Solvatochromic solvent polarity measurements, retention, and selectivity in reversed phase liquid chromatography

MISSING IMAGE

Material Information

Title:
Solvatochromic solvent polarity measurements, retention, and selectivity in reversed phase liquid chromatography
Physical Description:
xiv, 212 leaves : ill. ; 28 cm.
Language:
English
Creator:
Johnson, Bruce Philip, 1958-
Publication Date:

Subjects

Subjects / Keywords:
Liquid chromatography   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1986.
Bibliography:
Bibliography: leaves 201-211.
Statement of Responsibility:
by Bruce Philip Johnson.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000951379
oclc - 16952117
notis - AER3622
sobekcm - AA00004863_00001
System ID:
AA00004863:00001

Full Text














SOLVATOCHROMIC SOLVENT POLARITY MEASUREMENTS,
RETENTION, AND SELECTIVITY
IN REVERSED PHASE LIQUID CHROMATOGRAPHY






BY






BRUCE PHILIP JOHNSON


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


UNIVERSITY OF FLORIDA


1986



























This dissertation is dedicated
to my son, Garrett Chase, whose
arrival coincided with the
completion of this work.















ACKNOWLEDGMENTS


There are many people that I wish to acknowledge; in

his or her own way, each has contributed to my educational

progression. I would like to begin with my parents, Stan

and Connie Johnson, who constantly encouraged me to explore

my world and who instilled in me an unquenchable thirst for

knowledge. There are not many children who are fortunate

enough to grow up with a laboratory and a photographic

darkroom in their own basement!

My deepest gratitude is extended to Prof. Dr.

Christian Reichardt of Phillips-Universitat (Marburg, West

Germany), who so generously provided my advisor with

samples of the ET-30 dye, as well as providing helpful

comments during the preparation of two manuscripts.

I must also express my gratitude to the Eastman Kodak

Company, who funded 3 years of my graduate education

through the Kodak Fellow program, with no strings attached.

My advisor, Dr. John G. Dorsey, has been one of the

most enjoyable aspects of my graduate education. The many

hours we spent talking about everything from chromatography

to congealed desserts will always be treasured. He should

also be thanked for initiating my addiction to the










Wall Street Journal, though he was unable to turn me into

an oenophile.

I also want to acknowledge my fellow group members,

who, along with their superb sense of humor, have made

graduate school an experience I shall always cherish.

Lastly, without the love, patience, and support of my

wife, Bonnie, and her parents, Phil and Sylvia Reinstein,

the completion of this work would not have been possible;

this was especially so after the arrival of our son,

Garrett Chase, whose timely (?) arrival coincided with the

completion of this tome.















TABLE OF CONTENTS


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

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

LIST OF FIGURES .......................... ...............ix

ABSTRACT............................................... xiii

CHAPTERS


I


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

Mobile Phase Effects............................3
Stationary Phase Effects........................7
Empirical Measures of Solvent Polarity.........17
Analytical Application of the ET-30 Dye.........26

II SOLVATOCHROMIC SOLVENT POLARITY MEASUREMENTS...34

Experimental.................................. 34
Results ........................................ 42
Relationship Between Snyder's P' Polarity
Values and the ET(30) Scale...................59

II CORRELATIONS BETWEEN CHROMATOGRAPHIC
RETENTION AND MOBILE PHASE POLARITY..............65

Experimental..................................65
Results.. ................................... 67
Comparison with the "Carr Approach".............99

IV CORRELATIONS BETWEEN CHROMATOGRAPHIC
SELECTIVITY AND MOBILE PHASE POLARITY......... 104

Experimental.................................104
Introduction.................................105
Results ....................................... 108

V DISCUSSION AND CONCLUSIONS................... 136

Stationary Phase Effects...................... 150
Application of These Results..................156










Interfacial Tension Effects...................160
Suggestions for Future Research...............166

APPENDICES

A CHROMATOGRAPHIC RETENTION AND SELECTIVITY
DATA . ..................... ... .... 175

B MODIFICATION OF CURVE FITTER PROGRAM TO
ALLOW CALCULATION OF CONFIDENCE INTERVALS..... 196

C MODIFICATION OF CURVE FITTER PROGRAM
TO INTERPOLATE SPECTRAL PEAK POSITIONS........ 197

D SOLVATOCHROMIC SOLVENT POLARITY
MEASUREMENTS.................................. 198

REFERENCES. ............................. ...... .....201

BIOGRAPHICAL SKETCH................o ..........*...........212














LIST OF TABLES



Table Page

2-1 Effect of varying ET-30 concentration on
X and absorbance in 45/35/20 (v/v/v)
Me8 /ACN/H20. .. ......................... ..... ...... 36

3-1 Linear regression results for correlations
between log k' and either percent organic
modifier or ET(30) polarity.................... 78

3-2 Mean and median r2 values for correlations
shown in Table 3-1..............................95

3-3 Multiple linear regression between log k'
values and a, 8, and w*........................103

4-1 Squared correlation coefficients (r2) for
log a data with respect to percent organic
modifier (OM), mole fraction OM, and ET(30)
polarity.......................................118

4-2 Comparison of log a values as measured by
nitroalkanes and alkylbenzenes for a
Hamilton PRP-1 column..........................125

4-3 Correlations between log a and percent
organic modifier (OM), mole fraction organic
modifier (MF OM), or ET(30) polarity for
a Hamilton PRP-1 polymeric column..............126

5-1 Effect of increasing solute size upon
sensitivity to changes in ET(30) polarity
for alkylbenzenes.............. ....................159

5-2 Effect of increasing solute size upon
sensitivity to changes in ET(30) polarity
for halobenzenes.............................141

5-3 Comparison of slope and y-intercept values
for log k' versus ET(30) polarity for
phenanthrene................................... 145


vii










5-4 Comparison of slope and y-intercept values
for log k' versus ET(30) polarity for
ethylbenzene........................ ........... 145

5-5 Correlations between enthalpy of transfer
(AH) and ET(30) polarity values.................147

5-6 Ratio of slopes for a given solute and column
with methanol and acetonitrile as organic
modifiers......... ...... ............. 150

5-7 Intersection points for log k' versus ET(30)
for alkylbenzenes..............................154


viii














LIST OF FIGURES


Figure Page

1-1 Structure of the ET-30 dye molecule,
2,6-Diphenyl-4-(2,4,6-triphenyl-N-
pyridino)phenolate in the ground and
excited states..................................24

2-1 Beer's law plot for ET-30 dissolved in
45/30/10 methanol/acetonitrile/water (v/v/v)....37

2-2 Thermochromism of 4-nitroanisole in
33.3/33.3/33.4 methanol/water/acetonitrile
(v/v/v) ......................................... 40

2-3 Thermochromism of 4-nitroanisole in
33.3/33.3/33.4 methanol/water/acetonitrile
(v/v/v) . .. ............... 41

2-4 Representative UV/VIS absorbance spectrum of
4-nitroanisole in methanol.......................43

2-5 Measurements of f* dipolarity/polarizability
for methanol/water mixtures with respect to
percent methanol................................45

2-6 Measurements of f* dipolarity/polarizability
for methanol/water mixtures with respect to
mole fraction of methanol........................46

2-7 Measurements of w* dipolarity/polarizability
for acetonitrile/water mixtures with respect
to percent acetonitrile....................... 47

2-8 Measurements of w* dipolarity/polarizability
for acetonitrile/water mixtures with respect
to mole fraction acetonitrile..................48

2-9 Representative UV/VIS absorption spectrum
of the ET-30 dye dissolved in methanol..........51









2-10 Measurements of ET(30) polarity for methanol/
water mixtures with respect to percent
methanol ........................................ 52

2-11 Measurements of E (50) polarity for methanol/
water mixtures with respect to mole fraction
of methanol ..................................... 55

2-12 Measurements of ET(30) polarity for
acetonitrile/water mixtures with respect to
percent acetonitrile............................54

2-13 Measurements of ET(30) polarity for
acetonitrile/water mixtures with respect to
mole fraction of acetonitrile...................55

2-14 Comparison between Snyder's P' and Dimroth-
Reichardt's ET(30) polarity values for pure
solvents ........................................ 61

2-15 Comparison between ET(30) polarity values
predicted by equation 2-3 and actual
ET(30) polarity values reported by Reichardt
and Harbusch-Gornert (1983)...................... 64

3-1 Retention data for 4-nitrophenol plotted with
respect to percent acetonitrile.................71

3-2 Variation in mole fraction of methanol as
a function of volume percent....................72

3-3 Variation in mole fraction of acetonitrile as
a function of volume percent....................73

3-4 Retention data for 4-nitrophenol plotted with
respect to mole fraction of acetonitrile........74

3-5 Retention data for 4-nitrophenol plotted with
respect to n* dipolarity/polarizability for
the same solvent mixtures.......................75

3-6 Retention data for 4-nitrophenol plotted with
respect to the ET(30) polarity for the same
solvent mixtures..................................76

3-7 Histogram of r2 values for the 332 retention
data sets plotted with respect to percent
organic modifier ................................ 90











3-8 Histogram of r2 values for the 332 retention
data sets plotted with respect to ET(30)
polarity ................................ ........91

3-9 Modified histogram of r2 values for the
332 retention data sets plotted with respect
to percent organic modifier....................92

3-10 Modified histogram of r2 values for the
332 retention data sets plotted with respect
to ET(30) polarity............. ................93

4-1 Chromatographic selectivity measurements as a
function of percent methanol ....................110

4-2 Chromatographic selectivity measurements as a
function of mole fraction of methanol..........111

4-3 Chromatographic selectivity measurements as a
function of ET(30) polarity of methanol/
water mixtures.................................112

4-4 Chromatographic selectivity measurements as a
function of percent acetonitrile...............113

4-5 Chromatographic selectivity measurements as a
function of mole fraction of acetonitrile......114

4-6 Chromatographic selectivity measurements as a
function of ET(30) polarity of acetonitrile/
water mixtures......................... ... 115

4-7 Comparison between r2 values for plotting
methylene selectivity data with respect to
either percent organic modifier or ET(30)
polarity ............................. ... ....... 119

4-8 Comparison between methylene selectivity
results obtained with either 1-nitroalkanes
or alkylbenzenes as the homologous series......124

4-9 Example of the measurement of methylene
selectivity with nitroalkanes as the
homologous series..............................128

4-10 Chromatographic selectivity measurements as
a function of percent methanol................129

4-11 Chromatographic selectivity measurements as
a function of mole fraction of methanol.........130











4-12 Chromatographic selectivity measurements as
a function of ET(30) polarity of methanol/
water mixtures.............................. ..131

4-15 Chromatographic selectivity measurements as
a function of percent acetonitrile............... 132

4-14 Chromatographic selectivity measurements as
a function of mole fraction of acetonitrile.....133

4-15 Chromatographic selectivity measurements as
a function of E (30) polarity of
acetonitrile/water mixtures.................... 134

5-1 Slope of log k' versus ET(30) polarity as a
function of carbon number for methanol/
water mixtures................................. 142

5-2 Slope of log k' versus ET(30) polarity as a
function of carbon number for acetonitrile/
water mixtures... ............................ 143

5-3 Variation in surface tension as a function
of percent methanol........................... 161

5-4 Variation in surface tension as a function
of mole fraction of methanol...................162

5-5 Variation in surface tension as a function
of percent acetonitrile........ ............. 163

5-6 Variation in surface tension as a function
of mole fraction of acetonitrile...............164

5-7 Comparison between surface tension and
ET(30) polarity for methanol/water mixtures....167

5-8 Comparison between surface tension and
E (30) polarity for acetonitrile/water
mixtures .......................................168


xii














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

SOLVATOCHROMIC SOLVENT POLARITY MEASUREMENTS,
RETENTION, AND SELECTIVITY
IN REVERSED PHASE LIQUID CHROMATOGRAPHY

BY

BRUCE PHILIP JOHNSON

August, 1986


Chairman: John G. Dorsey
Major Department: Chemistry


The ET(30) polarity and '* dipolarity/polarizability

of binary acetonitrile/water and methanol/water mobile

phases used in reversed-phase liquid chromatography were

measured and compared with chromatographic retention and

selectivity. For the retention data, plots of log k'

versus the ET(30) polarity were generally found to be

better descriptors of retention than the more commonly used

plots of log k' versus percent organic modifier. A total

of 332 sets of retention data were examined, and the

overall average r2 values obtained for simple linear

regression of log k' versus either percent organic modifier

or ET(30) values were 0.9783 and 0.9910, respectively.

The slope and y-intercepts of plots of log k' versus

ET(30) polarity were found to be dependent on the solute


xiii










size, solvent system, and the column. Also, for a given

column and solute, the slope for the two solvent systems

examined was found to vary in systematic matter, with that

of methanol/water mixtures 1.43 times greater (on the

average) than that for acetonitrile/water mixtures. This

variation in slope is evidence of differential solvation of

the bonded phase alkyl chains by the two organic modifiers.

In addition, the variation in methylene selectivity as

a function of either percent organic modifier or ET(30)

polarity has been examined for various bonded phases, as

well as for a polymer-based column.

Solvatochromic solvent polarity measurements offer a

unique view of the retention process, by providing a means

of determining mobile phase polarity that is independent of

the chromatographic system, thus allowing de-convolution of

subtle stationary phase solvent effects, as well as the

prediction of chromatographic retention.


xiv















CHAPTER I
INTRODUCTION


The actual mechanism of retention in reversed phase

liquid chromatography (RPLC) has been the subject of much

controversy and debate since the first bonded phases for

chromatography became commercially available in the early

1970s. Despite its name, reversed phase liquid

chromatography is actually a more "popular" technique than

normal phase liquid chromatography (NPLC). Stationary

phases used in RPLC typically consist of a silica-based

supporting material to which nonpolar carbon chains are

bonded. These carbon chains are most commonly straight

chains of length 8 or 18 carbons (hence the terms "octyl,"

"octadecyl," C-8, C-18, etc. to describe the type of bonded

phase). The carbon chains are attached through a bonding

reaction in which the surface hydroxyl groups present on

the silica (silanols) are reacted with the appropriate

chlorosilane, leading to an Si-0-Si-C bond. For example,

to produce an octadecyl bonded phase, one could react

dimethyloctadecylchlorosilane with silica. The bonding

reaction is not exhaustive, however, so in a second step

trimethylchlorosilane or hexamethyldisilazine is typically

added in order to "endcap" the residual silanols that may









not be accessible to the larger, more sterically hindered

silane used in the first bonding reaction.

