MSU

LIBRARIES
—_——.

 

 

RETURNING MATERIALS:
PIace in book drop to
remove this checkout from
your record. FINES w111
be charged if book is
returned after the date
stamped be10w.

 

 

 

 

METHODS FOR IMPROVED DETECTION AND
SEPARATION IN HIGH PERFORMANCE
LIQUID CHROMATOGRAPHY

By

Siriwan Ratanathanawonge

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

1986

ABSTRACT

METHODS FOR IMPROVED DETECTION AND
SEPARATION IN
HIGH pERFORuANOE LIQUID CHROMATOGRAPHY
By

Siriwan Ratanathanawongs

This dissertation consists of two parts that are related by the
need for methods to facilitate the analysis Of complex samples. Two
different approaches have been taken. The first area of research
involves increasing the sensitivity and selectivity Of the detector
through post-column derivatization. The second approach has been to
improve chromatographic separation by recycle chromatography.

An on-line post-column reaction detector for phenols has been
developed. This system links together two highly versatile techniques -
high performance liquid chromatography and air segmented continuous
flow analysis. The resulting system contains the combined advantages
of each system. The reaction employed involves the coupling of phenol
and diazotized sulfanilic acid to form an azo dye. The optimum reaction
conditions were initially determined by a univariate method and
subsequently by simplex Optimization. Other studies performed included
compatibility checks between the post-column reaction detector and HPLC
systems. The application Of this detector to the determination of
phenols in various sample matrices has been described to illustrate its

versatility.

A new approach to recycle chromatography, recycle heartcut
chromatography, has been developed for applications involving complex
sample matrices. In this approach, a heartcut is taken from an eluting
peak which is then recycled onto the column. The sequence Of heartcut
and recycle is repeated until an 'adequate’ separation is Obtained. This
technique has been applied to a number of applications to demonstrate
its potential. The results show small improvements in resolution and
simpler chromatograms due to the isolation of the components of interest
for further separation. The amount Of band broadening was studied as
a function of the heartcut loop tubing dimensions. A microcomputer
controlled HPLC system with time and detector-based heartcutting

capabilities was employed.

To my family,

for all their love and support.

ii

ACKNOWLEDGEMENTS

A large number Of people have been instrumental in making my
stay at Michigan State University a very happy experience. I would
like to thank them all for their support and friendship throughout these
years. First, there is our fearless leader, Dr. Crouch (I mean, Stan), to
whom I am very grateful for his support, guidance and friendship.
Special thanks to my dearest friends, Arnie, Bob, Peter and Marguerite,
who have always been there when I needed them.

Of the Older students, I would like to thank Frank for teaching
me how to ride a bicycle (into trees, concrete benches and parked cars)
and Pat and Keith for their unique sense of humor. 0f the present
group members, there is Paul, who collaborated with Bob to conduct late
night chemistry shows e.g., N1: and laser light shows; Helen, who
accompanied me on a trip to Detroit Metro and almost didn’t make it
back to East Lansing; Tom and Linda, and their camaraderie and driving
lessons; Cheryl and Thanksgiving dinner; Mark, the victim of the booby
trapped dragons; Max and Emily Post; Jimmy and his kindness; the
legendary Nelson and his anti-Midas touch; Adrian and family, Scrabble
games and that Middle Eastern salad; Kris, who learned the hard way,
how not to connect the tubing from the HPLC detector to the column;
and Mayda and Steve, who were assigned various duties because of

their lack of group seniority.

iii

A second group of people that I am very grateful to is the
departmental support staff who have assisted me in numerous ways.
Finally, I would like to thank my wonderful roommates, Catalina, Regina,

Touran and Mireille.

iv

TABLE OF CONTENTS

 

CHAPTER PAGE
LIST OF TABLES viii
TABLE OF FIGURES ix

 

I INTRODUCTION

 

 

A. Reaction Derivatization in Liquid Chromatographymmwwm 1
1. Pro-Column Derivatization 2
2. Post-Column Derivatization 2
3. Post-Column Reaction Detection of Phenolsmm " ___fl_____ 4
B. Recycling Methods in Liquid Chromatographywmmmm 5
1. Resolution Enhancement in Liquid Chromatography_ 6
2. Recycle Heartcut Chromatography 10

 

II HISTORICAL BACKGROUND

 

 

 

A. Continuous Flow Analysis 12
B. Post-Column Reaction Detectors 20
C. Colorimetric Reactions for Phenols 27
D. Recycle Chromatography 30

 

III DESIGN AND CHARACTERIZATION OF A POST—COLUW REACTION
DETECTOR FOR PHENOLS BY AIR-SEGMENTED CONTINUOUS
FLOW ANALYSIS

A. Overview 36

 

 

 

 

 

 

 

 

B. Experimental

1. Apparatus ' 37

2. Preparation Of Solutions for the Determination of
Phenols
a. ,Reagents 40
b. Phenols Standards 41
c. HPLC Solvents 41

3. Sample Preparation
a. River Water Samples 41
b. Residual Fuel Oil Fractions 42

 

CHAPTER PAGE

C. Results and Discussion

 

 

 

 

 

 

 

 

 

 

 

 

1. Introduction 44
2. Reaction Characterization 44
a. Optimization via the Univariate Approach
1 . pH Effects * 45
i i . Influence Of Reagent-to-Substrate
Ratio 53
i i i . Effect of Reaction Time 54
iv . Effect Of Temperature 55
b. Simplex Optimization 59
c. Comparison With Other Workers ___‘_____- 67
3. Evaluation of the HPLC/PCR System
a. Mobile Phase Effects 68
b. Band Broadening Contributions 71
0. Flow Rate Compatibility 73
d. Statistical Figures of Merit 75
e. Performance Comparison with Other Systems 75
4. Applications 77
a. Determination of Phenols in River Watermwmw 79
b. Phenols Determination in Residual Fuel Oilmm. 82

 

IV THE DEVELOPMENT OF RECYCLE HEARTCUT CHROMATOGRAPHY
FOR ANALYSES IN COMPLEX SAMPLE MATRICES

A. Introduction 88

 

 

 

 

 

 

 

 

 

 

B. Experimental
1. Instrumentation 92
a. HPLC Apparatusmw . 92
b. Microcomputer Controlled HPLC Systemflwww_‘_____._ 93
3' Solution and Sample Preparation“ 97
C. Evaluation of the REC System
1. Reproducibility of Valve Switchings .. 98
2. Band Broadening in the RHC Systeni _ m. 101
a. Heartcut Loop Tubing Dimensions *« 102
b. Partial Filling of the Heartcut Loop: ____________________ 106
1" Applications ...... .c_*.. 111
1- Component<a> iaoiation 113
2- Verification 0f Peak Purityw.“ 117
3- Quantitative Rite 119
3- Conclusions 122

 

V FUTURE PROS PECTIVES

 

 

 

 

B. Post-Column Reaction Detectors
1. Tandem Analyses _ 124
2. Solvent Segmentation _.. 125

vi

CHAPT.

C. Recycle Heartcut Chromatographymm

LIST 01’ REFERENCES

 

 

vii

PAGE

126
128

TABLE

1.1
2.1
3.1

3.2

3.3

3.4

3.5

3.6

4.1

4.2

4.3

4.4

4.5

LIST OF TABLES

Factors that affect s, N and k'.

 

Applications of post-column reaction detectors.W"mnwmwmn _- ._

Sequential elution solvent chromatography.

 

The extent of reaction completion, X Abs-u,
for various reaction times.

 

Values Of the coefficients in the Skinhart-Hart
equation for various temperature ranges.

 

Boundary conditions for simplex optimization with
the composite modified simplex method.

 

Optimization of the phenol/sulfanilic reaction

system by the composite modified simplex method._"___W__w____~__

Comparison of the statistical figures of merit for
the determination of phenols through pre- and
post-column derivatization.

 

Determining the precision Of valve switchings
triggered by peak count, detector threshold
(10% full scale) and time.

 

Effect of tubing dimensions on the peak width at
half height, wm, of a recycled peak.

 

Comparison of the calculated and experimental

band broadening in tubing of various dimensions.WWWhmm

Value of peak width at half height, win, that
resulted from the recycling of heartcuts that
partially filled tubes of different i.d.s to varying
extents.

 

Linear regression analysis Of calibration curves
for quantitative RHC.

 

viii

PAGE

22
43

54

57

63

65

78

100

103

105

112

121

FIGURE

1.1

2.1

2.2

2.3

3.1

3.2

3.3

3.4

3.5

3.6

3.7

TABLE OF FIGURES

Routes for improving chromatographic resolution.

quu.u-a~umna

Continuous flow analysis output for three different

cases a.) ideal b.) nonsegmented and c.) air segmented.m

Flow profiles for a.) air segmented and
b.) nonsegmented flow streams.

 

Two configurations of recycle chromatography
a.) alternate pumping b.) closed loop.

 

Apparatus employed for the post-column detection

of phenols using an air-segmented continuous flow
analyzer. 1: solvent reservoir; 2: injection valve;

3: analytical column; 4: UV detector; 5: peristaltic
pump; 6: reaction coil; 7: calorimeter; 8: bubble

gate; 9: readout device. Flow stream path is

denoted by —-— and the electrical signal path by --—.___’

CFA manifold employed in this post-column reaction
work.

 

Effect of pH on PCR detector response for a
reaction time of 15 8. Conditions: 0.05 M sodium
borate buffer adjusted to appropriate pH with
NaOH or RC]; flow rates: 30 uM phenol in borax
buffer - 0.6 ml/min; DSA - 0.03 ml/min.

 

Effect of pH on PCR detector response for reaction
times of 35 s (A), 55 s (I) and 130 s (e).

 

Effect of pH on PCR detector response for reaction
times of 10 min (e) and 30 min (A).

 

Distribution diagram for diazonium, Aer‘, and
phenolste, Ar’O', ions at various pH values.

 

Influence of DSAzphenol ratios and reaction times
on the PCR detector signal. DSAzphenOl ratios:
0.7 (e); 1.2 (a); 2.8 (I); 6.7 (o); 12.4 (a); 27.8 (0)2

.....~.—...-..—..—.¢-¢——..

ix

PAGE

15

17

32

38

46

47

48

49

51

56

FIGURE

3.8

3.9

3.10

3.11

3.12

3.13

3.14

3.15

3.16

PAGE

Temperature studies. PCR flow rates: 80 11M

phenol - 0.30 ml/min; 5.9 mM DSA - 0.05 m1/min;

pH 10.1 borax buffer - 0.60 ml/min. Reaction time

Of 37 s. 58

 

Progression of a two dimensional simplex over a
response surface. Circles represent iso-response
conditions. 61

 

Scatter diagrams of absorbsnce vs. the parameters

studied. a.) pH b.) DSA concentration c.) time._______m_w_ 66
Influence of varying CHaCN mobile phase

composition on the diazo coupling reaction.

HPLC flow rate is 1.0 ml/min with a IX/min

gradient. ' 69

Influence of varying CHaOH mobile phase

composition on the diazo coupling reaction.

HPLC flow rate is 1.0 ml/min with a ZX/min

gradient. 70

Comparison of peak widths at half height, wm,
for the UV and PCR trace. 72

Flow rate compatibility between the HPLC and PCR

systems. Sample consists of phenol, 3,5-dimethylphenol

and 2,6-dichlorophenol, with elution in this order.

Mobile phase is 60/40 CHaCN/Hzo. Flow rates are

a.) 0.5 ml/min b.) 1.0 ml/min and c.) 2.0 ml/minmm-__-_m_ 74
Linear dynamic range of the phenol—sulfanilic

reaction. 76

Comparison of UV and PCR detection for the

separation Of phenols in a spiked river water

sample (Grand River, East Lansing, Michigan).

I.) resorcinol, 2.2 ppm 2.) phenol, 1.9 ppm

3.) o-cresol, 2.2 ppm 4.) 3,5-dimethylphenol,

2.4 ppm. HPLC: Spherisorb ODS column;

acetonitrile/water (50:50) mobile phase; flow rate

of 1.0 ml/min; UV detection at 254 nm; 0.04 AUFS.

PCR: manifold as shown in Figure 3.2; detection at

450 nm; 0.10 AUFS. 80

 

FIGURE PAGE

3.17 Comparison of selectivity of UV and PCR detection
for phenols in a spiked river water sample.
1.) phenol, 18.8 ppm 2.) aniline, 19.6 ppm
3.) o-cresol, 21.6 ppm 4.) 3,5-dimethylphenol, 24.4 ppm.
HPLC: Spherisorb ODS column; acetonitrile/water (50:50)
mobile phase; flow rate of 1.0 m1/min; UV detection at
254 nm; 0.04 AUFS. PCR: manifold as shown in
Figure 3.2; detection at 450 nm; 1.0 AUFS. 81

 

3.18 Reverse phase separation of RFO, fraction #7.
HPLC: Spherisorb ODS column; acetonitrile/water
(50:50) mobile phase; flow rate Of 1.0 ml/min; UV
detection at 254 nm; 0.64 AUFS. PCR: manifold as
shown in Figure 3.2; detection at
450 nm; 1.0 AUFS. 83

3.19 Chromatogram of RFO, fraction #7, spiked with
a.) 90 ppm phenol b.) 400 ppm phenol. 84

3.20 Chromatogram of RFO fraction #7 diluted 20x and
spiked with 32 ppm o—cresol. 86

4.1 HPLC system and valving configuration used to
implement recycle heartcut chromatography. 89

 

4.2 Valve switching sequence employed in RHC, .
a.) start-up b.) heartcut c.) column flush and
d.) recycle heartcut. 91

_ 4.3 Block diagram of microcomputer controlled HPLC
system used in BBC. 94

4.4 An example of s FORTH routine used to take a
heartcut based on peak count which is then
recycled through the column. 96

4.5 Examples Of FORTH commands used to initiate valve
actuation by a.) peak count b.) detector threshold
and c.) time after injection. 99

4.6 Heartcuts taken to fill a 60 cm x 0.01" i.d. loop to
various extents - 100% loop fill (---—), 75% loop fill (----),
50% loop fill (---) and 25% loop fill (—-).
450 uM naphthalene with a 100% CH30H mobile
phase and a 1.0 ml/min flow rate. Detection is
at 254 nm and 0.16 AU full scale. 107

xi

FIGURE

4.7

4.8

4.9

4.10

4.11

4.12

4.13

Heartcuts taken to fill a 60 cm x 0.02" i.d. loop to
various extents - 100% loop fill (---), 75% loop

fill (am), 50% loop fill (---) and 25% loop

fill (—-). 450 mM naphthalene with a 100% CH30H
mobile phase and s 1.0 mI/min flow rate. Detection
is at 254 nm and 0.16 AU full scale.

 

Chromatograms Of the recycled peaks that resulted
from the partial filling of the heartcut loop (60 cm
X 0.01" i.d.). 1

 

Chromatograms of the recycled peaks that resulted
from the partial filling Of the heartcut loop (60 cm
x 0.02" i.d.).

 

Isolation of 1,4-androstadiene-3,17-dione from a
spiked urine sample. Separation conditions: mobile
phase is 85/15 CH30H/H20; flow rate is 1.0 ml/min.
Heartcut is time-based and for a volume of 100 ”It“..-

Isolation of naphthalene from a mixture of _
polynuclesr aromatic hydrocarbons consisting of
naphthalene, biphenyl and phenanthrene.
Separation conditions: mobile phase 90/10
CHaOH/H20; flow rate is 1.0 m1/min.

 

Verification of peak purity a.) cycle 1 b.) cycle 2
and c.) cycle 3. Sample is 1.8 x 10" M in
naphthalene and 4.7 x 10" M in biphenyl.

 

Calibration curves for naphthalene standards of which
two heartcuts have been taken and recycled.

0.... ..............................._..........._.

xii

PAGE

108

109

110

~ 114

116

118

120

CHAPTER I

INTRODUCTION

The determination Of compounds in complex sample matrices Often
poses a formidable challenge to the chromatographer. Complications can
arise in both the separation and detection stages. They include
chemical and spectral interferences, even after an extensive sample
workup;<inadequate resolution, which requires the use of an elaborate
separation scheme; and a lengthy overall analysis time. Two solutions to
these difficulties are investigated in this work. In one approach,
methods for increasing detector selectivity and sensitivity are studied,
while in the other, a new technique for improving chromatographic
separation is investigated. ' The methods that are employed here to
achieve these objectives are post—column derivatization and a new form
Of recycle chromatography. The bases of these methods are discussed

in the following sections.

A. REACTION DERIVATIZATION IN LIQUID CHROMATOGRAPHY

A limitation encountered in modern liquid chromatography is the
lack of sensitive and specific detectors in applications involving complex
sample matrices. Conventional HPLC detectors, such as the UV-visible
and refractive index detectors, Often lack the selectivity and sensitivity
required for such samples. Fluorescence and electrochemical detectors

can be more selective and sensitive, but they are not as widely

1

2

applicable and are more susceptible to interferences. These problems
may be circumvented by reacting the analyte to form a derivative with
different and/or enhanced characteristics. Reaction derivatization may
be divided into two categories depending on the Objectives of the
derivatization. If the goal is to alter the separation characteristics,
pro-column derivatization is utilized. If the purpose is to enhance the
detection, the situation calls for post-column derivatization, which is the

methOd employed in this work.

1 . Pre-Column Derivatization

Pro-column derivatization involves a chemical reaction of the
sample components prior to separation. This method has been
extensively employed in gas chromatography (1,2) before its application
to liquid chromatography (1,3,4). The major advantage Of pro-column
derivatization over post-column derivatization is greater freedom in the
selection of reaction conditions. The primary disadvantage is the
possible formation Of multiple products, which would further complicate a
separation, especially one involving a complex sample matrix. Also, a
different separation scheme may have to be developed for these

derivatives.
2. Post-Column Derivatization
Post-column derivatization involves a chemical reaction of the

sample components after chromatographic separation and prior to

detection. Selective reagents can enhance detectability of the desired

3

compounds by forming derivatives which are more readily detected
and/or possess more distinct spectral characteristics. The presence of
different substituent groups on the reagent as well as the reactant can
impart different spectral characteristics to the derivatives, thus making
reaction detectors very flexible and readily tailored to meet the needs of
a particular situation. The reaction conditions such as pH and time, may
also be used to obtain an additional degree of selectivity. Other
advantages Of post-column reaction (PCR) detectors are: (1) in
comparison to pro-column derivatization, the reaction does not need to
go to completion, which results in reduced analysis time and decreased
band broadening; (2) different types of detectors can be used in series
with the post-column reaction detector and; (3) since the compounds are
separated in their original form, separation procedures from the
literature can be employed.

Disadvantages of the PCR detector arise mostly from the
incompatibilities between the separation and detection systems. The
solvents required for the optimum chromatographic separations are Often
not suitable as reaction media. This can result in precipitation or
decomposition Of the reagent, mixing problems arising from viscosity
differences and lower detector response due to altered reaction kinetics
and mechanisms. Other incompatibilities that may be observed are the
flow rates Of the two systems and the chemical inertness of the PCR
detector materials to the separation solvents. Thus, these potential
incompatibilities have to be investigated in the development of a PCR
detector. Despite these complications, the benefits of post—column

derivatization are numerous and it has become a widely used method.

4

3. Post-Column Reaction Detection of Phenols

Phenols are an important class of widely used chemicals consisting
of a variety Of substituted compounds, some Of which have undesirable
characteristics. They are present in many industrial effluents and
occur as a consequence of environmental transformation of both natural
and synthetic chemicals. Chlorinated phenols are considered to be water
pollutants as they impart a disagreeable taste to drinking water. The
presence of phenols in petroleum products such as paint and varnish
thinners causes a retardation in the drying rate of these products. On
the desirable side, phenols are used as antiseptics for cleaning purposes
as well as in medications. They have also been used extensively as.
antioxidants in fuels to prevent gum formation and in food products as
preservatives.

