,4-c]phenanthrene and tetrabenzonaphthalene, which exhibit

146

Figure 5.3: Representative graph of the retention factor versus pressure used to
calculate the change in molar volume according to Equation 1.7. Column:
polymeric octadecylsilica. Mobile phase: methanol, 283 K, 0.08 cm/s. Symbols

defined in Figure 5.1.

147

0000

.72.. 0:0000::

 

 

 

 

 

comm coon comm ooou com F coop com
o 61 0 o
91 191 11 1? 6
III l ill a
01 101 10

 

 

 

0... 050.

HOlOVfl NOLLNHLBH

148

positive changes in molar volume [12,16]. The molar volumes reported herein

are consistent with those in Chapter 4 [12].
5.3.2 Kinetic Behavior

5.3.2.1 Rate Constant

Although the thermodynamic data demonstrate the steady-state aspects
of chromatographic behavior, they do not fully explain the retention mechanism.
Using the equations and methods developed in Chapters 1 and 2, the pseudo-
first-order rate constants, activation enthalpies, and activation volumes were
calculated. These values help to quantify the kinetic aspects of PAH transfer
between the mobile and stationary phases as a function of the annelation
structure.

Representative values of the rate constants, calculated using Equations
1.16 and 1.17, are summarized in Tables 5.3 and 5.4. The most condensed of
the planar solutes, pyrene, has the largest rate constants, whereas the least
condensed, chrysene, has the smallest rate constants for transport from mobile
to stationary phase (ksm) and from stationary to mobile phase (km). However,
the nonplanar solute, benzo[c]phenanthrene, has larger rate constants than the
planar solutes. As for the solutes in Chapter 4, the transition from the stationary
to mobile phase is the rate-limiting step for all of the four-ring PAHs. As shown in
Table 5.3, the rate constants for all PAHs increase with increasing temperature.
This behavior is a consequence of the increased diffusion coefficients and the

enhanced fluidity of the stationary phase. As more kinetic energy is imparted,

149

Table 5.3: Rate constants for PAH isomers calculated by using the

exponentially modified Gaussian (EMG) model as a function of temperature

 

 

 

 

 

 

 

 

 

 

 

kms (5.1) A Km (5.1) A
S°'”t° 233 K 293 K 233 k 293 K
Pyr 193 1137 269 1179
BaA 74 521 241 1099
Chr 7 53 34 163
BcP 216 1360 262 1330

 

A Rate constants from stationary to mobile phase (km) and from mobile to

stationary phase (ksm) calculated at 3585 psi

150

 

Table 5.4:

Rate constants for PAH isomers calculated by using the

exponentially modified Gaussian (EMG) model as a function of pressure

 

 

 

 

 

 

 

 

 

 

 

k... (s") A k... (5‘) "
Solute 585 psi 3585 psi 585 psi 3585 psi
Pyr 313 198 f 411 269
BaA 73 62 224 202
Chr 14 7 61 34
BcP 330 216 395 , 262

 

A Rate constants from stationary to mobile phase (km) and from mobile to

stationary phase (ksm) calculated at 283 K

151

 

the alkyl chains become more labile and can more readily undergo rotation of the
carboncarbon bonds from the trans to gauche conformation. In fact, this
polymeric octadecylsilica stationary phase is known to undergo a phase
transition in the vicinity of 318 K [13,16]. This increased lability enables the
PAHs to diffuse in and out of the stationary phase more freely. As shown in
Table 5.4, the rate constants for all PAHs decrease with increasing pressure.
This behavior is a consequence of the compression of the alkyl chains, which

impedes the PAH diffusion in and out of the stationary phase.

5.3.2.2 Activation Enthalpy

A representative graph of the logarithm of the rate constant versus the
inverse temperature is shown in Figure 5.3. The activation enthalpy is calculated
from the slope of this graph, according to Equation 1.20. As shown in Table 5.5,
there are slight differences that indicate the most condensed solute, pyrene, has
the smallest activation enthalpies whereas the least condensed solute, chrysene,
has the greatest activation enthalpy. However, given the uncertainties in these
values, the differences are not statistically significant. The similarity in values
implies that the activation enthalpy is largely independent of annelation structure,
and suggests that the PAHs encounter a similar barrier at the interface between
the mobile and stationary phases. As the activation enthalpies for the stationary
to transition state (AHts), as the mobile phase to transition state (AHm) are

compared to one another and to the change in molar enthalpy in(AHsm) Table

5.2, a trend emerges: AH. > AHm >> AHsm. Thus, the enthalpic barrier for the

152

Figure 5.4: Representative graph of the logarithm of the rate constant (kms)
versus inverse temperature used to calculate the activation enthalpy according to
EqUation 1.20. Column: polymeric octadecylsilica. Mobile phase: methanol,

3585 psi, 0.08 cm/s. Symbols defined in Figure 5.1.

153

Figure 5.4

 

 

 

IIIIII I I [IIIIfI I I [IIIIII I I

O
O

‘—

10000
10 0

(,_s) .LNVLSNOO ELLVH

154

[I'VIIFTE

O ‘-

‘-

 

3.30 3.35 3.40 3.45 3.50 3.55 3.60 3.65 3.70

3.25

”TEMPERATURE ( x 10'3 K")

Table 5.5: Activation enthalpies for PAH isomers

 

 

 

 

 

Solute AH... (kcanol) A AH... (kcanol) A
Pyr 24 i 3 19 i 3
BaA 25 i 4 17 i 3
Chr 31 :1: 2 23 a: 2
BcP 27 i 3 23 i 3

 

 

 

 

 

A Activation enthalpies from stationary phase to transition state (AHts) and from

mobile phase to transition state (AHm) calculated at 283 K and 3585 psi

155

transition is significantly greater than the thermodynamic change in molar

enthalpy. This result is consistent with the data in Chapter 4.

5.3.2.3 Activation Volume

A representative graph of the logarithm of the rate constant versus
pressure is shown in Figure 5.4. The activation volume is calculated from the
slope of this graph, according to Equation 1.20. In contrast to the activation
enthalpies, the activation volumes are a statistically significant function of the
annelation structure. As shown in Table 5.6, the most condensed solute, pyrene,
has the smallest activation volume whereas the least condensed solute,
chrysene, has the greatest activation volume. The nonplanar solute,
benzo[c]phenanthrene has a smaller activation volume than the planar chrysene,
but can not be statistically differentiated from pyrene. Similar to the enthalpic
trends, the activation volumes for the stationary to transition state (Avg) and the
mobile phase to transition state (Avm), when compared to the change in molar
volume (AVsm) demonstrate the trend of AV.s > Av... >> AVsm. Hence, the
volumetric barrier for the transition is large, even though the overall change in
molar volume is relatively small. This conclusion conforms to the data discussed

in Chapter 4.

156

Figure 5.5: Representative graph of the rate constant (kms) versus pressure
used to calculate the activation volume according to Equation 1.20. Column:
polymeric octadecylsilica. Mobile phase: methanol, 283 K, 0.08 cm/s. Symbols

defined in Figure 5.1.

157

Figure 5.5

 

1000
j

ooov

oomm

P

ooom

oomN

L

0...0. 0:0000::
0000

oomv

ooow

-

oom

-

 

 

@0059.

[IIFI I I

O
,_

oow

 

' Ijir I l

coo .

(,-s)1Nv19Noo 31W

158

Table 5.6: Activation volumes for PAH isomersA

 

 

 

 

 

 

Solute AV“ (mL/mol) AV;m (mUmol)
Pyr 47 i 7 43 i 7
BaA NR '3 NR

Chr 69 :l: 9 57 i 9
BcP 38 i 13 37 i 13

 

 

 

 

A Activation volumes from stationary phase to transition state (Avg) and from
mobile phase to transition state (Av...) calculated at 283 K

8 Values for benz[a]anthracene are not statistically reliable (NR) owing to large

relative standard deviation (>100 °/o RSD).

159

 

th
nc
te;
re:
the
de'
fell
to e
we:
dfifl

EM]

mec
”0111
with

the S
“Flea;
USed
This r
lesidL

exhibi

5.3.2.4 Kinetic Deviations

In the studies presented above, the PAH zone profiles were fit very well by
the EMG equation, with typical R2 values of 0.99 or greater. However, it was
noted that the quality of fit was degraded slightly for some PAHs at the lowest
temperature (273 K), with R2 values of 0.98 - 0.99 and small nonrandom
residuals along the'tailing edge of the zone profile. Figure 5.5 illustrates a profile
that exhibits this high R2, but poor fit using the EMG equation. As these
deviations from the EMG model may potentially arise from a change in the
retention mechanism, more detailed investigation was warranted. First, in order
to ensure that this effect did not arise from limited solubility, the PAH solutions
were equilibrated at 273 K prior to injection. No statistically significant
differences were observed in the PAH zone profiles or the quality of fit to the
EMG equation.

To examine whether the low temperature causes a change in the retention
mechanism, two alternative models were examined. The first alternative was the
nonlinear chromatography model. The retention factors for the PAHs increase
with decreasing temperature (Table 5.1), thereby increasing the concentration in
the stationary phase. Hence, the isotherm might be expected to deviate from
linearity at low temperature. Accordingly, the NLC model (Equation 2.6) was
used to iteratively fit each of the PAH zone profiles that showed deviant behavior.
This model, too, gave R2 values of 0.98 — 0.99 and exhibited small nonrandom
residuals from the observed zone profiles. Figure 5.6 illustrates a profile that

exhibits this high R2, but poor fit using the NLC equation. As shown in Table 5.7,

160

Figure 5.6: Representative graph from the nonlinear regression of experimental
data using the exponentially modified Gaussian equation (EMG). The points
represent collected data, while the line results from the predicted values from the

EMG equation. Solute: chrysene, 273 K, 585 psi

161

Am. m2...
oooo oooo oooo oomo oooo oomo ooco oomo ooNo

» P p p r _ b r

005

oooo

 

 

 

 

0.0 0.00..

BONSOSEHOD‘H

162

Figure 5.7: Representative graph from the nonlinear regression of experimental
data using the non-linear chromatography equation (NLC). The points represent
collected data, while the line results from the predicted values from the NLC

equation. Solute: chrysene, 273 K, 585 psi

163

oooo

oooo

F

Am. mi:

coho oooo

_

oomo

L

oovo

P

oomo

1—

oomo

oowo

oooo

 

 

 

 

.0 0.00.:

BONSOSHHOD'H

164

Table 5.7: Rate constants for PAH isomers calculated by using

 

 

 

 

 

 

 

the EMG and NLC models

EMG A NLC A
S°'“‘° 10.0") 10. (8") RA A 10.0") k... 6") RA A
Pyr 33 69 0.996 12 23 0.999
BaA 12 64 0.994 7 37 0.999
cm 1 1 1 0.937 6 44 0.993
BcP 52 34 0.997 12 20 0.999

 

 

 

 

 

 

 

 

A Rate constants from stationary to mobile phase (km) and from mobile to
stationary phase (ksm) calculated at 273 K and 3585 psi

8 Square of correlation coefficient for nonlinear regression (R2)

165

the quality of fit for the NLC and EMG models was similar. The rate constants
are of the same order of magnitude, but those determined by the NLC model are
somewhat smaller than those determined by the EMG model for pyrene,
benz[a]anthracene, and benzo[c]phenanthrene but somewhat larger for
chrysene. This difference is a result of the peak shape of chrysene, which
demonstrates more asymmetry than the other solutes.

As neither the EMG nor NLC model described the data well, a third model,
the bi-exponentially modified Gaussian equation, was examined. This model is
appropriate if there are two sites in the stationary phase with significantly
different kinetic behavior. This behavior may arise from two partition sites or
partition and adsorption sites. Accordingly, the EMG model (Equation 2.9) was
used to iteratively fit each of the PAH zone profiles that showed deviant behavior.
As shown in Table 5.8, this equation also had difficulty in fitting the zone profiles.
For chrysene, the regression failed owing to overflow and underflow errors.
Figure 5.7 demonstrates the visual aspects of this failure that result in a
truncation of the regressed line. For the other PAHs, the EMG equation forces
T1 and 12 to assume values very close to one another. The rate constants are the
same for both sites, but are larger than those calculated using either the EMG or
NLC models. These values imply that the kinetic sites are equally distributed and
the same in their energetic natures. Thus, Tables 5.7 and 5.8 indicate that
neither the NLC nor the E2MG model provides a more accurate description of the

solute zone profiles than the EMG model.

166

Table 5.8:

Rate constants for PAH isomers calculated by using the bi-

exponentially modified Gaussian (EZMG) model for two sites

 

 

 

 

 

 

Site 1 A Site 2 32' '3
Solute kms (5-1) ksm (5,1) km (3“) ksm (5“)
Pyr 79 143 79 143 0.999
BaA 32 171 32 171 0.993
Chr NR ° NR NR NR -
BcP 24 37 24 37 0.993

 

 

 

 

 

 

 

 

A Rate constants from stationary to mobile phase (km) and from mobile to
stationary phase (ksm) calculated at 273 K and 3585 psi

8 Square of correlation coefficient for nonlinear regression (R2)

C Values for chrysene are not statistically reliable (NR) owing to overflow and

underflow errors in nonlinear regression.

167

Figure 5.8: Representative graph from the nonlinear regression of experimental
data using the biexponentially modified Gaussian equation (E2MG). The points
represent collected data, while the line results from the predicted values from the

E2MG equation. Solute: chrysene, 273 K, 585 psi

168

oooo

oooo

p

oooo

17

oowo

L

A0. m2...
oooo oomo

_ b

oovo

_

oomo

-

ooNo

p

oowo

_

oooo

 

 

ill.‘

//

 

 

0.0 0.00.00

30N30$380n1d

169

5.4 Conclusions

In this study, the thermodynamic and kinetic behavior of four-ring PAHs
was examined in reversed-phase liquid chromatography. Even though the
changes in molar enthalpy and molar volume are relatively small, the kinetic data
suggest that the barrier for transition between mobile and stationary phases is
large. For a solute like chrysene, the activation enthalpy is 23 kcanol for entry
into the stationary phase, and 31 kcanol for exit from the stationary phase.
There is a similarly large activation volume for each transition of 69 and 57
mL/mol, respectively. These relatively large values for activation enthalpy and
activation volume suggest that the transition is more likely to occur in a series of
discrete steps rather than in a single step. When the values for chrysene are
compared to those for pyrene, it is apparent that the annelation structure
influences the energy and volume contributions to the retention mechanism.

In addition, the anomalous behavior at 273 K suggests that the retention
mechanism may change from a partition to an adsorption or mixed-mode
mechanism. It is important to emphasize that this behavior is not apparent from
the thermodynamic (steady state) measurements. The kinetic measurements
are essential to make inferences about the mechanism. It is also important that
these measurements be derived concurrently from the same data, so that the
thermodynamic and kinetic measurements are internally consistent. From the
examined models, there is no conclusive evidence for a nonlinear isotherm or for

multiple sites with different kinetic behavior. The lack of data for a multiple site

170

model does not imply that such a mechanism is not present; simply that it is

immeasurable using the current mathematical model.