An ideal bonded phase would have no residual silanols

and would possess univariate pore and particle size

distributions. Practically speaking, no bonded phase can

be said to be free of residual silanol groups; one of the

major differences between competing commercial stationary

phases is in the degree of endcapping.

The presence of residual silanols is highly

undesirable since it leads to a second mechanism of

retention. Polar compounds and/or ones possessing hydrogen

bond donor/acceptor ability can interact with these

silanols (silanophilic solutes), leading to distorted peak

shape and/or greatly increased retention (Bij et al., 1981;

Nahum and Horvath, 1981). This problem was recently

reviewed, and the effects of both silanophilic and

metallophilic interactions were compared (Sadek et al.,

1985a). It was found that stainless steel inlet frits

commonly employed in LC columns also cause losses in

efficiency, due to both mechanical and chemical

interactions. Silanophilic interactions were found to be

the major factor in affecting the retention of basic amine

compounds.

While the mechanism of normal phase liquid

chromatography (NPLC) can be said to be fairly well

characterized in terms of adsorption at active sites upon










the silica or alumina surface, that of RPLC remains

controversial. In general, one can distinguish two broad

areas of study of this mechanism. These two areas are

referred to as "mobile phase effects" and "stationary phase

effects."



Aobile Phase Effects

In the first of these ("mobile phase effects"), one

observes or calculates the effects of changing mobile phase

composition on chromatographic retention. Typical mobile

phases used in RPLC consist of water to which an organic

modifier has been added. The most frequently used organic

modifiers are methanol, acetonitrile, and tetrahydrofuran.

One example of the approaches classified as "mobile

phase effects" is that of solubility parameter theory.

Hildebrand's solubility parameter has been shown to be

useful in the prediction of many solution properties and is

defined by



6 = (E/V)1/2 (1-1)


where E is the molar heat of vaporization of the solvent

and V is its molar volume. When applied to chromatography,

retention is viewed in terms of the relative solubility

parameters of the solute, mobile phase, and the stationary

phase (Karger et al., 1978; Schoenmakers et al., 1982).










Capacity factors can then be considered to be related to

these parameters, as shown in the following equation:



In k' = (v/RT)(6 + 6 26.)(6 6 )
m S 1 m S


+ In (n /n ) (1-2)



where v is the molar volume of the solute and 6,, 6s, and

Si are the solubility parameters of the mobile phase,

stationary phase, and solute, respectively. The (ns/nm)

term is the ratio of moles of the stationary and mobile

phases, respectively. If the solubility parameter for a

solvent mixture is approximated by assuming linear

additivity of volume fractions, the dependence of retention

on the volume fraction of one of the components becomes



In k' = A (4)2 + B (<) + C (1-3)



where A, B, and C are constants. Assuming linear

additivity of solubility parameters is questionable,

however. Particularly for aqueous mixtures, where hydrogen

bond forces have a very large effect on the heat of

vaporization, the solubility parameter is likely to be

complex function of the volume fraction of the

components. This equation also results from assuming that

the stationary phase solubility parameter is a constant,









regardless of the mobile phase composition, which is also a

questionable assumption (see discussion of stationary phase

solvation in the latter part of this section). Moreover,

it is impossible to measure the solubility parameter of

either the binary/ternary solvent mixtures used in RPLC, or

the alkyl chains of the stationary phase. While the theory

enables qualitative predictions to be made on the basis of

relative polarity of the solute and phases, quantitative

calculations are not possible.

Hafkenscheid and Tomlinson (1983) have recently re-

cast solubility parameter theory for RPLC. Semi-empirical

relationships were derived in order to allow more accurate

predictions of retention. Other workers have subdivided

the solubility parameter into individual contributions due

to dispersive forces, proton transfer, polar interactions,

etc., in an attempt to predict retention with greater

accuracy. An example of this approach would be the work of

Tjissen et al. (1976). Unfortunately, as the accuracy of

prediction increases the practicality of applying such

complex equations also decreases in an inverse fashion.

One of the most well-known approaches to

chromatographic retention is the hydrophobic theory of

Sinanoglu (1968), as applied by Horvath and Melander

(1977). This model (also referred to as the solvophobic

model) describes retention in terms of repellent forces

between the relatively nonpolar solute and the highly polar









aqueous mobile phase. This results in the formation of a

complex between the stationary phase ligands and the

relatively hydrophobic solute. Here the stationary phase

acts as a passive receptor to hydrophobic molecules that

are repelled by the aqueous mobile phase (cavity effect).

The stationary phase is treated as a constant, and specific

interactions between residual silanols and polar groups on

the solute molecules are not treated. Various mathematical

expressions were derived relating retention to solute

properties such as the hydrocarbonaceous surface area (HSA)

and solvent properties such as the dielectric constant or

surface tension.

Recently, Antle et al. (1985) compared various RPLC

columns with respect to solvophobic selectivity. Retention

differences seen among columns were ascribed to three

effects: differences in phase ratio, the polarity of the

bonded phase, and the dispersion solubility parameter of

the stationary phase.

Martire and Boehm (1980, 1983) have applied

statistical-mechanical theory to the description of

chromatographic retention. Using a lattice model,

predictions were made about the effects of either changing

mobile phase composition or length of alkyl bonded

groups. Though these derivations are quite rigorous,

practical application of the results is somewhat










difficult. Furthermore, because of several assumptions

made, again only qualitative predictions are possible.

Interaction indices (empirical measurement of I values

based on solute retention) have also been used by Jandera

et al. (1982) to predict retention. Here it is assumed

that the empirical interaction index of a solvent mixture

is a linear sum of the volume fraction contributions of the

solvents and again only qualitative predictions are

possible.



Stationary Phase Effects

In the second broad area of study ("stationary phase

effects"), the nature of the stationary phase is studied

through either direct physical measurement or through the

effects of changing the type of bonded phase (i.e., C-2,

C-8, etc.) on chromatographic retention. Direct physical

measurements of the stationary phase involve either

spectroscopic methods or actual chemical dissolution.

The most basic chemical analysis of a stationary phase

is the determination of percent carbon. The percentage of

carbon loading provides information about the extent of the

bonding reaction, as well as the degree of surface coverage

(assuming the surface area has been determined). Of

course, this provides no information about the conformation

or spatial distribution of the alkyl chains. Other

dissolution methods have been used in an attempt to examine









the chemical form of the bonded alkyl chains. For example,

Lullman et al. (1985) carried out studies in which bonded

phase packing were fused with potassium hydroxide. In

this manner, the alkyl ligands were cleaved from the silica

substrate, and subsequent IC analysis of the fusion

products revealed the presence of hexaalkyldisiloxanes and

trialkylsilanols for monomeric bonded phases. Another

approach is to digest the stationary phase in hydrofluoric

acid and then to subject this digest to analysis by gas

chromatography (Fazio et al., 1985). Based on GC analysis

of these digests, it was possible to distinguish between

the various methods used to derivatize the silica (i.e.,

whether mono-, di-, or tri-chlorosilanes had been used).

While the stationary phase is often treated as a

passive or invariant entity, there is much evidence that

the solvation of the alkyl ligands themselves changes in

response to varying composition of the mobile phase. This

is best exemplified by the re-equilibration necessary after

an organic concentration gradient. That is, the alkyl

chains that comprise the stationary phase surface are

preferentially solvated by the organic component of the

mobile phase. Because of this, the interfacial region

between the bulk mobile phase and the surface of the silica

base has widely varying physical properties, so that

chromatographic retention reflects the statistical mean of

these varying physical properties. The organic modifier









content of the stationary phase increases with the

concentration of modifier in the mobile phase (Yonker et

al., 1982a, 1982b). Among the three most commonly used

organic modifiers, tetrahydrofuran has been shown to

solvate the stationary phase to the greatest extent,

followed by acetonitrile and methanol (Yonker et al.,

1982a, 1982b). Thus, while the mobile phase may consist of

a 50/50 mixture, the stationary phase will be solvated by a

mixture with a significantly higher proportion of the

organic modifier.

Lochmuller and Wilder (1979) compared the selectivity

of various bonded phases with that of equivalent liquid-

liquid systems. For chain lengths greater than

approximately 12 carbons, the selectivity was found to be

comparable to the liquid-liquid system. Also, Lochmuller

et al. (1981) prepared bonded phases with either n-heptyl,

cyclohexyl, or bicycloheptyl alkyl chains. The n-heptyl

phase was found to have the highest selectivity and

capacity, though the cyclic phases were found to retain

cycloalkanes preferentially. Jinno and Okamato (1984)

prepared bonded phases with various aromatic moieties.

Capacity factors were measured for various polynuclear

aromatic hydrocarbons (PAHs). In this case, the pore size

of the silica matrix appeared to influence retention,

either because of its effect on the bonding reaction or









varying abilities of the PAHs to penetrate the interior

pores (steric effects).

One way to use a spectroscopic method in

characterizing the stationary phase is to sorb or

chemically bond a probe molecule to the surface and then

observe the electronic spectrum of the probe molecule.

Fluorescence spectroscopy lends itself to this type of

measurement, because of the type of sample and the inherent

lower detection limits possible. Since the sample is a

solid, it is very difficult to observe the adsorbed species

directly through absorption spectroscopy. That is, the

solid silica particles tend to scatter the incoming light

beam to a greater degree; using a lower amount of suspended

solid lowers the degree of light scattering at the expense

of lowered sensitivity to the presence of the probe

molecule. A secondary problem is the need to apply very

small amounts of the probe molecule to the stationary

phase. If too much is applied, more than a monolayer may

be formed, and thus the resultant information is of

questionable value. Also, one would not want to "overload"

the packing, i.e., operate at a concentration where the

sorption isotherm becomes nonlinear, which would also yield

results not applicable to the true conditions seen by the

stationary phase. For these reasons, fluorescence (for

UV/VIS) or diffuse reflectance (for infrared or UV/VIS)

spectroscopy are ideally suited to this type of









experiment. Of course, it is essential that for

fluorescence experiments, the excitation and emission

wavelengths be sufficiently separated to avoid interference

from the aforementioned scattered light.

In choosing a probe molecule to study the stationary

phase, the two most important criteria are spectral

response to changing solvent polarity and affinity for the

bonded alkyl chains. A probe molecule with insufficient

affinity (i.e., too low of a partitioning coefficient

between the mobile and stationary phases) will reside in

the mobile phase to such an extent that the fluorescence

cannot be attributed solely to that residing on the

stationary phase. For these reasons, the most commonly

used probe molecule has been pyrene, a 4-ring fused

aromatic compound. The fluorescence spectrum has vibronic

structure which is quite sensitive to the solvent environ-

ment. In fact, this molecule has been used to establish

the Py scale of solvent polarity (Dong and Winnik, 1984;

empirical solvent polarity scales are discussed in detail

in the latter part of this chapter). The fluorescence

spectrum of pyrene contains five major vibronic bands,

labeled I to V, beginning with the 0-0 band. The ratio of

the intensities of bands I and III has been shown to be

highly responsive to changing solvent environment. Being a

large, hydrophobic molecule, it has a very large affinity

for the alkyl chains of the stationary phase. Two recent









papers have reported on the variation in stationary phase

polarity (as seen by pyrene sorbed onto the column packing)

as a function of the mobile phase composition (Carr and

Harris, 1986; Stahlberg and Almgren, 1985).

It is interesting to note that these two groups

obtained data for complementary organic modifier

concentration ranges, as a result of the experimental

conditions used. Stahlberg and Almgren (1985) measured the

surface polarity of C-2 and C-18 surfaces in the presence

of 0-50% methanol/water and acetonitrile/water mixtures.

This was done by using a suspension of 2-3 mg packing per

mL of solvent. Sodium tetradecylsulfate was also added

(0.5 mg/mL) to prevent flocculation of the particles. In

the 0-30% acetonitrile range, the surface polarity of the

C-18 packing was found to be greatest at the extremes,

while in methanol it decreased steadily as the

concentration increased. At higher concentrations of

organic modifier (>30% v/v), the concentration of pyrene in

the solvent mixture became too great and thus obscured the

fluorescence spectrum of the sorbed material. The

interpretation of these results is complicated, however,

because of the presence of added surfactant (0.5 mg/mL;

0.0015 M), which is also likely to sorb onto the bonded

chains and modify the surface polarity. Carr and Harris

(1986) studied both polymeric and monomeric C-18 phases in

a similar manner, except that the sample consisted of a










flow-cell packed with the solid, through which the solvent

mixture with pyrene was passed. In this case, the

investigators were limited to concentrations greater than

20, 25, and 50% acetonitrile, tetrahydrofuran, and

methanol, respectively, because the entire packed particle

bed could not be fully equilibrated with pyrene. Here the

ratio of stationary phase to mobile phase volume was much

higher, leading to a much greater quantity of sorbed

pyrene.