Due to the presence of phenols, wanted or unwanted, numerous
analytical methods have been reported for the qualitative and
quantitative analysis of phenolic compounds (5—10). Some of these have
involved post-column derivatization of phenols with fluorescence and
electrochemical detection (8—10). Some Of the problems distinctive to
these detection methods are quenching, as witnessed by Wolkoff and
Larose (8), and electrode fouling. UV-visible absorption is an extremely
versatile and extensively used detection mode because a large number Of
compounds absorb in this region of the spectrum. Also, the detectors
needed for routine HPLC work are generally of a simple and inexpensive
design. For these reasons, it is surprising that there have been no
reports of on-line post—column reaction systems with colorimetric

detection for phenols.

5

The primary objective of this research has been to develop a
colorimetric based reaction detector for use in conjunction with HPLC in
the determination of phenols. The characteristics of the ideal PCR
detector include on-line detection, high sensitivity and selectivity, high
precision and accuracy, total compatibility with the separation system,
no band broadening and high sample throughput. With these goals in
mind, a post-column reaction detector was developed for the colorimetric
determination of phenols. A detailed account of this work is given in

Chapter III.

B. RECYCLING METHODS IN LIQUID CHROMATOGRAPHY

In the following discussion, a review Of the various aspects Of
chromatographic resolution is provided. The parameters that govern
resolution and the changes in experimental conditions that need to be
made to achieve an improvement in resolution are described. The
situations or types of separation problems that are best handled by the A
use of recycle chromatography are then presented along with a
systematic approach for obtaining improved resolution. Finally, this
discussion provides a basis for the introduction of recycle heartcut

chromatography.

6

1. Resolution Enhancement in Liquid Chromatography

Resolution is a quantitative measure of the degree of separation
between two adjacent bands. The fundamental relationship between
resolution and parameters which in turn can be related to experimental
conditions is given by the equation:

Re 2 (1/4) («I - 1) N1” “flak-{m Equation 1.1
’ + 1
The resolution, Rs, is expressed as a function of the separation
selectivity, a, the separation efficiency as expressed by the number Of
theoretical plates, N, and the capacity factor, k’. The experimental
conditions that can affect each of these parameters are listed in
Table 1.1.

A flow diagram outlining the routes for improving resolution is
shown in Figure 1.1. This diagram, representing one school of thought,
utilizes the capacity factor as the point Of origin. I This is due to the
ease with which k’ can be altered by changing the solvent strength. As
k’ is increased, R. would also increase as shown in Equation 1.1.
However, this is accompanied by an increase in retention time, Rt, and
peak width, in. Eventually a point is reached where the benefits of
improved resolution are offset by the longer separation times and the
increased peak width. Also, from Equation 1.1, it is observed that the
fraction, k’/(k’ + 1), approaches unity asymptotically as k’ increases.
Thus, the gain in R. becomes increasingly smaller. The optimum range
of k’ values, as suggested by Snyder (12), is between 1 and 10. The

first question of the flow diagram verifies this condition. If k’ is not in

Table 1.1. Factors that affect C, N and k’.

 

 

 

Parameter Symbol Method of Variation
Separation a 1. Mobile phase
selectivity 2. Stationary phase

3. Mobile phase pH

4. Temperature

5. Special chemical effects
Separation N 1. Solvent velocity
efficiency 2. Column length

3. Column packing
Capacity k’ l. Solvent strength

factor

 

 

OPTIMUM RANGE

.t

 

 

 

 

 

CHANGE k'

 

 

YES

 

 

 

A E N ADEQUATE
CHANGE a C" "9 SEPARNHON

 

 

 

 

 

Figure 1.1. Routes for improving chromatographic resolution.

9
this range, improvements in resolution should be achieved by first
changing the solvent strength. If k’ is already in the Optimum range
and an additional gain in resolution is still needed, the next parameters
to be considered are a and N. The only clear cut situation which calls
for a change in a rather than N is when («I - 1) is equal to zero.
Under this circumstance, Rs = 0, and any changes in N would prove
ineffective. The separation selectivity is a difficult parameter to vary,
especially when dealing with complex samples, because the prediction Of
the resulting effects on the separation is an intricate procedure. This
often results in a trial and error type of approach to Obtain satisfactory
separation. The number of theoretical plates, on the other hand,
produces a much more straightforward effect on the resolution and is
Often a simpler route to take.

The number of theoretical plates can be increased by employing
columns with smaller size packings, decreasing the flow rate and
increasing the column length. The first method may not be a feasible
choice especially if the column being used already contains the smallest
size particles that current column technology offers. The second
alternative, decreasing the flow rate, provides a limited gain in the
number Of theoretical plates. This can be explained by recalling the
equation, analogous to the Van Deemter equation for 60, that was
derived by Snyder (13) for L0

H=Au°43 +§+Cu+Du
u

where H is the theoretical plate height, 11 is the flow rate and A, B, C
and D are constants for a given column. A plot Of H vs u for a liquid
chromatographic column containing small porous packings (8 pm) shows

the characteristic plot with an optimum flow rate. Flow rates higher or

10
lower than this optimum result in an increase in theoretical plate height.
The third alternative, increasing the column length, may be accomplished
by using a longer column, by coupling of columns in series or by
recycling the sample through the same column. The coupling of multiple
columns in series results in an expensive system as well as one with
high column inlet pressures. The latter is not a desirable consequence,
particularly, when working with columns containing packing materials
that cannot tolerate high pressures. An example of this is the gel-
based packing used in size exclusion chromatography columns. Under
these circumstances, recycle chromatography is the method of choice.

To summarize, recycle chromatography is a technique that should
be used when: (1) the path chosen to achieve a gain in resolution is
through increasing the number of theoretical plates; (2) if changing the
column packing and mobile phase flow rate are not possible or show no
improvement in resolution; and (3) if multiple column lengths have to be

employed and an increase in column back pressure is undesirable.

2. Recycle Heartcut Chromatography

The analysis of complex samples can be facilitated through
modifications to the separation system that would result in an improved
chromatographic separation. As discussed in the previous section, one
route to improving resolution is through increasing the number Of
theoretical plates. This, in turn, can be accomplished with recycle
chromatography. However, a problem with recycle chromatography is
the eventual overlap of early and late eluting peaks as the sample plug

broadens to occupy an entire column volume. Thus, samples containing

1 1
components representing a broad k’ range (e.g., complex samples) may
not be successfully separated because of the resulting limitation on the
number Of recycles.

The Objective of this research was to develop a separation system
that incorporated the method Of increasing resolution used by recycle
chromatography and was applicable to the determination of specific
compounds in complex matrices. Some Of the desirable characteristics
are versatility, simplicity, ease of implementation and inexpensiveness.
Such a system, recycle heartcut chromatography, has been developed as

described in detail in Chapter IV.

CHAPTER II

HISTORICAL BACKGROUND

This chapter presents the historical background for the research
described in this dissertation. The first section provides a brief
discussion Of the development and basic principles of air segmented
continuous flow analysis, the technique employed in this post-column
detection work. An extensive literature review of post—column
derivatization from its early days to the present is then presented.
This is followed by a short survey of the colorimetric reactions which
have been used in the determination Of phenols. The last section
contains a description of the research that has been performed in the

area of recycle chromatography.

A. CONTINUOUS FLOW ANALYSIS

The purpose of the following discussion is to provide a brief
review of the historical development and basic principles of CFA. Due
to the breadth of the subject matter and the existence of a number of
reviews (14-18), this discussion will focus primarily on the early years
of CFA and the basic principles and developments that have made CFA
suitable for use in post-column reaction detection.

The need for methods that would alleviate the ever increasing load
Of samples to be analyzed in the clinical laboratory prompted the

introduction of continuous flow analysis in the 1950’s by Skeggs (19,20).
12

13

In this thesis, the term continuous flow analysis (CFA) is used to
describe the technique that uses a network Of tubing to perform the
sequential analysis of samples in a flowing liquid stream segmented at
regular intervals by air. Continuous flow analysis involves the
uninterrupted treatment and monitoring Of a train Of samples contained
in a network of tubing. These samples can be diluted, mixed with
various reagents, heated, dialyzed, extracted and delivered to the
detector without operator intervention. This sequential analysis Of
samples in a flowing stream provided the desired increase in sample
throughput without the use of complex instrumentation. In contrast to
the batch analyzers used at the time, the automated continuous flow
analyzers had very few moving parts, the liquid being the moving
component of the system. As a result, the latter was mechanically
simpler, easier to construct and less expensive. This form of automated
chemical analysis proved to be a flexible way to perform operations that
would be encountered in a routine chemical analysis. The concept and
prototype instruments that were built by Skeggs were purchased by
Technicon Corporation, in 1956. The following year, the first CFA
analyzer was marketed under the trade name, Autoanalyzer. This
marked the beginning of a line of instruments that were to become
widely used in clinical and industrial laboratories.

In order to present a complete picture, it is necessary to mention
the related technique- of flow injection analysis (FIA). Until the early
1970’s, air segments were considered necessary for successful analyses
in flow streams, and the term continuous flow analysis referred
exclusively to air-segmented CFA systems.‘ In the mid-1970’s, two

independent groups of researchers, Stewart and coworkers (21), and

l4

Ruzicka and Hansen (22) demonstrated the potential of nonsegmented
continuous flow analysis. It was shown that if the dispersion Observed
in CFA could be controlled and reproduced, nonsegmented streams could
be used for analytical purposes. This technique, presently known as
flow injection analysis (FIA), has grown tremendously in popularity in
the last decade. A book on this subject has been written by Ruzicka
and Hansen (23) and numerous reviews have been published (24-28).
The relative performance of CFA and FIA have been discussed in a
number of papers (16,29-33).

A major problem that was encountered by Skeggs in his early
experiments with nonsegmented flow streams was axial dispersion. This
spreading of the sample bands along the flow axis placed a limit on the
throughput rate because the samples had to be separated by a minimum
distance. The effect of air bubbles on axial dispersion was a
serendipitous discovery that resulted in an important breakthrough by
Skeggs. The output of the two types of flow streams are compared to
that of the ideal case in Figure 2.1. The sample and wash solutions are
alternately introduced into the flow stream, which produces the
rectangular input. Under ideal circumstances, the peak profile at the
output would be identical to that at the input (Figure 2.1a) i.e., the
introduction of a rectangular sample plug should yield an identical
rectangular output. When dispersion takes place, the rise and fall of
the sample peaks are no longer instantaneous and a Gaussian peak
shape is Obtained (Figure 2.1b) as in the case of a nonsegmented flow
stream. (The sample zone spreads as it moves downstream and changes
from the original square plug to an asymmetric shape and eventually to

the Gaussian form.) Figure 2.1c shows the characteristic 'flat top’

15

 

b.)

b
P% h

INPUT OUTPUT

Figure 2.1. Continuous flow analysisoutput for three different cases
a.) ideal b.) nonsegmented and c.) air segmented.

16

output for CFA. The dispersion Observed is less than that for the
nonsegmented case, but is significant when compared to the ideal case.

The differing peak shapes Obtained for the nonsegmented and air
segmented cases can be explained by examining the sample plug profile
as it flows through a narrow open tube (Figure 2.2). When the stream
is nonsegmented, (Figure 2.2b), significant axial dispersion of the sample
is observed. This is due to the parabolic velocity profile that is
characteristic of laminar flow (34). The liquid in the center of the tube
is moving at approximately twice the mean speed Of the fluid while the
liquid layers near the wall are moving at progressively slower velocities.
The resulting dispersion is counteracted by radial diffusion. (A radial
concentration gradient is created by this process, which in turn, causes
radial diffusion. The molecules along the walls diffuse into the center
of the tube and vice versa. Thus, the sample plug is not dispersed
throughout the tube.) When the flow stream is segmented with air, as
in Figure 2.2a, there is a significant decrease in dispersion because the
air bubbles act as physical barriers that break up the laminar flow
profile. In addition, the presence of the air segments introduces a
toroidal flow within each liquid segment as illustrated in Figure 2.2a.
These circular motions result in homogeneously mixed liquid
segments (35,36).

The use of air segmentation successfully reduces the axial
dispersion that is caused by laminar flow. However, in CFA, there is a
second source of band broadening which arises from the transference of
small amounts of sample from one segment to the next via the liquid
layer that wets the tube wall. This dispersion causes, in part, the

nonideality of the CFA output shown in Figure 2.1c. Equations have

17

a.)

 

AIR AIR

Q
C.)

fl

3

fl

 

b.)

 

Figure 2.2. Flow profiles for a.) air segmented and b.) nonsegmented
flow streams.

18

been developed to describe the dispersion observed in the flow stream
and to relate them to experimental parameters (35,36). Axial dispersion
can be minimized by the use of shorter lengths of tubing, high flow

rates and high segmentation frequencies. Slow mass transfer between
the stagnant liquid film and the liquid segments, on the other hand, is

fundamental to air segmented CFA, as it is currently practiced. The
liquid film cannot be eliminated without having detrimental effects on
the hydraulic stability of the segmented stream. Nord and Karlberg (37)
have studied the effect of different materials that are used for
constructing manifold components on dispersion.

Since the conception of CFA, there have been a large number Of
developments in the instrumentation used in this field. Patton (17) has
traced the development of continuous flow analyzers, in particular the
Autoanalyzer, from the initial one channel unit to the present day
computer controlled, multichannel, multifunction system. Furman (15)
gives a detailed account of the technological advances made for each
component of a continuous flow analyzer e.g., pumps, detectors, flow
cells, tubing materials, and dialyzers. One important advance that
should be specifically mentioned is the bubble gate. The feature of air
segmented CFA that is often referred to as a disadvantage or
complication is the removal of the air segments for the purposes of
spectroscopic detection. The early methods that employed physical
removal of the air bubbles from the flow stream prior to detection (38)
resulted in significant amounts of dispersion because of mixing in the
debubbler chamber and in the tube between the point of debubbling and
detection. In addition, a fraction of each liquid segment is lost with

each bubble removal. An alternative method to physical debubbling is

19

to remove the bubbles electronically by gating the signal from the
detector. Habig et al. (39) designed a bubble gating flow cell that
employed a conductance-activated relay to detect the presence of air
bubbles in the flow cell. When the conductivity was below a preset
value, the relay Opened and disconnected the power supply from the
recorder servo motor, thus inactivating the pen. In 1974, Neeley and
coworkers (40) introduced an electronic bubble gate, which they
modified as reported in a subsequent paper (41). This bubble gate
made use of the difference in the colorimetric detector output when an
air bubble enters the flow cell e.g., erratic spikes are Observed instead
Of a constant signal. Two measurements were made within a time
interval and their values compared by a window comparator circuit. A
difference between the two signals that is within the set limits Of the
comparator, indicates that the flow cell is filled with liquid. The real
time signal was then stored and output to a chart recorder.

In 1982, Patton and coworkers (17,43) described the design and
development of a simpler bubble gate unit that monitored the
transmittance of the flow stream for the presence of an air segment.
The periodic fluctuations of the detector signal as air and liquid
segments passed through the flow cell were used to synchronize data
acquisition and storage. When an air segment traverses the flow cell, it
reflects most of the incident light and the transmittance approaches
zero. A comparator with a set threshold is used to determine whether
air is present in the flow cell. Measurements were made only when the
transmittance was above the threshold value. This gating of the signal
from the detector eliminates the problems previously observed when the

air bubbles were physically removed.

20

In post-column reaction work, it is desirable to have a system
that introduces an insignificant amount of band broadening to the
chromatographic peak. A second desirable trait is good mixing between
the reagents and the sample. These features, characteristic of CFA, are

very important, especially for post-column reaction detectors that employ

reactions with multistep addition of reagents and long delay times.
The advantages in using CFA will become evident as the work done for
this dissertation involving post-column reaction detection is discussed in

Chapter III.

B. POST-COLUMN REACTION DETECTORS

Post-column derivatization has become an extensively used
technique for improving the detection characteristics of a selected
analyte. There have been numerous reviews assessing the current
status of reaction detectors in liquid chromatography (44—54) and
publications reporting new applications. Several comprehensive books
have been written on this topic including those by Krull (55), Lawrence
and Frei (56,57) and Blau and King (1). Jupille (58), in his article on
UV-visible absorption derivatization in liquid chromatography, provides a
listing of manufacturers who provide reagents and technical support for
derivatization techniques. More recently, reaction units constructed
specifically for post-column derivatization have become commercially
available (59). The chemistry developed for Technicon Autoanalyzers
(60) has often been combined with known or newly developed LC

separation techniques to produce new post-column reaction detectors.

21

The applications of post-column derivatization have been so
extensive that it is not practical to report, in this thesis, all the work
that has been done. Instead, some representative examples are given in
Table 2.1 to demonstrate the broad range of possibilities Of this
technique. Reviews have been published discussing the role Of chemical
derivatization in the areas Of pharmaceuticals (72), pesticides
analysis (73), food additives (74) and nuclear materials (75). Post-
column derivatizations have also been used in conjunction with a number
of detection modes. Kissinger et al. (76) and Krull and coworkers (77)
have reviewed the role of PCR detectors in electrochemistry.

Lawrence (78) and Rob and Schwedt (79,80) have presented overviews of
chemical derivatization techniques coupled with fluorescence detection.
The possibility of post-column derivatization with detection by mass
spectrometry, nephelometry and flame and plasma spectrometry are
discussed by Frei (81).

Numerous developments in the area of post-column reaction
detectors have resulted from the need for selective and sensitive
detectors and the efforts to overcome the disadvantages of this type of
detector. As previously mentioned, these shortcomings include band
broadening due to inappropriate choice of reactor and poor mixing
between sample and reagents, and dilution of the sample. One method of
reducing band broadening is through the use of an appropriate reactor
design based on the kinetics and complexity Of the reaction. The three
commonly used reactors are the Open tubular reactor, packed bed
reactor and segmented stream reactor. The theoretical aspects of band
broadening and the performance of each type of reactor have been

discussed in detail by a number Of authors (52-54,82-84)).

22

Table 2.1. Applications of post-column reaction detectors.

 

 

 

 

 

Analyte Reagent Reactor Detection Ref
Type Mode
Organic ‘ Ce“ PBR F 61,62
acids
Penicillin Imidazole SSR UV-Vis 63
Carbamates hydrolysis, OTR F 64,65
o—phthaldehyde
Phenolic Bra OTR EC 66
ethers
Sugars Copper OTR EC 67
phenanthroline
Phenolic B-glucoronidase SPR EC 68
glycosides
Cholesterol Cholesterol SPR UV 69
oxidase
Cocaine hv OTR EC 77
Enzymes various PBR F 71
OTR - open tubular reactor F - fluorescence
PBR - packed bed reactor UV - ultraviolet
SSR - segmented stream reactor Vis - visible
SPR - solid phase reactor EC - electrochemical

23

The open tubular reactor (OTR) is the simplest reactor and is
usually made from stainless steel or Teflon tubing. These reactors can
be straight, coiled or knitted. (A knitted OTR consists of a piece of
PTFE tubing that has been knitted into a series of figure sights. The
two loops of the 8 occupy planes that are at a 120° angle from each
other, thus yielding a three dimensional structure of half loops. Due to
the increased radial mass transfer, band broadening is decreased.) It
has been reported that the latter two configurations result in reduced
band broadening due to a secondary flow effect (85-87). Open tubular
reactor designs are usually recommended for fast (< 30 s) reactions.
However, Engelhardt (86) has shown that it is possible to use a knitted
OTR for a reaction lasting several minutes.

Packed bed reactors (PBR) have been recommended for reactions
with intermediate kinetics (0.5 - 4 min) (88-92). They are usually made
from glass or stainless steel columns (1 mm i.d.) that are packed with
small nonporous glass beads (40 - 150 pm). These reactors resemble
liquid chromatography columns with no retentive properties. The
sources Of band broadening in these reactors are similar to those
encountered in LC. However, because of the absence Of a retention
mechanism, the equations describing dispersion in PBR are much
simpler (88,93).