171

5.5 References

[1]

[2]

[3]

[4]

[5]

[5]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

SR. Cole, J.G. Dorsey, J. Chromatogr. 635 (1993) 177-186.

J. Chmielowiec, H. Sawatzky, J. Chromatogr. Sci. 17 (1979) 245-252.
K.B. Sentell, J.G. Dorsey, J. Chromatogr. 461 (1989) 193-207.

K.B. Sentell, A.N. Henderson, Anal. Chim. Acta 246 (1991) 139-149.

K.B. Sentell, N.l. Ryan, A.N. Henderson, Anal. Chim. Acta 307 (1995)
203-213.

A.M. Stalcup, D.E. Martire, S.A. Wise, J. Chromatogr. 442 (1988) 1-13.
L.C. Sander, S.A. Wise, J. Chromatogr. A 656 (1993) 335-351.

L.C. Sander, S.A. Wise, Anal. Chem. 67 (1995) 3284-3292.

L.C. Sander, M. Pursch, S.A. Wise, Anal. Chem. 71 (1999) 4821 -4830.

J. Wegmann, K. Albert, M. Pursch, L.C. Sander, Anal. Chem. 73 (2001)
1814-1820.

S.B. Howerton, C. Lee, V.L. McGuffin, Anal. Chim. Acta 478 (2003) 99-
1 10.

SB. Howerton, V.L. McGuffin, Anal. Chem. 75 (2003) 3539-3548.
V.L. McGuffin, C. Lee, J. Chromatogr. A 987 (2003) 3-15.

C.J. Orendorff, M.W. Ducey, Jr., J.E. Pemberton, J. Phys. Chem. A 106
(2002) 6991 -6998.

172

[15] MW. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5576-5584.

[16] V.L. McGuffin, S.-H. Chen, J. Chromatogr. A 762 (1997) 35-46.

173

Chapter 6: Thermodynamic and Kinetic
Characterization of Nitrogen-Containing
Polycyclic Aromatic Hydrocarbons in Protic
and Aprotic Mobile Phases

6.1 Introduction

As demonstrated in Chapters 4 and 5, the effect of ring number and
annelation are very important to the thermodynamics and kinetics of retention in
liquid chromatography for polycyclic aromatic hydrocarbons (PAHs). This
chapter preSents similar measurements for a series of nitrogen containing
polycyclic aromatic hydrocarbons.

At elevated temperatures, PAHs are capable of reacting to form
compounds that contain substituent moieties that alter their toxicity. Two
substitutions that lead to highly toxic analogues involve the exchange of a carbon
with nitrogen to form an azaarene, and the replacement of hydrogen with an
amino or nitro group. These solutes are collectively known as nitrogen-
containing polycyclic aromatic hydrocarbons (NPAHs). NPAHs are commonly
found in fossil fuels and their derivatives, though some form during combustion
related events.

Relative to their parent PAH, NPAHs are more soluble in aqueous
environments. Increased water solubility results in an increased potential for
harm. For example, 1-aminopyrene has been reported to demonstrate a fifty-fold
increase in mutagenic activity when compared to the parent PAH [1]. Similarly,
azaarenes demonstrate increased carcinogenicity when compared to their parent

compounds [2,3].

174

Studies have been carried out to determine the presence of NPAHs in
fossil fuels [4-8], coal substitutes [9], lake sediment [10], urban aerosols [11], and
polymer degradation products [12]. In order to identify and differentiate
nitrogenous-PAHs, many investigators have used chromatographic techniques
due to the high resolving power. These techniques have included gas [8.13.14],
thin layer [15], supercritical fluid [16], and liquid chromatography [4-7,12,13,17-
22]. However, in nearly all instances the chromatographic applications have
focused only upon the optimization of the separation, and not a detailed probe of
the molecular contributions to retention. Colin and co-workers noted the large
amount of asymmetry in reversed-phase systems, and postulated that this tailing
results from the presence of unreacted silanols or the slow equilibrium between
the protonated and nonprotonated forms of the NPAHs [17]. Although the
retention of these solutes has been studied extensively, no quantitative
explorations of these solutes in reversed-phase liquid chromatography have
been published. To overcome this dearth of information, the thermodynamics
and kinetics of retention in RPLC for a series of NPAHs are presented.

Using the previously established methodology for the analysis of PAHs
[23,24] described in Chapters 1,2, and 4, the thermodynamics and kinetics of
retention are studied as a function of temperature, pressure, and mobile phase.
By using protic (methanol) and aprotic (acetonitrile) solvents, the effect of mobile
phase can be quantitated. This analysis provides insight into the mechanism of

retention for these highly toxic compounds.

175

6.2 Experimental Methods

6.2.1 Solutes

As depicted in Figure 6.1, five NPAHs were chosen to study the effect of
position, annelation, and ring number on the thermodynamics and kinetics of
retention. 1-Aminopyrene (Sigma-Aldrich), 1-azapyrene, 4-azapyrene,
benz[a]acridine, and dibenz[a,j]acridine (lnstitiit fiir PAH Forschung) were
obtained as solids and dissolved in high-purity methanol and acetonitrile (Burdick
and Jackson, Baxter Healthcare) to yield standard solutions at a concentration of
10“ M. A nonretained marker, 4-methylhydroxy-7-methoxycoumarin,[25] was

added to each solution at a concentration of 10" M.

6.2.2 Experimental System

The system used to study the NPAHs is depicted in Figure 2.2. The
mobile phases were methanol and acetonitrile, and the stationary phase
polymeric octadcecylsilica (5.4 umol/mz). The temperatures for these
experiments were 283, 288, 293, 298 and 303 K for methanol, and 298, 303,
308, 313, and 323 K for acetonitrile. The inlet pressure for the methanol
experiments were 1000, 1750, 2500, 3250 and 4000 psi. These values
correspond to average pressures of 585, 1335, 2085, 2835, and 3585 psi,
respectively, which were calculated assuming a linear pressure drop along the

column. The inlet pressure for the acetonitrile experiments was 2500 psi, which

176

1 -aminopyre ne

 

Be nz[a]acridine Dibenz[a,/]acridine
Figure 6.1: Structure of several nitrogen containing polycyclic aromatic

hydrocarbons

177

corresponds to an average pressure of 2175 psi, which was calculated
assuming a linear pressure drop along the column. The data were analyzed
using the exponentially modified Gaussian (EMG) and the non-linear

chromatography (NLC) equations, which are described in section 2.3.1.

6.3 Results and Discussion

6.3.1 Peak Profiles

For all of the analyses described below, the EMG function was used.
The overall quality of fit for the data collected using a methanol mobile phase
was high with random residuals, large correlation coefficients (R2 > 0.98), and
large F-statistics (F > 1000). The zone profiles (Figure 6.2A) are similar to
those observed in previous studies [23,24]. However, the zone profiles for the
data collected using acetonitrile are different, demonstrating a much larger
degree of asymmetry (Figure 6.2B). This asymmetry likely results from a
change in mechanism from partition to either an adsorptive or mixed mode (i.e.
partition/adsorption). As a result, several zone profiles in acetonitrile were
analyzed by using the NLC model. However, the NLC model failed to produce
any improvement in the quality of fit (R2 > 0.95, F > 1000) and was not used for
comprehensive analysis of the data.

Figure 6.3 illustrates the difference between a peak fit with the EMG
(6.3A) and the NLC (6.3B) functions for 4-azapyrene in acetonitrile. As

illustrated, neither function demonstrates a high degree of visual correlation

178

Figure 6.2: Representative chromatograms of 4-azapyrene in A) methanol, 303
K 2085 psi, 0.08 cm/s and B) acetonitrile, 303 K, 2175 psi, 0.08 cm/s. Column:

polymeric octadecylsilica. The inset is an expansion of the methanol

chromatogram.

179

 

Figure 6.2

 

 

 

ooo

oom

3...... wsE.

oov com com oow

1» _ »

 

 

 

 

mm

 

on

mm

€25 m2...

oN

.

 

   

or or m o

. r b I 111114.

 

111

 

 

rill» l.1ll1lL

 

 

 

 

 

 

 

0.0 0.00.:

180

Figure 6.3: Representative chromatogram of 4-azapyrene in acetonitrile fit with

the A) exponentially modified Gaussian equation and the B) non-linear

chromatography equation.

181

 

 

FigUre 6.3

 

 

 

 

 

 

 

 

 

0.0 0.00:

182

along the entirety of the peak. Numerically, the statistics of fit for both models
are large: EMG (32 = 0.935, F = 52500), NLC (32 = 0.960, F = 13900). However,
a comparison of Figure 6.3 and the statistics of fit demonstrate the danger in
relying upon the statistics alone when analyzing peak profiles with either of these

models.
6.3.2 Methanol Mobile Phase
6.3.2.1 Thermodynamic Behavior

6.3.2.1.1 Retention Factor

Representative values of the retention factor for the NPAl-ls in methanol
are summarized in Table 6.1. The retention factors for the NPAHs are smaller
than the parent PAHs [23]. At 303 K, 1-azapyrene and 4-azapyrene are eluted
38% and 44% faster than pyrene, respectively. Similarly, the retention of
benz[a]acridine is 12% faster that benz[a]anthracene. This decreased retention
in methanol is due to the increased solubility of the NPAHs relative to the PARS.
Since a partition mechanism is driven by differences in solubility [26], the more
soluble NPAHs are less retained.

As demonstrated, the retention factor for all solutes decreases with
increasing temperature. However, the retention factor increases with increasing
pressure for benz[a]acridine and dibenz[a,j]acridine, but remains statistically
invariant for the substituted pyrenes. The reasons for this behavior are

discussed below in further detail.

183

Table 6.1: Retention factors for NPAHs in methanol

 

 

 

 

 

 

 

 

ka kb
Solute a
288 K 303 K 585 psi 3585 psi

1-Aminopyrene 0.94 0.22 0.35 0.35
4-Azapyrene 0.78 0.47 0.79 0.78
1-Azapyrene 1.12 0.52 0.95 0.94
Benz[a]acridine 1 .02 0.58 0.97 1 .03
Dibenz[a,/]acridine 3.26 1 .51 2.96 3.26

 

 

 

 

 

a Retention factor (k) calculated at 3585 psi

b Retention factor (k) calculated at 288 K

184

 

6.3.2.1.2 Molar Enthalpy

A representative graph of the logarithm of the retention factor versus the
inverse temperature is shown in Figure 6.4. The data for methanol are depicted
using solid lines. The graph for each solute is linear (R2 = 0.986 - 0.999) and the
slope is positive. A positive slope is indicative of a negative change in molar
enthalpy and suggests that the transfer from mobile to stationary phase is
enthalpically favorable. The change in molar enthalpy is calculated from the
slope of Figure 6.4, according to Equation 1.5.

As shown in Table 6.2, the changes in molar enthalpy are the least
negative for 1-aminopyrene. 1-Azapyrene and 4-azapyrene have changes in
molar enthalpy that are similar, while benz[a]acidine is more negative than both.
Dibenz[a,j]acridine, the NPAH with the most rings, demonstrates the most
negative change in molar enthalpy. 1-Aminopyrene has the smallest change in
molar enthalpy because it is more polar than the azapyrenes. This increased
polarity increases the solubility in the mobile phase resulting in a smaller
enthalpy. The changes in molar enthalpy for the other solutes follow the same
trends as observed for the parent PAHs [23,24]. These tends indicate that the
change in molar enthalpy becomes more negative with increasing ring number
and decreasing condensation.

When compared to the parent PAHs, the changes in molar enthalpy are 1
to 2 kcanol more negative for the azapyrenes, and ~0.5 kcanol more negative
for benz[a]acridine, even though the retention factors are smaller for the NPAHs.

The more negative molar enthalpy for the NPAHs likely results from interactions

185

Figure 6.4: Representative graph of the retention factor versus inverse

temperature used to calculate the change in molar enthalpy. Mobile phases:

methanol, 2085 psi, 0.08 cm/s (—), acetonitrile, 2175 psi, 0.08 cm/s (— — —).

Solutes: 1-aminopyrene ( O ), 1-azapyrene (O ), 4-azapyrene ( O ),

benz[a]acridine ( I ), dibenz[a,j]acridine ( A ).

186

 

 

 

 

 

Figure 6.4

 

100
/

 

ood

.3. 00. x. 0:02:0020: .

 

 

 

 

on... ovd omd oud o_..m oo.m
. . _ . _ Poo
...... 0 I-HO
..... 01.111.
0 nnnnnnnn o..-
\‘
\1 IP
0001.
..... o 1 .
..... 0... I111
....l.... ...... I. ............... 4
........ 4........
I. . .4 ........ 4....
4.. .0.
4.00%.“. oow
M 1...,

HOLOVd NOLLNBJSH

187

Table 6.2: Molar enthalpy and molar volume for NPAHs in methanol

 

 

 

 

 

 

Solute AHsm (kcanol) A AVsm (mL/mol) 5
1-Aminopyrene -5.0 :t 0.2 —1.7 :l: 2°
4-Azapyrene -5.6 i 0.2 -1.5 i 5°
1-Azapyrene -5.7 i 0.3 —1.0 :1: 4°
Benz[a]acridine —6.5 i 0.1 -6.8 1: 1°
Dibenz[a,j]acridine -8.9 i 0.1 —10.7 1: 1°

 

 

 

 

 

a Molar enthalpy (AHsm) calculated at 2085 psi
b Molar volume (AVsm) calculated at 283 K

° Molar volume (AVsm) calculated at 288 K

188

 

tr.

with the residual silanols [17]. The presence of the nitrogen within the NPAHs
results in an increased basic character relative to the parent PAHs. This basicity
allows the nitrogen to interact with silanols resulting in a more negative molar
enthalpy.

The changes in molar enthalpy, and the retention factor at varying
temperatures, are directly related to the depth to which a given solute penetrates.
Highly condensed solutes, such as 1-azapyrene and 4-azapyrene, probe the
regions of the octadecylsilica (ODS) that are closer to the distal terminus. As
discussed previously, this region is characterized by a high degree of disorder
and an increased density of gauche conformations [24.27.28]. Benz[a]acridine
and dibenz[a,j]acridine, having a more linear structure, can probe the regions of
the ODS closer to the proximal terminus, where the alkyl group is bound to the
silica surface. These regions are highly ordered with all trans carbon-carbon
bonds [27,28]. Since these regions are more ordered, the intermolecular
distances between the solute and stationary phase are smaller, resulting in a
more negative change in molar enthalpy. As temperature is increased, the ODS
becomes more fluid. This increased fluidity allows the smaller solutes to
penetrate further into the ODS, closer to the silanols.