This effect can be demonstrated quantitatively by the

following equation relating capacity factor (k') to the

thermodynamic distribution coefficient (K) and phase ratio





k' = ko (1-4)



The phase ratio, 0, is the ratio of the volumes of the

stationary and mobile phases. The capacity factor, k',

corresponds to the ratio of the moles of the sorbed

material present in the stationary and mobile phases at any

given instant. Thus, the packed bed used by Carr and

Harris (1986) has a much higher value than the slurry used

by Stahlberg and Almgren (1985), making the use of higher

organic modifier concentrations necessary (lower K

values). On the other hand, the upper limit of organic

modifier concentration is increased, since the higher value









compensates for the greatly decreased affinity of pyrene

for the stationary phase (lower K). Thus, Carr and Harris

(1986) were able to report surface polarities for up to 80,

45, and 70% methanol, tetrahydrofuran, or acetonitrile,

respectively. For a C-18 monomeric packing, the surface

polarity was found to increase with increasing organic

modifier concentration, with methanol systems having

consistently lower polarity than that of the acetonitrile

or tetrahydrofuran systems. This is a direct reflection of

the fact that much less methanol is absorbed by the alkyl

chains as the organic concentration is increased, so the

polarity remains closer to that of a pure alkane. On the

other hand, much greater amounts of acetonitrile and

tetrahydrofuran solvate these alkyl chains, leading to an

increase in apparent polarity (with respect to a pure

alkane). These results are fully consistent with those of

earlier workers who measured the adsorption isotherms of

organic modifiers onto various column packing. For

example, McCormick and Karger (1980a, 1980b) and Tanaka et

al. (1980) reported the organic modifier content of

reversed phase column packing under various

concentrations. Even at a concentration of 10% organic

modifier, acetonitrile was found to solvate the alkyl

chains to a much higher degree than methanol.

One way to get around the problem of lowered affinity

of the probe for the stationary phase at high organic









modifier concentration is to simply immobilize it by

bonding it to the stationary phase. Lochmuller et al.

(1985) measured the fluorescence of surface bonded

exciplexes (pyrene/N,N-dimethylaniline) to measure the

surface polarity after endcapping with either

trimethylchlorosilane (TMCS) or hexamethyldisilazine

(HMDS). Trimethylchlorosilane was found to yield a

stationary phase of lower polarity.

Another major area of spectroscopic examination of

stationary phases involves the use of nuclear magnetic

resonance (NMR). As with the sorbed probe fluorescence

experiments, the greatest wealth of information is derived

from those in which the packing is examined under "real"

conditions, i.e., in the presence of a mobile phase. Most

often, 13C is used in NMR experiments. One problem that

must be overcome is the low signal to noise ratio of 13C-

NMR, which is aggravated by the nature of the sample.

Also, alkyl chain C-atoms have nearly identical chemical

shifts, making it difficult to differentiate between

individual positions within the chain. Gilpin and Gangoda

(1984, 1985) have synthesized stationary phases in which

the terminal carbon atom is enriched with the 13C isotope,

thus overcoming some of these difficulties. The 13C spin-

lattice relaxation times (in either pure deuterated

chloroform or acetonitrile) were found to be fairly

constant for the various chain lengths studied. However,









at higher coverage densities, a decrease was noted. This

is evidence for the increasing interaction between the

neighboring alkyl chains. The effect of solvent viscosity

was also explored; an inverse relationship was found

between spin-lattice relaxation time and solvent

viscosity. Also, 29Si-NMR with magic angle spinning has

been used to differentiate between the various chemical

environments of the silicon atoms in dry samples of column

packing (Fyfe et al., 1985).

Fourier transform infrared spectroscopy (FTIR) has

been used to directly observe the stationary phase alkyl

chains. Again, experiments have been done with both dry

packing and in the presence of solvent mixtures. In this

case, a major difficulty arises from the strong infrared

absorption band of water and methanol (O-H stretching),

which tends to obscure the C-H stretching band of the

bonded alkyl chain. One solution to this problem is to use

deuterated solvents, as Sander et al. (1983) have done with

C-1 to C-22 column packing. The range of 70-100% methanol

was studied, and evidence of increasing chain order was

found at the higher organic concentrations. Also,

temperature studies were carried out on the dry packing,

and no phase transitions were observed at temperatures near

or below the corresponding alkanes. The degree of disorder

of the chains was found to be comparable to that of liquid

n-alkanes at room temperature; thus the surface of the









bonded phase behaves like silica with a thin oily

coating. Suffolk and Gilpin (1985) have made FTIR

measurements of a cyanoalkyl bonded phase. Here the

cyanoalkyl group could easily be observed with little

interefrence from the solvent. In hexane, there appeared

to be two distinct populations of bonded ligands (possibly

due to interaction with surface silanols), while in 1-

butanol ligand-solvent interaction was more apparent.

Other studies of the stationary phase have made use of

such diverse analytical techniques as differential scanning

calorimetry (Hansen and Callis, 1983), ESCA (Miller et al.,

1984), and photoacoustic spectroscopy (Lochmuller et al.,

1980; Miller et al., 1984). Recently, two general reviews

of stationary phase structural studies have been published

(Gilpin, 1984, 1985).

Despite the plethora of spectroscopic studies

published on the nature of the stationary phase, in no case

is a quantitative relationship derived between these

experimental results and actual chromatographic

retention. In all cases, the spectroscopic results are

interpreted in a qualitative manner.



Empirical Measures of Solvent Polarity

For any chemical process occurring in solution, the

polarity of the solvent plays a crucial role in determining

the outcome. While this has been known for many years,









only recently has the exact role of the solvent begun to be

clarified and quantitated. Solvent properties influence

not only the rates of chemical reactions, but also the

position of chemical equilibria. Many spectral properties

are affected by the nature of the solvent. It is well

known that both the intensity and absorption or emission

frequency of NMR, IR, UV/VIS, and luminescence spectra are

affected by the solvent. This is an example of

solvatochromism, in which the position, intensity, or shape

of a spectral peak is affected by the solvent. The

importance of solvatochromism is demonstrated quite clearly

by the Sadtler library of standard ultraviolet spectra

(Sadtler Research Laboratories, Philadelphia, PA), in which

the spectra are reported for solutes dissolved in methanol,

wherever solubility permits. In this way, peak positions

for different substances are easily compared, with no need

to correct for the effect of different solvents.

In the field of analytical chemistry, solvent effects

must be taken into account when developing a method of

analysis. Solvents will affect the position and intensity

of spectral peaks being measured in quantitative IR or

UV/VIS spectroscopy, the rates and extent of reactions used

in derivatization or titration, etc. Also, chemical

separations by liquid chromatography (either normal or

reversed phase), which are controlled primarily by the

nature of the solvent(s) used as the mobile phase, are a









direct result of the different polarities of the stationary

and mobile phases.

There are many ways to characterize the polarity of a

solvent. Bulk physical properties, such as dielectric

constant, viscosity, or refractive index represent the

simplest measures of solvent properties. However, no

single physical property can adequately characterize the

"polarity" of a solvent. The "polarity" of a solvent is

extremely difficult to define and represents the sum total

of all possible interactions that a solute may experience

when dissolved in a particular medium. Therefore, bulk or

macroscopic properties will only provide information about

the interaction between the solvent molecules themselves.

Interactions that a solute may experience include

dispersion, dipole-dipole, dipole-induced dipole, and

hydrogen-bond forces. Because of the difficulty of

characterizing the polarity of a solvent through bulk

physical properties, a number of empirical scales of

solvent polarity have been developed in the past 50

years. These empirical scales are based on the properties

of particular solutes dissolved in the solvent of

interest. In this way, specific, microscopic interactions

with the solvent are probed, since the test solute is able

to "see" these better than bulk, macroscopic properties

can. Extensive reviews of empirical measures of solvent









polarity have been published (Griffiths and Pugh, 1979;

Reichardt, 1979).

The earliest empirical scales of solvent polarity were

based on a kinetic measurement of some reaction carried out

in the solvent of interest. Perhaps the most well-known

scale of this type is the Y-scale, developed by Grunwald

and Winstein (1948). The Y-scale is based on the

solvolysis of t-butyl chloride. The rate constant for this

first order process is measured, and a Y-value is

calculated with the following equation:



log k log k0 = mY (1-5)



where k is the rate constant, kO is the rate constant in

80% (aqueous) ethanol, and m is the sensitivity of the

substrate (m = 1 for t-butyl chloride, by definition).

There are also many empirical scales of solvent

polarity based on a spectroscopic measurement. The fact

that a solvent will influence the spectral properties of a

solute is used as a way of characterizing solvent

polarity. These measurements are quite simple and involve

nothing more than dissolving the test solute in the solvent

and recording the absorption or emission spectrum (either

IR, NMR, or UV/VIS). One example of this type of scale is

that of Kosower's Z-values (Kosower, 1958), which are based

on the intermolecular charge-transfer absorption of










1-methyl-4-carbomethoxypyridinium iodide. The Z-values are

defined by



Z = 28592/X (1-5)



where Xmax is the position of the charge transfer peak (in

nm). The constant in equation 1-6 is the product of

Avogadro's number, the speed of light, and Planck's

constant. The Z-values have been reported for more than 50

pure solvents and solvent mixtures (Kosower, 1958). For

mixtures with a high water content, the charge transfer

peak merges with that of the aromatic ring and is unable to

be located.

The w* scale of solvent polarity was developed by

Kamlet et al. (1977). Its name derives from the fact that

it is based on the positions of the w to w* transitions of

a series of chromophores. Rather than a single solute, it

is based on a series of aromatic solutes, in which the w*

parameter was adjusted to give the most consistent

correlation among the various test solutes. The inventors

of this scale prefer to refer to it as the w* scale of

solvent dipolarity/polarizability. In fact, in solvents

with no potential for hydrogen bonding, there is a linear

correlation between the molecular dipole moment and the

measured r* value. Brady and Carr (1982, 1985) have

discussed this scale in terms of the Onsager reaction field










and Block and Walker dielectrically saturable reaction

field models. For a given solute, 11* values are calculated

with the following equation:



rr* = (v vO)/s (1-7)



where s is the sensitivity of the solute to the ir* scale,

and v and vO are the absorption maxima (X 10-3 cm-1) of the

solute in the solvent and cyclohexane, respectively. The

appropriate constants for this equation have been published

(Kamlet et al., 1977). In addition both a and 6 measures

of solvent hydrogen bond donor and acceptor ability have

been derived from these same solutes. These measures are

based on the enhanced solvatochromic shift of one indicator

relative to another in the presence of hydrogen bond

donor/acceptor solvents. For example, one way to measure

the 8 value is to compare the solvatochromism of 4-

nitroanisole with respect to 4-nitrophenol; solvents

capable of hydrogen bond acceptor interactions will cause

an enhanced solvatochromic shift for the 4-nitrophenol with

respect to 4-nitroanisole (Kamlet and Taft, 1976). In a

similar manner, solvent a values can be derived from the

enhanced solvatochromic shift of ET-30 with respect to 4-

nitroanisole videe infra).

The ET(30) scale of solvent polarity was developed in

the early 1960s by Dimroth et al. (1963a, 1963b) and









Reichardt (1979), who reported on the solvatochromism of a

series of 42 pyridinium betaine dye molecules. In the

original paper, derivative #30 was found to have the

greatest sensitivity to changes in solvent polarity. Thus,

the ET(30) scale was named as such because it is derived

from the molar energy of transition (ET) of the thirtieth

pyridinium betaine (30). The ET(30) scale of solvent

polarity is based on the intramolecular charge transfer

absorption of 2,6-Diphenyl-4-(2,4,6-triphenyl-N-pyridino)-

phenolate (structure shown in Figure 1-1). It possesses a

number of unique features, such as a 44-electron aromatic

ring system, a negatively charged phenoxide group and a

positively charged pyridine ring nitrogen atom. This

molecule undergoes one of the largest known shifts in Xmax,

amounting to some 357 nm in going from water (453 nm) to

diphenyl ether (810 nm). Since this dye absorbs within the

visible light region, it is possible to estimate visually

the polarity of a solvent. In methanol, the solution is

wine-red, while in acetonitrile the solution becomes deep

blue in color. Values of ET(30) polarity are calculated in

the same manner as are Z-values (equation 1-6) and have

been reported for over 200 solvents (Reichardt and

Harbusch-Gornert, 1983). Also, the range of the scale has

been expanded through the use of a more lipophilic betaine,

in which a t-butyl group is attached to each of the five




























0 ^ .^ h v I.;




101 101
e




Figure 1-1. Structure of the ET-30 dye molecule, 2,6-
Diphenyl-4-(2,4,6-triphenyl-N-pyridino)-
phenolate in the ground and excited states.









phenyl groups (para position). Recently, a normalized

scale of ET(30) polarity (EN) has been defined by Reichardt

and Harbusch-Gornert (1983), in which the polarity of water

is defined to be 1.0, while that of tetramethylsilane (TMS)

is 0. These values are calculated by using the following

equation:


ET(solvent) ET(TMS)
N (1-8)
T ET(H 2O) ET(TMS)


where ET(solvent) is the ET(30) polarity of the solvent in

question as calculated by equation 1-6, and ET(H20) and

ET(TMS) have values of 63.1 and 30.7, respectively. The
normalized scale is used for convenience in expressing a

polarity relative to water or tetramethylsilane and has no

actual effect on the types of correlations discussed

herein. All ET(30) polarity values reported here are in

kcal/mole, as calculated from equation 1-6.