Segmented stream reactors (SSR) have generally been
recommended for use'with slow reactions (up to 20 min) or in place of
packed bed reactors. However, Scholten et a1. (83) have shown, in a
comparison of the three post-column reactor designs, that not only can
segmented stream reactors be used for fast reactions, but they also

yielded the least amount of band broadening. Segmented stream

24

reactors are based on segmentation of the flow stream with either air or
an immiscible solvent. As previously discussed, the segmentation
reduces dispersion by inhibiting laminar flow and at the same time
enhances mixing within each sample containing segment. Dispersion in
an air-segmented reactor can be approximated by the equation derived
by Frei and Scholten (54). A more detailed equation, derived by
Snyder (35,36), takes into account various experimental parameters
specific to the reaction system, such as viscosity, liquid surface tension
and sample diffusion coefficient. The use of immiscible solvents for
segmentation has opened up a new area in PCR detection. Aside from
reducing sample dispersion, the immiscible solvent segments can be used
as reagent carriers, reaction media and as an extraction media (94-99).
Apffel et al. (100) have used a solvent segmentation system for post-
column cleanup in the analysis of urine for antineoplastic agents. The
column effluent was segmented with dichloroethane which served as an
extraction medium for the analyte. The two phases were separated on-
line, and the dichloroethane layer containing analyte was directed to the
detector. This system, which is not a reaction detector in the strictest
sense, allows the urine sample to be used without any other sample
pretreatment. Other areas to which these two phase reactors have been
applied are LC/MS (101) and LC/GC (102). The objective in both cases
was to make the analyte more compatible with the detection systems.

In using immiscible solvents for segmentation, the problem of
compressibility of gases is not encountered. Compressibility leads to a
noisy signal in air segmented systems because the pump pulsations are
transmitted down the length of the tube. Another problem Of air

segmentation, band broadening due to wetting of the tube walls, can be

25

reduced when using solvent segmentation. The key is to employ tubing
materials that are not wetted by the phase that contains the compounds
of interest. An alternative to this is to use extraction solvents that do
not wet the tubing material that is being used. The effectiveness Of

solvent segmentation in reducing band broadening was demonstrated by

Werkhoven (97) who utilized this reactor type in a 24 min derivatization
reaction. Lawrence et al. (96) have published some interesting findings
concerning the influence of glass versus Teflon on band broadening as
a function of flow rate for air segmented and solvent segmented
reactors.

A disadvantage Of post-column reaction detectors, especially when
using OTR and PBR is inefficient mixing. Mixing, which has to take
place in a short period of time and without significant band broadening,
is not readily achieved when the column effluent and reagents have
different flow rates and physical properties (e.g., viscosity and density).
Mixing units Of different geometries have been designed by several
workers (47,91,103,104). Frei et al. (105) have compared the performance
of some mixing units in a derivatization system using OTR and a 10 a
residence time. Use of the traditional 90° tee mixer resulted in layering
effects and thus band broadening. When a mixing unit that combined
the column effluent and reagents at 60° angles was used, band
broadening was significantly reduced. This was due to the turbulence
produced in the mixing area which resulted in enhanced mixing. Jonker
and coworkers (91,92) have proposed the use Of narrow diameter tubing
filled with large glass beads to enhance mixing. This is a practice

commonly employed in flow injection analysis (23).

26

A disadvantage of post-column reaction detectors is that the
detection limits are increased due to dilution by reagent addition.
Recently, reaction units have been developed that minimize or eliminate
the need for reagent addition. They include solid phase reactors (106-
110), electrochemical reactors (10,66,111), photochemical reactors
(70,112,113) and thermosensitized reactors (114). Solid phase reactors
are essentially the same as packed bed reactors. The difference is that
instead of using inert glass beads, a reagent or catalyst is immobilized
on the glass surface. One such system is the immobilized urease reactor
used to convert urea to ammonia which then reacts with o-phthaldehyde
to produce a fluorescent derivative (109). Electrochemical reactors have
been used to generate reagents needed for derivatization. King and
Kissinger (10) have described the use of an amperometric generator
electrode to generate bromine from the potassium bromide that had been
added to the mobile phase. The bromine undergoes reaction with
phenols in the space between the upstream generator electrode and
downstream detector electrode. The analyte concentration is
proportional to the difference in reagent concentration between these
two points. A thermosensitized reaction detection was reported by
Le Page and Rocha ( 114). Their work involved the ninhydrin reaction
commonly used for the determination of amino acids and other primary
amines. Since this reaction takes place extremely slowly at ambient
temperatures, the ninhydrin reagent may be added to the mobile phase.
When the column effluent, containing a homogeneous mixture of sample
and reagent, is subjected to an increase in temperature, the reaction

then proceeds at a rapid rate.

27

From this review, it is evident that post-column reaction detectors
have become widely used. This is reflected by both the number of
applications and the extent Of coupling of post-column derivatization to
different detection modes. With the development of new reaction units
(e.g., solid phase, electrochemical and two phase reactors), comes endless

possibilities for expansion into different application areas.
C. COLORIMETRIC REACTIONS FOR PHENOLS

Many colorimetric reactions for the qualitative and quantitative
determination of phenols have been developed. Veibel (115), Siggia and
Hanna (5), and Pesez (116) have summarized some of these methods.
Cheronis (7) and Knapp (117) have listed some important derivative
classes for phenols along with recommendations concerning the
derivatives that would be most suitable for different phenol types.
Reagents used in conjunction with various chromatographic methods for
the detection of phenols and organic acids were summarized by
Hanai (118). The reactions that have been used for the quantitative
determination of phenols by colorimetric means can be divided into two
categories based on whether a coupling reaction takes place or not.
Coupling reactions produce highly conjugated derivatives with
correspondingly high molar absorptivities. The derivatives most
commonly formed are azo, indophenol and antipyrine analogues.

Koppe et al. (119) have published a comparison of the reactivity Of four
different reagents with 126 phenols.

The classical method for the determination of phenols is the

coupling of phenol with a diazonium salt to form a colored azo dye. An

28

extensive amount of work has been done with this type Of reaction
because of the numerous possible diazonium-phenol combinations. Some
Of the diazonium reagents that have been used are p-aminobenzoic acid
analogues (20), p-nitroaniline (119,122-125), p-arsanilic acid (121) and
sulfanilic acid (119,122-125). Also, the importance of azo compounds in
the dye industry has resulted in thorough investigations Of the
formation mechanism Of these compounds (6,127-130). Diazo coupling
reactions have found wide analytical use because Of: (1) good
reproducibility, (2) high selectivity (due to the relatively weak
electrophilic nature Of the diazonium ion), (3) insensitivity to small
variations in reaction conditions, (4) predominantly monosubstituted
reaction products and (5) the aqueous reaction medium. Since
aromatic diazonium reagents are not highly reactive electrophiles,
coupling takes place only with activated aromatic compounds (primarily
phenols and anilines). The reactivity of the diazonium group can be
altered by ring substitutions (131). Electron withdrawing substituents
enhance the electrophilic nature of the diazonium group, whereas
electron donating substituents decrease the reactivity of the this group.
Indophenols can be formed by the oxidative coupling Of phenols
with various reagents such as diamines (132) and hydrazones (119,133)
or by reaction with quinoneimines (134-139). Some reagents that have
been used for the oxidative coupling of phenols are N,N-dimethyl-p-
phenylenediamine (131.) and 3-methyl-2-benzothiazolinone
hydrazone (119,132). The most well known quinoneimines are 2,6-
dichloro- and 2,6-dibromoquinone chlorimide, also known as Gibb’s
reagent (133,134). For many years, the latter was considered the

standard reaction for the quantitative determination of phenols.

29

Although it is very sensitive, Gibb’s reagent does not yield reproducible
results. Svobodova et al. (138,139) have studied the phenolic reaction
with this reagent extensively and have reported that a reagent to
sample ratio of 30-50 to 1 is essential for reproducibility. Other
reagents similar to Gibb’s reagent, such as benzenesulfonyl
quinoneimine, have also been used (140).

The third class of derivatives, antipyrines, are formed by the
coupling reaction of phenolic compounds with 4-aminoantipyrine (also
known as Emerson’s reagent) in the presence an oxidizing
agent (141-143). This reaction is currently the most commonly used and
accepted colorimetric reaction for quantitative phenol
determinations (60,145,146). Potassium ferricyanide is the oxidizing
agent most commonly used in this reaction. However, as reported by
Blo et al. (147) in their work on pre-column derivatization Of phenols,
this oxidizing agent causes the production of a large number of reaction
products that interfere in the analysis. These authors have
investigated the use of silver chloride as the oxidizing agent and
compared its performance to potassium ferricyanide. Their findings
show that, in comparison to the system in which potassium ferricyanide
was employed, the reaction time was greatly increased (40 min) and
there was a reduction in sensitivity. The advantages of the
4-aminoantipyrine reaction (with potassium ferricyanide as the oxidant)
are sensitivity, speed, and approximately the same has for all
derivatives. The latter may be a disadvantage if information is desired
for specific phenols, since the different phenol derivatives cannot be

spectrally resolved.

30

Several reactions are used for phenol determinations that do not
require coupling with the reagent to form derivatives with extended
conjugation. These include nitrosation (148) and halogenation (149,150)
of the phenols. In the former case, phenols are treated with a sodium
nitrite solution in sulfuric acid to form a nitrosophenol which rearranges
in the presence of excess alcoholic ammonium hydroxide to form a highly
colored quinoid radical. This reaction has been used in the analysis of
gasoline and allied products for phenols (151). Quantitative and
qualitative analyses for phenols have also been carried out by
halogenation with iodine or bromine (149,150). Since substitution may
take place at more than one position on the ring, experimental conditions

have to be carefully controlled when doing quantitative work.

D. RECYCLE CHROMATOGRAPHY

The idea of recycling was first suggested for use in gas
chromatography (152). In 1962, this concept was applied to liquid
chromatography for the first time by Porath and Bennich (153), who
showed that it was possible to achieve very large column plate numbers
without the use of additional column packing. Recycling has been a
successful alternative in complex separations when a change Of columns
and/or mobile phases have failed to provide the necessary resolution
and in cases where the coupling of columns in series is not a feasible
Option. Since its introduction, recycle chromatography has been widely
.used in size-exclusion chromatography (SEC) (154-165). The use of long
columns and the coupling of multiple columns to obtain an increase in

the number of theoretical plates is not a feasible option in SEC due to

31

the inability of the gel based column packing to withstand the resulting
increase in back pressure. ‘

Recycle chromatography has been carried out using two different
approaches - the closed loop (CL) principle (153) and the alternate
pumping (AP) principle (161), both of which are shown in Figure 2.3.
Most Of the research done in recycling has involved the use of one Of
these approaches, sometimes with slight modifications. The unique
characteristic of the CL configuration, illustrated in Figure 2.3a, is the
connection of the pump, injector, column and detector in a continuous
loop. The injected sample is separated on the column and detected as it
elutes from the column. The valve is used to divert the effluent to a
fraction collector or back to the pump for recycling. A major advantage
of the closed loop configuration when it is used to recycle the entire
sample is minimum dilution since the mobile phase that the sample is
eluting in is being fed back into the column. Thus, there is no
additional dilution with each recycle. Its major drawback is the band
broadening that occurs when the partially separated sample flows
through the pump chambers. This places a limitation on the type of
pump that can be employed with CL recycling. In order for negligible
band broadening to occur in the pump, the internal volume of the pump
that should be used has been calculated to be about 5 pl (166). In most
cases, this is not a realistic requirement as the routine LC pumps have
a much higher internal volume. Therefore, the CL configuration has
been used mainly for slow analyses performed with wide columns .and
has not been used in HPLC. A theoretical and experimental study Of the
closed loop principle was published by Martin et al. (167) in 1976. In

this work, the theoretical predictions and the experimental observations

32

.98— posO—o A... use 8595:—
35:3? .6 haagusaosao Echoes «O 3033:3300 0.5. .m.« assur—

 

 

 

._o?.ooe¢ L

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Eozom
afiaa
zeta»:—
-_ $3.0m
.33....
33: 33> aEam
— 5:25 a 55:8
Loaoouoo
5:23 .337:
Loauoauo couscous
a
To

 

 

 

 

i? A

 

To

33

of a number of factors were compared (e.g., the relationship between
resolution and the number of recycles and the effect of sample
concentration on resolution). The advantages of recycling over one
cycle systems that yielded equal resolution were noted.

The alternate pumping principle was initially introduced for use in
SEC by Biesenberger et al. (161) and later adapted to high performance
liquid chromatography (HPLC) by Henry et al. (168). The apparatus
used for alternate pumping recycle chromatography is shown in Figure
2.3b. Sample is introduced via the injection port and initially separated
on column 1. The six port valve is used to divert the effluent either to
waste or on to column 2. In this way, undesirable components can be
removed from the system. As the last peaks elute from column .1, the
valve switches so that the pump is now directly connected with column
2. The effluent from column 2 is then diverted onto the head of column
1 after passage through cell 2. This sequence of steps is continued
until the sample peaks have broadened to occupy the entire length of
the column or an adequate resolution has been achieved. The
advantages of AP over the closed loop approach are: (1) decreased band
broadening, (2) all types of pumps can be used with AP recycling since
the sample does not recycle through the pump and (3) a reference cell
through which the mobile phase flows is maintained throughout the
separation. The disadvantages, on the other hand, are the necessity for
two identical columns -and two high pressure flow cells. The number Of
times that a sample can be recycled is limited by the eventual
broadening of the sample peaks to occupy an entire column volume.
When this happens, the early eluting components start to overlap with

the late eluting components. Columns of the same type, but with

34

significantly different separation characteristics can result in a further
limitation on the number of recycles. In addition, the valve switching
sequence can become very complicated, especially if time is the
parameter used to initiate the valve switchings.

A third configuration for recycle chromatography was introduced
by Minarik et al. (166) in 1981. In this approach, the eluate from the
column outlet is repeatedly transported into the column inlet with a
circulation valve that is turned by a stepper motor. This approach
attempts to combine the advantages of the AP and CL recycling systems.
The dead volume in the valve is Of the same order as that encountered
in the capillary connections used in AP. The recycling path, however,
is similar to that used in the CL pumping principle. This system has
been successfully tested in the separation Of a mixture of isophthch
acid and its esters by gel permeation chromatography.

A technique that should also be mentioned in this section is
boxcar recycle chromatography (BRC). This technique which was
introduced by Snyder in 1981 (169) combines boxcar chromatography
with the alternate pumping method of recycling. Boxcar chromatography
is a form of column switching that enables increased sample throughput
for routine applications in CC and LC. This is accomplished by
continuously injecting samples which are serially separated on the first
column. The partially separated components Of interest for each sample
injection are diverted onto a second column via a switching valve. A
second pump is used to effect separation on the second column. In this
way, the second column is filled with a train of samples which is further
treated. ' The instrumentation requirement for BRC are two pumps, two

switching valves and three columns. The first and second columns are

35

used for performing boxcar chromatography. When the second column is
loaded with a train of sample fractions, the third column is switched in
and these last two columns are used for recycling.

Recycling has been used in gel permeation chromatography for the

separation of Oligomers and polymer additives (154), diastereomers (156)

and phenols (155). Henry et al. (168) applied the AP method to the
separation Of endO/exo isomers Of methyl-1,1-(spirocyclopropyl)-indene
by high speed liquid chromatography. Recycle chromatography has also
been used in microbore liquid chromatography as reported by Kucera

and Manius (170).

CHAPTER III
DESIGN AND CHARACTERIZATION OF A POST-COLUMN
REACTION DETECTOR FOR PHEOLS BY

AIR-SEGMENTED CONTINUOUS FLOW ANALYSIS
A. OVERVIEW

The development of an on-line post-column reaction detector for
phenols is described from the initial reaction characterization
experiments to the final evaluation and application stage. The technique
Of air-segmented continuous flow analysis (CFA) was chosen due to its
unique characteristics and suitability as a post-column reaction detector
system. The presence of air-segments has two major consequences -
reduced longitudinal dispersion due to the break down Of the parabolic
flow profile and enhanced mixing within each liquid segment due to the
toroidal flow in each Of these segments.

The reaction employed in this work involves coupling between a
phenol (Ar’OH) and a diazotized aromatic amine (AerC 1") to form an
azo dye. The general diazotization and coupling reactions are

summarized in Equations 3.1 and 3.2, respectively.

. ArNHz + NaNOz + 21101 PI! ArN2*Cl‘ + 2H20 + NaCl (3.1)

Aer’Cl" + Ar’OH ?* ArN=NAr’0H + 1101 (3.2)

36

37

The aromatic amine (ArNHx) used in this case is sulfanilic acid. The
performance of diazotized sulfanilic acid (DSA) and several other
coupling reagents have been evaluated for 126 different phenolic
compounds by Koppe et al. (119). Studies by Whitlock et al. (125) and
Baiocchi et al. (126) to characterize the reaction and to determine the
optimum reaction conditions for DSA have not produced consistent
results. It was therefore necessary to undertake an extensive study of
the effects Of experimental conditions on the reaction detector response.
Simplex optimization was employed to confirm the Optimum reaction
conditions. The potential of the post-column reaction detector for use
with reverse phase liquid chromatography was investigated by studying
the effects of organic solvents on the diazo coupling reaction. Other
aspects Of interest were the influence of flow rates on resolution and
the degree of band broadening introduced by the detector system. The
analytical figures of merit were determined and compared to that Of
other systems. TO evaluate the usefulness of the system, the PCR
detector was employed for the determination of phenols in fuel and

spiked river water samples.
B. EXPERIMENTAL
1. Apparatus
A schematic of the instrumentation employed in this work is shown
in Figure 3.1. The main components of the HPLC unit were a Spectra-

Physics SP8700 solvent delivery system (Santa Clara, CA), a Rheodyne

7120 injection valve (Cotati, CA) with a 20 311 sample loop and a

38

.1... as seen Beams 32.3030 05 and ..II an 630.3“. mm fies Scenes
33% 6325 .3025.» 6 “Sum @3933 "m “uses—3.330 "a “:00 someones "m
5539 3.33.3.3.” 6 830306 >3 3 E3300 Romain—use an "039, :ofioogm
"N vaccine: .2533 3 3.3.253 no... ceases—moons? so News: 30:23
no HOSOanp 5:381:09 one so» Ochoa—so 25939:. Ad 953...—

n

IJ

 

nanomeom

 

3nd;

 

5

 

 

 

_
L

 

 

 

 

 

 

 

 

 

39

Chromatronix 220 UV detector (Berkeley, CA). Separations were carried
out an a 5 pm Spherisorb ODS column (25 cm x 4.5 mm) purchased from
Alltech Associates (Deerfield, IL). The PCR system is a miniaturized
continuous flow (mCF’) analyzer designed in our laboratory (17,33,43).
The peristaltic pump (Model IP-IZ, Brinkmann Instruments, Westbury,
NY), used to proportionate reagents to the column effluent, was modified
from the original 8 roller assembly to one with 16 rollers for a smoother
flow. A‘ dual channel calorimeter (17) with narrow bandpass filters for
wavelength discrimination was used to monitor the absorbsnce of the
flow stream at 450 nm. Light was transmitted from the source to the
calorimeter through a bifurcated fiber optic. The signal from the
calorimeter was directed into a "bubble gate" circuit where the effect of
the air segments was electronically removed (43) and the resulting
signal observed on a digital recorder.

As shown in Figure 3.1, the sample is injected into the HPLC
system where it undergoes separation on an analytical column followed
by UV absorption detection. The column effluent is. then directed to the
PCR detector. The interface between the HPLC and PCR systems is a
short piece of 0.01" i.d. stainless steel HPLC tubing, which is connected
at one end to the output of the UV detector and at the other to a
stainless steel reagent addition block on the PCR manifold. The flow
stream is immediately air-segmented using the dual pump tube method
described by Habig, et al. (39). Reagents are added to the column
effluent via a 90° glass tee. Mixing and time delays are accomplished
by means of glass coils of various lengths. In the final stage, the

analytical stream containing derivatized products passes into a low

40

volume bubble-through flow cell (2 ul) mounted on a calorimeter and
then out to waste.