Although there have been no published studies on the thermodynamics of
NPAHs in RPLC, Woodrow and Dorsey studied the partitioning of acridine in
micellar electrokinetic chromatography [29]. They report the enthalpy of
transition as -6.75 kcanol for this solute. In addition, Shang et al. report the

molar enthalpy of pyridine in supercritical fluid chromatography [30]. Varying the

189

density of the 002 mobile phase, they report values ranging from -6.17 to -7.41
kcanol. When compared to the data in Table 6.2, these values are similar,
suggesting that the method employed herein is consistent with other

experimental measurements of these solutes.

6.3.2.1.3 Molar Volume

A representative graph of the logarithm of the retention factor versus
pressure is shown in Figure 6.5. For all solutes, the effect of pressure is small in
methanol. Although the graph for each PAH is linear, the correlation coefficients
are varied given the small influence that pressure has on retention (R2 = 0.594 —
0.998). Similar to the molar enthalpy graph, the slope is positive for all solutes.
A positive slope is indicative of a negative change in molar volume and suggests
that the solute occupies less volume in the stationary phase than in the mobile
phase. It should be noted that the smallest correlation coefficients correspond to
the azapyrenes, the solutes that are affected by pressure changes the least. The
change in molar volume is calculated from the slope of this graph, according to
Equation 1.7. As shown in Table 6.2, 1-aminopyrene, 1-azapyrene, and 4-
azapyrene have values that are statistically indistinguishable, and close to zero.
In contrast, benz[a]acridine and dibenz[a,j]acridine demonstrate changes in
molar volume that are statistically nonzero, with the five-ring NPAH
demonstrating a more negative change in molar volume than the four ring
analogue. The values for the azapyrenes and the benz[a]acridine are similar to

those for the parent PAHs [24].

190

 

 

Figure 6.5: Representative graph of the retention factor versus pressure used to
calculate the change in molar volume. Experimental conditions and symbols

defined in Figure 6.4.

191

 

10

oooV

:03 mewwwmn.

 

 

 

 

 

 

coon ooow coo _.
of :4 0i 1.0 4
Q 1 l9 9 0
T 1.11
F 1.“ Ii fl H-
1 law] lql 1T 4

 

 

 

0.0 0.00...

...o

ow

BOlGVd NOLLNBJSH

192

The differences in the molar volume are attributed to the depth that each NPAH
penetrates into the stationary phase [23,24]. As pressure is increased, the ODS
alkyl chains are forced closer together, increasing the overall order of the system.
Highly condensed solutes, which probe the regions of the octadecylsilica that are
closer to the distal terminus, demonstrate near-zero changes in molar volume
because these regions remain fluid even at high pressure. However, those
solutes that penetrate further into the stationary phase can probe regions that are
more tightly packed, reducing the final volume that the solute occupies. This
change in fluidity is analogous to the change imparted by temperature, except
that an increase in pressure results in the same phenomena as a decrease in

temperature.
6.3.2.2 Kinetic Behavior

6.3.2.2.1 Rate Constants

Although the thermodynamic data demonstrate the steady-state aspects
of chromatographic behavior, they do not fully explain the retention mechanism.
Using the equations and methods developed in Chapters 1 and 2, the pseudo-
first-order rate constants, activation enthalpies, and activation volumes were
calculated. These values help to quantify the kinetic aspects of mass transfer
between the mobile and stationary phases as a function of solute sthcture.
These data provide information about the retention that would be unavailable
from thermodynamic data alone.

Representative values of the rate constants, calculated using Equations

1.16 and 1.17, are summarized in Table 6.3. At constant pressure. 1-

193

aminopyrene undergoes the fastest rates of transfer, followed by 4-azapyrene
and benz[a]acridine. Dibenz[a,j]acridine undergoes slower rates of transfer, but
1-azapyrene is the slowest of all the solutes. The rate constants for the
azapyrenes and benz[a]acridine are smaller than their parent PAHs [24]. It
should be noted that at 288 K, the retention factor for benz[a]acridine is equal to
unity (k = 1). Thus, the rate constants are equivalent.

For 1-aminopyrene, 1-azapyrene, 4-azapyrene, and benz[a]acridine the
rate-limiting step is the transfer from mobile to stationary phase. By contrast, the
rate-limiting step for dibenz[a,j]acridine is the transfer from stationary to mobile
phase. The difference in the rate-limiting step is a consequence of the solubility
of the solutes. As noted above, the polarity of the solutes make them less
soluble in the ODS, thus increasing the rate of expulsion. Although
dibenz[a,j]acridine is also polar, the larger number of rings decrease the localized
electron density on the nitrogen. This delocalization of the lone pair of electrons
causes the molecule to behave in an analogous manner to the parent PAH.
These increased interactions lead to the slower rates of transfer out of the

stationary phase.

194

.v. ooN .0 00.0.3200 0.:0.0:00 0.0.... 0

 

 

 

 

 

 

 

 

 

 

 

 

.00 0000 .0 00.0.8.8
.60... 000.... b0:0_.0.0 0. 0.50:. 80.. 0:0 .06... 000.... 0.50:. 0. 00:30.0 E0... 0.:0.0:00 0.0.. 0
.0 00 0 0 00 .0 00 0 00.050.00.200...
3 t 1 0. 00 3 .0 3 005.600.0000
0.0 0.0 0.0 0.0 0 0.0 v 00 0000302..
3 3 0. 3 00 3 00 0. 0000.002...
00 00. 0E .00 00. 00 000 03 000580.02;
.00 0000 .00 000 .00 0000 .00 000 v. 000 v. 000 v. 000 v. 000 0.0.00
0 .00. e... e .00. 20. . .00. .5. . .00. .0...

 

 

 

 

 

 

0.0000... 0:0 0.0.0.00E0. .0 5.5:... 0 00 .0:0:.0:. :_ 0...<n.z .0. 0.:0.0:00 0.0..

”0.0 0.00.

195

As shown in Table 6.3, the rate constants for all solutes increase with
increasing temperature. This behavior is a consequence of the increased
diffusion coefficients and the enhanced fluidity of the stationary phase. As more
kinetic energy is imparted, the alkyl chains become more labile and can more
readily undergo rotation of the carbon-carbon bonds from the trans to gauche
conformation. .

Also shown in Table'6.3, the rate constants decrease, or remain invariant,
with increasing pressure for all solutes. This behavior is a consequence of the
compression of the alkyl chains, which impedes solute diffusion into and out of
the stationary phase. As the stationary phase is compressed, it becomes more
hydrophobic, and thus excludes the polar NPAHs. The net increase in van der
Waals forces that dibenz[a,j]acridine participates in overcomes this exclusion and
allow the solute to behavior in a manner more similar to the parent PAHs.

It should be noted that the rate constants for 1-azapyrene and 4-
azapyrene are statistically invariant with pressure. These data indicate that
these solutes are not affected by the compression of the octadecylsilica chains.
Rather, these solutes are interacting with adsorption sites within the stationary
phase. These adsorption sites may be at the silica interface, or may result from
the incomplete polymerization of the stationary phase. While it may be argued
that the observed effect results from the depth to which these solutes penetrate
the ODS, the likelihood that this is the only cause of the pressure independence

is unlikely. As demonstrated for pyrene, pressure has little effect on the rates of

transfer.[24] However, the rates are noticeably larger (~1000 s") than those in

196 ’

Table 6.3. The difference in the rate constants for these solutes indicates that
the azapyrenes are accessing different sites, and that these sites are lower in
energy than the partition ‘sites’ of the system (i.e. the molar enthalpies are more
negative). These data support the assertion by Colin and co-workers [17].
Finally, the difference between the rate constants for 1-azapyrene and 4-
azapyrene indicate that the location of the nitrogen is important when accessing
these adsorption sites. This order of magnitude difference implies that the 1-

azapyrene is configured preferentially to sorb at these sites.

6.3.2.2.2 Activation Energy

A representative graph of the logarithm of the rate constant versus the
inverse temperature is shown in Figure 6.6. The data for methanol are depicted
using solid lines. The graph for each solute is linear (R2 = 0.968 — 0.995) and the
slope is negative. A negative slope is indicative of a positive energy barrier. The
activation energy is calculated from the slope of this graph, according to Equation
1.18. As shown in Table 6.4, the activation energies are positive for all solutes.
The activation energy for the transfer between the stationary and transition state
(AE3) is larger than the activation energy for the transfer between the mobile and
transition state (AEm). These data indicate that it is easier for the solutes to exit
the stationary phase than it is for them to enter it. Using a student t test to
determine whether there are statistical differences, it was impossible to discern

any definitive trend for these solutes.

197

Figure 6.6: Representative graph of the rate constant (kms) versus inverse
temperature used to calculate the activation energy. Column: polymeric

octadecylsilica. Mobile phases: methanol, 2085 psi, 0.08 cm/s (——),

acetonitrile, 2175 psi, 0.08 cm/s (— - -). Symbols defined in Figure 6.4.

198

 

000”

and"

.L. 00.... 003200020...
00.0 00.0 00.0

0.0" cog"

 

 

- «I h h

p .005.

000..

. fio

5':
(,-s) lNViSNOO 31w

.oo.

0000.

 

 

0.0 050:

coco?

199

v. wwm .0 00.0.3200

.fi.><. 0.0.0 00...000.. 0. 000.... 0.30.... 0.0.. 000 10>... 0.0.0 000.000.. 0. 000.... 0002.0.0 0.0.. 00.03.? 02.02.04.

.00 0000 .0 8.0.8.00

.fimS 0.0.0 000.000.. 0. 000.... 0.50... .00.. 000 .39.. 0.0.0 00...000.. 0. 000.... 0002.00 .00.. 00.9000 00..0>..o<

 

 

 

 

 

0. 0 .0 0. .0 00 . .0 0. . 0 00 00.200.00.203
00 .0 00 00 0 00 0 0 0. 0 0 .0 8.250.008...
00 .0 0.. .0 0 .0 0 0 0. 0 H 00 009.000..-.

.00 000 .00. .00. 000.302-..
0. .0 00 00 .0 .0 . .1. 0. . 0 .0 009.08%...

 

 

0 .6055. §><

 

. 0005.... 0.><

 

0 29509.. 5.00

 

0 005.09.. 0.00

 

 

0.3.00

 

6000.0... 0. 0...<n.z .0. 000.20., 00..0>..o0 000 00.9000 00..0>..0<

3.0 0...0...

200

6.3.2.2.3 Activation Volume

A representative graph of the natural logarithm of the rate constant versus
pressure is shown in Figure 6.7. The activation volume is calculated from the
slope of this graph, according to Equations 8 and 9. As shown in Table 6.4, 1-
aminOpyrene has the largest activation volume while 1-azapyrene and 4-
azapyrene have the smallest. Given the large standard deviation, the values for
the azapyrenes are statistically indistinguishable and may in fact be zero. The
values for the two azapyrenes are analogous to pyrene (AV;s = 47 i 7 mL/mol,
AV”, = 43 :l: 7 mUmol) [24]. Benz[a]acridine and dibenz[a,j]acrdine demonstrate
activation volumes larger than the azapyrenes, but smaller than the
aminopyrene. Given the large deviation abut the values, definitive correlation
between the activation volumes and structure are currently unavailable.
However, the activation volumes and the change in molar volume demonstrate
the trend of AVE;s > AV;m >> AVsm. Hence, the volumetric barrier for the transition

is large, even though the overall change in molar volume is relatively small.

201

Figure 6.7: Representative graph of the rate constant (kms) versus pressure
used to calculate the activation volume. Column: polymeric octadecylsilica.

Mobile phase: methanol, 283 K, 0.08 cm/s. Symbols defined in Figure 6.3.

202

 

ooov

080

90... mewwmmn.
Doom

coo.

 

 

 

 

 

 

‘l

 

 

 

 

..0 050..

_..o

,9
(,-s) .LNVLSNOO 31w

.. 00..

coo ..

203

6.3.3 Acetonitrile Mobile Phase

As noted above, Colin and co-workers postulate that the nitrogen in an
NPAH can interact with the silica surface, leading to high degrees of asymmetry
[17]. In order to test this hypothesis, acetonitrile was used to determine how an
aprotic solvent would affect retention and the concomitant thermodynamic and
kinetic parameters. Since acetonitrile is incapable of hydrogen bonding, it does
not compete for adsorption sites that are capable of participating in such bonds.
Although most solutes are retained via a partition mechanism in RPLC when
acetonitrile is used, we hypothesized that the NPAHs could be retained either by
a partition or adsorption mechanism.

Given that the data from the methanol studies exhibited small changes
with pressure, the acetonitrile study was conducted at a single average pressure
of 2175 psi. In contrast to the effect of pressure, the effect of temperature was
significant in methanol. Thus, a large temperature range was also used for the

acetonitrile (T = 293 to 333 K).

6.3.3.1 Thermodynamic Behavior

6.3.3.1.1 Retention Factor

Representative values of the retention factor are summarized in Table 6.5.
As demonstrated, 1-aminopyrene is eluted first, followed by 4-azapyrene and
benz[a]acridine. Dibenz[a,j]acridine is retained the most of the solutes. When
compared to the retention factors in methanol (Table 1), all solutes except for 1—

aminopyrene demonstrate an increase in retention. Given that acetonitrile is less

204

Table 6.5: Thermodynamic data for NPAHs in acetonitrile a

 

 

 

 

 

k k AHsm
Solutes (303 K) (313 K) (kcal/mol)
1-Aminopyrene 0.23 0.16 -6.3 :l: 0.7
4-Azapyrene 4.54 2.74 -9.5 i 1.1
Benz[a]acridine 4.47 4.15 —8.3 d: 2.1
Dibenz[a,j]acridine 6.43 5.49 -6.1 i 2.1

 

 

 

 

 

 

a All data calculated at 2185 psi

205

polar than methanol, the retention factor is expected to decrease if a partition
mechanism is in effect. However, the increase in retention factor for four of the
solutes implies that an adsorption mechanism is contributing to the observed
behavior. By contrast, 1-aminopyrene exhibits behavior that is consistent with a
partition mechanism. Similar to methanol, an increase in temperature results in a

decrease in the retention factor for the reasons described above.

6.3.3.1.2 Molar Enthalpy

A representative graph of the logarithm of the retention factor versus the
inverse temperature is shown in Figure 6.4. The data for acetonitrile are graphed
using dashed lines. The graph for each solute is linear (R2 = 0.724 — 0.986) and
the slope is positive. A positive slope is indicative of a negative change in molar
enthalpy and suggests that the transfer from mobile to stationary phase is
enthalpically favorable. The change in molar enthalpy is calculated from the
slope of Figure 6.4, according to Equation 1.5. i

As shown in Table 6.5, the changes in molar enthalpy are the least
negative for 1-aminopyrene and dibenz[a,j]acridine followed by benz[a]acridne
and 4-azapyrene. When compared to the methanol data, the changes in molar
enthalpy are similar for all solutes except 4-azapyrene. This difference as a
function of mobile phase suggests that the 4-azapyrene is accessing sorption
sites that are less available to the more linear solutes. .