The ET(30) scale has been shown to be sensitive to

both solvent dipolarity/polarizability as well as solvent

hydrogen bond donor ability (HBD). Taft and Kamlet (1976)

calculated that 68% of the shift in Xmax in going from

cyclohexane to n-butanol is due to HBD stabilization of the

ET-30. This stabilization is a direct result of the

presence of the negatively charged phenoxide group on the

ET-30 molecule. The phenoxide group acts as a hydrogen

bond acceptor, so that protic solvents may function as









suitable H-bond donors. In fact, Taft and Kamlet (1976)

have used the enhanced solvatochromic shift of ET-30 with

respect to 4-nitroanisole to measure the solvent hydrogen

bond donor acidity (a-scale). In both protonic and

nonprotonic solvents, the ET(30) scale can be related to

the T* and a scales by use of the following equation

(derived from equation 7 of Kamlet et al., 1976):



E.(530) = 30.51 + 14.6w* + 16.53a (1-9)




Analytical Applications of the ET-30 Dye

Owing to its extreme sensitivity to changes in overall

solvent polarity, ET-30 may be used to determine the

composition of binary solvent mixtures. However, ET-30 is

particularly sensitive to the presence of small amounts of

water in aprotic solvents. For protic solvents, the

presence of water has a smaller effect, as illustrated with

tert-butyl hydroperoxide. Langhals et al. (1980) have

reported that the presence of 5.2 moles/liter water changes

the apparent solution color from blue (Xmax = 575 nm) to

red (Xmax = 532 nm). Thus, the color of the solution

serves as a visual indicator of the water content, and

measurement of Amax for ET-30 dissolved in a given solvent

can be a rapid and precise alternative to Karl-Fischer

water determinations. Of course, quantitation of the water









content for a given solvent requires that the ET(30) value

be known for each composition. Values of ET(30) polarity

for many binary solvent systems have been reported

(Balakrishnan and Easteal, 1981a, 1981b; De Vijlder, 1982;

Dimroth and Reichardt, 1966; Jouanne et al., 1978; Koppel

and Koppel, 1983a, 1983b; Krygowski et al., 1985;

Maksimovic et al., 1974). If the variation in ET-30 as a

function of composition is monotonic, i.e., no maxima or

minima occur, this determination is fairly

straightforward. On the other hand, if there are any

maxima or minima, this is not possible, since a given Amax

value will correspond to more than one concentration. This

would be the case, for example, for mixtures of

acetonitrile with isopropanol, as reported by Koppel and

Koppel (1983a). Langhals (1982a) has proposed the

following equation to follow changes in ET(30) polarity

values in binary solvent mixtures:



ET(30) = Ed ln(C /C* + 1) + ET(30) (1-10)


where Cp is the molar concentration of the most polar

component, Ed and C* are constants determined for each

binary system, and ET(30) is the ET(30) polarity for the

least polar solvent. The appropriate constants for a total

of 46 binary solvent systems have been reported, as well as

for an organic co-polymer (Langhals, 1982a, 1982b). This










equation is discussed in further detail in Chapter III.

Alternatively, the change in absorbance of a solution of

ET-30 at a fixed wavelength has been used to determine

mixture composition. For example, Kumoi et al. (1970) have

reported that water concentrations of 60 pg/mL can be

detected in acetonitrile. In this case a major dis-

advantage of the method is that the ET-30 concentrations

must be precisely controlled, and a calibration curve must

also be constructed for each determination.

In the examination of the polarity of aqueous micellar

media, ET-30 has also been shown to be useful. Use of

micellar solutions in analytical chemistry has increased in

recent years and has been reviewed by Cline-Love et al.

(1984). Since ET-30 is essentially insoluble in pure

water, the hydrophobic interior of aqueous micelles

provides an ideal site for solvation. Its insolubility in

water means that partitioning between the micelles and

surrounding water will not occur to a significant extent,

and thus interpretation of the spectral results is

simplified. Zachariasse et al. (1981) have reported the

use of ET-30 as a polarity probe for micelles,

microemulsions, and phospholipid bilayers. Changes in

micelle conformation (e.g., sphere-to-rod transition) were

easily detected by the discontinuity in measured ET(30)

polarity as the concentration of sodium chloride was

increased. Also, Plieninger and Baumgartel (1983) have









studied the NMR spectrum of ET-30 in various surfactant

media to determine the position in which the molecule

resides in the micelles. In cationic micelles the

phenoxide group was found to be located in the rigid region

of the electrical double layer, while in anionic micelles

it is found in the diffuse layer, with the pyridinium

nitrogen atom in the rigid layer.

In addition to being used as a probe of micellar

environments, ET-30 also provides useful information about

the structure of binary solvent mixtures. For example,

Kohler et al. (1969) compared the NMR absorption spectrum

for the water proton in aqueous/organic mixtures with the

ET(30) polarity. Binary mixtures of water with either

acetone, dioxane, or tetrahydrofuran were studied, and a

linear relationship was found between the water proton

absorption peak and the measured ET(30) polarity for the

same mixture. It must be noted, however, that the

concentration range examined was fairly small (50-95%

organic component by volume), so it is possible that

outside this range the relationship is not linear.

Balkrishnan and Easteal (1981b) have also discussed the

variation in ET(30) polarity in binary acetonitrile/water

mixtures (see Chapter II).

Heats of solution at infinite dilution have been

correlated with the ET(30) polarity scale by Ilic and

coworkers (1984). A linear relationship was found between









a solute's ET(30) polarity and its heat of solution.

Solutes that were studied included n-alkyl ketones, n-

alcohols, and di-n-alkyl ethers. The heats of solution of

these solutes were measured in solvents such as n-hexane,

carbon tetrachloride, benzene, etc. None of the solvents

were capable of hydrogen bonding with the solutes, however,

and thus the results cannot be generalized to include every

solute/solvent system. Also, heats of solution were

measured only in pure solvents, rather than mixtures. In

addition, this type of correlation would not be possible

for solid compounds, since it is not possible to measure

their ET(30) polarity. Of course, it might be possible to

estimate the ET(50) polarity for solid compounds by using

heat of solution measurements, as Fuchs and Stephenson

(1983) have done for the 7* dipolarity/polarizability of

solid compounds.

The ET(50) scale of solvent polarity has been applied

to chromatographic systems in a number of ways. The

applications discussed here include supercritical fluid

chromatography and normal phase liquid chromatography

(NPLC). These types of investigations can provide

information about either the mobile phase (solvent

polarity) or the stationary phase (surface polarity).

For example, the ET(30) polarity of a mobile phase

used in supercritical-fluid chromatography (SFC) has been

reported (Hyatt, 1984). Typical mobile phases used in SFC









are compressed gases such as carbon dioxide or ammonia, at

a temperature greater than their critical point. Hyatt

calculated the ET(30) polarity of both sub- and

supercritical carbon dioxide to be 33.8 kcal/mole, by using

the more lipophilic penta(tert-butyl) derivative of ET-50

(Reichardt and Harbusch-Gornert, 1983). It was necessary

to use the more lipophilic compound due to the low

solubility of ET-30 in supercritical CO2. An ET(30)

polarity of 33.8 kcal/mole is comparable to that of either

toluene or tetrachloroethylene. However, the strength of

the mobile phases used in SFC is controlled by the pressure

(and resultant density). The ET(30) polarity of

supercritical CO2 was reported for only one pressure (1000

PSI) and temperature (420C), and thus it is likely that a

different polarity would result for different pressures

(densities). For example, Sigman et al. (1985) measured

the n* dipolarity/polarizability and a (hydrogen bonding

basicity) for supercritical C02, which were found to be

highly dependent on the density. Since these measurements

are also based on the use of solvatochromic dyes, it is

likely that the ET(30) polarity would also be greatly

affected by a change in the CO2 pressure. Thus, useful

information would be provided by performing the same

experiments with the more lipophilic, t-butyl derivatized

betaine. Levy and Ritchey (1985) have reported on the

effects of adding small amounts of additives such as










methanol or acetonitrile to the mobile phase in SFC. In

theory, ET(30) polarity of these binary mixtures could also

be measured.

The polarity of silica surfaces used in normal phase

liquid chromatography was examined (Lindley et al.,

1985). In this case, the diffuse reflectance spectrum of

the betaine adsorbed onto the silica was measured. The

peak corresponding to minimum reflectance was used, in

conjunction with that of 4-nitroanisole, to calculate a, a

measure of the acidity of the silica surface. The silica

surface was found to be a strong hydrogen bond donor. The

degree of the dye loading also influenced the measured

values, which decreased at higher levels, apparently as a

result of the formation of more than a monolayer of the

test solutes on the silica surface. None of these

experiments were done in the presence of a mobile phase,

however, most likely because the ET-30 is quite soluble in

typical mobile phases used in NPLC (such as those

containing dichloromethane).

Another interesting application of ET(30) polarity

measurements involves Snyder's eluent strength parameters

for solvents used in normal phase liquid chromatography.

Krygowski et al. (1981) compared the ET(30) polarity of

various pure solvents with Snyder's eluent strength

parameter (6o). In this case, it was necessary to










incorporate a second parameter, BKT (Kamlet/Taft basicity),

in order to predict the eo values. Also, only pure

solvents were treated, rather than the binary or ternary

mixtures typically used in normal phase liquid

chromatography.

To date there have been no comparisons made between

empirical measurements of mobile phase polarity and

chromatographic retention or selectivity. In reversed

phase chromatographic experiments, it is often assumed that

the strength of the mobile phase varies linearly with the

percentage of organic modifier. In this dissertation, the

results of empirical solvent polarity measurements of the

most commonly used mobile phases are discussed, as well as

the correlation between these measurements and

chromatographic retention and selectivity.














CHAPTER II
SOLVATOCHROMIC SOLVENT
POLARITY MEASURE.lENT3


Experimental

E_T(30)-Value Measurements

A sample of the ET-30 was kindly provided by Professor

Christian Reichardt of Philipps-Universitat Marburg,

Federal Republic of Germany. The synthesis of ET-30, which

is not commercially available, is reported elsewhere

(Dimroth et al., 1963b). Binary solvent mixtures were

generated by a Spectra-Physics Model SP8700 ternary

proportioning LC system. Degassing was achieved by

sparging the solvents vigorously with helium. Both HPLC

grade methanol and acetonitrile (Fisher Scientific, Fair

Lawn, NJ) were used as received. Water was first purified

with a Barnstead Nanopure system (Boston, MA) and then

irradiated with UV light in a Photronix Model 816 H.P.L.C.

reservoir (Photronix Corp., Medway, MA) for at least 24

hours. The water was then filtered through a 0.45

micrometer Nylon-66 membrane filter (Rainin Instruments,

Woburn, MA) prior to use.

After collecting 3 mL of a given solvent mixture in a

1 cm path length quartz cell, approximately 0.8 mg of ET-30










was added, and a spectrum was obtained with a Hewlett-

Packard Model 8450A diode array spectrophotometer.

Wavelength accuracy of the instrument was checked with a

Holmium Oxide interference filter. Spectra were acquired

at 2520C. In pure methanol, the change in Xmax is 10 nm

for a temperature change from 25 to 550C (Dimroth et al.,

1963a). Thus, a 2 degree variation leads to an

uncertainty in ET(30) values of approximately 0.1

kcal/mole. The "peak-find" function of the instrument was

used to determine Xmax. For each solvent mixture, ten

spectra were acquired at one second integration time, and

the resultant Xmax values averaged. The pooled standard

deviation for 620 Amax measurements was found to be

1.16 nm. Values of ET(30) polarity were calculated from

Xmax data by using equation 1-6.
As with many dye molecules, the possibility of

dimerization of the ET-30 exists. Since ET(30) exists in a

zwitterionic form, dimerization would be favored through

interaction between oppositely aligned molecules.

Dimerization would lead to a dependence of Xmax upon its

concentration, as well as nonlinearity of a Beer's law

plot. It has been reported that Beer's law is obeyed for

concentrations in the range of 10-05 to 10-03 M (Dimroth

and Reichardt, 1966); all sample concentrations were in

this range. As a further check, the concentration of ET-30









in 45/35/10 MeOH/ACN/H20 was varied, and Xmax and

absorbance at Xmax measured. These data are shown in Table

2-1 and are plotted in Figure 2-1. No dependence of Amax

on ET-30 concentration was observed. Also, Beer's law was

obeyed over the concentration range studied.


Table 2-1.
Effect of varying ET-30 concentration on ax and
absorbance in 45/55/20 (v/v/v) MeOH/ACN/H2 .

Conc. (mg/mL) Xmax Absorbance

0.13 508.5 0.2757
0.26 508.3 0.6901
0.38 508.3 1.111
0.52 509.5 1.498
0.65 508.4 1.914




Values of ET(30) have been previously reported for

these same solvent mixtures (Dimroth and Reichardt, 1966;

Krygowski et al., 1985); however, in these cases mixtures

were prepared by adding water to the organic solvent to

attain a fixed total volume. In contrast, LC pumps

typically mix solvents on the basis of additive volume.

For example, 100 mL of 50/50 (v/v) mixture of

methanol/water (as delivered by an LC pump) is comprised of

50 mL methanol to which 50 mL water is added. Excess

volumes of mixing lead to slight differences in solvent

composition and resultant ET(30) polarity values, so these



























Absorbance 1 -




0 I l
100 200 300 400 500 600 700
ET-30 Concentration (mg/mL X 103)




Figure 2-1. Beer's law plot for ET-30 dissolved in
45/30/10 methanol/acetonitrile/water (v/v/v).










measurements were made with solvent mixtures generated with

the LC pump system itself.

T*-Value Measurements

Measurements of '* values were made with 4-

nitroanisole (Aldrich Chemical Co., Milwaukee, WI) and

using the following equation from Kamlet et al. (1977):



= (vmax 0)/2.343 (2-1)



where vmax is the observed maximum in wavenumbers (X 10-03

cm-1), and vo is the value for the solute in cyclohexane

(IT* = 0 in cyclohexane, by definition). This reference

(Kamlet et al., 1977) lists a number of solutes (for

example, 4-ethylnitrobenzene) that can be used to measure

the fr* dipolarity/polarizability; 4-nitroanisole was chosen

because of its low sensitivity to hydrogen bond

donor/acceptor effects. In this case the 4-nitroanisole

was added to the water at a concentration of 5 ug/mL, and

the resulting solvent mixture + solute was passed through a

0.25 mL Hellma flow cell thermostatted at 400.10C with a

Haake Model D1 water bath (Haake, Saddle Brook, NJ). Flow

was stopped while acquiring spectra to equilibrate the

temperature of the mixture and reduce the effect of

refractive index variability in the sample.