Temperature studies were performed with a 30 k9. thermistor
(Unicurvefl', Fenwal Electronics, Inc., Framingham, Mass) and a
thermastatted water bath. The simplex optimization program (OPTIMA)
obtained from Dr. Adrian P. Wade (British Petroleum Research Centre,
Middlesex, Great Britain) was run with an IBM compatible microcomputer

(PC Designs, Tulsa, Oklahoma).

2. Preparation of Solutions for the Determination of Phenols

a. Reagents

Commercially available chemicals were used without further
purification. Diazotized sulfanilic acid was prepared using the
procedure described by Whitlock et al. (125) with some modifications. A
5 mmol amount of the sodium salt of sulfanilic acid was dissolved in
50 ml of distilled water. A 10 ml aliquot of a 50 mM sodium nitrite
solution was transferred to this solution followed by cooling to 0°C.
After the addition of 2 ml of 2 M HCl, the final volume was adjusted to
100 ml with chilled, distilled water. The diazotized sulfanilic reagent
was tested for the presence of excess nitrous acid using starch-
potassium iodide paper. If the test was positive, either sulfamic acid or
urea was added to decompose the nitrous acid. The reagent was
transferred to an amber bottle and stored at 5°C.

The pH of the 0.05 M sodium borate buffer was adjusted by the

addition of 1101 or NaOH. A surfactant, Brij' 35 (Fisher Scientific

41

Company, Fair Lawn, NJ), was added to all solutions in quantities of

0.5 ml per liter of solution.

b. Phenol Standards

Standard solutions of phenol were made from a 1 mg/ml phenol
stock solution (Sigma Chemical Company, St. Louis, MO). Solutions of
higher concentrations were made from phenol crystals. All phenolic
compounds used were purchased from Aldrich Chemical Company
(Milwaukee, WI), Eastman Organic Chemicals (Rochester,NY), and K & K
Laboratories (Cleveland, OH). Brij' (0.5 ml/l) was added to all the

standards.

c. HPLC Solvents
Spectral grade HPLC solvents (Burdick and Jackson, Muskegon,
MI) were filtered through a 0.45 pm filter and degassed with helium

prior to use.

3. Sample Preparation

8. River Water Samples

Water samples collected from the Grand River (Lansing, M1), were
filtered through a 0.45 pm filter prior to injection into the HPLC system.
The samples were stored at 4°C with no added preservatives. The
recommended technique (171) for the preservation of phenols in waste
waters is the addition of copper sulfate and phosphoric acid to samples,
followed by storage at 4°C. These conditions inhibit the breakdown of

phenols due to microbiological activity. As the effect of this

42

preservation method on organic moieties has not been studied and the
water samples were used within 48 hours of collection, no preservatives
were added. Ettinger et al. (172) have reported a 15% loss of phenolic
content in an unpreserved river water sample that had been stored at

4°C for four days.

b. Residual Fuel Oil Fractions

The residual fuel oil (RFO) samples were obtained from the
National Aeronautics and Space Administration (NASA). Sample
preparation consisted of an initial fractionation step by means of
sequential elutian solvent chromatography (173,174). This method
involves the use of nine solvent mixtures of varying polarities to elute
compounds containing different functional groups. The solvent mixtures
and the functionalities to be found within each fraction are summarized
in Table 3.1. This step was carried out using a flash chromatography
apparatus (J. T. Baker 00., Phillipsburg, NJ) which consisted of a 2 cm
column packed with 17 cm of 40 um 'Flash’ silica gel. A 20 g quantity
of silica gel on which 0.5 g of sample had been deposited was placed
over the packed bed. A 300 ml amount of each solvent mixture was
sequentially added to and eluted from the column. A back pressure of
nitrogen (2-5 psi) was used as the driving force to speed up the
eluting process. Each fraction was collected and the solvent evaporated
to yield a concentrated sample. An RFO fraction containing phenols,

fraction #7, was employed in this work.

43

Table 3.1. Sequential Elution Solvent Chromatography.

 

 

Fraction no.

Solvent Mixture

Compound Types

 

Hexane

Roxane/15X Benzene
Chloraform
Chloraform/10X Ether

Ether/3% Ethanol

Methanol

Chloraform/3%Ethanol

Tetrahydrafuran/
3% Ethanol

Pyridine/3% Ethanol

Saturates,
Alkenes

Aromatics

Polar aromatics
Monophenals
Basic nitrogen
heterocycles,

(diphenols)

Highly functional
molecules

Polyphenols

Phenolics,
O - compounds

Heterocycles

 

 

44

C. RESULTS AND DISCUSSION

1. Introduction

The results of the work done in the various developmental stages
of a post-column reaction detector for phenols are presented and
discussed in this section. The effect of changes in variables such as
p11, reaction time, reagent-to-substrate ratio and temperature on the
diazo coupling reaction was investigated. Studies were carried out via
both univariate and simplex optimization. These findings have been
compared with those of other researchers. The extent of compatibility
between the separation (HPLC) and PCR detection (air-segmented CFA)
systems was studied. The aspects of interest were the compatibility
between the media and the flow rates of the separation and detection
systems as well as the degree of band broadening introduced by the
PCR detector. The statistical figures of merit are reported along with a
comparison of this method to other methods for determining phenols.
This section concludes with some applications to actual samples; these
help to demonstrate the advantages of post-column reaction detection of

phenols separated by HPLC.
2. Reaction Characterization
The reaction characterization and optimization studies were carried

out on the mCF reaction system without coupling to the HPLC unit. The

manifold employed throughout these studies, unless otherwise stated, is

45

shown in Figure 3.2. Phenol was used in all of the characterization

experiments.

8. Optimization via the Univariate Approach

The following optimization with the univariate method was
performed by changing one reaction condition at a time (keeping all
others constant) and recording the resulting absorbsnce value.

i. pH Effects. Borate buffers of different pH values were used to
attain the necessary solution pH. Pump speeds and coil lengths were
varied as needed to produce the desired reaction times. The effect ‘of
pH on the detector response at varying reaction times is shown in
Figures 3.3 - 3.5. For all the curves, an optimum can be seen at '
approximately pH 10. The general peak shapes can be explained by

considering the following equilibria:

Ar’OIl i‘é Ar’O" + 11* pk; = 9.89 (3.3)

Aer“ F3 AerOH “6* AerO‘ + H" pK1 + pKz = 20.96 (3.4)

The reaction is an electrophilic aromatic substitution with the reactive
species being the diazonium, Aer, and the phenolate, Ar ’0', ions.
These two species predominate at conflicting pH conditions. Thus, the
response increases to an optimum and then decreases as the ratios of
the reactive species change with pH. The plateau regions which extend
over a range of pH values in Figure 3.4 can be varied by changing the
amount of DSA added. Higher precision should be obtained in this

region as small fluctuations in pH would cause no changes in absorbsnce

46

.5753 someones can—00:38 25 5 “530380 Bahama <m0 .~.m 0.53m

 

 

 

 

 

 

 

 

 

 

 

 

 

Ec omv o_.:z<.._._:m
- no. N45
0 no
u 3 oo. x<mom
no.
OF. m_<
A£E>EV
04 Each. 30.. 302

35802

47

 

0.7 -
0.8 '-

005 -

0.4L //
0.3 :- / °

0.2 "
0.1 '-

Abaarbonce

 

 

 

0.0 *-
4 1 ~ . L . L . l . l L
7.0 8.0 9.0 10.0 11.0 12.0
pH

Figure 3.3. Effect of pH on PCR detector response for a reaction time
of 15 s. Conditions: 0.05 M sodium borate buffer adjusted to
appropriate pH with NaOH or RC]; flowrates: 30 11M phenol in borax
buffer - 0.6 ml/min; DSA - 0.03 nil/min.

48

 

0.7 -

Absorbonce

 

 

 

 

Figure 3.4. Effect of pH on PCR detector response for reaction times of
35 s (A), 55 s (I) and 130 s (O).

49

 

A '\s
0.0
, r //’ :\F
0.5 - ‘
0.4 P

0.3 '-

 

Absarbonce

0.2 b

0.1 - \‘
0.0 '- l n L L I 1 I L l L

S 7.0 ‘ 0.0 9.0 10.0 11.0 12.0
0“.

 

 

 

Figure 3.5. Effect of pH on PCR detector response for reaction times of
10 min (a) and 30 min (A).

50

readings. The various shapes of the curves arise from the different
reaction times used. This aspect is discussed in a later section.

A distribution diagram for the diazonium and phenolate ions as a
function of pH is shown in Figure 3.6. The standard Henderson-

Hasselbach equation,

pH : pK + log [5L
[HA]

was used for computation of the mole fraction of phenolate. A derived
expression was used for calculations involving the diazonium ion. The
necessity of a derived equation arises from the differences between the
hydrolysis characteristics of the diazonium ion and those of the 'classic’
dibasic acids, in that, for the former case, K2 ) K1 . Neutralization
curves of aqueous solutions of diazonium salts do not yield two -
inflection points with an intermediate pH region as is normally observed
for dibasic acids (175). Instead, one step that extended over two
equivalents of base per diazonium ion was observed. This anomalous
behaviour was first reported by Schwarzenbach (176) who proposed
possible explanations for this phenomenon. Since K2 ) K1 , individual
constants were not available and the overall equilibrium constant was
used in the calculation of the mole fraction of diazonium ions at

different pH values. Shown below is the derivation of an equation

51

 

 

 

LO —

0.8 -

O.6 *-

Mole Fraction

0.2 -

 

 

 

 

Figure 3.6. Distribution diagram for diazonium, Aer', and phenolate,
Ar’O", ions at various pH values.

52

which makes possible the use of the overall equilibrium constant,

pKi + p82 , in this calculation. Referring to Equation 3.4,

P.
N
p

II

[Aer OH] [11*]

 

 

 

[ArNa’]
K: = [AerO'][H°]
[MNzOH]
KiKz = [3*]2[ArN20']
[Aer’]
pKi 4‘ pKz = 21:11 - log LArN20;]_

cat-mm-

[ArNi’

pH : pK1 + pKz + 1 log _[_A_EN30'I
2 2 [ArNs’]

Calculations employed the equilibrium constants given alongside
Equations 3.3 and 3.4 (177,178). The resulting distribution diagram,
Figure 3.6, shows that the amounts of diazonium and phenolate ions are
maximized relative to each other at pH 10.2. (The intersection of the
lines mark the point of optimum pH.) It should be noted that the
presence of an optimum pH does not dictate that the reaction will
proceed as other aspects have to be taken into account. These include
the types and positions of substituents on the aromatic ring. In
general, electron withdrawing substituents do not promote the diazo
coupling reaction. Substitutions occur at the para and artho positions,
with preference for the former. Thus, phenols with substituents at both
the ortho and para positions do not undergo reaction with diazotized
sulfanilic acid. Large and bulky substituents may also inhibit the

reaction due to steric hindrance.

53

ii. Influence of Reagent-to-Substrate Ratio. The influence of the
reagent-to-substrate concentration ratio was investigated by the
addition of diazotized sulfanilic acid solutions of different concentrations
to the buffered flow stream containing phenol. The absorbance values
obtained at different reagent-ta-substrate ratios and reaction times are
shown in Figure 3.7. Pump speeds and coil lengths were varied as
needed to produce the desired reaction times. Theoretical calculations
of the mole fraction of diazonium and phenolate ions, Figure 3.6, show
that the two species are present in equal amounts (0.7 mole fraction) at
a pH of 10.2. Thus, working at this pH, a DSA:phenol ratio greater than
one should produce no increase in the maximum absorbance. However, a
definite kinetic effect is observed as demonstrated in Figure 3.7. There
was a decrease in the time required for the reaction to go to completion
as the amount of reagent added was increased. Results indicate that
the mole ratio of DSA to phenol should be greater than or equal to
three for the reaction detector to attain a maximum response in less
than 60 s.

iii. Effect of Reaction Time. The dependence of detector response
on reaction time when different DSA:phenol ratios are used is
demonstrated by Figures 3.3 - 3.5 and 3.7. From the data presented in
Figures 3.3 - 3.5, the degree to which the reaction has proceeded to
completion, given by X Abs-u, is calculated for the different reaction
times. These values are summarized in Table 3.2. The reaction was 97%
complete when a delay time of 35 s was employed and 100% complete for
a 55‘s and 130 s delay. When a reaction time of 10 min and greater was
used, secondary reactions (179) may be taking place as evidenced by

the increase in maximum absorbsnce and the appearance of a second

54

Table 3.2. The extent of reaction completion, 2 Abs... for

various reaction times.

 

 

 

Time (8) Abs 3 Abs-a x ‘
15 0.48 83
35 0.56 97
55 0.58 100
130 0.58 100
600 0.63 109
1800 0.70 121

 

 

' Assuming that the reaction is complete at 130 s.

55

peak. These observations are supported by the results shown in Figure
3.7. The time needed for the reaction to go to completion is less than
45 s when DSAzphenol ratios greater than 2.8 are used. The results
shown in Figures 3.3 - 3.5 and Table 3.1 were obtained with a reagent-
to-substrate ratio of 3.3. Based on these results and considerations of
throughput and sensitivity, the optimum reaction time selected was in
the range of 35 s to 55 s.

iv. Effect of Temperature. Temperature studies were carried out
with glass reaction coils built into plexiglass water jackets. A
thermistor was used to measure the temperature of the thermostatted
flow stream as it exited the flow cell. The thermistor was calibrated via
the Skinhart—Hart equation:

T

ml 2 A + B(ln R) + C(ln R)’

where T is temperature and R is the resistance. The coefficients A, B
and C were experimentally determined for different temperature ranges
using the following procedure (180). The thermistor was submerged in a
thermostatted bath of known temperature. Resistance readings were
recorded at three different temperatures in 10°C intervals. This yielded
a solvable set of three equations with three unknowns. The values for.
A, B and C at the different temperature intervals are given in Table 3.3.
To prevent erroneous readings as air and liquid segments flowed pass
the thermistor, the flow stream was debubbled prior to measuring
temperature. The previously established optimum reaction conditions
were used in this study. A plot of the results shown in Figure 3.8
indicates little change in detector response between 21.0 °C and 28.0°C.

Therefore, normal fluctuations in room temperature should have little

56

 

 

 

 

 

 

 

l l I I
0.6 - q
.1
0.)
0 ——A—— ‘—
C uni
O
.0
L- ..
O
(I)
.0
< c-
i
O. J J 1 1 l 1 1 1 J_ l 1 1 1 L l 1 1 L L l L 1 1
0.0 0.5 1.0 1.5 2.0

Reaction time (min)

Figure 3.7 . Influence of DSA:phenol ratios and reaction times on the
PCR detector signal. DSA:phenol ratios: 0.7 (s); 1.2 (A); 2.8 (s); 6.7 (a);
12.4 (A); 27.8 (s).

Table 3.3.

57

Values of the coefficients in the Skinhart-Bart
equation for various temperature ranges.

 

 

 

Temperature A B C
Interval (°C)
21 - 30 0.1326 -0.05954 0.002789
31 - 40 0.05351 -0.02261 0.001594
41 - 50 0.03578 -0.01446 0.001373
51 - 60 0.01001 0.001121 0.0005273

 

 

58

 

 

 

 

0.8 1 I I I l
0.6 - ‘
Q) C
U 1-
C
O
D 0.4 ~ "
i...
O
U) ..
.0
< -—1
0.2 '-
1 l 1 l 1 l l l ‘ 1 1
080.0 30.0 40.0 50.0

Temperature (0C)

Figure 3.8. Temperature studies. PCR flow rates: 80 uM phenol - 0.30

mein; 5.9 mM DSA - 0.05 ml/min; pH 10.1 borax buffer - 0.60 ml/min.
Reaction time of 37 s.

59

effect on the absorbance signal, and thermostatting is not essential. A
decreased signal observed at temperatures above 28.0 °C can be
attributed to the increased thermodynamic and kinetic feasibility of

secondary reactions.

b. Simplex Optimization

A univariate optimization in which one variable is changed while
all others are held constant is the scheme most commonly employed for
establishing the best reaction conditions. The assumption made in this
approach is that all the variables are independent of each other. A
false optimum may be found if this assumption is incorrect. Therefore,
simplex optimization was used to test the optimum conditions established
by the univariate approach.

For a better understanding of the work done in this section, it is
necessary to present a brief discussion on the historical progression
and the basic rules of simplex optimization.

Simplex optimization was first presented by Spendley et al. (181).
This method suffered a number of limitations, some of which were the
inability to accelerate a simplex over the response surface when
searching for an optimum and the lack of verification procedures to
confirm that the optimum found is a true optimum. In 1965, Nelder and
Mead (182), proposed the 'modified simplex optimization’ which introduced
new operations that would overcome some of the limitations of the
original procedure. These operations included expansion and contraction
of the simplex. Since these early works, a large number of papers have
been published concerning applications of simplex optimization and

further modifications to the basic method (183-190). In 1985, Betteridge

and coworkers (191) introduced the 'composite modified simplex' (CMS)
method which incorporated many of the modifications previously
proposed by other workers. A second paper by this group (192)
evaluated and compared the performance of the CMS method to that of
other modified simplex methods. The program developed by Betteridge
et al., OPTIMA, was the one employed in this work.

A simplex is a geometric figure whose number of vertices is equal
to one more than the number of variables (or dimensions) being
investigated. A simplex in two dimensions is a triangle. In three
dimensions, it is a tetrahedron. Higher dimensions are often used, but
the figures are not as easily visualized. For the purpose of introducing
the principles of simplex optimization, a two dimensional simplex will be
used. Figure 3.9 shows the progress of the two dimensional simplex on
a contour map as it moves towards the optimum. The contour lines
(circles) on the response surface represent different combinations of the
two parameters being studied that would give equal responses. The
objective is to rapidly move the simplex towards the region of optimum
response. This progression is guided by a set of rules, of which only
the basic ones are discussed here. First, after the evaluation of the
initial simplex, a move is made after each response. Second, the point .
that gives the worst response is discarded and its mirror image across
the face of the remaining points becomes the new vertex of the simplex.
In Figure 3.9, the initial simplex is a triangle defined by vertices 1, 2
and 3. Point 1 gives the worst response of the three and is reflected
across the face 2-3 to generate point 4. The new simplex has as its
vertices, 2, 3 and 4. The response at point 4 is obtained and the next

move is made based on this value. This process continues until the

61

 

 

Figure 3.9. Progression of a two dimensional simplex over a response
surface. Circles represent iso-response conditions.

62

simplex starts to circle which signifies that an optimum has been found.
This optimum may be a local optimum rather than a global optimum.

This could be verified by restarting with a larger simplex. A third rule
is used to prevent the simplex from oscillating when the reflected point

yields the lowest value in the new simplex. When this happens, the

second to lowest response in the new simplex is reflected to obtain the
new simplex.

A large number of rules not mentioned above are used in the
OPTIMA program to handle the pitfalls of the original simplex method.
These include the expansion, contraction, polynomial fitting and
redirection of the simplex. However, rather than going into extensive
explanations, this part is best left to the interested reader to pursue.
The three rules mentioned above should be sufficient information for the
reader to follow the progression of the three dimensional simplex used
in this work.

Before starting the simplex optimization, the boundary conditions
listed in Table 3.4 were established for the variables to be tested. This
was based on a general knowledge of the system. For instance, the
diazo coupling reaction with phenol does not proceed in acidic solutions.
This places a lower limit on the pH region that was studied. Sodium
borate has a buffering range between pH 8.0 and 11.8. Thus, the pH
interval to be investigated was set at 7.0 to 12.0. The peristaltic pump
has an arbitrary speed setting which ranges from 1 to 99. The lower
boundary was set based on the flow rates at pump settings ( 20 being
too slow. The upper limit of the DSA concentration was established by

the decomposition observed at high concentrations.