It should be noted that the changes in molar enthalpy are similar in both
methanol and acetonitrile, even though the retention factors are different. These

data imply that adsorption sites become available in the presence of an aprotic

206

solvent, but that the bonded ODS affects the accessibility of these sites. Since
the NPAHs must diffuse through the ODS to reach the adsorption sites, the ODS
acts as a barrier, preventing complete access to the surface. Thus, the effect of
temperature on ODS plays an important role for these solutes in both the protic
and aprotic mobile phases. 4-Azapyrene, being the most condensed solute, can
diffuse more easily through the ODS chains, reaching the surface more quickly.
Thus, the change in molar enthalpy for this solute is exaggerated. Since the
other solutes are larger, it is energetically more difficult for them to reach the

adsorption sites. Thus, their main mode of interaction still resides with the ODS.
6.3.3.2 Kinetic Behavior

6.3.3.2.1 Rate Constants

Representative values of the rate constants, calculated using Equations
1.16 and 1.17, are summarized in Table 6.6. At constant pressure, 1-
aminopyrene undergoes the fastest rates of transfer, followed by
dibenz[a,j]acridine, 4-azapyrene and benz[a]acridine. As with methanol, the rate-
limiting step for 1-aminopyrene is the mobile to stationary phase transfer The
other three solutes demonstrate the inverse trend, with the stationary to mobile
phase transfer taking more time. When compared to the methanol data, the rate

constants are two to four orders of magnitude smaller in acetonitrile.

207

.00 00.0.0 00.03200 0.00 __< 0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 0 0. 0 0 00 00. x 0.. 0-0. x 0.0 00. x 0.0 0-0. x .0 00.200.00.050...
0 .. 0 0 0 N.0. x 0.. N.0. x 0.. 00. x 0.0 0... x 0.0 00.008.00.00
o e 0 0 0 0.0. x 0.0 N.... x 0.. 0... x 0.. 0-... x 0.0 000.300....
o a 0 z 0... x 0.0 0... x 0.0 .00 .00 000.0055...
005.09.. 2.00 005.00.. .000 V. 0.0 0.000 v. 0.0 0000 00.200
.00. a... .00. a...
0 0.0.5.000 c. 0.202 .0. 0.00 2.0.5.

0 0.0 030..

8

20

The differences in the rate constants result from an increase in the ability
of these solutes to interact with the silica support. Although very few molecules
diffuse to the proximal terminus of the ODS, those that do bind strongly once
they are within the minimum distance for interaction with the underlying silica
support. This argument should hold true for dibenz[a,j]acridine as well.
However, the nitrogen in dibenz[a,j]acridine is partially occluded by a phenyl unit.
In addition, for the nitrogen to interact with the surface, dibenz[a,j]acridine must
diffuse toward the surface with its longest axis parallel to the underlying silica
support. Such diffusion is energetically less favorable than diffusion with its
shortest axis parallel to the underlying support, and results in larger rate

constants than might be predicted.

6.3.3.2.2 Activation Energy

A representative graph of the logarithm of the rate constant versus the
inverse temperature is shown in Figure 6.6. The data for acetonitrile are
depicted using dashed lines. It should be noted that the data for 1-aminopyrene
are not represented. The zone profiles for 1-aminopyrene were nearly Gaussian,
and as such had very small values of T. The small values of 1 result in calculated
values of kms and ksm that are highly scattered as a function of temperature.

As shown, the graph for the other three solutes is linear and the slope is
negative. Unlike the methanol data, the correlation coefficients for the linear
regression vary significantly (R2 = 0.038 - 0.960). As such, the reported values

should be used only to draw general conclusions.

209

The activation energy is calculated from the slope of this graph, according
to Equation 1.18. As shown in Table 6.6, the activation energies are positive for
all solutes. The activation energy for the transfer between the stationary and
transition state (AE3) is larger than the activation energy for the transfer between
the mobile and transition‘state (AEm). These data indicate that it is easier for the
solutes to exit the stationary phase than it is for them to enter it. Given the
standard deviations on the values, it is impossible to discern any definitive trend

as a function of solute structure.

6.4 Conclusions

In this chapter, the thermodynamic and kinetic behavior of NPAHs was
examined in reversed-phase liquid chromatography using a protic and aprotic
mobile phase. The thermodynamic data indicate a noticeable effect of
temperature for both mobile phases. An increase in temperature results in a
decrease in the retention factor in both mobile phases. However, pressure was
shown to have a small effect on the solutes in methanol, with the reported values
being similar to those of the parent PAHs. As a result, a single pressure was
examined for the acetonitrile data. The effect of mobile phase was more
profound than that of pressure. The data demonstrate that the retention factors
for all solutes, except 1-aminopyrene, increase in acetonitrile relative to
methanol. Such an increase implies a change from a partition mechanism
(methanol) to an adsorption mechanism (acetonitrile). However, the changes in
molar enthalpy are similar, suggesting that the octadecylsilica may hinder the

acess of the larger solutes to the silica support.

210

Similar to the thermodynamic data, the kinetic data indicate a noticeable
effect of temperature for both mobile phases. Once again, pressure was shown
to have a nominal effect on the solutes in methanol. More importantly, the
aprotic nature of acetonitrile was shown to alter the rate constants by a factor of
two to four orders of magnitude. However, the fluctuations imparted by the fitting
process inhibit our ability to discern definitive correlations between the activation
energies and structure in acetonitrile. Regardless of this deficiency, these data
represent the first quantitative comparison of NPAH retention in reversed-phase

liquid chromatography.

211

6.5 References

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

C.-H. Ho, B.R. Clark, M.R. Guerin, B.D. Barkenbus, T.K. Rao, J.L. Epler,
Mutat. Res. 85 (1981) 335-345.

GM. Seixas, BM. Andon, P.G. Hollingshead, W.G. Thilly, Mutat. Res. 102
(1982) 201-212.

A. Matsuoka, K. Shudo, Y. Saito, T. Sofuni, M. lshidae, Jr., Mutat. Res.
102 (1982) 275-283.

J.M. Schmitter, H. Colin, J.-L. Excoffier, P. Arpino, G. Guiochon, Anal.
Chem. 54 (1982) 769-772.

8.0. Ruckmick, R.J. Hurtubise, J. Chromatogr. 321 (1985) 343-352.
C. Borra, D. Wiesler, M. Novotny, Anal. Chem. 59 (1987) 339-343.

N. Motohashi, K. Kamata, R. Meyer, Environ. Sci. Technol. 25 (1991) 342-
346.

J.M. Schmitter, l. Ignatiadis, G. Guiochon, J. Chromatogr. 248 (1982) 203-
216.

MR. Guerin, C.-H. Ho, T.K. Rao, B.R. Clark, J.L. Epler, Environ. Res. 23
(1980) 42-53.

8.6. Wakeham, Environ. Sci. Technol. 13 (1979) 1118-1123.

~H.-Y. Chen, M.R. Preston, Environ. Sci. Technol. 32 (1998) 577-583.

M. Wilhelm, G. Matuschek, A. Kettrup, J. Chromatogr. A 878 (2000) 171-
181 .

K. Kamata, N. Motohashi, J. Chromatogr. 319 (1985) 331-340.

212

[14]

[15]

[15]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

K. Kamata, N. Motohashi, R. Meyer, Y. Yamamoto, J. Chromatogr. 596
(1992) 233-239.

K. Kamata, N. Motohashi, J. Chromatogr. 396 (1987) 437-440.
M. Ashraf-Khorassani, L.T. Taylor, J. Chromatogr. Sci. 26 (1988) 331-336.
H. Colin, J.M. Schmitter, G. Guiochon, Anal. Chem. 53 (1981) 625-631.

AM. Siouffi, M. Righezza, G. Guiochon, J. Chromatogr. 368 (1986) 189-
202.

TR. Steinheimer, M.G. Ondrus, Anal. Chem. 58 (1986) 1839-1844.

J.W. Haas, lll, W.F. Joyce, Y.-J. Shyu, P.C. Uden, J. Chromatogr. 457
(1988) 215-226.

K. Kamata, N. Motohashi, J. Chromatogr. 498 (1990) 129-135.

H. Carlsson, C. Cstman, J. Chromatogr. A 715 (1995) 31 -39.

SB. Howerton, V.L. McGuffin, Anal. Chem. 75 (2003) 3539-3548.
S.B. Howerton, V.L. McGuffin, J. Chromatogr. A (2003) In press.
V.L. McGuffin, S.-H. Chen, J. Chromatogr. A 762 (1997) 35-46.

J.H. Hildebrand, R.L. Scott, The Solubility of Nonelectrolytes, Reinhold
Publishing Corporation, New York, NY, USA, 1950.

MW. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5576-5584.

M.W. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5585-5592.

213

[29] EN. Woodrow, J.G. Dorsey, Environ. Sci. Technol. 31 (1997) 2812-2820.

[30] D.Y. Shang, J.L. Gransmaison, S. Kaliaguine, J. Chromatogr. A 672
(1994) 185-201 .

214'

Chapter 7: Retention Mechanisms in Whole Soil

7.1 Introduction

Using the techniques described in Chapters 1 through 4, an investigation
into the molecular contributions to retention for a fraction of soil was conducted.
Two polycyclic aromatic hydrocarbons, a series of nitroalkanes and a series of
alkylbenzenes were used as test solutes. These experiments were designed to
validate or refute the current theories of soil retention. This chapter contains a
chemical description of soil, the current theories of soil retention, as well as the

experimental methods used to study this complex material.

7.1.1 Composition of Soil

By its nature, soil is both physically and chemically heterogeneous.
Physically, soil particles range from the sub-micrometer up to the millimeter, and
sometimes centimeter sizes. The particles are composed of a wide range of
materials that include plant, animal, and mineral matter. As a result, the
chemical composition of soil can include plant and animal lipids, carbohydrates,
humic and fulvic acids, as well as a large number of inorganic materials (i.e., iron
and aluminum oxides, soluble salts, etc.). In addition, the environment in which
the primary materials are deposited can greatly affect the final composition of the
soil [1,2]. Despite the large amount of information regarding types of soils and
the constituent elements, there is no exact structure for soil organic matter. This

lack of information is due primarily to the large number of chemical moieties

215

found within soil (i.e. carboxylic acids, amines, etc.) and the environmental

variables in which the soil is formed (i.e. temperature, humidity, flora, etc.) [1 ,2].

7.1.2 Retention Mechanisms in Soil

While the mechanism of retention in synthetic materials has been
investigated using a wide range of experimental and theoretical techniques, the
mechanism of soil retention has been probed using a smaller number of
techniques. From the published literature, two theories of soil retention have
come to the fore over the past twenty years. While the most recent focuses on
soil organic matter, both recognize the importance of minerals in the sorption

process.

7.1.2.1 Partition with Adsorption

The first investigation to postulate the retention mechanism of nonionic
organic compounds (NOC) was published by Chiou et al. in 1979 [3]. In this
seminal work, the isotherms of seven NOCs were measured. The investigators
found that the isotherms were linear and exhibited no detectable level of
curvature. The lack of curvature suggests that retention occurs via a partition
mechanism. However, further experimentation was necessary in order to
validate this hypothesis.

Chiou continued his investigations into the mechanism of retention by
probing the effect of water on the sorption capacity [4,5], as well as the effect of
dissolved organic matter (DOM) [6,7] and surfactants [8,9] on the solubility of

NOCs. From these studies, water was found to alter the sorption capacity of the

216

soil. Above 90% relative humidity, Chiou and Shoup reported that the sorption
capacities approached those found in completely aqueous systems [4]. Below
this humidity, the soil exhibited an adsorptive mechanism. In addition, Chiou and
coworkers noted that DOM, as well as surfactants, could increase the solubility of
organic solutes in aqueous solutions. The authors also demonstrated that the
source of the DOM was important, with commercially available materials
increasing solubility by factors of four to twenty over that of samples from the
field [7]. Using these data, Chiou and coworkers hypothesized that in the
absence of water, the mineral matter in soil was the primary site of retention.
This hypothesis has been supported by investigations of volatile organic
compounds on silica, soils, and clays [10,11]. Furthermore, Chiou and
coworkers postulated that in the presence of water, soil organic matter (SOM)
becomes the primary site of retention. In contrast to the mineral matter, SOM is
thought to exhibit a partition mechanism. Given that most naturally occurring
soils are in water rich environments, Chiou and co-workers concluded that SOM
was the most important factor for assessing the sorption capacity of soils.
Because of Chiou’s work, the partition model was advanced as the
primary mechanism of sorption for NOCs in water-rich environments. However,
several investigators reported data that challenge the hypothesis of a partition
mechanism. Significant differences in the sorption and desorption of organic
compounds [12,13], competitive sorption [14,15], and nonlinear isotherms[16-18]
were reported, suggesting that a partition mechanism was insufficient to fully

explain retention in soil. The differences in the adsorption and desorption curves

217

for a series of PCBs reported by DiToro and Horzempa suggest that adsorption
contributes to retention [13]. Similarly, Stuart et al. [14] and Abdul and Gibson
[15] presented evidence of competitive sorption, which implies that soil has some
adsorptive character. Since a partition mechanism is presumed to be a non-
competitive, the reports of competitive sorption detract from Chiou’s initial
hypothesis. Lastly, the identification of nonlinear isotherms is also indicative of
adsorption, since there are a limited number of sites for interaction. Once again,
the partition mechanism is not predicated upon a limited number of sites, as is

the adsorption mechanism [19].

7.1.2.2 Dual Reactivity Model (DRM)

Using the observed deviations as a stimulus, Weber and co-workers have
hypothesized that the chemical nature of soil organic matter is not unlike a co-
block polymer [20-33]. Dubbed the dual-reactivity model (DRM), Weber and co-
workers postulate that many of the observed inconsistencies with Chiou’s
hypothesis are indicative of glassy and rubbery regions within soil organic matter.
The glassy regions are typified by high structural rigidity due to aromatic moieties
and cross-linking. These regions are hypothesized to result from longer periods
of diagenesis. According to conventional polymer theory, glassy polymers are
characterized by high glass transition temperatures and inflexibility. ’Rubbery’
regions are said to be less rigid and likely composed of aliphatic moieties with
minimal cross-linking. Rubbery regions are said to result from relatively small
amounts of diagenesis and are characterized by low glass transition

temperatures and higher degrees of molecular flexibility.

218

While Weber has been supported by other investigators [34,35], the only
direct experimental evidence to demonstrate the possibility of two mechanistic
sites come from differential scanning calorimetry data [27]. Although the authors
demonstrate the presence of a glass transition temperature for a humic acid, the
data are far from definitive, with some authors challenging the utility of the data
[36].

Chiou and co-workers contend that the DRM is in fact erroneous and that
the nonlinear effects arise from interactions with high-surface-area carbonaceous
material (HSCAM). This material has been postulated to be a charcoal, or soot
like material that is only a very tiny fraction of ordinary soil [37-39].