Because of the very small wavelength shift observed

with this substance in going from pure organic to pure










water ( max of 9 nm between water and acetonitrile), the

following algorithm was used to evaluate Imax. Spectra

were acquired, and the absorbance recorded at each

wavelength (1 nm readout resolution). Next the absorbance

data were fit with a 3rd degree polynomial using the

program "Curve Fitter" (see Appendix C; this algorithm was

suggested by Savitizky and Golay, 1964). The first

derivative (dy/dx) of the resultant polynomial was then

used to evaluate Xmax (by setting this equal to zero and

solving for Xmax with the quadratic formula). Repeated

calculations with either the entire data set (30 points;

30 nm wide) or only five points (max -10 max~-5 xmax

max +5 x +10) showed that only five were needed to

define the spectral peak accurately. By interpolating the

spectral peak position in this manner, the precision of

Xmax measurement was greatly improved. As an illustration
of the utility of this algorithm, in Table 2-2 and Figures

2-2 and 2-3, the effect of temperature on the peak position

of 4-nitroanisole in 33.3/33.3/33.4 (v/v/v) MeOH/ACN/H20 is

shown. Data of Xmax provided directly by the instrument or

that from interpolation (of the same spectral data set) are

plotted in Figures 2-2 and 2-3, respectively. The data

clearly indicate that thermochromism of 4-nitroanisole is

not observable without the use of this algorithm.

























312.2

312.0

311.8

311.6

311.4

311.2

311.0


Temperature (oC)


Figure 2-2.


Thermochromism of 4-nitroanisole in
33.3/33.3/33.4 methanol/water/acetonitrile
(v/v/v). Wavelengths obtained directly from
the diode array spectrophotometer.


Absorbance
Maximum
(nm)
























311.4

311 .2

311.0

310.8

310.6


Temperature (oC)


Figure 2-3.


Thermochromism of 4-nitroanisole in
33.3/33.3/33.4 methanol/water/acetonitrile
(v/v/v). Wavelengths obtained by
interpolation of absorbance data from the
diode array spectrophotometer.


Interpolated
Absorbance
Maximum
(nm)










Table 2-2.
Thermochromism of 4-nitroanisole in
33.3/33.3/33.4 (v/v/v) MeOH/ACN/H20.

Temperature max max
(0C) directly (interpolated)

40.0 312 311.4
45.0 311 311.2
50.0 311 311.1
55.0 312 310.9
60.0 312 310.8


Spectra of 4-nitroethylbenzene and 4-nitrophenol were

also acquired in the same manner in methanol/water and

acetonitrile/water mixtures, in connection with the

measurement of solvent a and B values (not utilized in the

present discussion; results tabulated in Appendix D).



Results

r*-Values

While the primary purpose of this research was to

investigate the ET(30) polarity scale in regard to

chromatographic retention, measurements were also done for

the i* scale of solvent dipolarity/polarizability in binary

hydro-organic mobile phases.

In Figure 2-4, a representative spectrum for 4-nitro-

anisole in methanol is shown. One advantage of the use of

this scale is that the spectral peak of interest (due to

the nitro group) is widely separated from that of the






43













4.0 1 -_ __ I








4. I
3.0 IV




W/





1.0








WAVELENGTH (nui)



Figure 2-4. Representative UV/VIS absorbance spectrum of
4-nitroanisole in methanol. Concentrations
for the two curves are top, 0.1 nig/mL; bottom,
0.02 mg/mL.










aromatic n-electron system. As discussed in Chapter I,

overlap of peaks can be a problem, as best exemplified with

Z-values (Kosower, 1958), in which the charge transfer peak

merges with that of the pyridine ring in highly aqueous

mixtures.

The results of 7* dipolarity/polarizability

measurements for binary methanol/water mixtures appear in

Figures 2-5 and 2-6. In terms of percentage methanol

(Figure 2-5), the r* values are seen to decrease steadily,

in a highly nonlinear fashion. However, when the data are

plotted versus mole fraction of methanol (Figure 2-6), a

nearly straight line results (r2 = 0.9959, s = 0.0119).

In Figures 2-7 and 2-8, the corresponding measurements

for the acetonitrile/water solvent mixtures are depicted.

Here the variation is much more complex; this is especially

true when compared to percentage of acetonitrile (Figure

2-7), where there are at least two points of inflection at

approximately the 30 and 70% concentrations. In contrast

to methanol/water mixtures, the variation with respect to

mole fraction of acetonitrile (Figure 2-8) is seen to be

highly nonlinear.

The n* scale of solvent dipolarity/polarizability is

distinctly different from the ET(30) scale in that it is

specifically intended to exclude hydrogen bond

donor/acceptor effects. As such these results then show
























1.1 1
1.0

0.9
TT* 0.8-
0.7

0.6
0.5 I I I I
0 20 40 60 80 100

% Methanol



Figure 2-5. Measurements of w* dipolarity/polarizability
for methanol/water mixtures with respect to
percent methanol.






































0.2 0.4 0.6 0.8


Methanol


Figure 2-6.


Measurements of w* dipolarity/polarizability
for methanol/water mixtures with respect to
mole fraction of methanol.


1.1

1.0

0.9

0.8

0.7

0.6

0.5






























0 20 40 60 80 100

% Acetonitrile


Figure 2-7.


Measurements of ir* dipolarity/polarizability
for acetonitrile/water mixtures with respect
to percent acetonitrile.


TT1*


























1.2

1.1

1.0

TI 0.9
0.8

0.7

0.6
0.0 0.2 0.4 0.6 0.8 1.0


XAcetonitrile


Figure 2-8. Measurements of v* dipolarity/polarizability
for acetonitrile/water mixtures with respect
to mole fraction of acetonitrile.









the variation in polarity due only to dipole/dipole,

dipole/induced dipole, and dispersion interactions. Thus

it is not surprising that at 100% organic concentration,

methanol is actually less. polar than acetonitrile (w*

values of 0.57 and 0.67, respectively). This is a direct

reflection of the fact that the nitrile bond of

acetonitrile is much more dipolar in nature than either the

C-0 or 0-H bonds of methanol. Kamlet et al. (1983) have

reported 7r* values of 0.60 and 0.75 for methanol and

acetonitrile, respectively. These compare with the values

reported here of 0.57 and 0.67 for the corresponding

solvents. This discrepancy between the values reported by

Kamlet et al. and shown here is not significant, however.

In the present work, w* values were calculated from

measurements obtained with one solute (4-nitroanisole),

while those reported by Kamlet et al. (1983) are actually

the values that lead to the most consistent result from

several test solutes. In fact, in the original paper

describing the r* scale, Kamlet et al. (1977) reported

values of 0.58 and 0.71 for methanol and acetonitrile,

respectively.

Based solely on the r* scale, one would conclude that

methanol should be a stronger (less polar) organic modifier

for RPLC. However, this conclusion does not agree with the

known properties of the two organic modifiers, since

acetonitrile behaves as a more nonpolar, hence stronger,










solvent in RPLC. Also, one would expect (based on v*

values) that methanol would solvate the stationary phase

alkyl chains to a greater extent, which, again, is simply

not consistent with the known properties of the two

modifiers (as discussed in Chapters I and V).

ET(350)-Values

A representative spectrum for ET-30 dissolved in pure

methanol is shown in Figure 2-9. The very large absorption

at wavelengths less than 400 nm is due to the aromatic i-

electron system. In pure water, Xmax decreases to 453 nm

(Dimroth et al., 1963a; a 10 cm path length cell was

used). It was not possible to obtain spectra of ET(30) in

pure water (due to its extremely low solubility; <10-06 M),

so this value has been used in the following figures.

In Figures 2-10 and 2-11, the ET(30)-values are

plotted with respect to percent and mole fraction of

methanol, respectively. In Figures 2-12 and 2-13, the

corresponding results for acetonitrile/water mixtures are

shown.

With both organic modifiers, the ET(30) polarity is

clearly a nonlinear function of composition; this is not

surprising since none of the solvents form thermo-

dynamically ideal solutions. For a thermodynamically ideal

binary solvent mixture, any bulk physical property, such a

dielectric constant or viscosity, is expected to be a






























0.50





0.40 co

tC


0.30





0.20





0.10





0. 0


CU in to PCC

WAVELENGTH (rn)





Figure 2-9. Representative UV/VIS absorption spectrum of
the ET-30 dye dissolved in methanol.























ET(30)


0 20 40 60


80 100


% Methanol




Figure 2-10. Measurements of ET(30) polarity for methanol/
water mixtures with respect to percent
methanol.























ET (30)


0.0 0.2 0.4 0.6 0.8 1.0

Methanol


Figure 2-11. Measurements of E (30) polarity for methanol/
water mixtures with respect to mole fraction
of methanol.
























ET(30)


60-

50


% Acetonitrile


Figure 2-12. Measurements of ET(30) polarity for aceto-
nitrile/water mixtures with respect to
percent acetonitrile.


















70


ET (30)


60

50
r15I


40 4-
0.0


Figure 2-13.


0.2 0.4 0.6 0.8

XAcetonitrile


Measurements of ET(30) polarity for aceto-
nitrile/water mixtures with respect to mole
fraction of acetonitrile.









linear function of the mole fraction of either component.

This is also the case for empirical solvent polarity

measurements. For example, this was reported to be true

for the ET(30) polarity of binary mixtures of 1,2-

dibromoethane and 1,2-dibromopropane, whose mixtures obey

Raoult's law, demonstrating ideal solution behavior

(Balakrishnan and Easteal, 1981a).

That methanol/water and acetonitrile/water mixtures

are not ideal solutions is also evidenced by the nonlinear

variation in viscosity and dielectric constant (Horvath and

Melander, 1977). Thus, it is not surprising that the

measured ET(30) polarity varies in a highly nonlinear

manner versus either percent or mole fraction of organic

component. The nonlinearity of these diverse properties

also illustrates the danger of assuming strictly additive

solvent properties, as is done in the derivation of both

liquid chromatographic retention models and gradient

elution schemes. It should also be pointed out here that

in gas chromatography, blending of stationary phase

materials can be done with this assumption in mind. This

is a reflection of the fact that these phases are almost

always nonpolar or weakly polar, nonhydrogen bonding

materials, and thus mixtures are nearly ideal in a

thermodynamic sense (Chien et al., 1980). Also, the mobile

phases used in gas chromatography are nearly inert gases










(hydrogen, helium, or nitrogen) and do not solvate the

stationary phase.

It is apparent that the variation in polarity of the

two systems is quite different. The different character of

these curves is a reflection of the differing hydrogen

bonding abilities of the two organic solvents. In the case

of acetonitrile, it is obvious that for concentrations

greater than 80% (by volume), the measured polarity

decreases rapidly.

Balakrishnan and Easteal (1981b) have discussed the

variation in ET(30) polarity in acetonitrile/water mixtures

and have found it to be consistent with the Naberukhin-

Rogov model (1971) for binary mixtures of water with a

nonelectrolyte. The Naberukhin-Rogov model describes the

structure in terms of two microphases (a and 8) at

concentrations of greater than 0.15 (mole fraction) of

acetonitrile. The a phase consists primarily of highly

structured water, while microphase a contains mostly

acetonitrile. At concentrations of greater than 0.6,

Balakrishnan and Easteal (1981b) postulated that the a

microphase predominates, and the water exists as single

molecules coordinated to these "globules" of

acetonitrile. Further evidence of the existence of

microphases is the phase separation that occurs in this

system, at a critical temperature and concentration of

272 K and 38 mole % acetonitrile, respectively.









Unlike methanol, acetonitrile is a very weak hydrogen

bond donor solvent and thus as the concentration is

increased, the remaining water becomes specifically

associated with the ET-30 due to the presence of the

negatively charged phenoxide group. As the water is

completely removed, and the ET-30 is no longer stabilized

through this hydrogen-bonded network, the apparent polarity

plunges 46.0 kcal/mole. This large change in ET(30)

polarity is not mirrored by the changes seen in log k'

retention measurements. Retention data for concentrations

greater than 80% have not been included in the present data

analysis. There were 61 cases among these data sets where

the 90% acetonitrile point was not included; these were all

from one reference (Hanai and Hubert, 1983). It must also

be pointed out, however, that at these concentrations the

retention time will be very short for most solutes, so that

the resultant k' (approaching zero) and log k' values

(approaching minus infinity) will have the highest relative

uncertainty of the entire retention data set. In fact, the

average log k' for the 61 (90% acetonitrile) points not

included was 0.049 (s = 0.089), corresponding to an average

k' of 1.12.









Relationship Between Snyder's P' Polarity Values
and the ET(30) Scale



Snyder (1974, 1978) has devised the P' scale of

solvent polarity for use in characterizing solvents used in

liquid chromatography. These values are based on

gas/liquid partition coefficients for various solutes and

solvents reported in the literature (Rohrschneider,

1973). For each solvent, the logarithm of the corrected

partition coefficients (K"; corrected to account for

differences in molecular volume and concentration units)

for ethanol, dioxane, and nitromethane are summed together

to calculate a P' polarity as shown below:



P' = log K"(1,4-dioxane) + log K"(ethanol)



+ log K"(nitromethane) (2-2)



In this manner, the solvent's ability to undergo three

types of interactions (proton donor/acceptor, polar) with

solutes is measured, and P' values then represent the total

of these potential interactions. Snyder also reported the

fractional contribution of each of the three test solutes

(Xe, Xd, X, parameters) to the overall P' value. Using

these partial contribution values, Snyder classified all

solvents into eight possible categories. This

classification is often referred to as Snyder's solvent








selectivity triangle, in that three characteristics (proton

donor, proton acceptor, and polar) are assigned to each of

the three vertices of a triangle). Each solvent can then

be placed into a unique position within this triangle on

the basis of its Xe, Xd, and Xn values.

Since Snyder's classification scheme is intended to be

useful for the measurement of solvent selectivity in liquid

chromatography, it is worthwhile to examine briefly the

relationship (if any) between the P' and ET(30) scales of

solvent polarity. The relationship between Snyder's eluent

strength parameters for NPLC and ET(30) polarity has

already been discussed in Chapter I.

The easiest comparison that can be made is between the

P' (summed polarity) values reported by Snyder (1978) and

ET(30) polarity values for pure solvents reported by

Reichardt and Harbusch-Gornert (1983). There were 48 cases

in which tnis comparison could be made; the resultant

comparison plot is shown in Figure 2-14. While there is a

statistically significant correlation between the two sets

(r = 0.7986), there is also a great deal of scatter around

the line (s = 1.1146), so ET(30) values cannot be used to

accurately predict P' values or vice versa. The line drawn

through the data (using linear regression) in Figure 2-14

has a slope of 0.18610.04 and y-intercept of -3.54f1.84.