63

Table 3.4. Boundary conditions for simplex optimization with the
composite modified simplex method.

 

 

 

 

 

Variable Lower Upper

Limit Limit

pH 7.0 12.0
Pump speed 20 99

[DSA] (mM) 1.5 16.6

 

 

64

The reaction manifold was thermostatted at 26.5 °C throughout
these experiments. The phenol solution employed was 30 11M and the
sampling and wash times were 40s and 20s, respectively. The blank was
borax buffer and DSA.

The parameters varied were reaction time, pH of the reaction
medium and the reagent concentration. The response optimized was
absorbance. The readings obtained for each set of conditions were
entered‘into the microcomputer and the next set of conditions was
generated. The results of the simplex optimization are summarized in
Table 3.5.

The final simplex has as its corners, experiments that have yielded
the highest absorbsnce readings, namely, numbers 3, 7, l3 and 14. The
largest absorbsnce value of 0.31 was measured for simplex no. 3. The
response obtained when the reaction was carried out using conditions
determined univariately, 0.30 AU, was not significantly different from
that obtained at the optimum conditions established by simplex
optimization. The precision of the absorbance readings is £0.01 AU.

The four points of the final simplex were all obtained in the pH interval
of 10.2 10.5 indicating an optimum pH region. This observation coincides
well with that made in the univariate experiments as well as the value
predicted using dissociation constants. The magnitude of this pH
interval will, once again, depend on the amounts of DSA added.

The major factor influencing the response of the PCR detector is
pI-l as demonstrated by the scatter diagrams shown in Figure 3.10. A
definite optimum is observed for this variable. However, there appears
to be no clear trends in the effects of reaction times and DSA

concentrations on the PCR response in the intervals studied.

Table 3.5.

65

Optimization of the phenol/sulfanilic reaction system
by the composite modified simplex method.

 

 

 

Expt Simplex Time' pH” [DSA] Abs°

no. Vertices (s) (mM) (AU)
1 1 16 8.06 1.50 0.09
2 49 8.06 1.50 0.06
3 67 10.08 1.50 0.31
4 67 8.69 7.70 0.25
5 3,4, 1,2 127 10.67 4.27 0.28
6 3,5,4,1 56 11.40 5.58 0.20
7 3 ,5 ,4,6 67 10.58 4.56 0.30
8 3,7,5,4 83 11.61 1.50 0.05
9 3,7 ,5,4 72 9.44 5.30 0.29
10 3,7,9,5 48 9.40 2.07 0.29
1 1 3,7 ,9,10 127 10.48 4.27 0.26
12 3,7 ,9,10 56 9.66 2.62 0.30
13 3,7,12,9 60 9.78 2.82 0.30
14 3,7,13,9 60 10.64 1.50 0.30
15 3,7,13,14 72 9.62 3.65 0.29.
Univariate Method 45 10.10 5.90 0.30

 

' Time was measured for each pump setting used.
” pH of solution measured after reaction has proceeded.

C Average of triplicate measurements.

66

as: 3

someones—00:00 (mm #0
03030 9.3080ch 05 .m> 00:00.3st «0 030.530 003.com .25 0.53m

 

 

 

 

:0 A...

 

3 an:
o9 2. 8. as 2 o .
1 a a q u 00
0
To
v
a
s
O
o «.o u
v
N
M . ...._
om on no
3 To
3.5 :33 1..
O m h 0 n v n p o . e— n— N— Z a— a a
q q q a q . q a q oo . . . 4 a \1
m o o
o I. F-o o
v
1 a
s
o
J «.o u 0
V
N o
a o . «.0. 0o
0
00 m .. n6 0 0&0
106

A...

 

od

'6

v6

BONVBUOSBV

A6.

67

0. Comparison With Other Workers
It should be noted that the optimum reaction conditions reported
by other workers for this reaction are not entirely in agreement with

those established here. Whitlock et al. (125) reported that, for phenol,

the optimum pH was 8.5. A sodium bicarbonate buffer was used, and the
reaction required 2 min for completion. Preliminary experiments
performed in our laboratory to characterize the reaction also employed
sodium bicarbonate as the buffer. The optimum pH determined with this
buffer agreed with that reported by these workers. However, the
buffer was changed to sodium borate due to the poor buffering capacity
of sodium bicarbonate at pH 8.5. The two different optimum pH values
obtained indicate that one or both buffers are not inert.‘ Other
observations made by Whitlock et al., such as the effect of excess
reagent and the stability period for the DSA reagent are in agreement
with our findings.

Baiocchi et al. (126), reported an aptimum pH of 11.0 in their work
on pro-column derivatization of phenols by sulfanilic acid. In this case,
NaOH was added to obtain the desired pH. Their reported reaction time
was 15 min, and the optimum DSAzphenol ratio was 40:1; both of these
. are much larger than values determined in this work. In addition, the
reagent had to be used within 10 minutes of its preparation. These
observations may be due to improper preparation of the diazotized
sulfanilic acid. If the starting compound is sulfanilic acid, rather than
its sodium salt, an indirect diazotization needs to be carried out (130).

That is, prior to diazotization, sodium carbonate is added to the solution

68

containing sulfanilic acid to convert it to the water soluble form. This

step was not reported by Baiocchi and coworkers.

3. Evaluation of the HPLC/PCR System

The optimized PCR detector for phenols was used in conjunction
with an HPLC apparatus with UV detection for system characterization
studies. The performance of the PCR detector was evaluated with
respect to its compatibility with the chromatographic conditions, band
broadening contributions, detection limit, linear dynamic range and

precision.

a. Mobile Phase Effects

The effect of varying amounts of common reverse phase solvents
on the detector response was" investigated. A 75 11M phenol solution
prepared in pH 10.1 borax buffer was continuously added to and mixed
with the column effluent. The manifold employed was identical to that
shown in Figure 3.2. The UV and PCR detector responses were
monitored while running a gradient of water/acetonitrile (Figure 3.11) or
water/methanol (Figure 3.12). A drift of 1% in the detector signal is
observed as the acetonitrile composition of the mobile phase spanned 0%
to 80%. The upper limit was set by the solubility of sodium borate in
acetonitrile. A decrease in the signal was observed when the
percentage of methanol was increased beyond 60%. This effect may be
attributed to a combination of different factors among which are solvent
polarity and proton donating ability. The blank signal shows no change

for either solvent. From these results, it was concluded that

69

000620 £53:

0 50: 58>:— oé am 300 30: 0.75 6030000 0:20:00 0030 05 :0
sawawaaasou 000:0 0:000 2050 05.300, uo 00:03.03 .36 0.305

2:): msE.

 

 

 

 

 

.\.o.n 30.0
00050 0.306 5 20.310 o\..

0.0

m0

0..

EDNVBHOSBV

70

a as... 55:... o; a 3.: .6: 0.80

4030000 :w6\xw
503000.. 359300 0030 05 :0

cowuwaoasao 000:0 "Janos ~3an 800.300, a0 0000335 .NA.« 0.305

 

 

 

 

 

0.0

0.0

 

 

“2.5: 05:...
e 0.. Lo. 0.. 0 m.
n p h b i? n
0 ON 00 0m 00 00.

$23 232: s Ionzo .x.

0..

EDNVBHOSBV

71

acetonitrile is the more versatile solvent, especially for separations
requiring gradient elution. Also, a slightly higher response was
obtained in acetonitrile (0.56 AU) than in methanol (0.51 AU). This may
be due to a shift of has from 430 nm for acetonitrile to 420 nm for

methanol.

b. Band Broadening Contributions

Laminar flowing streams in open tubes have a characteristic
parabolic flow profile. The particles that are adjacent to the walls of
the tubing travel at a slower velocity than those towards the center of
the tube (due to viscous drag along the walls). The result of this
phenomenon is longitudinal dispersion in the tube. When air segments
are added to the flowing stream, this parabolic flow profile is broken
down and thus dispersion is reduced. This reduction of dispersion is
particularly dramatic when long sections of tubing have to .be used.

The degree of band broadening introduced by the PCR detector
was examined. For the most accurate results, pyrogallol was employed.
This trihydroxy phenol is weakly retained by the ODS column. A typical
peak width comparison for the two detectors is shown in Figure 3.13.
From the ratio of the peak widths at half height of the UV and PCR
responses, the amount of band broadening was found to be 51-20%.
This value demonstrates the effectiveness of air-segmentation and the

miniaturized CFA instrument in limiting dispersion.

72

uv -----

PCR —

 

 

 

for the UV

Comparison of peak widths at half height, wm.

Figure 3.13.
and PCR trace.

73

c. Flow Rate Compatibility

Studies were done to determine whether there was a limitation
placed on the flow rate of the HPLC system by the PCR system or vice
versa. The flow rates were varied in the range normally used for HPLC,

that is, 0.5 ml/min to 2.0 ml/min. A sample mixture of phenol,

3,5-dimethylphenol, and 2,6-dichlorophenol was injected in 20 ul
quantities. The responses for flow rates of 0.5 ml/min, 1.0 ml/min and
2.0 ml/min are illustrated in Figure 3.14. The mobile phase was 40%:60%
water:acetonitrile.

The resolution of the separated peaks is maintained by the PCR
system within this flow rate interval. This demonstrates the
efffectiveness of air-segmentation in limiting dispersion. ,The reason for
the varying relative peak heights at different flow rates originates from
the difference in reaction kinetics for each phenolic compound. Since
the manifold was not changed, the reaction time used decreased with
increasing flow rates. Thus, the coupling reaction of _ each compound
was reaching a different degree of completion.

The conclusion to be drawn from these results is that there is a
compatibility in flow rates between the two systems for this particular
reaction. This conclusion should be extended to other systems with
some caution. This is because the flow rate and the reaction time
needed are the factors that determine the coil length that has to be
used. For instance, for a flow rate of 2 ml/min and a reaction time of 5
min, the coil length needed would be 330 cm (coil i.d. of 0.1 cm). This

is not a 'practical’ length.

74

o.) b.) c.)

m L Ami

Figure 3.14. Flow rate compatibility between the HPLC and PCR systems.
Sample consists of phenol, 3,5-dimethylphenol and 2,6-dichlorophenol,
with elution in this order. Mobile phase is 60/40 CH3CN/Hz0. Flow
rates are a.) 0.5 mein b.) 1.0 ml/min and c.) 2.0 ml/min.

 

 

75

d. Statistical Figures of Merit

The detection limit (S/N = 2) for phenol with this PCR detector is
0.75 pH. The molar absorptivity of the phenol derivative (22500 1 mol'1
cm") (calculated) is 16 times that of phenol itself (1400 1 mole"1 cm")
(193). The linear dynamic range of the reaction, shown in Figure 3.15,
spanned over 2 orders of magnitude, from 0.75 1111 to 150 Md. The
upper limit was established by deviations from Beer’s law which resulted
from instrumental effects rather than chemical. Reproducibility, based

on 15 injections of 100 ”M phenol standards into the HPLC, was $0.64X.

e. Performance Comparison with Other Systems

As mentioned earlier, no colorimetric-based post-column reaction
detectors for phenols have been developed. For this reason, it is not
possible to compare the performance of the developed detector to other
post-column colorimetric systems. Instead, the following discussion will
focus on a comparison of the post-column derivatization of phenols with
fluorescence detection and precolumn derivatizations with colorimetric
detection to the system developed in this work.

The systems that were compared included two post-column
derivatization schemes with fluorescence detection that have been
repoer by Cassidy et al. (194) and Wolkoff and Larose (8). The PCR
detector reported by the first set of investigators measured the
fluorescence of derivatives formed from the reaction between
hydroxyphenyls and dansyl chloride (1-dimethyl-aminonaphthalene-S-
sulfonyl chloride). The second PCR system, described by Wolkoff and
Larose, involved the indirect detection of phenols through the reduction

of cerium (IV) by phenols to the fluorescing species, cerium (III). An

76

 

4.00 '-

3.00

2.00

ABSORBANCE

1.00

 

0.000
0.000

Figure 3.15.

 

4 1 1 J 1 L . 1
50.0 100. 150. 200.

CONCENTRATION (um)

Linear dynamic range of the phenol-sulfanilic reaction.

 

77

air-segmented continuous flow analyzer was used as the PCR system.
Two other systems included in this comparison involve precolumn
derivatization of phenols with 4-aminoantipyrine (147) and sulfanilic acid
(126) to form derivatives that absorb in the visible region. The
reported analytical figures of merit for the fluorescent and colorimetric
systems are summarized in Table 3.6. The figures quoted in this table
for the fluorescent systems do not represent an accurate comparison
with the colorimetric based detectors because different compounds were
used. For example, phenol was used for the colorimetric systems,
hydroxyphenyl for the dansyl chloride reaction and o—chlorophenol for
the cerium system.

The analytical figures of merit that are listed in Table 3.6 indicate
that post-column derivatization with diazotized sulfanilic 'acid, when
compared to the other systems, shows the best overall results. The
linear dynamic range spans over two orders of magnitude, and the
precision (determined by multiple injections of phenol and measurement
of the peak height) is a factor of ten better than that achieved by pre-
column derivatization with 4-aminoantipyrine (4-AAP). The higher
detection limit observed for post-column derivatization with sulfanilic
acid, in comparison to the other colormimetric procedures, is due to
dilution "of the separated sample as reagents are added.

It should be noted, that although fluorescence in general, is a
more sensitive detection mode than (JV-visible absorption, it suffers to a
larger extent from interferences. As a result, fluorometric detection may
not be suitable for analyses of complex samples without extensive sample
pretreatment. Fluorescence also suffers from a number of inherent

problems, one of which is self absorption. This is the factor that limits

78

0E:_00I0.a I 0.5 6050—50200 I 0 60000000005 I m "000 00030>0an<

83085 8: u

 

 

 

S: a u and 0 95 05.4
mm“ a a med 0 0.5 05003.5
0: $0 02 I 2.0 3.0 0 ~60 £05030
mfi . 3 I do 3.0 m 000 53.80

83.820
X: 0; 0mm .. o; u 0 000 7850

:3 80.80 :3 050 8385.5 503083.80

80 005 008200 508380 50880 .8 on»... 05050

 

 

5.85 8 :35838 20 .80 0.8.- 8 8.500 50803.. 20 8 5.0.8050 .3" £5...

00300550001 538$qu .000 I003 Aha—0.3..

79

the linear dynamic range of the cerium PCR detector. Self absorption

has been observed at high concentrations of cerium (IV) sulfate (196).

4. Applications

The performance of the reaction detector as a sensitive and
selective HPLC detector for phenols was compared with that of the UV

detector in the analysis of real samples.

a. Determination of Phenols in River Water

Grand River (Lansing, MI) water, collected downstream from a coal
driven power plant, was analyzed for the presence of phenols. No
phenols were detected. The sample was then spiked with 1.9 ppm
phenol, 2.2 ppm o-cresol, 2.4 ppm 3,5-dimethylphenol and 2.2 ppm
resorcinol. The detector responses, both prior to (UV) and following
derivatization (PCR) are shown in Figure 3.16. The chromatographed
peaks were barely discernible by UV detection but clearly
distinguishable by PCR detection despite the sensitivity of the former
being set at a lower scale than the latter. The .figure clearly
demonstrates the increased sensitivity obtained when using post-column
derivatization.

The diazo coupling reaction is an electrophilic aromatic
substitution reaction. Thus, the aromatic compounds that are most likely
to undergo this reaction would be those with a strongly activating
substituent such as phenols and aromatic amines. The class selectivity
of the PCR detector is demonstrated by the chromatogram shown in

Figure 3.17 where the UV and PCR responses are compared for a water

 

 

uv
1 2 1.00m;
3 4
. 2
I
PCR
3 IDIAU
4

 

 

 

 

 

 

 

 

f I r I I I I I I I
O I ' 2 3 4 5 6 7 8 9
TIME (MIN)

Figure 3.16. Comparison of UV and PCR detection for the separation of
phenols in a spiked river water sample ( Grand River, East Lansing,
Michigan). 1.) resorcinol. 2.2 ppm 2.) phenol, 1.9 ppm 3. o-cresol,
2.2 ppm 4.) 3,5-dimethylphenol, 2.4 ppm. HPLC: Spherisorb ODS
column; acetonitrile/water (50:50) mobile phase; flow rate of 1.0 ml/min;
UV detection at 254 nm; 0.04 AUFS. PCR: manifold as shown in Figure
3.2; detection at 450 nm; 0.10 AUFS.

I 2 I004 AU
W
3
4
I

PCR

3 IJ AU

4

TIME (MIN)

Figure 3.17. Comparison of selectivity of UV and PCR detection for
phenols in a spiked river water sample. 1.) phenol, 18.8 ppm

2.) aniline, 19.6 ppm 3.) o-cresol, 21.6 ppm 4.) 3,5-dimethylphenol,
24.4 ppm. HPLC: Spherisorb ODS column; acetonitrile/water (50:50)
mobile phase; flow rate of 1.0 ml/min; UV detection at 254 nm; 0.04 AUFS.
PCR: manifold as shown in Figure 3.2; detection at 460 nm; 1.0 AUFS.

82

sample that had been spiked with phenol, aniline, o-cresol and 3,5-
dimethylphenol. UV detection of the chromatographed sample showed the
presence of all four compounds, with aniline as a major peak. The PCR
detector showed only three peaks which corresponded to the phenols.
This is a result of the different optimum reaction conditions needed for
DSA coupling to phenol and to aromatic amines.

It may be observed in Figures 3.16 and 3.17, that the column void
volume peak was not present in the PCR detector trace. This is due to
the toroidal flow within each liquid segment that is characteristic of air-
segmented flow streams. The result is enhanced mixing which leads to

homogeneity within each liquid segment.

b. Phenols Determination in Residual Fuel Oil

A residual fuel oil sample was separated into fractions of different
functional group types by sequential elution solvent chromatography.
The fraction containing phenols was chromatographed under isocratic
conditions with a 50:50 acetonitrile:water mobile phase. A series of
peaks were observed in the UV trace as shown in Figure 3.18. Only two
peaks were evident in the PCR detector recording. These peaks have
been tentatively identified by retention times and spiking experiments to
be phenol and a cresol. Figure 3.19 shows the responses obtained when
the sample was spiked with 90 ppm and 400 ppm phenol. At the lower
concentration, there is no measureable change in relative peak heights
(ratio of phenol containing peak, peak b, to peak a on the
chromatogram) for the UV detector response, whereas an 11 fold
increase is observed in the PCR trace. When the sample was spiked

with 400 ppm phenol, a two fold increase in the relative peak height is

83

 

 

 

 

 

0 UV

_L l l l L l l l l l

O | 2 3 4 5 6 7 8 9
TIME (MIN)

Figure 3.18. Reverse phase separation of RFC, fraction #7. HPLC:
Spherisorb ODS column; acetonitrile/water (50:50) mobile phase; flow rate
of 1.0 ml/min; UV detection at 254 nm; 0.64 AUFS. PCR: manifold as
shown in Figure 3.2; detection at 450 nm; 1.0 AUFS.

84

 

 

 

 

 

 

 

 

 

 

 

 

i JL’; ) V—~ PCR

Figure 3.19. Chromatogram of RFC, fraction #7, spiked with a.) 90 ppm
phenol b.) 400 ppm phenol.

observed in the UV trace and a 46 fold increase is obtained for the PCR
response. This further demonstrates the greater sensitivity of the PCR
detector over the UV detector. In the identification of cresol by
spiking, the sample was diluted 20 times and the resultant solution

spiked with 32 ppm o-cresol. The detector responses are shown in

Figure 3.20. From the change in relative peak heights (UV) and the
corresponding changes in the PCR response, it was apparent that other
compounds were coeluting with phenol and o-cresol and were not

detected by the derivatization detector.
D. Conclusions

The advantages of using post-column derivatization in HPLC are
apparent. The gain in selectivity and sensitivity have been
demonstrated with applications to phenols determination in water and
fuel samples. The latter application also shows that a potentially
complicated separation may be greatly simplified by using a detector of
this type. In addition, the advantages of using air-segmented
continuous flow analysis such as minimized band broadening due to
longitudinal dispersion in the PCR detector and enhanced mixing within '
each liquid segment have been illustrated. Other positive points,
particular to this system, include on-line detection and compatibility
between the separation and reaction systems with respect to flow rates
and varying acetonitrile/water mobile phase composition.