Given the discourse in the literature about the mechanism of retention in
soil, a series of experiments were developed to probe the retention mechanism

of a reference soil (CLN Soil-3 RTC Corp.).
7.2 Experimental Methods

7.2.1 Stationary Phase

Unlike the previous experiments that used a synthesized stationary phase,
the stationary phase for these experiments was a reference soil (CLN Soil-3,
Resource Technology Corp.). Since this material contains a large particle size
distribution, it is unsuitable for packing within a fused-silica capillary as received.
As a result, the soil was fractionated using a series of nylon mesh screens to
yield a fraction from 5-125 pm. In order to confirm that the sorption of the

characteristics of the material did not change with fractionation, batch isotherm

219

studies were carried out using naphthalene, phenanthrene, anthracene,
fluoranthene, pyrene, and chrysene (Sigma) with equilibration times that ranged
from 5 minutes to 748 hours and concentrations that ranged from1.4 to .75 M.
Each reaction vessel was filled with ~1.0 g of the soil and 4 mL of the PAH
solution. The supernatant from each vessel was analyzed in triplicate.

Using the data from the batch isotherms as a guide, this fraction of soil
was packed into a series of fused-silica capillaries (320 um i.d., 1 m long,
Polymicro Technologies) using the both the slurry pack method (Chapter 2) [40]
and a manual procedure. The manual procedure involved filling a syringe with a
small amount of slurry and introducing it into the fused-silica capillary in small
aliquots. Once a sufficient amount had been injected, the soil was compressed

at low pressures to remove any voids in the column.

7.2.2 System Parameters

Using the system described in Section 2.2.3.1 (Figure 2.1), the soil column
was used to study the retention of two polycyclic aromatic hydrocarbons (PAHs),
pyrene and benzo[a]pyrene (Sigma), which were dissolved in high purity
methanol (Baxter Healthcare) at concentrations ranging from 10‘ to 10‘3 M. In
addition, a homologous series of nitroalkanes and alkylbenzenes was studied as
a function of mobile phase (methanoVwater, 90-100%). The nitroalkane series
contained nitromethane, nitroethane, nitropropane, nitrobutane, and nitrohexane
(Aldrich) dissolved to 2%, by volume, in a solution that matched the mobile

phase. The series of alkylbenzenes contained toluene, ethylbenzene,

220

butylbenzene and hexylbenzene (Sigma) dissolved to 2%, by volume, in a
solution that matched the mobile phase.

The separations were carried out at 298 K in pressure control mode at a
nominal flow rate of 0.15 uUmin. Detection of the PARS was achieved using
laser-induced fluorescence as described in Chapter 2. Detection of the aliphatic
compounds was achieved using the UV-Visible absorbance spectrometer at 210

nm for the alkylbenzenes and 260 nm for the nitroalkanes.
7.3 Results and Discussion

7.3.1 Batch lsotherms

Table 7.1 contains data from the batch isotherms of whole and
fractionated soil. A minimum of twenty-four hours was necessary in order to
reach equilibrium within the batch reactors. Using this equilibration time, the
batch isotherms were studied using a range of concentrations that fell within the
limits of detection for the UV-Visible absorbance spectrometer. The equilibrium
constants were calculated by first determining the amount of each PAH in the
supernatant. This was accomplished using a calibration curve. The equilibrium

constants were calculated for both the whole and fractionated soil

_[_)_(_]5_ = ([chon ‘[X]SUp)
[xlm [xlsup X Msoil

 

K6,, = (7.1)

221

 

 

 

 

 

 

 

 

 

 

 

a? a 3% Se « 0.08 on .. we 2828
«.8 a 32.. 4.8 a 32 in. l 3 825
QB a 3m 3 a 3m 3 1 ca 22298.”.
3. a 39 E. a 3m em 1 3 28955
5. a o? o.» a 8.8 3 .. 3 mceecmcmi
em a 3 we a was 5K .. 2: 9.222%:
223 09.3.
f: x V 5322 A -2 x :23. 832880 928

 

 

.__om uoficozomt

new 20:3 co 209893... 25:63 2.2623 Co wmtmm m .8 9:938 82.9.53 new mcozgcmocoo m...» min...

222

where [X]s is the amount of solute X in the stationary phase and [le is the
amount of solute X in the mobile phase. Since sorption to the reaction vessel
was possible, control samples were examined concurrently with the soil samples.
The difference between the concentration in the control vessel ([X]co.,) and the
supematant ([X]sup) is equivalent to the amount sorbed by the soil. Since the
same mass of soil (Msou) could not be accurately reproduced, the equilibrium
constants were normalized to the amount for each reactor to compensate for any
differences. The equilibrium constants were compared using a Smith-
Satterthwaite test [41], which uses the t-statistic. Using a null hypothesis that the
equilibrium constants in the whole soil and the fractionated soils are equivalent
(i.e. Ao=0), the test indicates that for all solutes, except naphthalene, there is a
statistical difference between the whole and fractionated soil. However, these
differences ranged only from a factor of ~03 to 2 for all solutes. Although the
difference was measurable, the small degree of the difference suggests that the
fractionated soil could be used without dramatically altering the retention

characteristics.

7.3.2 Soil Column Characteristics

In order to study the retention characteristics of soil, the manufacture of a
capillary soil column was necessary. As detailed in Chapter 2, the fractionated
soil was packed into a fused-silica capillary via the slurry method and the manual
method described above. A number of columns were packed and then

characterized via visual inspection.

223

For both methods, the fractionated soil was found to pack in a nonuniform
manner. The region closest to the quartz wool frit was found to have a large
number of very small particles (< 5pm) that were densely packed. The
remainder of the column was packed with larger particles (>75urn), which were
loosely packed. In time, though, the voids between the larger particles were also
filled in with small particles. When the soil columns were pressurized, they
generated larger back pressures than their ODS counterparts. For example, a
25 cm soil column at 2500 psi evinced a 0.15 uL/min flow rate. By contrast, a 1m
long ODS column at 800 psi evinced a 1 uL/min flow rate. This large
backpressure in the soil columns prevented flow rates of 1 uL/min from being

achieved.

7.3.3 Retention Mechanisms

Using a soil column packed via the slurry method, the investigation of soil
retention began with a study of pyrene and benzo[a]pyrene. As demonstrated in
previous chapters, these solutes are retained via a partition mechanism in
octadecylsilica (ODS). Given that the experiments on ODS required only small
injection volumes for detection, 10uL injections of 10'5 M solutions were used
initially. However, this volume was undetectable even when the concentration
was increased to 10‘3 M. These results suggested that the sorption capacity of
the soil was much greater than that of the ODS. However, the slow flow rates
could also contribute to the inability to detect the solutes. At very slow flow rates,

the zone profiles would eventually become very broad and shallow due to

224

diffusional processes. Thus, it is possible that the flow rate contributed to the
lack of detection using the small injection volumes.

Given the difficulties with the smaller volumes, a larger injection loop was
constructed (0.25 mL) so that the solutes could be studied via a breakthrough
curve. Figure 7.1 is illustrative of the breakthrough curve for benzo[a]pyrene. As
shown, the retention time for the solute is in excess of five hours at room
temperature. This retention time is much larger than that observed for the solute
on the synthetic phases (~1 h at comparable temperature and pressure) [42].
Note once again the flow rate is slower in the soil column. In order to
compensate for the slow flow rates, an attempt was made to determine the
retention factor for benzo[a]pyrene. However, the unretained marker (4-
methylhydroxy-7-methoxycoumarin) that had been successfully employed in the
ODS system was undetectable. Although several other solutes were used in an
attempt to determine the true void time, no detectable solute was ever found.
Regardless, pyrene was also injected at similar concentrations and volumes for a
comparison. Unlike benzo[a]pyrene, the retention time of pyrene was found to
be ~2 hours. These data suggest that the soil is retaining the solutes via a
partition mechanism, and are consistent with investigations carried out on
chemically immobilized humic and fulvic acids [43]. As shown in Chapter 4, an
increase in ring number results in a logarithmic increase to the retention time
when a partition mechanism is present. Thus, the observed behavior on the
fractionated soil conforms to this observation, suggesting the presence of a

partition mechanism.

225

Figure 7.1: Representative breakthrough curve of benzo[a]pyrene on a soil

column. Mobile phase: methanol, 0.15uL/min.

226

 

oooom

 

 

Doom _.

P

 

Doom _.

p

@ wsE.
coon _. ooom _.

[r P

 

coon _.

_

ooov P

»

coon _.

..

 

 

E 9:9“.

asoi

EONBOSEHOR'H

227

mechanism in soil. Given the long retention times, the alkylated homologous
series were used to better study retention since their elution time was shorter.

The nitroalkanes and the alkylbenzenes were separated under similar
conditions as the two PAHs. If a partition mechanism were present, the solutes
would be expected to elute in order from the shortest alkyl chain solute to the
longest alkyl chain solute. This behavior has been demonstrated in previous
investigations [44-46] . However, the alkylbenzenes eluted in order from the
longest chain (hexylbenzene) to shortest (toluene) in methanol. Similarly, the
nitroalkanes eluted in a reversed order (i.e. longest to shortest). These data
imply that both series are separated based upon an exclusion mechanism.

In size exclusion chromatography, smaller solutes are able to access
regions of the stationary phase (e.g. pores) that larger solutes cannot. As a
result, smaller solutes spend a larger amount of time in these regions, away from
the flowing mobile phase. Thus, the solutes elute in an inverse order.

In an effort to verify the presence of an exclusion mechanism, the mobile
phase was altered to include 5 and 10% water. The presumption was that if any
partition mechanism were present, it would be enhanced with increasing water
concentration since the solutes would be less soluble in the mobile phase.
Previous studies as a function of mobile phase for these solutes indicate an
increase in retention factor with increasing water content [46-49]. Figures 7.2
and 7.3 illustrate the retention time of each series as a function of mobile phase
composition. It should be noted that for these data there is a significant amount

of scatter due to small changes in the linear velocity of the mobile phase. For

228

Figure 7.2: Graph of retention factor versus carbon number for a series of
nitroalkanes as a function of methanol/water mobile phase. (0) 100% methanol,

(I) 95% methanol, (A) 90% methanol.

229

 

 

mum—232 20mm<0
v n N v o

 

 

- h h n mm

. cow

.. Nov

.0

rvow

. wow

0

. wow

0

to:

IN:

1v:

 

 

mg 95m: m:

SWIJ. NOIlNELLSH
230

Figure 7.3: Graph of retention factor versus carbon number for a series of
alkylbenzenes as a function of methanol/water mobile phase. (0) 100%

methanol, (I) 95% methanol, (A) 90% methanol.

231

 

 

 

 

% 20mm<o

 

 

cow

e For

- Nov

32

. vow

- mo?

floor

 

 

m8 2:9“.

now

SWLL NOLLNELEH

232

example, multiple separations on the same day resulted in a variation of 5-7% in
the retention time.

As shown, the data indicate that at all mobile phase compositions the
solutes maintain the reversed elution order, indicative of an exclusion
mechanism. However, the effect of water on the retention is not as obvious. For
the nitroalkanes (Figure 7.2), the retention times are statistically indistinguishable
as a function of mobile phase composition. For the alkylbenzenes (Figure 7.3)
the retention time increases with increasing water concentration. It should be
noted that the overall differences in retention time between the alkylbenzenes are
very small. Using multiple injections at each mobile phase composition, the
Student t-test results indicate that the retention times are statistically
indistinguishable as a function of mobile phase composition. Thus, the
alkylbenzene data that depicts the effect of mobile phase composition suggests
the presence of a partition mechanism. However, the variance in the
measurements prevents definitive identification of this mechanism. Regardless,
the reversed elution order at all mobile phase compositions indicates the

presence of an exclusion mechanism.

7.4 Conclusions

The experiments within this chapter demonstrate that the use of soil as a
stationary phase presents some practical limitations that are not present with
ODS. The large particle sizes and the large particle size distribution results in
nonuniform packing of the capillary column. This nonuniform packing leads to

very high pressures and very low flow rates. The retention of the PAHs suggests

233

a partition mechanism. The partition mechanism likely results from interactions
with the organic carbon in the soil. The data from the alkylbenzenes could
support this observation if the standard deviations for the retention times could
be reduced. However, the data for the nitroalkanes and the alkylbenzenes
demonstrates that size exclusion is more important for these solutes. The
exclusion mechanisms likely results from the micro- and mesoporous structure of
the mineral matter within the soil. This mechanism is not observed for the PAHs
due to their large size. Although the molecular level contributions could not be
studied as with ODS, the data contained in this chapter still provide a glimpse
into the mechanism of soil retention, indicating that solute stmcture is an

important variable in the retention mechanism.

234

7.5 References

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

NC. Brady, R.R. Weil, The Nature and Properties of Soil, Simon &
Schuster, Upper Saddle River, NJ, USA, 1999.

SW. Buol, F.D. Hole, R.J. McCracken, Soil Genesis and Classification,
Iowa State University Press, Ames, Iowa, USA, 1989.

CT Chiou, L.J. Peters, V.H. Freed, Science (Washington, D. C., 1883-)
206 (1979) 831 -832.

GT. Chiou, T.D. Shoup, Environ. Sci. Technol. 19 (1985) 1 196-1200.
C.T. Chiou, T.D. Shoup, P.E. Porter, Org. Geochem. 8 (1985) 9-14.

C.T. Chiou, R.L. Malcolm, T.l. Brinton, D.E. Kile, Environ. Sci. Technol. 20
(1986) 502-208.

C.T. Chiou, D.E. Kile, T.l. Brinton, R.L. Malcolm, J.A. Leenheer, Environ.
Sci. Technol. 21 (1987) 1231-1234.

D.E. Kile, C.T. Chiou, Environ. Sci. Technol. 23 (1989) 832-838.

D.E. Kile, C.T. Chiou, R.S. Helbum, Environ. Sci. Technol. 24 (1990) 205-
208.

R.D. Rhue, P.S.C. Rao, R.E. Smith, Chemosphere 17 (1988) 727-741.

R.D. Rhue, K.D. Pennell, P.S.C. Rao, W.H. Reve, Chemosphere 18
(1989) 1971-1986.

S.W. Karickhoff, K.R. Morris, Environ. Toxicol. Chem. 4 (1985) 469-479.

DM. DiToro, LM. Horzempa, Environ. Sci. Technol. 16 (1982) 594-602.

235

[14]

[15]

[15]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[25]

[27]

B.J. Stuart, G.F. Bowlen, D.S. Kosson, Environ. Prog. 10 (1991) 104-109.
A.S. Abdul, T.L. Gibson, Haz. Waste Haz. Mater. 3 (1986) 125-137.
W.P. Ball, P.V. Roberts, Environ. Sci. Technol. 25 (1991) 1223-1237.

G.P. Curtis, P.V. Roberts, M. Reinhard, Water Resour. Res. 22 (1986)
2059-2067.

M.D. Piwoni, P. Banerjee, J. Contam. Hydrol. 4 (1989) 163-179.