That there is such a poor correlation is not

surprising, since the P' values represent the summation of





61




















10-- a

8--





0-

30 40 50 60 70

ET(30)




Figure 2-14. Comparison between Snyder's P' and Dimroth-
Reichardt's ET(30) polarity values for pure
solvents.









the three interactions in proportions that will not

necessarily be similar to the responsiveness of the ET-30

probe. Also, it should be noted that according to the P'

scale, methanol is a stronger organic modifier for RPLC (P'

= 5.1) than acetonitrile (P' = 5.8), which is not in

agreement with the known chromatographic properties of

these two solvents.

Perhaps a better way to compare these scales is to

compare ET(30) polarity values with the partial

contribution values (Xe, Xd, and Xn) by using multiple

linear regression. This should allow the various

contributions to be more properly weighted. However, it

must be remembered that these partial values represent the

fraction of the total P' value for each solvent and will

always add up to one. Thus, the true magnitude of each of

the three interactions is masked, and to make a valid

comparison, one must first multiply each partial

contribution value by the total P' value for each

solvent. Using multiple linear regression, an attempt was

made to correlate each ET(30) value with the three

corrected partial contributions for the same solvent. For

the 48 cases, the multiple correlation coefficient was

found to be 0.8912, with a standard deviation of 3.685.

The equation relating the ET(30) to the three interactions

was











E (30) = 29.93.8 + 7.831.8 X' + 2.82.6 X'
T e d


1.792.9 X' (2-3)
n


The regression coefficients indicate that the ET(30)

scale is significantly related to only the terms derived

from the partition coefficients for ethanol and dioxane.

These results are shown graphically in Figure 2-15, where

the ET(30) values predicted by equation 2-3 are plotted

with respect to actual reported ET(30) polarity values

(Reichardt and Harbusch-Gornert, 1983). It is interesting

to note that this multiple linear regression leads to a

poorer standard deviation (s = 3.86) than obtained by

plotting the original P' versus ET(30) values (s = 1.11).






















ET(30) 50o
(predicted)
40


30


ET(30) (Actual)


Figure 2-15. Comparison between ET(30) polarity values
predicted by equation 2-3 and actual ET(30)
polarity values reported by Reichardt and
Harbusch-Gornert (1983).













CHAPTER III
CORRELATIONS BETWEEN CHROMATOGRAPHIC
RETENTION AND MOBILE PHASE POLARITY


Experimental

Retention measurements (other than those reported in

the literature) were obtained with a Spectra-Physics SP8700

ternary proportioning LC system (Spectra-Physics, San Jose,

CA). Columns were an Altex Ultrasphere ODS (5 micron

particle size; Altex Scientific, San Ramon, CA) and a

Hamilton PRP-1 (10 micron; Hamilton Company, Reno, NV).

Both columns were of size 15 cm X 4.6 mm I.D. Test solutes

were obtained from Aldrich Chemical Co. (Milwaukee, WI) and

the Eastman Kodak Co. (Rochester, NY). Sample introduction

was achieved with either an Altex injector equipped with a

5 microliter sample loop (Altex Scientific, San Ramon, CA)

or a Rheodyne Model 7125 injector equipped with a 20

microliter sample loop (Rheodyne, Inc., Cotati, CA). Flow

rates were either 1.0 or 2.0 mL/min. The column was

thermostatted at 40O.10C with a Haake Model D1 water bath

(Haake, Saddle Brook, NJ). Solvents were obtained as

described previously (Experimental, Chapter II). A fixed

wavelength, 254 nm, Beckman Model 153 UV detector (Altex

Scientific, San Ramon, CA) was used.









The retention times for an unretained species (to)

were evaluated with injections of the pure organic solvent

(either methanol or acetonitrile). For the Hamilton PRP-1

column, this proved to be difficult at low organic modifier

concentrations due to actual retention of the acetonitrile

or methanol. Other supposedly unretained solutes (such as

urea and uracil) exhibited similar behavior. Therefore,

the to obtained from injections of pure organic modifier at

60% organic modifier concentration was used, since at this

concentration the retention time reached a minimum in each

of the two solvent systems.

Simple linear regression calculations were done with

the program "Curve Fitter" (Interactive Microware, Inc.,

State College, PA) run on an Apple II Plus 43K

microcomputer (Apple Computer, Inc., Cupertino, CA). The

program was modified to allow calculation of 95% confidence

intervals for slope and y-intercept values. This program

was also used to interpolate ET(30) values for solvent

compositions that had not been measured (e.g., 45%

methanol/water).

When curve fitting the data to either a linear or 2nd

degree polynomial, the resultant standard deviations

(s-values) were used to calculated an F-ratio as


F = s(linear)/s(2nd degree polynomial)


(3-1)









The significance level (a% values reported in Table

3-1) of a given F-ratio was then determined by using the

program "F Distribution" (public domain software provided

by Computer Learning Center, Tacoma, WA). In this way,

much more accurate estimates of the significance level were

obtained than those from published F-distribution

statistical tables.

Multiple linear regression calculations were done by

using the program "Statworks" (Datametrics, Inc., and

Heyden and Son Limited, Philadelphia, PA), run on a

Macintosh 512K computer (Apple Computer, Inc., Cupertino,

CA).



Results

In RPLC, retention of solutes decreases as the

concentration of organic modifier is increased. That is,

as the overall polarity of the mobile phase is decreased,

solutes will spend less time in the stationary phase. Of

course, there are many ways to express this decrease in

polarity; the simplest measure of this is the proportion of

the organic modifier. Traditionally, chromatographers have

measured capacity factors at various organic modifier

concentrations and then plotted the logarithm (base 10) of

the capacity factor as a function of this concentration.

In the present discussion, the abbreviation "log" shall

denote the base ten logarithm. Plotting the logarithm of









capacity factor is quite logical, owing to its dependence

on the free energy of transfer (AG) of the solute between

the mobile and stationary phases. This relationship is

expressed by the following equation:



log k' = (-2.303 AG/RT) + log(p) (3-2)



where < is the phase ratio for the particular column.

Thus, plotting log k' versus percent organic modifier gives

a sense of the change in the energetic of chromatographic

retention as the composition of the mobile phase is

changed. Whether or not this type of plot is linear in

nature has been the subject of much debate. In terms of

concentration, the only reason that percent organic

modifier is usually used is that all chromatographic

instrumentation has been built to deliver mixtures by

volume percentage. While plots of log k' versus percent

organic modifier often appear to be linear, they will

always exhibit some curvature if a wide enough

concentration range is investigated and are best fit by a

quadratic equation (Schoenmakers et al., 1983). To

illustrate this point, in Figure 3-1, retention data for 4-

nitrophenol have been plotted with respect to percent

organic modifier. If the data are fitted with a straight

line, a squared correlation coefficient of 0.9803 is found,

while a quadratic curve-fitting leads to an increase to









0.9985. Clearly, the variation in the log k' values is

best accounted for by an equation containing a quadratic

term.

From a physical standpoint, it would be much more

logical to plot the retention data with respect to the mole

fraction of organic modifier. That is, solution properties

(of which reversed phase chromatographic retention can be

considered to be a result of) are best expressed by

observing the property as a function of mole fraction. In

such cases, deviations from linearity are then (by

definition) deviations from nonideal solution behavior.

The extent to which plotting log k' values versus

either volume percent or mole fraction or organic modifier

can affect the curve shape is illustrated in Figures 3-2

and 3-3. For both methanol and acetonitrile, the mole

fraction has been plotted with respect to percent of

organic modifier. In both cases, the actual mole percent

is significantly lower than the percent by volume at all

concentrations (except, of course, at the 0 and 100%

points). Using the same log k' values shown in Figure 3-1,

the data have been re-plotted with respect to the mole

fraction of acetonitrile in Figure 3-4. Here the curvature

has been accentuated, and a straight line fit of the data

yields a r2 of 0.933552. Since the variation in mobile phase

strength is not necessarily directly related to the percent

(by volume) or the mole fraction, a more logical approach









would be to compare retention with experimentally derived

measures of mobile phase polarity, such as the *ir and

ET(30) values. In Figure 3-5, log k' values for 4-
nitrophenol (same values as used in Figures 3-1 and 3-4)

are plotted with respect to the 7r* values for the same

composition (n*-values discussed in Chapter II). In this

case, there is a point of inflection, and the data are best

fit by a 3rd degree polynomial. As discussed in Chapter

II, one would not expect the Tr* scale to correlate well

with chromatographic retention, owing to its insensitivity

(by design) to hydrogen bonding effects in solution. This

is clearly reflected by the data shown in Figure 3-5. A

straight line fit of the data results in a squared

correlation coefficient of 0.9667.

The ET(30) values discussed in Chapter II can also be

compared with chromatographic retention. The ET(30) scale

has been shown to be sensitive to both hydrogen bonding and

dipolarity effects (as discussed in Chapter I) and thus may

serve as a better indicator of the strength of the mobile

phases used in RPLC. In Figure 3-6, retention data used in

previous figures have been plotted with respect to the

measured ET(30) polarity for the same mobile phase

composition. In this case, the linearity is much greater,

yielding a square correlation coefficient of 0.9950 when

fitted to a straight line model. This is in great contrast




















log k'


10 20 30 40 50 6


% Acetonitrile




Figure 3-1. Retention data for 4-nitrophenol plotted with
respect to percent acetonitrile. Ultrasphere
ODS (C-18) column; flow rate 1.0 mL/min.



















Mole
Fraction


% Methanol


Figure 3-2. Variation in mole fraction of methanol as a
function of volume percent.



















Mole
Fraction


% Acetonitrile


Figure 3-3. Variation in mole fraction of acetonitrile as
a function of volume percent.





















log k'


0 0.1 0.2 0.3

XAcetonitrile


Figure 3-4. Retention data for 4-nitrophenol plotted with
respect to mole fraction of acetonitrile.
Ultrasphere ODS (C-18) column; flow rate 1.0
mL/min.



























log k'


-1 -.
0.8


I I I


TT1*





Figure 3-5. Retention data for 4-nitrophenol plotted with
respect to Tr dipolarity/polarizability for
the same solvent mixtures. Ultrasphere ODS
(C-18) column; flow rate 1.0 mL/min.


0-




















2




log k'
0-


-1 -i i i ii
56 57 58 59 60 61 62

ET (30)




Figure 3-6. Retention data for 4-nitrophenol plotted with
respect to the ET(50) polarity for the same
solvent mixtures. Ultrasphere ODS (C-18)
column; flow rate 1.0 mL/min.








to the other three values for Figures 3-1, 3-4, and 3-5, of

0.9803, 0.9332, and 0.9667. The best fit is obtained when

the log k' values are plotted with respect to the measured

ET(30) polarity for the same mobile phase mixture.

Of course, the previous figures pertain to only one

individual set of retention data generated for this

research; in order to make any generalizations about the

correlations between the various variables, it is necessary

to examine a large body of chromatographic data. A total

of 332 sets of chromatographic retention data (log k'

versus percent organic modifier) have been examined.

Retention data reported in the literature, as well as data

generated exclusively for this study, have been included in

these correlations. Hereafter the discussion will be

confined to two types of correlations: those between log

k' and either percent organic modifier or ET(30) polarity.

Linear regression was carried out for all retention

data sets with the log k' data compared to both percent

organic modifier and the ET(30) polarity. The results of

these correlations are compiled in Table 3-1. The data

have been sorted in a hierarchical manner, using the

following sequence: organic modifier, column, and

solute. Squared correlation coefficients (r2) for both log

k' versus organic modifier and ET(30) polarity are

reported, as well as the regression coefficients for the



















Cli


to
0


*H
.4-)


r-




cu
0




a
&0






















c
ca


r-
0












t-P
Uo




















0
4.-






















.4-)
a)


a>
.1











cu













.02
CI,
C


0

E-4
c








()

0
0


-p
CU



*-t
^ -4


CO


o c
H
(U
Ll














.0 *
3-p



3r-(

O 0






a,
L2
.MO

C 0)




as~
(l Li
* so o


0
..

C 9 CJ CC0 0 0 C
c c o .4c40000 a ce 4 c 0 r o
c a. c0 c c C. = .- 5s 1j 14
c c.4 44 ) ) 0 x 4 0.4 o. = 'aoo40..C0..4o 4
4.0 SOC X CO ...O COCO C
4Cc C c 0 0 o o c a p c.a c
*44414 J *4 Ll @0 44 44 jC. h4 h< k 6JS J.X! CjJ4C.C
C C C C N N.. 044 t C CV'' sO) .C: '* J 44 0 .. *0 *. J0 ..J

rj j r rr o. -
'o .0.6 C c N0a 0a0 N..4 0. 0
00000.40.044 r 0.4.0.0.00 1" CCC 0, -c).0 CS 0.4

3 ~ ~ ~ ~ o<^- ^N> a. '3. ^ I|I> 4C I I I W W | I W K c
*I z zz4 44 c .P0 OC 4 Z4 C a *C 4 44 0 z a v '-0 v 0 u .
"a 44 14 10J 4 .0 C = g. T4 U. I4 41
"A0 I t IN V 3 C C .4 O W0 | C 13 | -N Z | f"

*l

ii ~0 1 N1,4 4 1 "4 4'104 4 0 ,4 4
4 4 % 4 N N N N N v 4,


78








0
m


4)1
f4.