Post—column reaction detection of phenol derivatives formed from
the diazo coupling reaction with sulfanilic acid does have its limitations,

however. The first of these disadvantages is dilution, which is inherent

I .032 AUFS

[W‘—

 

 

I .OI AUFS

 

 

W

Figure 3.20. Chromatogram of RFC fraction #7 diluted 20x and spiked
with 32 ppm o-cresol.

87

to PCR detector systems that involve the addition of reagents to the
column effluent. Therefore, the volume of reagents added should be
kept to a minimum. A second limitation is that the diazo coupling does
not occur for all phenols and therefore, cannot be used for the
determination of total phenols. As with other colorimetric reactions for
phenols that involve electrophilic aromatic substitution, in general no
reactions occur for phenols containing electron withdrawing groups.
Phenols with substituents occupying both the ortho and para positions
to the hydroxyl group also yield no response to this reaction as the
positions at which electrophilic aromatic substitutions would take place
are blocked. This lack of response from all phenols may be regarded as
a disadvantage or an advantage depending on the desired information.

A survey of the reactivity of 126 phenols to diazotized sulfanilic
acid, as measured by the presence of a measureable absorbance, was
reported by Koppe et al. (119). In general, this work provides a good
guide to the types of phenols that undergo this reaction. However,
detectable responses have been obtained in our laboratory for some
phenols that have been listed by Koppe et al. as having no responses
is. 2,4-dichlorophenol and p-tert-butylphenol. This may be a result of
the reaction conditions that were used by Koppe et al. being different
from that employed by us e.g., NaOI-I was added to obtain a basic

solution. As previously noted, the buffering systems are not inert.

CHAPTER IV
THE DEVELOPMENT OF RECYCLE HEARTCUT
CHROMATOGRAPHY (REC) FOR

ANALYSES IN COMPLEX SAMPLE MATRICES

A. INTRODUCTION

The development of a new valve switching technique, which we
call recycle heartcut chromatography (RHC), is described. Preliminary
results are presented that characterize the REC system. Several
potential applications are explored.

The method of recycle heartcut chromatography involves the
repetitive isolation of specific components and the recycling of these
components through the column for an improved separation. The HPLC
components used to implement RI-IC are shown in Figure 4.1. In addition
to the two six port valves, the system consists of a single detector, a
pump and one column. The recycled sample does not flow through the
pump so that the problem of peak dispersion in the pump cavity is not
observed. Such dispersion is encountered in the direct pumping
approach to recycle chromatography. The use of the same column for
recycling eliminates any complications that may arise from employing
more than one column as with the alternate pumping principle. Such

complications originate from variations in column characteristics.

89

.hsnsshodsaeso 33.30: 20.30.—
.anoaam 3 pee: canvas—amass ‘52.; use Seash- oan :3. 0.53m

_:o>.

omom

EoZom

 

_.

_

 

 

m

m

 

 

 

 

 

 

.o _ ,
w m MWW=_ kartam

 

 

 

 

 

 

 

uouooaoo vllllllt
c5330

 

90

The valve switching sequence used in RHC is demonstrated in
Figure 4.2. When the sample is first injected, valves 1 and 2 are both
in the counterclockwise (COW) position as shown in Figure 4.2a. The
column effluent is directed to waste until the desired fraction begins to
elute. At this point, valve 2 switches clockwise (CW) (Figure 4.2b) to
divert this fraction to a heartcut loop that is made from stainless steel
tubing. After an appropriate delay, valve 2 switches back
counterclockwise (Figure 4.2c) to allow the column to flush. Figure 4.2d
shows the final stage when valve 1 is turned clockwise to divert the
mobile phase flow through the heartcut loop and sweep its contents onto
the head of the column. This valve switching sequence is repeated in
order to obtain a 'cleaner’ and more resolved fraction with each recycle.

There are a number of advantages that arise from the positioning
of the detector in this configuration. First, only one detector is needed
to monitor the column effluent continuously. The detector output can be
used to trigger a valve switch, as discussed in a later section. Third, a
high pressure flow cell is not needed because one end of the detector is
always connected to ambient air.

In the following sections, the instrumentation used in RHC is
described. The methods used to trigger valve switchings are
illustrated. The various contributions to band broadening are discussed
for the specific heartcut loops that were used in this work. Finally,

some applications of EEO are presented.

91

a.)

 

Waste
Pump

 

 

 

 

 

 

 

 

 

b .)
Uoslc
Pump
c.)
Waste
Pump
do)
Waste
Pump

 

 

Figure 4.2. Valve switching sequence employed in REC, a.) start-up
b.) heartcut c.) column flush and d.) recycle heartcut.

92

B. EXPERIMENTAL

1. Instrumentation

The description of the instrumentation used to implement RHC is
divided into two sections. First, details are given of the HPLC
apparatus that was employed. This is followed by a discussion of the

role of the microcomputer in experiment control.

a. HPLC Apparatus

A block diagram of the apparatus employed for recycle heartcut
chromatography is shown in Figure 4.1. The HPLC system is assembled
from a Spectraphysics SP8700 ternary solvent delivery system (Santa
Clara, CA) with a 280 microprocessor, a Rheodyne 7120 injection valve
(Cotati, CA) with a 20 111 sample loop, a 4.6 mm x 25 cm Zorbax ODS
column (Dupont, Wilmington, DE), a Chromatronix 220 UV absorbance
detector (Berkeley, CA) and two Rheodyne valves (7010 and 7040) with
pneumatic actuators for diverting the solvent flow stream along various
paths. The pneumatic actuators are operated by 2-port and 2-position
solenoid valves (Scovill Corporation, Wake Forest, NC) which direct the
flow of air or nitrogen to either side of the actuator piston. The gas
pressure (50 psi) provides the driving force that turns the valves. For
details on the circuitry employed, the reader is referred to the original

work (197).

93

b. Microcomputer Controlled HPLC System

The HPLC system is controlled by an Intel 8085 based
microcomputer that was built in our laboratory (197). As shown in
Figure 4.3, the microcomputer can be used to control the pump, the
chart recorder drive and the valve switchings as well as to facilitate a
number of other functions.

Communication between the pump and the microcomputer is
achieved through the use of a keypad emulator (198) which allows the
computer to simulate a keystroke on the keypad of the SP8700 solvent
delivery system. This set-up enables the option of a synchronized
separation and data acquisition scheme consisting of changing separation
conditions as well as valve switchings and data acquisition based on the
same time clock. The AM9513 timer (Advanced Micro Devices), used for
this purpose, is installed as part of the microcomputer system. Data
taken by the microcomputer are stored and processed with a DEC L81
11/23 minicomputer. The chromatogram is displayed on a chart recorder
which may also be placed under microcomputer control. The valves are
switched as directed by the TTL signal sent by the microcomputer to an
optically isolated relay that is connected to the solenoid valve. Valve
positions are sensed by the microcomputer via the absence or presence
of a current through a reed switch that is mounted on the valve
housing. A magnet mounted on the valve shaft causes the switch
contacts to close as it moves near the switch when the valve is turned
clockwise. The state of this reed switch, sensed by the microcomputer,
enables it to determine the position of the valve and whether the valve

turning had been successfully carried out.

94

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

used in REC.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DEC LSl 11/23 KEYPAD SP8700
EMULATOR
muncoupures zao MICRO- SOLVENT
PROCESSOR RESERVOIRS
J TERNARY
INTEL 3035 PROPORTIONING
MICROCOMP T ' PUMP VALVE
U ER W 8255
ARV“) PIO
. y
U — lNJECTOR
AM9513 VALVE J,
1"ng CONTROL "'7 43-0
l
; ti
\/ '
A0558 A0574 : o-o
I
DAC ADC """l obj
CHART UV
RECORDER DETECTOR
Figure 4.3. Block diagram of microcomputer controlled HPLC system

 

 

95

This system, aside from the initial sample injection, is completely
automated. Its construction also allows for the manual control of each

of the functions mentioned above.

2. Software

The software developed by P. M. Wiegand (197) employed
polyFORTH as the operating language at the microcomputer level. FORTH
is a threaded language that allows a new word to be defined either by
stringing together existing words or by building the word from the
basic kernal commands. A single word may be used to carry out a
complex sequence of events which makes this language particularly
suitable for instrument control. In addition, the FORTH routines are
easy to read, write and alter. An example of a FORTH routine that is
used to take a cut when a'specific peak is sensed and to recycle it is
shown in Figure 4.4. When the command " lCUT " is entered, the

following events take place:

1.) Valves 1 and 2 are turned counterclockwise (CCW).

2.) When the mobile phase flow has stabilized, a message is
displayed on the terminal prompting the operator to inject
a sample.

3.) When the sample is injected, the microcomputer clock starts
timing and the recorder chart drive is turned on.

4.) At 2 min 30 s, the data acquisition and peak counting
routines are initiated.

5.) When the peak count = 1, valve 2 is switched CW for a
heartcut.

6.) After a 12 s delay, valve 2 is turned CCW to allow the
column to flush. '

7.) At 4 min 30 s, valve 1 is turned CW and the heartcut is
recycled. _

8.) At 8 min 30 s, the message " END OF RUN " is displayed on
the terminal.

9.) The pump, chart drive and peak finding and data taking

routines are ended.

96

Block Number: 85

0

(One cut)

2 lCUT 1 CCW 2 CCW

FLOW? ." FLOW IS READY, INJECT SAMPLE."
INJECT DIGREC ON

2 30 EVENT TAKEDATA Z#PEAKS ?PEAK

BEGIN 0 RFLAG C! PAUSE #PEAKS O 1 2 END 2 CW
1200 WAIT 2 CCW

4 30 EVENT 1 CW

8 30 EVENT ." END OF RUN."

STP PRESS OFF ISTOP 4STOP ENDDATA ;

Figure 4.4. An example of a FORTH block used to take a heartcut based
on peak count which is then recycled through the column.

97

3. Solution and Sample Preparation

Polyaromatic hydrocarbons were purchased from Chem Service
(West Chester, PA). Steroids were obtained from Steraloids, Inc. (Wilton,
NH). All solutions were filtered through a 0.45 pm filter prior to
injection into the HPLC system.

Urine samples were pretreated using a Cu Sep-Pak cartridge
(Waters Associates, Milford, MA) which had been prewetted with 2 ml of
methanol followed by 5 ml of water. The cartridge was then charged
with 20 ml of urine and washed with 5 ml of water to remove salts. A
3 ml amount of methanol was used to elute the organic components off
the cartridge. This fraction was collected and spiked with 10 uM of
1,4 androstadiene-3,17-dione. As a final step, the urine samples were
filtered through a 0.45 pm filter.

Spectral grade HPLC solvents (Burdick and Jackson, Muskegon,
M1) were filtered through a 0.45 pm filter and degassed with helium

prior to use.

C. Evaluation of the RHC System

1. Reproducibility of Valve Switchings

The microcomputer software has been developed so as to allow
valve actuation to be initiated based on the fulfillment of a variety of
conditions. These conditions are peak number, detector threshold and
time. Heartcuts that are initiated by the first two methods are referred

to as detector-based cuts since the detector output is monitored to

98

determine whether the conditions for a heartcut are met. Excerpts of
FORTH routines used to implement valve switchings in the determination
of valve switching precision are listed in Figure 4.5. The first case,
Figure 4.5a, is set up to count the number of peaks, with instructions
to turn valve 2 CW for a heartcut when the first peak is sensed. Valve
actuation can also be initiated by setting a threshold based on detector
output as shown in the Figure 4.5b. In this example, a cut is taken
when the detector response has reached 10% of the full scale value.
The third type of heartcut is one that is based on the time after
injection. As demonstrated by the third case, Figure 4.5c, this is
accomplished by specifying a time (e.g. 4 min 30 s).

The reproducibility of the valve switchings was determined by
injections of a 300 ”M naphthalene solution followed by a 12 s heartcut
that corresponds to a 200 pl volume. This heartcut was sent back
through the column, and the peak height of the recycled- peak was
recorded for use in precision calculations. The mobile phase was 100%
methanol at a flow rate of 1.0 ml/min. The detector was set at
0.16 absorbance units full scale (AUFS). Table 4.1 lists the peak height
measurements made for the first and second cycles through the column.
The percent relative standard deviation (%RSD) was calculated from the
ratio of peak heights of cycle 2 to that of cycle I to eliminate the
contribution from injection inconsistencies. Ideally, the sample should
be injected as a sharp plug. However, mixing of the sample and mobile
phase can occur due to imperfect injector design, badly made connectors
and poor injection technique. Any of these factors could lead to a

variation in peak height.

99

a. 2 30 EVENT TAKEDATA Z#PEAKS ?PEAK

BEGIN 0 RFLAG C! PAUSE #PEAKS 0 l = END 2 CW

b. 2 30 EVENT TAKEDATA

10 XFSCHG 2 CW

6. 2 3O EVENT TAKEDATA

4 30 EVENT 2 CW

Figure 4.5. Examples of FORTH commands used to initiate valve
actuations by a.) peak count b.) threshold and 0.) time after injection.

100

Table 4.1. Determining the precision of valve switchings triggered
by peak count, detector threshold (10% full scale) and

 

 

 

time.
Peak Height Relative peak )4 R80
(A/D units) height
Cycle 1 Cycle 2
Peak Count
792 304 .384
849 311 .366
829 325 .392 4.54
825 318 .385
819 340 .415

Detector Threshold

824 483 .586
830 475 .572_
809 501 .619 3.23
821 496 .604
818 501 .612
'Phne
840 423 .504
840 448 .533
830 373 .449 7.25
868 428 .493

875 397 .454

 

 

 

101

The retention time, R3, of a compound can differ from day to day
as a result of small fluctuations in separation conditions such as
temperature and flow rate. Over an extended length of time, these
variations may arise from the deterioration of the analytical column.
Regardless of the source, this change in R2 causes a different portion of
the chromatogram to be isolated when time-based cuts are employed.
This results in a variability of the peak height and area of the recycled
peak. The advantage of detector-based heartcuts over time-based cute
is evident. The variations in retention times do not affect the detector-
based cuts as they do a time-based system. This is reflected by the
XRSD values that are listed in Table 4.1 which shows better precision
for detector-based cuts. Even though detector-based cuts are

preferred, certain applications will require the use of time-based cuts.
2. Band Broadening in the 12110 System

Band broadening is a crucial factor in RHC. Improvements in
resolution may be lost if significant amounts of dispersion are
introduced in recycling the heartcut. Band broadening can originate
from components of the HPLC system such as the injection valve, tubing,
column, detector and any of the numerous connectors.

In this section, the dispersion that occurs in the tubing used as the
heartcut loop is described. The peak widths at half height of the

recycled peak are compared when tubing of different dimensions are
used. The feasibility of using a heartcut loop of larger volume than

needed and of partially filling it is discussed.

102

A 450 pl! naphthalene solution was used in these studies. The
mobile phase was 100% methanol and the flow rate employed was

1.0 ml/min. Time-based heartcuts were utilized throughout.

a. Heartcut Loop Tubing Dimensions

The heartcut loops were made from stainless steel tubes of various
lengths and internal diameters (i.d.). A cut of the eluting naphthalene
peak was diverted into the heartcut loop. This cut was then recycled
through the column, and the peak width at half height, wm, measured.
The values for win for both the first and second cycle through the
column are summarized in Table 4.2. Note that tripling the length of the
0.01" i.d. tube (cases 1 and 3) produces a 15% increase in HUI whereas
tripling the length of the 0.02" i.d. tube (cases 4 and 5) produces a 65%
increase. Equal volume cuts with the two different i.d. tubes (cases 2
and 4) yield approximately equal win values. This leads to the
conclusion that long lengths of 0.02" i.d. tubing should not be used as
the heartcut loop. This is confirmed when the win values of the
recycled peaks are compared for the cases in which long lengths of the
two tubes are used (cases 3 and 5). The 0.02" i.d. tube yields a
recycled peak width that is 60% higher than that obtained for the 0.01"
i.d. tubing. Therefore, if large volume cuts need to be made, the 0.01"
tubing should be employed.

The trends observed between peak width and tubing dimensions

are as expected. The degree of band broadening in tubing is directly

103

Tabls4.2.Effectoftubingdi-ensbnsonthepeskwidthathalf
heighgwm, ofarecycledpssk.

 

I _—T “J
i r

 

 

 

 

Heartcut Tubing Dimensions wm (i 0.03 cm)
Length i.d. Volume Cycle 1 Cycle 2
(cm) (in) (m1) (cm) (cm)
60 0.01 34.0 2.45 2.20
72 0.01 38.8 2.45 2.19
183 0.01 92.6 2.45 2.49
18 0.02 36.8 2.45 2.32

60 0.02 121.6 2.45 3.93

 

 

104

proportional to the tube length and proportional to the fourth power of
the tube radius as expressed in the equation derived by Scott and

Kucera (199).

6.53 = 11 r‘ 1 Equation 4.1

nm.....m.—nh....—...

The symbols in the above equation denote the following:

up is the standard deviation of the Gaussian peak that
represents band broadening due to the tubing (8),

r is the internal radius of the tubing (cm),
1 is the length of the tubing (cm),

D. is the diffusitivity of the solute in the mobile
phase (cma/s) and

Q is the flow rate of the mobile phase (cm3/s).

A second expression, Equation 4.2, can be derived to calculate the
relative amount of band broadening introduced by heartcut loops made
from tubes of different lengths and diameters. This equation assumes

identical separation conditions for the cases being compared.

(£1923 ...... = {£3.23 ...... 13 .. Equation 4-2
The calculated and experimental values for band broadening are
compared in Table 4.3. The calculated values were higher than those
obtained experimentally. A possible explanation for this observation lies
in the necessary coiling of the long lengths of tubing used as the
heartcut loop. In continuous flow analysis systems, coiling a piece of
tubing will result in a smaller amount of dispersion than that observed

for the same length of straight tubing (87).

105

Tabls4.3.Compsrisonofthecslculstsdandexperi-sntslbsnd
brosdeningintubingdvsrisusdi-nsions.

 

t -' -—
I L J

 

 

Hesrtcut Tubing Dimensions % Band Brosdeningt
Length i.d. Calculated Experimental

(cm) (in)

60 0.01 0 0
72 0.01 110 0
183 0.01 170 113
18 0.02 220 106
60 0.02 400 179

 

 

_
‘—

8 calculated relative to values obtained for the case of
60 cm x 0.01" i.d.

106

b. Partial Filling of the Heartcut Loop

With the configuration described above, the heartcut volume is
fixed and determined by the dimensions of the tubing used as the
heartcut loop. One method that would allow different heartcut volumes
and increase the flexibility of the system is to use a large volume
heartcut loop which is only partially filled. Studies were performed to
determine whether this approach would be analytically useful. Two
pieces of tubing, 0.01" i.d. x 60 cm and 0.02" i.d. x 60 cm, were used as
heartcut loops. A 450 uM naphthalene solution was injected and cuts of
different sizes (Figures 4.6 and 4.7) were taken so as to fill the loop to
various extents. The peaks resulting from recycling the partial loop
fills of two heartcut loops are shown in Figures 4.8 and 4.9. The same
behaviour was observed for both the 0.01" and 0.02" i.d. loops. First,
the peak maximum occurs at a progressively longer time as the
percentage of the loop volume that is used, Vmp, decreases. In
partially filling the heartcut loop, the sample displaces the mobile phase
present in the loop beginning with the section that is closest to the
column outlet. The remaining portion of the loop that is not occupied
by sample (i.e., the side attached to the column inlet) remains filled with
mobile phase. The distance between the center of the heartcut sample
plug and the column inlet increases when the percentage of the loop
volume that is used, V1.69, decreases. This causes the small offset in
retention times that is observed (Figures 4.8 and 4.9) when the
heartcuts of different Vloop values are recycled. Second, a
disproportionality is observed between peak height and Vmp, i.e., the
peak height for the 50% loop fill is not half that of the 100% loop fill.