L. Snyder, in E. Heftmann (Editor), Chromatography, Van Nostrand
Reinhold Company, New York, NY, USA, 1975, p. 46-76.

W. Huang, M.A. Sclautman, W.J. Weber, Jr., Environ. Sci. Technol. 30
(1 996) 2993-3000.

W. Huang, TM. Young, M.A. Sclautman, H. Yu, W.J. Weber, Jr., Environ.
Sci. Technol. 31 (1997) 1703-1710.

W. Huang, W.J. Weber, Jr., Environ. Sci. Technol. 31 (1997) 2562-2569.
W. Huang, W.J. Weber, Jr., Environ. Sci. Technol. 32 (1998) 3549-3555.

M.D. Johnson, W. Huang, 2. Dang, W.J. Weber, Jr., Environ. Sci.
Technol. 33 (1999) 1657-1663.

M.D. Johnson, W. Huang, W.J. Weber, Jr., Environ. Sci. Technol. 35
(2001) 1680-1687.

M.D. Johnson, TM. Keinath, II, W.J. Weber, Jr., Environ. Sci. Technol. 35
(2001 ) 1688-1695.

E.J. LeBoeuf, W.J. Weber, Jr., Environmental Science and Technology 31
(1997) 1697-1702.

236

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

PM. McGinley, L.E. Katz, W.J. Weber, Jr., Environ. Sci. Technol. 27
(1993) 1524-1531.

W.J. Weber, Jr., PM. McGinley, L.E. Katz, Environ. Sci. Technol. 26
(1992) 1955-1962.

W.J. Weber, Jr., W. Huang, Environ. Sci. Technol. 30 (1996) 3130-3131.
W.J. Weber, Jr., TM. Young, Environ. Sci. Technol. 31 (1997) 1686-1691.
TM. Young, W.J. Weber, Jr., Environ. Sci. Technol. 29 (1995) 92-97.
TM. Young, W.J. Weber, Jr., Environ. Sci. Technol. 31 (1997) 1692-1696.
J.J. Pignatello, B. Xing, Environ. Sci. Technol. 30 (1996) 1-11.

G. Xia, J.J. Pignatello, Environ. Sci. Technol. 35 (2001) 84-94.

E.R. Graber, M.D. Borisover, Environ. Sci. Technol. 32 (1998) 3286-3292.
C.T. Chiou, D.E. Kile, Environ. Sci. Technol. 32 (1998) 338-343.

C.T. Chiou, D.E. Kile, D.W. Rutherford, G. Sheng, S.A. Boyd, Environ. Sci.
Technol. 34 (2000).

O. Gustafsson, F. Haghseta, C. Chan, J. Macfarlane, PM. Gschwend,
Environ. Sci. Technol. 31 (1997) 203-209.

J.C. Gluckman, A. Hirose, V.L. McGuffin, M. Novotny, Chromatographia
17 (1983) 303-309.

J.L. Devore, Probability and Statistics for Engineering and the Sciences,
4th ed., Duxbury Press, New York, NY, USA, 1995.

SB. Howerton, V.L. McGuffin, Anal. Chem. 75 (2003) 3539-3548.

237

[43]

[44]

[45]

[45]

[47]

[48]

[49]

K. Kollist-Siigur, T. Nielsen, C. Gron, P.E. Hansen, 0. Helweg, K.E.
Jonassen, O. Jorgensen, U. Kirso, J. Environ. Qual. 30 (2001) 526-537.

H. Colin, G. Guiochon, J. Chromatogr. Sci. 18 (1980) 54-63.
G.E. Berendsen, L. de Galan, J. Chromatogr. 196 (1980) 21 -37.
E. Grushka, H. Colin, G. Guiochon, J. Chromatogr. 248 (1982) 325-339.

W.R. Melander, C.A. Mannan, C. Horvath, Chromatographia 15 (1982)
61 1-615.

R.B. Sleight, J. Chromatogr. 83 (1973) 31-38.

R.P. Ranatunga, P.W. Carr, Anal. Chem. 72 (2000) 5679-5692.

238

Chapter 8: Characterization of Polycyclic Aromatic
Hydrocarbons in Environmental Samples
by Selective Fluorescence Quenching

8.1 Introduction

Given the ubiquitous nature of polycyclic aromatic hydrocarbons (PAHs) in
the environment (Chapter 1), a new technique is employed to study these
molecules from natural sources. This new technique takes advantage of
selective fluorescence quenching for the qualitative characterization of altemant
and nonaltemant PAHs. Using a novel experimental design (Figure 2.3), three
standard reference materials were analyzed and correlated using statistical
methodology. This treatment allows environmental samples to be fingerprinted in
a new way, providing a method for analyzing samples from contaminated waste

sites and oil spills.

8.1.1 Fluorescence and Fluorescence Quenching

Fluorescence is the process by which molecules absorb radiation and are
promoted to a higher energy excited state, then return to the ground state via the
emission of a photon. Most PAHs are natively fluorescent because the electrons
within the aromatic rings are delocalized and can be readily excited. The planar
structure of PAHs restricts vibrational relaxation from the excited state, thus
increasing the probability of relaxation through emission of a photon. Depending
on the number and arrangement of the aromatic rings, the resulting fluorescence
spectrum is highly characteristic of the PAH. For example, many altemant PAHs

exhibit fluorescence spectra with well-defined vibrational fine structure, whereas

239

most nonaltemant PAHs have broad spectra with few structural features (Figure
8.1) [1 ,2].

Fluorescence quenching can be defined simply as any process that
causes a reduction in the observed fluorescence power [3-5]. This reduction can
be the result of trivial processes, such as primary or secondary absorption,
refractive index effects, etc. These trivial processes can be minimized by using a
cell with a small optical path length, such as a capillary flow cell [6]. Of more
interest for environmental applications are the quenching processes that involve
specific interactions between the fluorophore and another molecule known as the
quencher. Such quenching can occur by two mechanisms: static and dynamic.
Static quenching occurs when the ground-state fluorophore and ground-state
quencher form a stable complex [3,4]. This new complex may exhibit different
spectral characteristics than the fluorophore, resulting in a reduction in the
fluorescence power at the selected excitation and emission wavelengths.
Dynamic quenching occurs when an excited-state fluorophore collides with and
transfers energy to a ground-state quencher [3,4]. The excited-state quencher
subsequently returns to the ground state via a nonradiative path (i.e. vibrational

relaxation). Dynamic quenching is quantified by the Stem-Volmer equation:

0 O
EL=Z=1+k t°C =1+K C (81)
(1)1 Pf d f q q q '

240

Figure 8.1: Fluorescence spectra of (A) pyrene and (B) fluoranthene, which
demonstrates the spectroscopic differences between altemant and nonaltemant

polycyclic aromatic hydrocarbons

241

 

 

nfim

afim

n:m

 

n5?

Es zeozm4m>§>
o:

D

owv

b

00v

D]

“3*

aim

nKm

aim

 

 

 

 

 

 

 

 

 

E 2:9“.

EONBOSBHORWA

242

where (b? and (b, are the quantum efficiencies of the fluorophore in the absence

and presence of a quencher, respectively. The quantum efficiencies are directly
proportional to the measured fluorescence powers (P19 and Pg, respectively)
provided that the source power, efficiency of optical irradiation and collection,
and fluorophore absorbance remain constant. Under these conditions, the ratio
of the powers can be related to the molar concentration of the quencher (Cq), the

bimolecular rate constant for dynamic quenching (kd), the fluorescence lifetime of
the unquenched fluorophore (12?), and the Stem-Volmer quenching constant (Kq).

The Stem-Volmer quenching constant is characteristic of the fluorophore-
quencher interactions and is a direct measure of the efficiency of the
fluorescence quenching process. A graph of the ratio of powers versus the
concentration of the quencher should yield a straight line with a slope equal to
the quenching constant and an intercept of unity.

Although a large number of quenchers for PAHs have been identified, very
few of them have been characterized in sufficient depth and detail to permit their
routine use in environmental applications [5]. Initial studies by Sawicki et al. [7]
showed that nitromethane, which acts as an electron acceptor, selectively
quenches the fluorescence of altemant PAHs. Subsequent studies by Acree et
al. [8-10] demonstrated that this so-called “nitromethane selective quenching
rule” is broadly applicable with only a few exceptions. A quantitative study by
Ogasawara et al. [11] revealed that the Stem-Volmer quenching constants of
nitromethane are 33 to 100 times greater for altemant than for nonaltemant

isomers. In contrast, recent investigations by Goodpaster and McGuffin [12]

243

demonstrated that amines, which act as electron donors, are selective quenchers
for nonaltemant PAHs. The Stem-Volmer quenching constants of
diisopropylamine are typically 15 to 45 times greater for nonaltemant than for
altemant isomers. In the present study, selective fluorescence quenching by
nitromethane and diisopropylamine are combined with high-efficiency capillary
liquid chromatography for detection of altemant and nonaltemant PAHs in
environmental samples. This approach is advantageous for the improvement of
qualitative and quantitative analysis by removing potential interferences. In
addition, PAH profiling may aid in the identification of sample origin, either for
environmental and health-hazard documentation or for geological studies on soil

sedimentation [1 3-15].

8.2 Experimental Methods

8.2.1 Materials

Three certified reference materials which contain a series of PAHs have
been chosen for analysis (Table 8.1). The first sample is a standard (EPA 610,
Supelco) that contains sixteen PAHs classified as priority pollutants by the United
States Environmental Protection Agency (US. EPA). The second sample is an
extract of Wheeling Pittsburgh medium crude coke oven tar (SRM 1647, National
Institute of Standards and Technology), which is a natural combustion-related
mixture of PAHs [16]. Both of these samples were used as received.

The third sample is a contaminated soil/sediment from the southern
branch of the Elizabeth River near Norfolk, VA (CRM104-100, Resource

Technology Corporation). This sample required extraction of the PAHs

244

 

 

m_nm__m>m «O: fl (2 n

mm:_m> vatoo .0: 65 385.228 c_ :96 mcozmzcmocoo .

 

 

 

 

 

 

2.” new odom 22525328
03, F2. 32 ocmsasoésiocmuc.
so: ..<2 38 2295233225
2.... man 32 mcmseaficmm
6th 6.3 3:: 222225622
$0.9 “Arne odow 2229822028
82 0.8 92: 2225
8s «.2 0.2: 22952328
3: 3m m8 225
new EN 38 22228.2
31 , v.3 32 2295.2
33 3533 3.53
832.25 32 2% So Em IE
:8 8.22280 22“. 828.20 222%

 

 

.m_m:9mE 8:222 coEtoo E 31.1% 309823: oszoLm o__o>o>_oa ho cozmbcoocoo

n...» 033.

245

according to US. EPA Method 3540c (Soxhlet Extraction) prior to analysis. The
solvents used in the Soxhlet apparatus were pesticide-grade acetone and
hexane (Burdick and Jackson Division, Baxter Healthcare) mixed in a 1:1 (v/v)
ratio. After reflux for 24 h, the yellow extract was transferred to a Kudema-
Danish evaporator (Kontes) and the volume was reduced from 400 to 8 mL over
a 2 h period. The resulting sample was then transferred, via a Pasteur pipette, in
small aliquots to a conical vial. The sample was evaporated to dryness with a
nitrogen stream, resulting in 0.63 g of solid material from 10.23 g of soil. The
solid was reconstituted in 1.0 mL of pesticide-grade methylene chloride (Burdick
and Jackson Division, Baxter Healthcare).

Two quenchers were chosen for these studies based upon their previously
reported selectivity for altemant and nonaltemant PAHs. Nitromethane (EM
Science) was volumetrically diluted with pesticide-grade methanol (Baxter
Healthcare, Burdick and Jackson Division) to yield a 2% (v/v) solution.
Diisopropylamine (Aldrich) was volumetrically diluted with pesticide-grade
acetonitrile (Baxter Healthcare, Burdick and Jackson Division) to yield a 50%

(v/v) solution.

8.2.2 Experimental System

Each of the samples was analyzed on the system shown in Figure 2.3,
which was constructed and characterized in house [17]. For these experiments
the sample was split 1:42 to provide an injection volume of approximately 24 nL.
The sample constituents were then separated on a fused-silica capillary (Hewlett-

Packard, 200 um i.d., 320 um o.d., 1.5 m length) that was packed with a 5 pm

246

 

octadecylsilica stationary phase (Shandon, Hypersil C18, 115,000 theoretical
plates), as described previously [18]. The column was immersed within a water
bath maintained at 24 °C to minimize the effect of temperature fluctuations on the
separation. The column effluent was combined and thoroughly mixed with the
quencher solution, which was delivered by a syringe pump (Model 140,
PE/Applied Biosystems) at a nominal flow rate of 1.0 pL/min.

The PAHs were then detected by laser-induced fluorescence in a fused-
silica capillary flow cell (75 um i.d., 360 um o.d, 1.0 m length, Polymicro

Technologies) using the system described in Section 2.2.3.3.

8.2.3 Data Analysis

Data from the chromatographic separation of each sample were integrated
over the wavelength range of 350 to 564 nm by using the Spectramax software
(Spectramax for Windows, v3.1, Instruments SA). These chromatograms were
used for visual display, for discernment of fluorescence quenching behavior and,
where possible, for identification of the individual PAHs. For correlation, it was
necessary to normalize the abscissa such that the known PAHs had the same
retention times in each chromatogram. The resulting chromatograms were then
exported as ASCII files into the statistical analysis software (SigmaStat, Version
1.02, SYSTAT). The chromatograms were correlated with one another, in a
point-by-point manner, using the product-moment correlation method [19,20].
This method is useful to establish the extent of association or similarity between
two chromatograms, both of which are regarded as independent variables. This

parametric method assumes that the association (if any) is linear and that the

247

residuals are normally distributed with constant variance. The resulting scatter
plot shows the relationship between the relative fluorescence power or
concentration of the PAHs in the two samples. The correlation coefficient (r)
quantifies the degree of similarity, and the corresponding P-value expresses the
statistical reliability of the results. This same approach can be used to examine
the correlation of fluorescence spectra in order to verify the identity and purity of

PAHs.

8.3 Results and Discussion

This experimental approach described above provides a wide range of
information that can be used to identify the individual PAHs, including
chromatographic retention time, fluorescence emission spectra, and Stem—
Volmer quenching constants. In addition, it provides many ways to uniquely
profile the distribution of PAHs in the sample, including chromatograms at
individual fluorescence wavelengths, chromatograms at integrated fluorescence
wavelengths, chromatograms with fluorescence quenching of altemant PAHs by
nitromethane, and chromatograms with fluorescence quenching of nonaltemant
PAHs by diisopropylamine. The results for each sample are discussed in detail

in the following sections.