0 I























00000'00,0 0
* x
SOoOo oooooo o c





























10100 1! M
w4 A0. 0 W W 6 w-o zw. .00wm0m
-ss ~ a 3 3 *

f- ^ U- H




04 0 C, 0

| eon fn -t~ro n~Qmrt {Lo-




















0~~~~~~~ 41u 41 U U UU U U u uu u u u
~0~0n~N00-ON 4104=9 ~C0~0N00~4,'01




















41 -c141- 4 41 0 0K 0 0 4 4 1- I -c 0 -c -C d 4= 0 Nc 41 4= 0c -c 4= 41 -c Nc
0000000000000000000000000000000000
*| 00000000000000000d000000000000000000




CI w .N^ .N r l( N 4 ,0 0 -~

4,~C =419N41II~~04,NWN40 .. 0~4==0.0Uu-r.00
















0
0
444> 44 444444444444 4 444 4c 444o4 4o O4 L 0040 44 4 4 444 4 44 4 4 4434 4444 44
4,000000 N00000001000000000000000410000000,
NN4,0444 000~0 -Ned4.44,PIN 4,4,NN4 4.444,4,4








4' 00 44 .4 04 ., Or 414414, 444ONO m~PS4N= ,.04SN -44N 4004004,44





0 00000000000000000000000000000000000
00c = 414= .- 00044 4,4145 0041 04100





Nf 000001 I 0000~ l0000000000000000000O00000




000--OOOO^'"-0000000000000OOOOOOOOOO
341 44 (44f44f44r





4i 4
Ur

















*J


I
C



0I M M M a M M <
S- = = 4 t t t0 .' .: .t : S. .,.:.
*
S


_* 1% 1, 1g o'' D' o' Da aw a a at -a a D ae Q a o A ^ih* -A e 0
o[ 0 : "2 00 02 M M0g V0"S v0 0 'D '0 :! M M "




0, cc coN o '0 v a w N 04 'a 'a 0w 0, vr 'o 404 44'D 'ao
4,t 4-lW^ 44 '044w 44f '04 ^0 44~ 44'0. '040 44o 44 4f 40 54 ^- '0 44^ 4-^ '0 40 ^'0 44u 404440
0ooo0oooooooooooo00000000000000000000000
*I 000000000000000000000000000000000000000



4-W* 444 40 040H 4H'H .' 4-H 4M-M 41 4M -- '00 '0W 4 l 441 44W '0 4H 0M 4H-H 04 H 4 0 ^i'0W '0

0 404 0-0 04- 0 A




41
o F. 04 0' 0C0! !! !n!nn0C.n0nn n f 22
a C* 0.1 1 0M 4 M 0M -M 0
o -, O 40 0- 4-0 04a 0O- 040 4 -0 c o 00 o 440 C 44 0 C 40 o 00 '00 4004 440 'C, 0*o-06 0 4o 0om


00 w 4- ow Mw 4D 4W'a






v -C -4 040 '44W 0440 '4 4 0 4c 44 K0 o c' 4 -C I 0-f 44 -C '04 40 4 -






u
00Q00000 0000.00 00000000000000 00 00 000









0 00 00 00 00 0 00 00 00 00 0 0 0 0 0 0 00 00 0 0 0
04004 004040 044004 0404 040 040 or 0 040 040 00 010404 040 '00 040 '0 a '04 00 04
~~~ >.:" =p 1. u^^ a^ k w"^ w- l- ^l n ^1r><~ ^'* ww^*^ w 0.6l
0 0 0 0 0 0 0 0 c 0 0 0 0 -s 00 0 0 0 0 0 0 0 0 1 1 o 41 2 W4 0 0 0 0 0.0 0 -0 0
J 3Z Z2ZZSZ2 ZZZZZ2!ZZ22ZZ 2Z'ZZ2ZZaZZ Z2ZZXZZZ
e UUU UUtUIUU*UIUU UUU t~J tUtU UItU~*U*UUUUU JtU UU
4444444444444444444444444444444444444444




















c rW 00 0 M0 000 z0L
IS



















M.C0. 0000
n a 000064o00r k



116 .CC.W 444 0-6 80 0- :30 S..0S-
0 00 0 56 a 0 wk.0IO. I. I0I



.0 0 010 a40 .0a0a a .0.00N40vIn010 .400
5044- O <<<<<.<<<<<<<<<<<440<<<<<<<<<<<<<<<<< 0<




-p I






0I .5 I I I I. .4 0
A .10 A A 4 0 4444.V0 00 -W, 4 4 4 .0 N
ni~~ ~ ~ co M c0 jj
4.04. 0*i~ I.0U 00.004 00444 40 O44E.'',O Q I4. 4.0(
A0 vwvvvvvv 4 0 4' 4O 0 44 4 .4. c 444440o M
i.4000.144oS00'aW'0444NC 44044'*i 0 44








) o o W '0 OWq '0 W' o o a 4o o44' 0 4 4o 4 n 4- 4 a
C Be e l kj>.j:u O 4 4>. ou >si b ) i U c) jS4J j;

j= B nO u e 0 *J aB o oj a o Cir oomrmr a.i u t< u


>* 00 00 00 f4 UO B0 00^ 00 u b U 0 B0 J3 u0 a OQ Oin^ *0^ oo
J:0 0 1m IOQ0m~0 o481> 4w ai- ^-4f ^*1 4* 4000m k4~mm e
*f1 11 1 1 1, 1 *'11 0 Dn t OQ QQ iITaQ- 30'^^nIrtO' ^
I 3 -4W >" | | l | l l l | lJC > l X:X: C >>* - *

N^no~no~~n~nnnno~ 91





(0(
E-i







80








61
















0
n









CC
4. q" ''l ;mM <>









99 9 9 .9 9 9 9 o 9 99 9

C 0! r:o











op-~~~~~~ 90~ ov m:: 00 o10 -in 010n MO No<" 0- Cc r1i0 1 -nowrt i N MMnr.
-2










w 0 19 0 v a v N ~~~~9N 0, Q M-0 CD Om 0
> 010 v M 0 W M M : N 1 0 0 '00oon
*( ooooooooooooooqoooooooooooooooooooooooooooo
oooooooooooooOOOOOQOO ooodooooooooodddddddddd
o o





















il0 zz z z z z z z zz z z a z z z 6 u u- uu-. u^- ''- u -- u u u u,^-M u u *r|u-"" I
61 -C N NI I 40 -C N 40 61 4 N N c 61 0 N 00 4 N N 4 0 1 N 0 40 40 40 4 61 61 40 W N -C4
Q4!.N 0~~OOn NO NN NNN N 4f40IA 6140WWO













0 0000000000000000000000000 000 00000000000000000















OC O1 ON











N>
-C cN N W I c N ** *N -r N0' Q .0.0 ..
61NN NWc~ ~











No -* P' o3-) N -on- o n 610 -M (DPW io s c-mrat- NT NNN-p~f


N o oo oooo ooooooooooooooooooooooooooooooooooo

c N
NW61-N61~~nNN~NfNINNNN9NNNWOlNNCNNWWWWNNNW





























006WA wN uAN u 0 0 IA40NNO NIA40NN N s a NN0N000 0 0
6161NN61 WWAIA .-400 ~ ~ ~ ~ ~~N-N N644NNAN1N4461 661IN-4r U















o -NNIA NN NNanlN N w NNNNN NNNNN Nwf4 NN ,.aI
0







































61Z1 61616U6 Q166116614 06 W zA W w z 1 'Dim, 16 661
N Nr W 6 1O I co 6N mW00 nO m61NN 0610!. 4'44














61 61166 61466 616 616646 616 6166146. 61111611666 N 616 61111666 61 66
,D ~ O -i D >,.< o o -I -< ^-< 'o y o ^ t rt -< x e~ e-n o c x ij c

^- 11 aO >* E M ~ c u u o C o r o dr c, \o o >* i a au c, to i' ? 0 2 2
0000000000000~00000000000000000~000000000000

00N000000000000000a000000000000000000000000000
..!.NN W...N.N N NN- - NN.-4NNNW

ZZZZZZ rZ ZTZZf ZZZZZZZ0 zzzrZZZZ".TZZW".Z~ZS
o 0 0 0 0 U 0 Q 0 < QQOUU 4O.4 UUUUUQQ N UU4. JUO'


444


~~4~44444~44 44






0, 6. 44 CCC CC
CO ~~ C'~ C4 CC 444.4
44444444-C-440 444 *4400C44444444 444 CCC C
C4444p)C44C C~ 00CC 004444 CC44. O O .444 --4)li a~ ~ (~l
44~ ~ ~ ~ ~ e~~ 44-----44~ 00444- V444'4444 CC C~ 04..4
C 2 ...C 34r4SX .".4OCCC OO 444 4444a4z 2 444

5 ( I X C
4400.4044C 444.0444-4 00C am 0.4.4 00 0 44
F BY 4OO044044044CC.0.0. Or .---~-- 44.C.CC CCCCC
CT044J44V.4-4..J4.''S~~~~ 440044400O'J44
44. j4 ; r4 4
442 TZZ T~rUUPZZZN W W N W NW N .NNWW ',44 TrflWII

-.4I

.0

I ~~ONnWIA4004-6161m~0NNW.A404~~-61600000-----N
c -j ~






81












a

0 c








rl
0


P ai f* .3. ft a anm aca::.:: a^ aaauuam ir



I -0%Nr -N W OrM n. 0%i0'a -Vvn 0
raON'aa' n wg NW=an ininN







oooo'oooooooooooooooooooo 1010 v 1 0 1-M





I N flU a' a 0 1 1 nWW0 W v vo WW IcaWWW
S0M' M 00 w Wm1000000000000 0 0 0DM 00IMM W 00 00
*I oodooooOOOooooooooooooOooooooooeooooodo











ovm a1 o m o m0
an 0 M M





lot
M-M M 01 0! cNCN a01 i aC a^ 0 a M M M Min M W
CI Ncl in WOC CNIn'aOWN NON O'a.WN~N0%000%N~in0 N n~
N0%i nnm00 %inNN%0F~~ 0 %NinX09N'0O W0 C N iN % %n 0%nn% % O %







o lw v N 1n Wn WW a 1 1 W Q 10
-^ N'ai 0N'aV M^ r ^* ^* *fiit ti fo'C u '' o ou i r nC













0 c4444 ~N44 44 44 4M 444 44 4~0%4 ~4444 4 4 4 4 c 4 044








r
"-4 r~r~o s
i W Wg WD Nr N N In N NN N N W in W W N N 'W in N W r N N 'a N in N N





















0
N 0%' NC< Nr0 NWO 'aNff NNf C 0t N ff( in 0%0f 0%in NOt 'Dinf N0%T NCffC NW f 'Din 7 NW O in T WW '



S000000 0000000000000000000000000000000
Nin'ai ~ ~nNCN~ n0 W %.n0W %nOmNW0%0 0%~NN0%0%N'aN'a0%N






















0%0%0%0% % 0% 10 0% 0%0 0 % %0 0% 0%0 %i$ % 0 0 0 N % a 0% %0 %0 0 0 0 $ % 0% c % 0 0% a 0%0
N
ha 00000000000 0 mc0000000000000 00000000000


0









0 .N inW 0rNNN,...'NN. -W A- P: o V g N. N. N
p u v r- N = a 0 0 k c W a o Z Jo 5o o















0
'^i 00m0000000000%0000000%000000000000000000000




I I



030












0 0 0
60I .% a 0Cc .0 =2 6U WW-
00 .00 N


-- -- - -- -

ca
E-4- g5.3 I
o 0;.e >. A 0 f

O QO h in O4 0050 00 O j *:* CN 4- 00 000
00.40 H UO |I BNO 00 00 00 00 0 00 00a00


I a -asi. *~ o0 e ( c j ooj i o B^o oo u t 00 000 o0
I 1 O4fl Win0uio^BMo c wa.o .o watwo. ..xx..m^ c Oi.i 4 3 a0
I % t i i i 0 ~O 04 -^f-^ i.: 0- i .0 N. SO j a ) > 00 0




(0.
U'
~~~~~~I ~~~CL~6~0~~~~D
I ----------------------------""
CIOO OOQQ

C',







82















.
u
r

I . .. . . .


I I





oI -W '4D '4D 'D a 1 0D 1010 W a a0 W0 .0 0'D 4' 4' 0 '4 0 V4 V V v

N00oNN |N4o a'NOwo 0..N
~~000000000000000000000000000000000000000
SI 0000000000000000OOOOQOOOOOOOOOOOOO




|I NOOO OOOO OOO 0 NO CNOC !OOOOOO OO! OONO





^ t ...... . . .
n4'N 0%MOOn nnnH NnnnN0%0%Nn-N4'00 ON O'N O.NN 0 N.N






4'0 M N4' .0 -N N* 4f 0* NM NM- 4' N M 4 0 O W 0 4' N* OW 0* H N N44 NM NH 0 0H Nt Nf> NH N4 4( 04 %
-nOCCCIOC10 0 w M I0~n~~Nn0



.OO0%0 4- O C '-0%N C 0 % C 0'% 4N
0%NN N^ N-p 00 Cp' 0- %O ff0 N^t 0t 4'a C 0%o C i/ 0 op 43 ao N0 0%~ 0 n 0 "* 0 a N %N




N No N/io 0 t I' N 0 4 N C0 0 N ~ a 4M 0 0 N 0 N 0 N 4'0 0 0 c 0.0 0 N 000 N 10 N
W ON N NO 4' 00 NN4 'NN '0!0 0! 0 0% NON 0 % N Ni OONC a







0 4'D TI 0% 0%4 % 0% 1010 0% cN 0 0w '0 4 000 0~ %0 0 CCN4











a O
0~nQ00~00000000 0000000000000 00000000000







*1




C ooooooooooo oooo oooooooooo ooooooooC -00C




















0
'0.000000 ...0 0 1 0 0 0 00 00 00 00 0000 00 N N N N N
0. 0 14 NI rN If or
N NNN NNN N NNNN NNN NN0 000 0 0 0.3 0 0b 0 4' 4'
0 Ok A 0 a'O mCO U U OU O UO U








'00




0) T
N N
0 0
444 44) 0 444

0 e N04 N O 4 4 4 N 0C 44 .0 NC
o CO 444 UCGCS -N C S S 00505"cu H n

I.0440s 4044>. 0 ce44o)* 00 0 0 4 0 444444444444044 >.>.0 044.444
-- 3 >.044 0 0. ).44 -N. B 44 .0 .4X 3 ... >^ >.0 0S4 0440004 .c N

I9 00 9 i O l liI 44 U J J0 *10, l i C. 4 4. )I J44.4 j > l 0. *C 4
i 20.I4o. B E: oe-nflBL z ^a x b f&ee -z <




----_---------------------N
-0
^ ~ h C~~~~~~~~~~~~~~~~~







83
















0 a:
U .