The reason for this becomes clear when Figures 4.6 and 4.7 are

107

0.20
F

0.10 -

Absorbance

0.05 -

 

 

175.0 195.0 215.0 235.0
Thne (s)

Figure 4.6. Heartcuts taken to fill a 60 cm x 0.01" i.d. loop to various
extents - 100% loop fill (--), 75% loop fill (~---), 50% loop fill (---)

and 25% loop fill (-—). 450 1111 naphthalene with a 100% 08803 mobile
phase and a 1.0 ml/min flow rate. Detection is at 254 nm and 0.16 AU

full scale.

108

0.20 r-

0.15 -

 

0.10 -

Absorbance

0.05 -

 

 

 

ssess-eeaeessesI-sss- essesseseessIssslslease-es-susssseeseseeeseeslessee-ease
------------------

 

 

0.00 -
175.0

 

 

l 1 1

215.0 235.0
Thne (s)

 

Figure 4.7. Heartcuts taken to fill a 60 cm x 0.02" i.d. loop to various
extents - 100% loop fill (---), 75% loop fill (um), 50% loop fill (---)

and 25% loop fill (—). 450 mM naphthalene with a 100% 03808 mobile
phase and a 1.0 Isl/min flow rate. Detection is at 254 nm and 0.18 All

full scale.

109

 

 

 

 

 

 

0.05 , , ,
i L
' XV
0.04 h 1 - 10.0,.” ‘
. 2 2 - 75 *
3 3 - 50.
g 0.03 " 4 - 25 .‘ _
.0 .. 3
3
h 0.02 -
'9 .
V
0.01 " . 4
0000 :Zx'r’-'“..“- I l , \\"‘..’ " ‘A‘- -‘ W;-‘_«_‘ 1%-
440.0 460.0 480.0 500.0
Time (s)

Figure 4.8. Chromatograms of the recycled peaks that resulted from the
partial filling of the heartcut loop ( 60cm x 0.01" i.d.).

110

 

007 1 . .

006

0.05

0.04

0.03

Absorbance

 

 

 

 

Figure 4.9. Chromatograms of the recycled peaks that resulted from the
partial filling of the heartcut loop ( 60cm x 0.02" i.d.).

111

reexamined. The cuts were taken on the slope of the peak. Thus, the
amount of sample within each heartcut is not proportional to the volume
of the cut. Third, the 181/: values of peaks that result from recycling
heartcuts of different volumes that partially filled a tube are
approximately equal for each tube. The ma values of the recycled
peaks in the cases where tubes of different i.d.s are filled to various
extents are summarized in Table 4.4. In the case of no sample
dispersion, the expected result is a decreasing wm value from

100% V1... to 25% Vleep for each piece of tubing. The fact that the
values for win listed within each set of peaks are not significantly
different indicates that dispersion is taking place. Work needs to be
done to determine whether this dispersion is occurring predominantly in
the heartcut loop or in other parts of the separation system. Despite
the dispersion, the plots show that partial loop fills are a feasible

alternative.

D. Applications

Preliminary experimental work on the application of RHC to various
problems has been carried out. The following discussion illustrates the
experimental approach used in RHC and shows the potential of the

technique.

112

Table 4.4. Values d peak width at half height, win, of
recycledheartcutsthatpartiallyfillsdtubescf
different i.d.s to varying extents.

‘j:
J

 

V1... (WI/3)“! (wmlm
(X) (cm) (cm)
100 1.29 2.28
75 1.29 2.25
50 1.27 2.20

25 1.27 2.20

 

 

113

1. Component(s) Isolation

There are numerous instances in which the purpose of the
analysis is to verify the presence of specific compounds and to provide
quantitative information about them. Examples of such cases are the
testing of body fluids for specific drug metabolites and the analysis of.
body fluids and tissue for certain chemicals that are produced by the
body as a sign of disease. In both cases, the sample matrices are very
complex. One way to address this type of analysis is to extensively
pretrsat the sample with hopes of removing all possible interferences
before subjecting the sample to a final analytical separation. The
complication often encountered is the presence of interferences which
have not been removed in the sample pretreatment step because of their
similar chemical and physical characteristics to the compounds of
interest. An intricate and time consuming separation scheme may have
to be used to circumvent this problem. An alternative method of
analysis is to continuously isolate and separate the compounds of
interest from the matrix. This can be accomplished with recycle
heartcut chromatography. Only minimal amounts of pretreatment are
needed prior to injection into the HPLC system. The heartcut step
provides any further fraction isolation that may be necessary. The
recycling step should provide an increase in the number of theoretical
plates and thus, improved resolution.

The analysis of a urine sample for the steroid, 1,4-androstadisns-
3,7-dions, is illustrated in Figure 4.10. The term, cycle 1, is used when
referring to the first pass through the column and cycle 2, when

referring to the second pass. The portion of the chromatogram

114

0.072 -1

0054' CYCLE I

0.056‘

0.048‘

0.040“

 

ABSORBANCE

0932‘ * CYCLE 2

0.024‘

0.016‘

   

 

 

0.008“

 

 

 

 

0 2.0 410 6T0 810 10.0 12.0 14.0
TIME (min)

Figure 4.10. Isolation of 1,4-androstadiene-3,17-dione from a
spiked urine sample. Separation conditions: mobile phase is
85/15 CHaOH/HaO: flow rate is 1.0 ml/min. Heartcut is time-based
and for a volume of 100 pl.

115

corresponding to cycle 1, shows the elution of a number of compounds,
some of which are unresolved. The heartcut that was taken is denoted
by the dotted lines shown in Figure 4.10. The position of this heartcut
was based on the retention time of a previously injected steroid
standard. The chromatogram for cycle 2 shows a baseline disturbance
due to the solvent front at 12 min and the isolated steroid component
eluting at 13 min 10 s (indicated by arrow). The single peak obtained
does not show any indication of other components being present. In
this example, a minimum amount of sample pretreatment (e.g. the urine
was passed through a Sep-Pak cartridge as described in the
experimental section) and an isocratic separation (85/15 CHaOH/H20)
were employed in the isolation of a specific compound present in a
complex matrix. These are some of the advantages that are offered by
RHC.

The technique of recycle heartcut chromatography also has many
potential uses in the area of preparative liquid chromatography. The
analytical column can be replaced by a preparative LC column with only
minor modifications to the separation system. Since such a column was
not available, an analytical column was used instead in the demonstration
of potential applications in this area. RHC can be used either to remove
a contaminant present in a final product or to isolate a desired
component. An example of the latter case is demonstrated in Figure
4.11. A mixture of polynuclesr aromatic hydrocarbons is partially
separated as shown in the cycle 1 portion of the chromatogram. .' The
desired compound is naphthalene which corresponds to the peak eluted
at 210 s. A heartcut was taken of this peak as denoted by the lines

drawn in cycle 1 and the elution of the recycled sample is shown in

116

 

 

 

 

 

 

 

I l f r
0.7 -
0 Cycle 1 1

0.00 "' fi‘ -
O 1.
o 0.50 _
§

0.00 I. —1
e Cycle 2
8 0.30 b _
Q
‘ 0.20 +- ..

0.10 1- ..

 

   
   
 

 

 

 

 

0.00 1 - ._
1 20.0 220.0 320.0 420.0 520.0
Tim 9 (s)

Figure 4.11. Isolation of naphthalene from a mixture of polynuclesr
aromatic hydrocarbons consisting of naphthalene, biphenyl and
phenanthrene. Separation conditions: mobile phase 90/ 10 ethos/11.0;
flow rate is 1.0 m1/min.

117

cycle 2. The heartcut step has been used, in this example, as a
fractionation step to expedite the isolation of a single component that is
surrounded by closely eluting impurities. Note that the column has
been somewhat overloaded to demonstrate a practice commonly employed
in preparative work to maximize the amount of purified sample per unit

time (especially when the desired compound is a minor component).

2. Verification of Peak Purity

The potential of employing RHC as a verification technique fer
chemicals listed as ‘chromatographically pure’ is demonstrated in Figure
4.12. The chemical system that was employed is a model mixture of
biphenyl and naphthalene. The chromatogram that results from the first
pass through the column, Figure 4.12s, shows a peak with a shoulder.

A 430 pl cut is made, as shown by the dotted lines, and recycled
through the column to obtain the chromatogram shown in Figure 4.12b.
This step is repeated and the chromatogram that resulted from this
second recycle is illustrated in Figure 4.12c. With the second recycle, it
is clear that the suspect peak consisted of more than one component.
The resolution of the two sample bands can be estimated via a visual'
comparison with the set of standard resolution curves for different
band-size ratios that have been introduced by Snyder (200). A small
increase in resolution (0.5 to 0.6) is observed after two recycles. The
final resolution is lower than the expected value of 0.85. (Resolution, as
defined by Equation 1.1, is proportional to the square root of the
number of theoretical plates, N. If the sample is cycled through the

column three times, there would be a three fold increase in N, assuming

4590.33
5 2 .12 s 56 use 0:035:93. 5 I 73 u m4 3 e—naem .n 3050 «.0
use N 0350 A;— — e—uho fie 32:.— xeen no 90335.33, .Nui ensur—

118

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

«a» 65.:
g. .80 .8. .08 g .005 .85
d a d d d
. eQe
V.
a.
s
e
- 86 W
a
u
a
s
1 «9e
b p p 8.0 A e U
«a» 2:2 «my 2:2
.85 98 93 .93 69" as. 934 .98. 69" as» gas .9x .93
q q q 11 4 d E 1 d d
u . 1
I 1 8.0 1 8.0
1 . m L 86
T 1.ee e a
1 w 1 86
A e
w «90 w
1 0 1 no.0
r 1.66 149°
F b h h F 36° A O a b b b h! b 80°

sausqaesqy

Tm

119

that N is additive. Resolution should then increase by a factor of (3)1”
or 1.7.). This smaller gain in resolution can be attributed to void
volumes in the RHC system or the use of poorly designed HPLC
components that cause mixing of the partially separated sample. In the
worst case, any separation that has been achieved by the first pass
through the column is lost. The sample that is recycled is then
equivalent to the homogeneous sample plug that was initially injected;
thus no improvement in resolution will be observed with each recycle.
Further studies need to be done to locate the major sources of

dispersion in the REC system.

3. Quantitative RHC

The feasibilty of employing RHC as a quantitative technique was
investigated by plotting calibration curves for naphthalene standards
that had undergone a series of two heartcuts. The two successive
heartcuts were taken at 10% detector full scale and were 234 pl in
volume. Figure 4.13 demonstrates the calibration curves obtained when
the absorbsnce at maximum peak height for each of the three cycles and
each standard was plotted. Linearity was observed for all three cases.
The statistics related to each calibration curve are summarized in Table
4.5. A problem that is inherent to the technique of RHC becomes
evident with this data. With each recycle, the sample becomes more
diluted by the mobile phase and with each heartcut, a smaller portion of
the peak is taken. This results in a decrease in the slope of the

calibration curves and thus, a decrease in sensitivity from cycle 1 to

120

 

 

 

 

_ O — Cycle 1
A — Cycle 2
0.8 _ Cl — Cycle 3 _
r
8
c 0.6 +- -
D
Q .
g
8
Q 0.4 "' "‘
V
0.2 r . -
0.0 1 . ll 1‘ l 1 L 1
150.0 250.0 350.0 . 450.0 550.0

Concentration (,aM)

Figure 4.13. Calibration curves for naphthalene standards of which
two heartcuts have been taken and recycled.

121

Table 4.5. linear regressbn analysis of calibration curves for
quantitative RHC. -

 

 

 

 

 

 

Cycle 1 Cycle 2 Cycle 3
Slope 1.7 x 10" 1.0 x 10" 6.9 x 10"
Std. error 0.0001 0.0003 0.0004
of estimate
Correlation .9992 .9914 .9822
coefficient

y-intercept -0.002 -0.016 -0.005

122

cycle 3. The observed increase in the standard error of the estimate is

a manifestation of the decrease in S/N ratio with each heartcut.

E. Conclusions

An automated HPLC system which integrates a heartcut step for
fraction isolation with a recycling step for increasing the separation
efficiency has been described. The performance of this recycle heartcut
chromatography system has been evaluated with respect to the precision
of the various methods that can be used to initiate a heartcut. The
method of partially filling a large heartcut loop to provide a more
flexible system was examined and determined to be a viable alternative.
The versatility and advantages of RHC were demonstrated with a number
of general applications. First, in the analysis of complex mixtures for
specific compounds, it was shown that an extensive pretreatment of the
sample prior to introduction to the HPLC was not necessary. Nor was it
essential to employ an intricate separation scheme. Second, the fraction
isolation capabilities of RHC make its extension to preparative LC a
logical move. This system would be capable of isolating fractions of
very high purity. The gain in purity with each heartcut will be offset
by the loss of sample. A balance between the two factors will have to
be established for each specific case. Third, RHC can be used as a
quantitative technique.

In addition to the advantages mentioned above, the RHC system is
simple and easy to implement as most of the components needed are
standard accessories used in liquid chromatography. The automated

features are desirable, but not essential for all applications. The cost

123

of an RHC system is minimal as only one pump, one column, one detector
with a low pressure flow cell and two valves are needed. (The first
three components are readily available in most laboratories that
routinely utilize liquid chromatography.) It should also be mentioned
that even though RHC has been discussed in this work as pertaining to
HPLC, this valve configuration may be extended with ease to other liquid
chromatographic techniques.

The RHC system has its limitations, a major one being the loss of
sensitivity with each recycle, due to a smaller fraction of the initial
sample plug being used and dilution by the mobile phase. This problem
could be somewhat alleviated with the use of gradient elution. A
gradient can be employed either in the last recycle to sharpen the final
peaks or in each recycle to compress the broadened peaks back to the
same volume as the peak that resulted from the first pass through the
column. Using this approach, there should be a minimal ‘loss of the
desired component with each recycled heartcut. A second problem that
has been encountered is band broadening in the system. As a result,
the expected gain in resolution has not been achieved. As mentioned
earlier, work needs to be done to determine the major source of
dispersion and if possible, to remedy it. In this chapter, some
preliminary results obtained using recycle heartcut chromatography have
been presented. Further work also needs to be done to characterize
and investigate the advantages and limitations of the method. In
particular, the trade-off between resolution enhancement and decreased

S/N ratios should be studied.

CHAPTER V

FUTURE PROSPECTIVES

A. OVERVIEW

The recent developments in post-column reaction detectors have
enabled the use of this type of detector in conjunction with a variety of
different techniques (e.g., miniaturization of PCR systems which has led
to the first work with microbore columns (201)). There is still room for
growth, both in the applications area and the development of improved
manifold components and instrumentation. The preceding statement also

applies to the area of recycle heartcut chromatography.
B. POST-COLUMN REACTION DETECTORS
1. Tandem Analyses

Even though most of the reactions that have been used with PCR
detectors are very selective, often more than one class of compounds is
able to react. The reaction conditions can be used to favor the reaction
of one class of compounds over another, thus providing an additional
degree of selectivity. It is Conceivable to develop a detector system
that would employ a change of reaction conditions after the first set of

compounds has been detected to favor reaction with a second set of
124

125

compounds. In this way, a tandem analysis of different compounds
could be carried out. A specific example is the formation of azo dyes
which, for phenols, take place in basic media, but for anilines, under
acidic conditions. If the reaction medium was initially alkaline, phenols
would undergo the coupling reaction with the diazonium reagent. After
the absorbance of the phenol derivative formed is measured, acid is
added to the flow stream and aniline would be able to react with the
diazonium reagent. The absorbance of the aniline derivatives would
then be measured on a second detector. It should be noted that in a

tandem analysis, different detection principles could be employed.

2. Solvent Segmen tation

The use of solvent segmentation as the post-column reaction
system provides a powerful and flexible detector. As discussed earlier,
advantages of segmentation with solvents over that. with air include
reduced sample interaction and background noise. The use of the
solvent segment as an extraction medium imparts an additional degree of
selectivity to the reaction detector. The possibility of choosing an
optimum solvent with respect to viscosity, solubility and spectral
characteristics is another factor that makes this system very versatile.

In order for this system to reach its full potential, the behaviour
of sample and solvent molecules and the flow dynamics in a solvent
segmentation have to be well understood. There is a need for such
investigations. The extent to which the band broadening equations that
have been developed for air segmentation can be used in predicting the

dispersion in a solvent segmented system remains to be determined.

126

Also needed, is the derivation of equations that can be used to calculate
the extraction efficiency between solvent segments in a flow stream.
This should then lead to a study of the effects of ceiling on dispersion

and extraction efficiencies.
3. Biological Detectors

Solid phase reactors have recently been introduced in PCR
systems. This was a natural progression because of the existing
methodology concerning immobilization of reagents onto surfaces. In
particular, the techniques for enzyme immobilization have been well
researched (202,203). Enzymes are among the most selective reagents
known. This degree of selectivity may be achieved through the use of
reactors containing materials that are coated with biological reagents
such as antibodies and proteins. These specific reaction detectors have
many potential applications, especially with the present interest in
biotechnology and the increasing number of syntheses of biological

substances that need to be separated and detected.
C. RECYCLE HEARTCUT CHROMATOGRAPHY

Future research in recycle heartcut chromatography can take
place in many areas. One possibility is the use of the commercially
available multiport, multiposition valves for analyses that require
multiple heartcuts to be taken. This system could also be employed in a
similar fashion to boxcar chromatography. Samples could be

continuously injected and the fraction of interest isolated and deposited

127

on the different heartcut loops. When all the heartcut loops are filled,
sample injection is put on hold, and the fractions on the heartcut loops
are recycled. The emptying out of each loop is spaced by a short time
interval. The result would be a train of sample peaks, similar to that
observed in boxcar chromatography. A second area would be in
reducing dispersion in the heartcut loop. Long lengths of tubing are
not recommended in HPLC because of the band broadening due to
laminar flow. In flow injection analysis systems, single bead string
reactors have been employed to break up the parabolic flow profile.
The use of an equivalent system (to a single bead string reactor) in
recycle heartcut chromatography could reduce the band broadening that
occurs in the heartcut loops. Finally, the nature of the dispersion and
diffusion in the heartcut loop require further investigation as that the
full potential of recycle heartcut chromatography can be evaluated.
Presently, computer simulations are being done (204) to clarify this

aspect.

LIST OF REFERENCES

8.

9.

10.

11.

12.
13.

14.

15.

16.
17.
18.

19.

LIST OF REFERENCES

K. Blau and 6.8. King, "Handbook of Derivatives for
Chromatography", Heyden, London (1977).

C.F. Poole and A. Zlatkis, J. Chromatogr. Sci., 17, 115 (1979).

J.F. Lawrence, in "Chemical Derivatization in Analytical Chemistry",
Vol. 2, R.W. Frei and J.F. Lawrence Eds., Plenum Press, New York,
1982, Ch. 5.

R.W. Frei, J. Chromatogr. Rev., 165, 75 (1979).

S. Siggia and LG. Hanna, "Quantitative Organic Analysis via
Functional Groups", John Wiley 8. Sons, Inc., New York, 1973.

K.A. Connors, "Reaction Mechanisms in Organic Analytical
Chemistry", John Wiley & Sons, Inc., New York, 1973.