8.3.1 Standard (EPA 610)

A chromatogram of the standard (EPA 610) is shown in Figure 8.2A. The
identity of each PAH was confirmed by comparison of the retention time and

fluorescence spectrum with authentic standards [1,21]. Of the sixteen known

248

components in this sample, only eleven are fluorescent with excitation at 325 nm
and emission at 350 — 564 nm. Several of the smaller PAHs, including
naphthalene, acenaphthylene, acenaphthene, fluorene, and phenanthrene, are
not excited efficiently by the helium-cadmium laser. The PAHs ranging from
anthracene to benzo[ghr]perylene are readily detected in spite of the relatively
small mass injected (2.4-4.8 ng). The fluorescence power of each PAH is a
function of the concentration (Table 8.1), as well as the molar absorptivity and
quantum efficiency at the selected wavelengths for excitation and emission. A
chromatogram of the standard with fluorescence quenching by nitromethane is
shown in Figure 8.2B. It is immediately evident that the nonaltemant PAHs
(fluoranthene, benzo[b]fluoranthene, benzo[klfluoranthene, and indeno[1,2,3-
cd]pyrene) substantially retain their original fluorescence power. In contrast, the
altemant PAHs are significantly quenched. This observation is consistent with
the previously reported Stem-Volmer constants of 0.07 and 0.64 M'1 for the
representative nonaltemant PAHs fluoranthene and benzo[b]fluoranthene, and
94 and 61 M'1 for the representative altemant PAHs pyrene and benzo[a]pyrene
[17]. It is also noteworthy in Figure 8.1 B that benzo[kjfluoranthene appears to be
more highly quenched than the other nonaltemant PAHs. This is consistent with
differences in the electron-donating ability of the aromatic system to the
nitromethane quencher [6,11]. The gas-phase ionization energy [22] of
benzo[quluoranthene (8.167 eV) is substantially less than that of fluoranthene
(8.466 eV) and benzo[b]fluoranthene (8.410 eV), which suggests that it is a

better electron donor. It is, in fact, more similar to the altemant PAHs

249

Figure 8.2: Chromatogram of standard PAHs (EPA 610) without (A) and with
fluorescence quenching by nitromethane (B) and diisopropylamine (C). Column:
1.5 m x 200 um i.d. fused-silica capillary, packed with 5 pm Shandon Hypersil
C18. Mobile phase: methanol, 1.0 uL/min, 24 °C, with post-column addition of
(A) methanol, 1.0 uL/min, (B) 2% (v/v) nitromethane in methanol, 1.0 uL/min, (C)
50% (v/v) diisopropylamine in acetonitrile 1.0 uL/min. Laser-induced
fluorescence detection: 325 nm excitation, 350—564 nm emission. Solutes: (1)
anthracene, (2) fluoranthene, (3) pyrene, (4) benz[a]anthracene, (5) chrysene, (6)
benzo[b]fluoranthene, (7) benzo[k]fluoranthene, (8) benzo[a]pyrene,

(9) dibenz[a,h]anthracene, (10) indeno[1,2,3-cd]pyrene, (11) benzo[ghr]perylene.

250

 

 

 

3 m2:

 

 

 

: ow

 

h

 

m.» 2:9".

l a . r l . . .. 5.4
an a . 1.} all .. AP». r f
In M... Di . GI: .. .. I Fllf
n- . P . Auir .mrllm Wu... MM." ”.va Ar...
R C T...» ..l .11 ..:il 5r.“ (1'. Fly)“
w . 11 .MW ll. Ll- Fl. M.Fll.
11) ‘ I at . H- -HWN ..HI.. I 14!:

251

 

benz[a]anthracene (8.111 eV) and chrysene (8.261 eV). As a result, the
fluorescence of benzo[k]fluoranthene is quenched better by nitromethane, than
by diisopropylamine (vide infra).

A chromatogram of the standard with fluorescence quenching by
diisopropylamine is shown in Figure 8.2C. In general, the nonaltemant PAHs are
moderately quenched and the altemant PAHs are unaffected. This observation
is consistent with the previously reported Stem-Volmer constants of 17.1 and
21.2 M" for the representative nonaltemant PAHs fluoranthene and
benzo[b]fluoranthene, and 1.2 and 0.47 M" for the representative altemant PAHs
pyrene and benzo[a]pyrene [12]. Benzo[k]fluoranthene is an interesting
exception to this general trend, as it is relatively unquenched by
diisopropylamine. Its behavior, again, is more similar to the altemant PAHs
benz[a]anthracene and chrysene than to the other nonaltemant PAHs

fluoranthene and benzo[b]fluoranthene

8.3.2 Coal-Derived Fluid (SRM 1597)

A chromatogram of the coal-derived fluid (SRM 1597) is shown in Figure
8.3A. The identity of each PAH was again confirmed by comparison of the
retention time and fluorescence spectrum with authentic standards [1,21]. The
concentrations of the known PAHs are summarized in Table 8.1. In addition,
there are a large number of unidentified PAHs that elute in the time range from
0.75 to 1.5 h. As the fluorescence spectra of many of these PAHs have
vibrational fine structure similar to altemant PAHs, these may be methylated or

other alkylated analogs. This is consistent with the combustion-related origin of

252

Figure 8.3: Chromatogram of PAHs in a coal-derived fluid (SRM 1597) without
(A) and with fluorescence quenching by nitromethane (B) and diisopropylamine
(C). Solutes: (12) unknown, possibly dibenzofluoranthene or
naphthofluoranthene isomer, (13) dibenzo[def,mno]chrysene, (14) unknown,
possibly dibenzofluoranthene or naphthofluoranthene isomer. Other

experimental conditions and solutes as described in Figure 8.2.

253

 

3 m2:

 

 

 

 

 

 

 

 

 

IGI

a; w; v; N; P md 06 c o Nd
3 Z N F
N_. v m
\.
J’I] J] [<4] ”ll 1’111'5’1 1 ]
3 F
m: \. o
m
2 / mx
3 5 m \
w v m F
9
n
m
me 26E

254

this sample and is supported by a prior detailed analysis by gas chromatography
with mass spectrometric detection, which was able to identify many methylated
PAH isomers [16]. There are two additional large peaks that can be observed
between 1.5 and 2.0 h. On the basis of the fluorescence spectra and the prior
analysis [16,23] these may be dibenzofluoranthene or naphthofluoranthene
isomers. In addition, dibenzo[def,mno]chrysene (anthanthrene) can be identified
unambiguously from the fluorescence spectrum [1].

A chromatogram of the coal-derived fluid with fluorescence quenching by
nitromethane is shown in Figure 8.3B. All of the known nonaltemant PAHs are
relatively unaffected and the altemant PAHs are quenched in the same manner
as for the standard sample in Figure 8.23. The many unidentified PAHs are
highly quenched by nitromethane, which is consistent with their tentative
assignment as alkylated altemant isomers. Acree et al. [8,9] have reported that
both altemant and nonaltemant PAHs with alkyl substituents preserve the
inherent quenching behavior of the parent PAHs. In addition, the fluorescence
power of the two PAHs tentatively assigned as dibenzofluoranthene or
naphthofluoranthene isomers is largely retained, as would be expected for
nonaltemant PAHs. However, the partial quenching by nitromethane indicates
that these PAHs are more similar to benzo[k]fluoranthene, which has significant
altemant character. This observation is in agreement with the previous results of
Tucker and Acree [24] for selected naphthofluoranthenes. Finally, note that the

qualitative and quantitative analysis for all of the altemant PAHs has been

255

substantially improved by fluorescence quenching with nitromethane in Figure
8.3B.

A chromatogram of the coal-derived fluid with fluorescence quenching by
diisopropylamine is shown in Figure 8.30. All of the known altemant PAHs are
relatively unaffected and the nonaltemant PAHs are quenched in the same
manner as for the standard sample in Figure 8.2A. The unidentified PAHs that
have been tentatively assigned as alkylated altemant isomers are not
significantly quenched by diisopropylamine, as would be expected. The PAHs
tentatively assigned as dibenzofluoranthene or naphthofluoranthene isomers are
slightly quenched. Although this behavior is unexpected for nonaltemant PAHs,
it is fully consistent with the behavior of benzo[k]fluoranthene. Finally, because
of the complexity of this sample with many altemant PAHs, the chromatogram is
not greatly simplified by fluorescence quenching with diisopropylamine.
However, some minor improvement in qualitative and quantitative analysis may

be obtained.

8.3.3 Contaminated Soil (CRM104-100)

A chromatogram of the contaminated soil (CRM104-100) is shown in
Figure 8.4A. The identity of each PAH was again confirmed by comparison of
the retention time and fluorescence spectrum with authentic standards [1,21].
The concentrations of the known PAHs are summarized in Table 8.1. Like the
coal-derived fluid, this sample has numerous PAHs in the time range of 0.75 to

1.5 h that appear to be alkylated altemant isomers. Moreover, the PAHs that

256

Figure 8.4: Chromatogram of PAHs in a contaminated soil (CRM104-100)
without (A) and with fluorescence quenching by nitromethane (B) and
diisopropylamine (C). Other experimental conditions and solutes as described in

Figures 8.2 and 8.3

257

 

 

0...

EV NEE.
v... NP F

P I I

 

NF

 

 

NF

NP

 

 

258

 

 

 

 

swan.

have been tentatively identified as dibenzofluoranthene or naphthofluoranthene
isomers are also present.

Chromatograms of the contaminated soil with fluorescence quenching by
nitromethane and diisopropylamine are shown in Figures 848 and 8.40,
respectively. The altemant and nonaltemant PAHs are quenched in the

expected manner, as described above for the standard and coal-derived fluid.

8.3.4 Statistical Correlation Analysis

In order to demonstrate the utility of fluorescence and fluorescence
quenching for profiling PAHs in environmental samples, the chromatograms were
correlated by using the product—moment method [19,20]. The complete
chromatograms were correlated, point by point, which provides a more detailed
comparison than the use of peak maxima or peak areas alone. The correlation
graphs for three representative cases are shown in Figure 8.5. When samples
are derived from exactly the same origin, the relative fluorescence power or
concentration of PAHs in each sample is identical. The resulting scatter plot
(Figure 8.5A) shows a high degree of correlation. Accordingly, the correlation
coefficients (r) for identical or replicate samples are typically 0.99 or higher.
When samples are of similar or related origin, many of the same PAHs may be
present but at different concentrations. This results in an intermediate degree of
correlation (Figure 8.58), with typical values of r in the range of 0.90 to 0.50.
Finally when samples are of distinctly unrelated origin, the disparate distribution
of PAHs will result in little or no correlation (Figure 8.50), with typical values of r

less than 0.50. In all cases, valid conclusions can be drawn about the identity or

259

 

 

0.9 4
0.8 r
0.7 '1
0.6 4
0.5 r

0.4 - .,

0.3 7 /

0.1 -/ ‘

o , . . . 1

0 0.2 0.4 0.6 0.8 1
COAL-DERIVED FLUID

COAL-DERIVED FLUID
o

 

 

Figure 8.5A: Scatter plot demonstrating three differing degrees of product-
moment correlation. (A) Coal-derived fluid (Figure 8.3A) versus coal-derived fluid

(Figure 8.3A), r = 1.000, P = 000 x 104930

260

 

0.9 t

0.8 '

0.7 ‘

0.6 “

0.5 ‘

CONTAMINATED SOIL

 

 

0.4 0.6 0.8 1
COAL- DERIVED FLUID

 

Figure 8.53: Scatter plot demonstrating three differing degrees of product—
moment correlation. (B) Contaminated soil (Figure 8.4A) versus coal-derived fluid

(Figure 8.3A), r = 0.877, P = 2.83 x 10‘113

261

 

 

 

0.9 ‘
0.8 “
0.7 ‘
0.6 1

0.5 J

 

0.4

COAL-DERIVED FLUID

0.3 . 0
0.2 '

0.1 Q 0. . . .

 

0 0.2 0.4 0.6 0.8 1
STANDARD

Figure 8.50: Scatter plot demonstrating three differing degrees of product—
moment correlation. (C) Coal-derived fluid (Figure 8.3A) versus standard (Figure

8.2A), r = 0.120, P = 0.0241

262

 

Table 8.2:

Correlation coefficient (r) of the product-moment method for

chromatograms obtained by using laser-induced fluorescence detection.

 

 

 

 

 

 

 

 

 

SAMPLE Standard Coal-Derived Fluid Contaminated Soil
EPA 610 SRM 1597 CRM104-100
:‘F‘j‘xdgg 1.000 0.120 0.214
Coats-33814331010 0.120 1.000 0.877
“agmfifigm' 0.214 0.877 1.000
Table 8.3: Correlation coefficient (r) of the product—moment method for

chromatograms obtained by using laser-induced fluorescence detection with

quenching by nitromethane.

 

 

 

 

 

 

 

 

 

SAMPLE Standard Coal-Derived Fluid Contaminated Soil
EPA 610 SRM 1597 CRM104-100
:‘S‘Rdgg 1.000 0.1 1 1 0.144
“aggafiesgf'um 0.1 1 1 1.000 0.977
00321710831030" 0.144 0.977 1.000
Table 8.4: Correlation coefficient (r) of the product-moment method for

chromatograms obtained by using laser-induced fluorescence detection with

quenching by diisopropylamine.

 

 

 

 

 

SAMPLE Standard Coal-Derived Fluid Contaminated Soil
EPA 610 SRM 1597 CRM104-100
23205? 1.000 0.104 0.146
Goals-231117953; '"id 0.104 1 .000 0.884
60333103131130" 0.146 0.884 1.000

 

 

 

 

263

 

 

 

origin of the samples when the P-value for the product moment correlation is less
than 0.05 (95% CL.) Table 8.2 summarizes the results of the product moment
correlation for the three samples examined with fluorescence detection alone. It
is apparent that there is little correlation between the standard and the coal-
derived fluid or contaminated soil, despite the common PAHs found in each
sample. As many of the PAHs in the more complex samples are not found in the
standard sample, these samples present the unique challenge of profiling with
limited information. This correlation method is well suited for such an analysis
since no a prior information about the samples is necessary, allowing for
specialists and non-specialists to employ the technique. The coal-derived fluid
and contaminated soil show an intermediate degree of correlation (r = 0.877),
which is consistent with their combustion-related origin and the similar
appearance of their chromatograms in Figures 8.3A and 8.4A.

This approach for PAH profiling becomes even more versatile and
powerful when combined with selective fluorescence quenching. The results of
the product moment correlation for the three samples with fluorescence
quenching by nitromethane are'summarized in Table 8.3. As the altemant PAHs
are selectively quenched, this correlation discriminates on the basis of the
distribution of nonaltemant PAHs in the samples. When viewed on this basis, the
standard is still distinctly different from the coal-derived fluid or contaminated soil.
However because of their common combustion-related origin [16], the coal-
derived fluid and contaminated soil are highly correlated in nonaltemant

character (r = 0.977). The results of the product—moment correlation with

264

 

 

 

fluorescence quenching by diisopropylamine are summarized in Table 8.4.
These results confirm that the dissimilarities between the coal-derived fluid and
the contaminated soil lie in the distribution of altemant PAH isomers (r = 0.884).
This approach has been successfully applied to a variety of other petroleum-

based samples, including gasoline, motor oils, petrolatum jellies, etc. [25].