C !



2' !" D' t

c~cc
I .



0 w 4S mm
j3




a: aa a v 0 v *










Nm mlU 0, S
C!aa a'a' 0! G! 0! a! C! 0! 'a 0! C! 0! 0! 0! 'a 0! a 0! 0! 0! 'a 0: 0! 0! '! C! V! 0!' a' a a' a
w 0 N NN-NNNNN'TN--emm MN.















M0 m0 m0 (70c m000ma000C0C n0000n0m 000 0,m0 w 00mm0 0000 00 00
000000000000000000000000000000000000000



















6zzzzzzzzzzzzzzzzzzzzZZZZ=ZZ=ZZZZZZZZZZO
U3




























ON C
NU,00'...00inin0000N00N'a ccnmona.'a SON

'a'annn 'aUU,'a'NNN'aC'NU,'aaUCaUaNU,,0'NN'a,'N ..




























'0 N a w,, N 0 'aa0
0 c 0 UUa .'a0U a v -' mO. N0s
o c r c 0
In F^ prfl tf'^ ao r u mL ^ o fl ^o l to 0M C^ M n ^ n l >oo ^ool c ff c m ms~\



4) *o = *o2 .0 oou, a :
























N N .0.0o00 0 0 i O 5. 0 N U m0 iC,00wNNw
S00 0000000 000000000000000000 000000 00 000


*000000000000r00000000000000000000~C~~m~0 0




a
p *
rn ooo oo ooo oo o '*o-ic oo oooooooooooo o
3
N O~C~~0000000000 00000 ~ 0 00000000 0000000

I.l


0
0000000000000000000000000000000000 .
I I: II I Ii IiII II II I I III I I ~,II II I III I U,
000000000000000000000000000000000000000 N

2Z2Z ZZZZZZZ ZZZZ ZZZZZZZZZZZZZZ 0
0 UO 0000000000000000000000000000000000 41U







0



aa
T3


a 2

P S 4

CO MO~O J5 0
0~2 .* .0h l(

0o( O C J 0 J
C 6.* N *N Bo
I4 24 a N N 24 e 62402c Ch44*E 024 O4N c B C *
IC C 24 NC1 C C3 eC c o *c o Q 6
2462 .0 C2 9 e.O0 NU.t soawv a a c ow .0 U
> NC~ 40). ^ ^il> C 4) 0k4 C J O jC 2442 U O 4.04 0. ->OOf
0 0) .C i0 C 0 44J e O CC.0240 4.0140 0 0 o 24l 0 240O .4I C Q 00 E



*~ ~~ 00 0 0 0 -- N N NN N N N N NNNN'a' aOOX
K-\ ~ ~ N N N N. N N N N N) N N.s* N N N N N N N N N N N. N : N N N N N N N 1 N N N N N -
~ ~ ~ ~ ~ ~ ~ ~ S~E &wj ,u- axiX M' j >u30SOQ tJ r
0 -4*r-Y~EX ~


r"^








84














y
0'


5

a
0



0W a V I m 10 w


W A 1 cc cc l` 00 a0 r0 a) W 1 co 44

cl *rWWWn WW4WWr WW~ WW4444444444444 4444IflW WWWWWWW


SD ~ m~O'SD WSDW 0'44 -044 rrmo'eoNnoao
w~~Cm0omwC~rCClvwv ONN-a~~NNN
wN N 44 N N N NW I 4 I I f I 1 10 'o N SD SD N N N I'4N
00 000000000 00000a000 00000000
ml QOOOOOO00000000000QOOOO000000000000000000


ONNN4 OSDN NNW0 00SD DN W'ONS
N.-NNOnNnNN OONNNNNN NNWNNnN90
moo 44 04 0444 ::! =. OWW4 W4w
g --^~C~C-^-~-P ~ ~ ~ n ~c-c-*" -- ii^^ 'f'^ -^ -' -o^i

W N 0NO IM0 44MM 0 NNW 40NM NSVM D D0 40 D0N 'N-4
-^ mrO CNN~ i 4 B4 NO N> N N No Nv -Wf N-^>- o o No N~nf NO N~f Nt N Nm 4



.0 N W O .S N t1
NNNSDNNNNNNNNN~~NN00N0N..N~OONNNNN...N0N0
i, v A .us A' -r w C Zfo*< w* v- Z M cr .r^ v^ ^ 0 -IDW< O ( f' -r M o 11 < 1
SDN DN0NNO N.S WO d44C0'000NON NNNO0 0OWSDm


Nlmwmwomo N !`oo
v*tO SD W W o'^ tN4N --N N< SD rsD 4 Nr 3 44 P 44r^ W r4 W WM S N 4 4 4 4 0 N 0 r N N SD 4



N444 SD -N SD NM N* N4 NM 0' SD 44 NM 0' 44 Nt 44 SD ON -c -4- 44 N NON NON ON NH 41- N SD 4 D 04 NO1 SD-H
NN0SDN S S NN N WN0 NS 4440NN4 0 0 0 N 00ON



0' 440 oN 0 N SD 00 SD 00 SD!i 00 ON N! 0!00 0! C! 00 0 0' 0' 0 0' a' C0 00' 01 '
Nj 0000000000000000000~~9~~~0000







1~~0 0' 04400 0'0 0 !
44 ~0 N ~~NO44Wm~~N~r0~0'SDS0'0'N 0S 4N D D'D0N00000000''
) 0'0040000N00'440oa) '0 'N 4 4 0'ip 0iou'^'tir ow -c 0 '040040000'0'0 0 '0n ( 0 0 0'F0' a 0o










00'040 00 04' 0' 0 0'040 04' 0 0' 0 0'0' 0' 0 0 0' a' 0' 00' 0,0 0' 00' a0
4' ~~0 0O 0 O O O O O O O O O 0 O O O 0 OOO0 E5


0 5 000 0000 5 00 30 a 0 50 0 0000 44 44 4 44 44 44 4 4 4444444440
W^i


M 0 C~^ omm Mo' 0 ^ 0 1 0 M 0 M m01 Cocol In~oc* u r u ( u u uo u< u~ u uu vv vu u o u ^ u uu
SDSDS SDSD DSDSD DSDS SDSDS SDSD DSDSNNNNNNNN NNNNNN NNNN S




















No
~~O 0 rOO O 0 O O0 0 O O O 00 0 00 0 00 0 0
S 0)00000000000000004a000000000000000000000













0
(S












~ ~ 0 )^ W^ 0r 0. x^ xOn cD n^ 'a M* ~ 0 .D a? co a. xDf*WC *t 'O O O^C
vi 0t u u\ W3 WO a. E. a^ D ou omoooooomooooUooooUUUU UUUUUooooUoUUUUUUU



Ul





S, 4,=S: 3= E: = = BS B S S = ~ E E S: S 4=4

y; 44 40000 00040000000000 0000 0000 000 0000












o w 0 C)a w u Cu Ca ) 4>







E--4
-I 1 N NW 44 S D eNE 0s0' 0-N NW 4SD0 N S 0 0SD N 0
a1
dddddddddddiddddddddd

e
aI




.I
*E
iiooooooooo0~~~n~~~~n~
3 ~~~~~~~~~O00000
c snonno~n 0~~0111~000~
*r-t g~l

-^ 5






c B *


0 oc c >*c o cr ,
c" 2 c *e 9 ** aee *O CO
I Ne c a *
I qU N D NU C CO 41 4> ^CU S 0 ) IC
| : C: C O C N U J < NC NC *C N C kj rI(I.
I ~ ~ ~ P -tC UC N PB C >jQ^ Y C PI C 0 N ) C o ^l 0) r >^ c^
~ cu~u > i3 Ac ^a ) *^ j= ^ >. c c 6'a> ~f ^ >nc u c *' h:>.c
: t...,,.~ :l c uioi l pc^^.^oic aoomOccf teo^^-i
| Q OW ^M fl.CO lC B ^O -f-^^'f CO W' Q.iOi tl BCVt'O-*


Ul| < C CQ fflUO UU bPi- ZZ& -(SCC0 t riy* Z ZE-Q Pm & &. a C ; 400
0,




Ei





















5,)



















0; 0; C 0;ON NMI
t a





















OmomcC P MOMMOM MMMOM, ~~m r00 1

co (a cc w mw.0w w- 0C f ,a ,m0mma la
*| o = =" r4 3 4 5 2 2 2
3 5,I O 0C










000000000000000600000Q00000000QOP00000



0I I ,0, 0 1 0 a ,
















o
all




ci
4e 41 0N40 .H4-0 !. -N t4 >44 044M H 4, 0 40 0 40 4 4, Nt 04 44- 0*140 441 NH 04 04 -M 01 40 NH 4 4 N* 4N 0 M 04





NP 0 On n o 0 n N cc 0 M o f s o o

























I 44 1, 441 0- 4
o B -* B r*^ n ON4IN





























40a1N41 4 14 0 0 04 4, 421 a a4 a 44 4 --N 0 0 a4 40 4414 04


nne ~ 0 c wm o
or 1 0 0 M .0 0 .0 w c -cu c 0 >.-.. a4,4,M0


























0.c .6. NV.0 cc 4 0 a- 4 w 44C
T 0 0 a c 0.
=A 0 J3 I > *0 a 0 0 m a A NNNa




0 N ~ N I400 1 44 V 0 N V n 0 N 4 N1 o 0 0 44 o oD a 0 O 0 ,, n V a. N 04 0 c4 c Iw


( -40A A 04 04 u 4 N 4, 400 40 N MN 0 0 7 4 ra N 00 0 N2 0 I44 N N a 41 -u N a a
00 N 1 10 M M0 a 3 N 4 04 0 N 0404 04v4 N 444 0 404 441w m N 04 N 04 N N 0

c 0 00000000 00000000 000000 00000 00







N oooeoooooooooooooooeooeeoooooooeooooooooo





- o4 oooooo ooooo ooooo04040404010oooo 04
0CCCO000O00ON~O~n~h00000O00OOOOO~O000
OcoOOOOOOnoOOOOOOOOOOOOOOOOOOOOOOOOOOOOO













> ZZEZZZ EZZSXESS ZSS S E EZSEZZESEZZSSSEZZZSZZ
0 U UC0000000000000 O00000 2.2.2.2.2.4.4.444.4.444 fOr. 44444









0
(Q













N
e


44) uS a. 24o-
M44 444444e ccE C oc 0.444.4ic4 40 44 o4MO
4)14 N4N4442 44 a~ o 2 o N44 0 0: 4 44044444
~ 4424644400.004444 c44*4240.444 240.-40..00004 04.4444044
o ,4V2454404..240.0o"5,c 40.44044005,0000444.24440.44

c-~g0404 004U42.z0.D0.U44 U44~04404404U044..a&ca













S1 5 S S4S 404 -a 04 0 N e N V 444 N 040 0N V 444 N 044

SS9 u B *> IN N N N N N N N> N N Nl N4 -N
o N X, 4, 4,O 4, 4,U 4 4 4*4 4 ce
gc-sas ss e- a s s a 5 as o= oi
3| r~frsr2rr ~2EI~Z5Ir

B 8 2i' 2 Z 3 0 <~ O nnO^o o n i~l ~ ~ o
"ir s s a s K 0s s aa ^ ^ ^ ^ ^ ?




















4D
0%
0 0-







cl
m 10 o
0 0000
0000000
. .1



| n

















'00
CQI w. ..



*F 0-C 4 % 0 ir*















m w '0 C, m0 0 .00
ns r, w 0 u
00000 00000



N -. N- 'Nfl

1 r .0no oo
Zn. NN. S o N0%



0 -- 00 0 a o E5 Fa







0 o
Sto 04 O C N
-0 ^v00r Nf (nON-- 4,























Cc0
NOW rm xo ro Ua
0% *'00 0% 0% 0n 00 NO 00M l f*u







0 c0
NF 0 .00
Sooooooooooo 'M



14 0

No~~oc N 0 0 11


l .L ). u .0 'w

















caa
00000000000 -
-


I- li e 14 U U 0 0
(s p-' f.- r- B. )k0j2
0 '. .4 .0 0. -
-ooooo -. -. a o 1 44


O5O0O0 0 00 0 12 1 1 1 04 t S1
:: : : : : 4, -I B -. 12 0 U -
N 44 0 12
4, 0 0O 4,OC 0~e CDDC-n ^ t 0
.4 0.^^ ^'^ '-^ 2'. 2'. 2'. 2' 0= 0 4t
3 55555o oSm 35 ^S lS m c f




0 SZZE -ES4EZ m < 0 0 4 .4
0. 0 0^ 0- 0 14 0K*
~~a~o o 0. 0. 0. 444

^ g 0 44y 8
SU Q U U U 0
0 =fl 0 0 0 I 0 0





w 3 ., ^ ^ -; .
00~...hb .. '3 '3 3 3 2 '
a u u b lu u t



9.0u


S f. O O. O. -l 12.0
B *. e > >*







*4 12 0 '0 o o,
-~~ ^ > = = ; = 5




0 2.0 012 0 ~ C 0


o' 120 2 o 5 o o o aua> i
<" >. 3 '" -. "






o3 *' 00 0 0 12 0 0 0 4f a
U 12 B a a B In C a In
-"* ~ ~~ ~ q) *' H)C ^ C





-- 0 3 0- 14. 12 12 -O In In In I O -
0p *2'. 4,.4 .4, ^ 3
S 0 11 4 ,
O e o Q. e IQ o c u u





I' 12 1222424.4.. 0 ^u u o o we hf 4. = .0
0 .00.4, 4,2104
'" : S S"S S g^ r < B U Q M h. S



- C NC NN N N N N S N N0



-ol i CD Ot O -
0000*^U U Y~
(0 *~W0 9
E-i ~ ~