8.4 Conclusions

In summary, fluorescence and selective fluorescence quenching appear to
provide complementary information for profiling PAHs in complex samples. The
single wavelength or total fluorescence emission offers broad-based information
about the PAH distribution. In contrast, fluorescence quenching by nitromethane
allows selective discrimination of the nonaltemant PAHs and quenching by
diisopropylamine allows selective discrimination of the altemant PAHs. Only
when all of these profiles indicate a high degree of correlation can it be

confidently concluded that two environmental samples are of the same origin.

265

8.5 References

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

W. Karcher, R.J. Fordham, J.J. Dubois, P.G.JM. Glaude, J.A.M. Ligthart,

Spectral Atlas of Polycyclic Aromatic Hydrocarbons, Reidel Publishing,
Boston, MA, USA, 1985. '

J. Malkin, Photophysical and Photochemical Properties of Aromatic
Compounds, CRC Press, Boca Raton, FL, USA, 1992.

R. Badley, Fluorescence Spectroscopy, Plenum Press, New York, NY,
USA, 1983.

JR Lakowicz, Principles of Fluorescence Spectroscopy, Plenum Press,
New York, NY, USA, 1983.

V.L. McGuffin, J.V. Goodpaster, in RA. Meyers (Editor), Encyclopedia of
Environmental Analysis and Remediation, Wiley, New York, NY, USA,
1998, p. 3815-3831.

S.-H. Chen, C.E. Evans, V.L. McGuffin, Anal. Chim. Acta 246 (1991) 65-
74.

E. Sawicki, T.W. Stanley, W.C. Elbert, Talanta 11 (1964) 1431-1439.

S.A. Tucker, J.M. Griffin, W.E. Acree, Jr., J.C. Fetzer, M. Zander, O.
Reiser, A. De Meijere, I. Murata, Polycyclic Aromat. Compd. 4 (1994) 141
161.

SA. Tucker, J.M. Griffin, W.E. Acree, Jr., M.J. Tanga, J.E. Bupp, T.K.
Tochimoto, J. Lugtenburg, K. Van Haeringer, J. Comelisse, P.C. Cheng,
L.T. Scott, Polycyclic Aromat. Compd. 4 (1994) 161-172.

W.E. Acree, Jr., 8. Pandey, S.A. Tucker, J.C. Fetzer, Polycyclic Aromat.
Compd. 12 (1997) 71-123.

F.K. Ogasawara, Y. Wang, V.L. McGuffin, Appl. Spectrosc. 49 (1997) 1-7.

266

 

[12] J.V. Goodpaster, V.L. McGuffin, Anal. Chem. 72 (2000) 1072-1077.

[13] D. Broman, A. Colmsjo, B. Ganning, C. Néf, Y. Zebuhr, Environ. Sci.
Technol. 22 (1988) 1219-1228.

[14] T. Vo-Dinh, Chemical Analysis of Polycyclic Aromatic Hydrocarbons,
Wiley, New York, NY, USA, 1989.

[15] L. Nondek, M. Kuzliek, S. Krupicka, Chromatographia 37 (1993) 381 -386.

[16] SA. Wise, B.A. Benner, G.D. Byrd, S.N. Chesler, R.E. Rebbert, M.N.
Schantz, Anal. Chem. 60 (1988) 887-894.

[17] J.V. Goodpaster, V.L. McGuffin, Appl. Spectrosc. 53 (1999) 1000-1008.

[18] J.C. Gluckman, A. Hirose, V.L. McGuffin, M. Novotny, Chromatographia
17 (1983) 303-309.

[19] J.L. Devore, Probability and Statistics for Engineering and the Sciences,
4th ed., Duxbury Press, Pacific Grove, CA, USA, 1995.

[20] RM. Thorndike, Correlational Procedures for Research, Gardner Press,
New York, NY, USA, 1978.

[21] LC. Sander, S.A. Wise, J. l-ligh Resolut. Chromatogr. 8 (1985) 248-255.

[22] RA. Hites, W.J. Simonsick, Calculated Molecular Properties of Polycyclic
Aromatic Hydrocarbons, Elsevier, Amsterdam, The Netherlands, 1987.

[23] SA. Wise, B.A. Benner, H. Liu, G.D. Byrd, A. Colmsjo, Anal. Chem. 60
(1988) 630-637.

[24] SA. Tucker, W.E. Acree, Jr., Appl. Spectrosc. 46 (1992) 1388-1392.

[25] J.V. Goodpaster, S.B. Howerton, V.L. McGuffin, J. Forensic Sci. 46 (2001)
1358-1371 .

267

 

 

Chapter 9: Conclusions and Future Directions

Over the past fifty years, liquid chromatography has been characterized
via theoretical treatments, as well as empirical measurements. However, the
molecular contributions to retention have been limited primarily to investigations
of the molar enthalpy associated with retention. Virtually no experimentation has
sought to characterize the change in molar volume. The change in molar volume
is important because it provides insight into how molecules interact with the
stationary phase. In addition to the molar volume, the kinetics of retention have
also been largely overlooked. Although some investigators have sought to probe
the kinetics of retention, the only author to publish extensively on the subject has
been Guiochon. The kinetics of retention are important because they provide
insight into the rate at which transitions occur, as well as the energy barriers that
exist for a molecule to move from one phase to another. By quantitating both the
thermodynamics and kinetics together, a better description of retention is
possible, providing for a more accurate comparison of solutes and stationary
phases to one another.

The research presented in this dissertation has attempted to overcome
some of the past theoretical and experimental difficulties by studying the
retention event in situ. Through the use of newly adapted mathematical analysis
and experimental design, solute transfer in reversed-phased liquid
chromatography has been studied as a function of several molecular variables
for a series of polycyclic aromatic hydrocarbons and their nitrogenous

derivatives.

268

9.1 Experimental Design and Theoretical Development

Chapters 2 and 3 detail the experimental design and the choice of
mathematical functions necessary to study retention in reversed-phase liquid
chromatography. The use of on-column detection at multiple locations allows
measurements to be made in situ. Measurements made in this manner allow for
the thermodynamics and kinetics to be studied without the error that arises from
extra column effects.

As demonstrated in Chapter 3, the extraction of the peak profiles from the
chromatogram can contribute to error during the fitting process. From the
simulation studies, statistical moment analysis was shown to be more sensitive to
the integration interval, number of points, and the noise when compared to the
exponentially modified Gaussian (EMG) equation.

The most pressing limitation in the analysis of peak profiles is the lack of a
multiple site model. In order for materials that have multiple retention sites to be
characterized, a more robust mathematical model is necessary to distinguish one
site from another. As illustrated in Chapter 5, the current form of a bi-
exponentially modified Gaussian equation provides for an inadequate analysis of
a two-site system. The primary cause for this limitation is the implicit assumption
that the sites are equal in number. As shown, there is no weighting factor for the
relative number of sites. As a result, the software used to fit the peak profiles
fails to distinguish different exponential contributions. This failure manifests as
an underflow or overflow error and ultimately leads to rate constants that are the

same for both sites.

269

A user-defined function that convolutes the requisite functions in either
Fourier or Laplacian space would be a straightforward method to circumvent this
problem since current software relies completely on algebraic notation. Barring
this approach, a second method to improve the analysis of multiple site data
would be the development of a completely new mathematical function from first
principles. Taking into account both the distribution of the sites (i.e. relative
abundance) as well as the physical characteristics of these sites (i.e. energies of
interaction, diffusion coefficients, etc.) could provide an improvement over the

currently derived function.

9.2 Retention in Octadecylsilica

Chapters 4 through 6 detail the thermodynamics and kinetics of a series of
polycyclic aromatic hydrocarbons (PAHs) and nitrogen containing PAHs (NPAHS)
in monomeric and polymeric octadecylsilica (ODS) using methanol and
acetonitrile mobile phases. As shown, the ring number, annelation, and planarity
of the solutes influences the thermodynamic and kinetic values. In addition, the
inclusion of a nitrogen into the PAH backbone allows for adsorption on the
underlying support. These adsorptive contributions are exacerbated in
acetonitrile, an aprotic mobile phase. The rate constants for four NPAHS are
smaller by two to four orders of magnitude in acetonitrile than in methanol. In
addition, the bonding density of the stationary phase was shown to alter the
thermodynamics and kinetics of retention. As illustrated, polymeric ODS (5.4

umonz) causes the retention factors to increase up to a factor of six, relative to

monomeric ODS (2.7 umol/mz). In addition, the changes in molar enthalpy and

270

volume are more negative for the planar PAHs in the polymeric ODS. Thus,
these studies demonstrate how the solute structure, mobile phase, and stationary
phase affect both the thermodynamics and kinetics of retention.

In order for a complete and quantitative description of retention to be
presented, a series of experiments beyond those presented herein is still
necessary. The first experiment should evaluate the effect of mobile phase
velocity on the thermodynamics and kinetics of retention. Except in the case of
the NPAHs, an implicit assumption in this dissertation is that a single kinetic site
contributes to the tailing of peak profiles. When the kinetic events are slow,
relative to the separation time, they contribute to asymmetric tailing. By
systematically altering the velocity, and observing how the asymmetry changes,
the assumption of a single kinetic site could be directly tested. If the rate
constants change as a function of velocity, then the system likely contains
multiple kinetics site, of which a limited number have been probed.

A second series of experiments that could lead to a more comprehensive
explanation of retention would be a systematic study of the effect of mobile
phase composition on the derived values. Methanol was used for the majority of
experiments, but as demonstrated in Chapter 6, acetonitrile has a dramatic effect
on the retention of the NPAHs. A series of alcohols with varying chain length
could be used to make comparisons between mobile phases and study their
effect on the stationary phase.

Similar to varying the mobile phase, a series of stationary phases could be

used to study the effect of chain length, synthetic method, and polarity on the

271

derived parameters. While preliminary work by Berendsen and de Galan used a
series of materials for a similar study [1], they focused on alkylated homologues
as probes of a partition mechanism. A study with the NPAHs using the varying
chain lengths for the alkyl-silica bonded phases could provide information about
the depth to which these solutes penetrate. Such an investigation would help to
identify the minimum length necessary to minimize adsorptive interactions for
commercial separations.

In addition to changing the mobile or stationary phase, modifications to the
instrument could also be undertaken. Such modifications would allow the
retention event to be probed on an even finer scale. For example, the addition of
near field optics or surface plasmon resonance could allow molecules to be
studied as they interact with the surface. Such measurements would likely
require the use of smaller grafted chains than octadecylsilica given the distance
limitations of these techniques. In addition, an on-line Raman system could
supplement the current fluorescence systems and expand upon the

investigations conducted by Pemberton and co-workers [2-5].

9.3 Retention in Soil

The study of pollutants, such as PAHs, in the environment is important
since it can lead to better models for risk assessment. The goal of using soil as a
stationary phase was to test whether current theories of retention are accurate,
and to gather enough data to present a first approximation of the molecular

parameters that lead to solute transport.

272

As shown in Chapter 7, the retention of two PAHs suggests that a partition
mechanism is present in soil. However, the experiments were carried out in pure
methanol with large concentrations (~10’3 M). These concentrations are well
above what would be found in an aqueous environment and may not represent
retention in a real world environment.

In addition, the data for the nitroalkanes supports the presence of a
partition mechanism. However, the data for the nitroalkanes and the .
alkylbenzenes also indicate the presence of an exclusion mechanism. The data
demonstrate that over the range of tested mobile phases the exclusion
mechanism predominates. Although the systems could not be characterized in
the same manner as the octadecylsilica phase, the data do provide a first
glimpse into the molecular retention of these solutes in soil, demonstrating a
difference in mechanism relative to synthetic phases.

The most promising direction for this research into soil retention would be
an improvement in the sieving and packing method for capillary columns. Since
many experimental difficulties arose from the heterogeneous nature of the soil,
isolating a fraction with a smaller size distribution would allow columns to be
more uniformly packed. For this improvement to be seen, though, swelling
materials such as clays would need to be removed or controlled to prevent
blockage of the quartz wool frit. However, the removal of the clay could
significantly alter the retention characteristics and thus may need to be carefully

analyzed through a more exhaustive series of batch isotherms.

273

A second approach to study retention in natural materials is to synthesize
a stationary phase that mimics soil, but is not hindered by the heterogeneous
size distribution. One approach that has been used involves tethering Aldrich
humic acid to silica supports [6]. However, this material is still far from
representative of most soils and may not yield information about younger soils. A
more reasonable approach would be to tether natural humic and fulvic acids to
silica supports. However, this method fails to truly represent the complex
chemical nature of soil, and once again provides only a limited amount of
information.

Perhaps the best method of synthesis would be to use organisms that
contribute to the decomposition of animal and plant matter (i.e. white rot fungi).
These organisms could be used to process mixtures of lipids, amino acids, and
other naturally occurring materials. Using such mixtures as a food source, the
by-products of degradation could be isolated and tethered to chromatographic
supports, resulting in a material that is better than a truly synthetic phase, but
easier to manipulate than soil. However, the reproducible synthesis and
purification of such materials would be challenging. If identification of the
materials were important before chemical immobilization, such a process would

be laborious requiring significant mass spectral analysis.

9.4 Selective Fluorescence Quenching

A final use of capillary chromatography for environmentally relevant
samples is the development of selective fluorescence quenching detection. As

demonstrated in Chapter 8, qualitative information can be gathered about

274

unknown samples that contain PAHs based upon the choice of the quencher. As
shown, several environmental samples were differentiated using post-column
addition of the quencher with Pearson product moment correlation. The resulting
data demonstrate how this method can be used to compare samples for
similarities and differences. Such qualitative comparisons could help in
identifying the source of a given mixture.

This method could be adapted to aid in the identification of solutes that are
important to security experts (i.e. explosives, chemical weapons). In order to
apply this technique to these solutes, though, selective quenchers would have be
identified and characterized. To date, no reports of selective quenchers for
pesticides or chemical weapons have been published. However, work by
Goodpaster and McGuffin have investigated nitrated explosives for indirect

detection [7].

275

9.5 References

[1]

[2]

[3]

[4]

[5]

[6]

[7]

GE. Berendsen, L. de Galan, J. Chromatogr. 196 (1980) 21-37.

M.W. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5576-5584.

M.W. Ducey, Jr., C.J. Orendorff, J.E. Pemberton, L.C. Sander, Anal.
Chem. 74 (2002) 5585-5592.

C.J. Orendorff, M.W. Ducey, Jr., J.E. Pemberton, L.C. Sander, Anal.
Chem. 75 (2003) 3028-3036.

C.J. Orendorff, M.W. Ducey, Jr., J.E. Pemberton, L.C. Sander, Anal.
Chem. 75 (2003) 3037-3043.

L.K. Koopal, Y. Yang, A.J. Minnaard, P.LM. Theunissen, van Riemsdjik,
W.H., Colloids Surf. A 141 (1998) 385-395.

J.V. Goodpaster, V.L. McGuffin, Anal. Chem. 73 (2001) 2004-2011.

276

   

IIIIIIIIIIIIIIIII