N.D. Cheronis, J.B. Entrikin and E.M. Hodnett, "Semimicro Qualitative
Organic Chemistry", Robert E. Krieger Publishing Company, New
York, Inc., Florida, 1983. .

A.W. Wolkoff and R.H. Larose, J. Chromatogr., 99, 731 (1974).

M. Frei-Hausler, R.W. Frei and O. Hutzinger, J. Chromatogr., 84, 214
(1973).

W.P. King and P.T. Kissinger, Clin. Chem., 26, 1484 (1980).

L.R. Snyder and J.J. Kirkland, "Introduction to Modern Liquid
Chromatography", John Wiley 8 Sons, Inc., New York, 1979, p. 36.

ibid., p. 50.
ibid., p. 170.

L.R. Snyder, J. Levine, R. Stay and A. Conetta, Anal. Chem., 48,
942A (1976).

W.B. Furman, Ed., "Continuous Flow Analysis, Theory and Practice",
Marcel Dekker, New York, 1976.

L.R. Snyder, Anal. Chim. Acta, 114, 3 (1980).
C.J. Patton, Ph.D. Thesis, Michigan State University, 1982.
H.A. Mottola, Anal. Chem., 53, 1313A (1981).

L.T. Skeggs, Jr., Clin. Chem., 2, 241 (1956).

128

20.

21.

22.

23.

24.
250
26.

27.

28.

29.

30.

31.
32.
33.
34.
35.
36.
37.

38.

39.

40.

41.

129

LeTe Skeg‘sp Jre, Ame Jo Cline P801101" 28, 31]. (1957’s

K.K. Stewart, G.R. Beecher and P.E. Hare, Fed. Proc., 33, 1439
(1974).

J. Ruzicka and E.H. Hansen, Anal. Chim. Acta, 78, 145 (1975).

J. Ruzicka and E.H. Hansen, "Flow Injection Analysis", John Wiley 8:.
Sons, New York, 1981 and references therein.

D. Betteridge, Anal. Chem., 50, 832A (1978).
J.C. Ranger, Anal. Chem., 53, 20A (1981).
J. Ruzicka and E.H. Hansen, Anal. Chim. Acta, 114, 19 (1980).

J.T. Dobbins, Jr., and J.M. Martin, Spectroscopy, September, 20
(1986).

B. Karlberg, in "Chemical Derivatization in Analytical Chemistry",
Vol. 2, R.W. Frei and J.F. Lawrence, Eds., Plenum Press, New York,
1982, Ch. 1.

M. Margoshes, Anal. Chem., 49, 17 (1977).

J. Ruzicka, E.H. Hansen, M. Mosbaek and F.J. Krug, Anal. Chem., 49,
1858 (1977).

M. Margoshes, Anal. Chem., 54, 678A (1982).

B. Rocks and C. Riley, Clin. Chem., 28, 409 (1982).

C.J. Patton and S.R. Crouch, Anal. Chim. Acta, 179, 189 (1986).
C.J. Patton, Ph.D. Thesis, Michigan State University, 1982, p.36.
L.R. Snyder and H.J. Adler, Anal. Chem., 48, 1017 (1976).

L.R. Snyder and H.J. Adler, Anal. Chem., 48, 1022 (1976).

L. Nord and B. Karlberg, Anal. Chim. Acta, 164, 233 (1984).

L.R. Snyder, in "Advances in Automated Analysis, Technicon
International Congress 1976", Vol. 1, Mediad, Inc., New York, p. 76.

R.L. Habig, B.W. Schlein, L. Walters and R.E. Thiers, Clin. Chem., 15,
1045 (1969).

W.E. Neeley, S. Wardlaw and M.E.T. Swinnen, Clin. Chem., 20, 78
(1974).

W.E. Neeley, S.C. Wardlaw, T. Yates, W.G. Hollingsworth and M.E.T.
Swinnen, Clin. Chem., 22, 227 (1976).

43.
44.

45.

46.
47.
48.
49.
50.
51.
52.

53.

54.

55.

56.

57.

58.
59.
60.
61.
62.

63.

64.

65.

130

C.J. Patton, M. Rabb and S.R. Crouch, Anal. Chem., 54, 1113 (1982).

R.W. Frei, H. Jansen and U.A.Th. Brinkman, Anal. Chem., 57, 1529A
(1985).

H. Jansen, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr. Sci., 23,
279 (1985).

R.W. Frei, J. Chromatogr., 165, 75 (1979).

R.W. Frei, L. Michel and W. Santi, J. Chromatogr., 142, 261 (1977).
L.R. Snyder, J. Chromatogr., 125, 287 (1976).

R.W. Frei and W. Santi, 2. Anal. Chem., 277, 303 (1975).

J.T. Stewart, TrAC, 1 (7), 170 (1982).

LS. Krull and E.P. Lankmeyer, Am. st., May, 18 (1982).

R.W. Frei, Chromatographia, 15, 161 (1982).

R.S. Deelder, A.T.J.M. Kuijpers and J.H.M. van den Berg,
J. Chromatogr., 255, 545 (1983).

R.W. Frei and A.H.M.T. Scholten, J. Chromatogr. Sci., 17, 152 (1979).

LS. Krull, Ed., "Reaction Detection in Liquid Chromatography",
Marcel Dekker, New York, 1986.

R.W. Frei and J .F. Lawrence, "Chemical Derivatization in Analytical
Chemistry", Vol. 1, Plenum Press, New York, 1981.

R.W. Frei and J .F. Lawrence, "Chemical Derivatization in Analytical
Chemistry", Vol. 2, Plenum Press, New York, 1982.

T. Jupille, J. Chromatogr. Sci., 17, 160 (1979).

Kratos Analytical Instruments, Ramsey, New Jersey.
Bibliography Service, Technicon, Inc., Tarrytown, New York.

S. Katz, W.W. Pitt, Jr. and G. Jones, Clin. Chem., 19, 817 (1973).
S. Katz and W.W. Pitt, Jr., Anal. Lett., 5, 177 (1972).

D. Westerlung, J. Carlqvist and A. Theordorsen, Acta Pharm.
Svecics, 16, 187 (1979).

H.A. Moye and T.W. Wade, Anal. Lett., 9, 801 (1976).

R.T. Krause, J. Chromatogr., 185, 615 (1979).

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.
79.
80.

81.

82.

83.

84.

85.

131

W.Th. Kok, U.A.Th. Brinkman and R.W. Frei, Anal. Chim. Acts, 162, 19
(1984).

N. Watanabe and M. Inoue, Anal. Chem., 55, 1016 (1983).

L. Dalgaard, L. Nordholm and L. Brimer, J. Chromatogr., 303, 67
(1984).

L. Ogren, I. Csiky, L. Risinger, L.G. Nilsson and G. Johansson, Anal.
Chim. Acta, 117, 71 (1980).

C.M. Selavka, LS. Krull and LS. Lurie, Forensic Sci. Int., 31, 103
(1986).

T.D. Schlabach, S.R. Chang, K.M. Gooding and P.E. Regnier,
J. Chromatogr., 134, 91 (1977).

S. Ahuja, J. Chromatogr. Sci., 17, 168 (1979).

W.P. Cochrane, in "Chemical Derivatization in Analytical Chemistry",
Vol. 1, R.W. Frei and J.F. Lawrence, Eds., Plenum Press, New York,
1981, Ch. 1.

B.D. Page and H.B.S. Conacher, in "Chemical Derivatization in
Analytical Chemistry", Vol. 2, R.W. Frei and J.F. Lawrence, Eds.,
Plenum Press, New York, 1982, Ch. 6.

R.M. Cassidy and S. Elchuk, J. Liq. Chromatogr., 4(3), 379 (1981).

P.T. Kissinger, K. Bratin, G.C. Davis and L.A. Pachla, J. Chromatogr.
Sci., 17, 137 (1979).

LS. Krull, C.M. Selavka, C. Duda and W. Jacobs, J. Liq. Chromatogr.,
8(15), 2845 (1985).

J.F. Lawrence, J. Chromatogr. Sci., 17, 147 (1979).
R. Reh and G. Schwedt, Chromatographia, 14(4), 249 (1981).
R. Reh and G. Schwedt, Chromatographia, 14(5), 317(1981).

R.W. Frei and J.F. Lawrence, "Chemical Derivatization in Analytical
Chemistry", Vol. 1, Plenum Press, New York, 1981, p. 294.

J.F.K. Huber, K.M. Jonker and H. Poppe, Anal. Chem., 52, 2 (1980).

A.H.M.T. Scholten, U.A.Th. Brinkman and R.W. Frei, Anal. Chem., 54,
1932 (1982).

Carr and L. Bowers, "Immobilized Enzymes in Analytical and Clinical
Chemistry, John Wiley & Sons, New York, 1980.

M. Uihlen and E. Schwab, Chromatographis, 15, 140 (1982).

86.
87.
88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

100.

101.

102.
103.

104.

132

H. Engelhardt and U.D. Neue, Chromatographis, 15, 403 (1982).
R. Tijsssn, Anal. Chim. Acta, 114, 71 (1980).

R.S. Deelder, M.G.F. Kroll, A.J.B. Beeren and J.H.M. van den Berg,
J. Chromatogr., 149, 669 (1978).

R.S. Deelder, M.G.F. Kroll and J.H.M. van den Berg, J. Chromatogr.,
125, 307 (1976).

T.D. Schlabach, S.H. Chang, K.M. Goodring and T.E. Regnier,
J. Chromatogr., 134, 91 (1977).

K.M. Jonker, H. Poppe and J.F.K. Huber, Z. Anal. Chem., 279, 154
(1976).

K.M. Jonker, H. Poppe and J.F.K. Huber, Chromatographia, 11, 123
(1978).

J .W. Hilby, in "Proceedings of Symposium on Interaction Between
Fluids and Particles", P.A. Rottenburg, Ed., Institution of Chemical
Engineers, London, 1962, p. 312.

A.H.M.T. Scholten, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr.,
205, 229 (1981).

J.F. Lawrence, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr., 171,
73 (1979).

J.F. Lawrence, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr., 185,
473 (1979).

C.E. Werkhoven, U.A.Th. Brinkman and R.W. Frei, Anal. Chim. Acta,
114, 147 (1980).

R.J. Reddingius, G.J. DeJong, U.A.Th. Brinkman and R.W. Frei,
J. Chromatogr., 205, 77 (1981).

F. Smedes, J.C. Kraak, C.E. Werkhdven—Goewie, U.A.Th. Brinkman and
R.W. Frei, J. Chromatogr., 247, 123 (1982).

J.A. Apffel, U.A.Th. Brinkman and R.W. Frei, Chromatographia, 18, 5
( 1984).

J.A. Apffel, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr., 312, 153
(1984).

U.A.Th. Brinkman and F.A. Maris, TrAC, 4, 55 (1985).
K. Zach and W. Voelter, Chromatographia, 8, 350 (1979).

E. Oelrich and D. Theuerkauf, J. High. Res. and Chromatogr.
Commun., 2, 256 (1979).

105.
106.
107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.
121.
122.
123.
124.

125.

133

R.W. Frei, L. Michel and W. Santi, J. Chromatogr., 126, 655 (1976).
J.F. Studebaker, J. Chromatogr., 185, 497 (1979).
K.H. Xie, S. Colgan and LS. Krull, J. Liq. Chromatogr., 6, 125 (1983).

L. Nondek, R.W. Frei and U.A.Th. Brinkman, J. Chromatogr., 282, 141
(1983).

H. Jansen, R.W. Frei, U.A.Th. Brinkman, R.S. Deelder and R.P.J.
Snelling, J. Chromatogr., 325, 255 (1985).

H. Jansen, U.A.Th. Brinkman and R.W. Frei, Chromatographia, 20, 453
(1985).

W.Th. Kok, G. Grosnendijk. U.A.Th. Brinkman and R.W. Frei,
J. Chromatogr., 315, 271 (1984).

M.F. Lefevere, R.W. Frei, A.H.M.T. Scholten and U.A.Th. Brinkman,
Chromatographia, 15, 459 (1982).

M.S. Gandelman, J.W. Birks, U.A.Th. Brinkman and R.W. Frei,
J. Chromatogr., 282, 193 (1983).

J.N. LePage and E.M. Rocha, Anal. Chem., 55, 136011983).

S. Veibel, "The Determination of Hydroxyl Groups", Academic Press,
New York, 1972.

M. Pesez, P. Poirier and J. Bartos, "Practique de L’Analyse
Organique Colorimetrique", Masson st Cie, Paris, 1966.

D.R. Knapp, "Handbook of Analytical Derivatization Reactions", John
Wiley 8. Sons, New York, 1979.

T. Hanai, Ed., "Phenols and Organic Acids", Vol. 1, CRC Press, Inc.,
Boca Raton, Florida, 1982.

P. Koppe, F. Dietz, J. Traud and Ch. Rubelt, Z. Anal..Chem., 285, 1’
(1977).

W.L. Koltun, J. Am. Chem Soc., 79, 5681 (1957).

L. Wofsy, H. Metzger and SJ. Singer, Biochemistry, 1, 103 (1962).
R.L. Hossfeld, J. Am. Chem. Soc., 73, 852 (1951).

R.L. Hossfeld, J. Am. Chem. Soc., 74, 5766 (1952).

J.J. Fox and J.B. Gauge, J. Chem. Ind., 39, 306T (1920-22).

L.R. Whitlock, S. Siggia and J.E. Smola, Anal. Chem., 44, 532 (1972).

126.
127.

128.

129.
130.
131.

132.

133.

134.
135.
136.

137 .
138.
139.
140.

141.
142.

143.

144.
145.

134

C. Baiocchi, M.C. Gennaro, E. Campi, E. Mentasti and K. Aruga, Anal.
Lett., 15(A19), 1539 (1982).

S. Patai, Ed., "The Chemistry of Diazonium and Diazo Groups", John
Wiley a. Sons, New York, 1978.

LM. Kolthoff and P.J. Elving, " Analytical Chemistry of Inorganic
and Organic Compounds: Treatise on Analytical Chemistry, Part II",
Vol. 15, John Wiley & Sons, New York, 1976.

K. Schank in "Methodicum Chimicum", Vol. 6, F. Zymalkowski, Ed.,
Academic Press, New York, 1975, Ch. 7.

H. Zollinger, "Azo and Diazo Chemistry", Interscience Publishers,
Inc., New York, 1961.

L.P. Hammett, "Physical Organic Chemistry", McGraw Hill, New York,
1940.

D.N. Kramer and L.U. Tolentino, Anal. Chem., 43, 834 (1971).

11.0. Friestad, D.E. Ott and F.A. Gunther, Anal. Chem., 41, 1750
(1969).

H.D. Gibbs, J. Biol. Chem., 72, 649 (1927).
J.C. Dacre, Anal. Chem., 43, 589 (1971).
E.F. Mohler, Jr. and L.N. Jacob, Anal. Chem., 29, 1369 (1957).

D. Svobodova, J. Gasparic and L. Novakova, Coll. Czech. Chem.
Commun., 35, 31 (1970).

D. Svobodova, P. Krenek, M. Fraenkl and J. Gasparic, Mikrochim.
Acts, [Wien], 251, I, (1977).

D. Svobodova, P. Krenek, M. Fraenkl and J. Gasparic, Mikrochim.
Acts, [Wien], 197, II, (1978).

0.6. Guiltbault, D.N. Kramer and E. Hackley, Anal. Chem., 38, 1897
(1966).

E. Emerson, J. Org. Chem., 8, 417 (1943).
E. Emerson and L.C. Beegle, J. Org. Chem., 13, 532 (1943).

M.B. Ettinger, C.C. Ruchhoft and R.J. Lishka, Anal. Chem., 23, 1783
(1951).

Technicon, Inc., Tarrytown, New York.

"American Chemical Society for Testing and Materials", Method D-
1783-70, Pennsylvania, 1970.

146.

147.
148.

149.

150.

151.

152.

153.

154.

155.

156.

157.

158.
159.
160.

161.

162.

163.

164.
165.

166.

135

"Standard Methods for Examination of Water and Wastewater",
American Public Health Association, New York, 1971.

G. Blo, F. Dondi and C. Bighi, J. Chromatogr., 295, 231 (1984).
R.W. Stoughton, J. Biol. Chem., 115, 293 (1936).

W. Scott, "Standard Methods of Chemical Analysis", Vol. 2, N.H.
Furman, Ed., Van Nostrand Co., New York, 1939, p. 2253.

R.D. Krause and D. Kratochvil, Anal. Chem., 45, 844 (1973).

L. Lykken, R.S. Treseder and V. Zahn, Ind. Eng. Chem., 18, 103
(1946).

A.J.P. Martin, "Gas Chromatography", V.J. Coates, H.J. Noebels and
LS. Fagerson, Eds., Academic Press, New York, 1958, p. 237.

J. Porath and H. Bennich, Arch, Biochem. BiOphys., Suppl. 1, 152
(1962).

S. Nakamura, S. Ishiguro, T. Yamada and S. Moriizumi,
J. Chromatogr., 83, 279 (1973).

H. Kalasz, J, Nagy and J. Knoll, J. Chromatogr., 107, 35 (1975).

J. Lesec, F. Lafuma and C. Quivoron, J. Chromatogr. Sci., 12, 683
(1974).

K.J. Bombaugh, W.A. Dark and R.F, Levangie, J. Chromatogr. Sci.,
7, 42 (1969).

J.L. Waters, J. Polym. Sci., A2, 8, 411 (1970).
K.J. Bombaugh, J. Chromatogr., 53, 27 (1970).
K.J. Bombaugh and R.F. Levangie, Separ. Sci., 5, 751 (1970).

J.A. Biesenberger, M. Tan, 1. Duvdevani and T. Maurer, Polym. Lett.,
9, 353 (1971).

J.A. Biesenberger, M. Tan and I. Duvdevani, J. Appl. Polym. Sci.,
15, 1549 (1971).

I. Duvdevani, J.A. Biesenberger and M. Tan, Polm. Lett., 9, 429
(1971).

K.E. Conroe, Chromatographia, 8, 119 (1975).
J. Lesec and C. Quivoron, Analusis, 4, 20 ( 1976).

M. Minarik, M. Pop] and J. Mostecky, J. Chromatogr. Sci., 19, 250
(1981).

167.

168.

169.

170.
171.

172.

173.
174.
175.
176.

177.

178.

179.

180.

181.

182.
183.
184.

185.

186.

187.

188.

136

M. Martin, F. Verillon, C.E. Ben and G. Guichon, J. Chromatogr., 125,
17 (1976).

R.A. Henry, S.H. Byrne and D.R. Hudson, J. Chromatogr. Sci., 12, 197
(1974).

L.R. Snyder, J.W. Dolan and Sj. van der Wal, J. Chromatogr., 103, 3
(1981).

P. Kucera and G. Maerius, J. Chromatogr., 219, 1 (1981).

M.J. Carter and M.T. Huston, Environ. Sci. and Technol., 12, 309
(1978).

M.B. Ettinger, S. Schott and 0.0. Ruchhoft, J. Am. Water Works
Assoc., 35, 299 (1943).

M. Farcasiu, Fuel, 56, 9 (1977).

M. Danna, Ph.D. Thesis, Michigan State University, 1985.

C. Wittwer and H. Zollinger, Helv. Chim. Acta, 37, 1954 (1954).
G. Schwarzenbach, Helv. Chim. Acta, 26, 418 (1943).

R.C. Weast and M.J. Astle, Eds., "CRC Handbook of Chemistry and
Physics", 60“ Ed., CRC Press, Boca Raton Florida, 1980, p. D-166.

H. Zollinger, "Azo and Diazo Chemistry", Interscience Publishers,
Inc., New York, 1961, p.51.

R.T. Morrison and R.N. Boyd, "Organic Chemistry", Allyn and Bacon,
Inc., Toronto, 1976, p. 773.

C.L.M. Stults, Ph.D. in progress, Michigan State University.

W. Spendley, G.R. Hext and F.R. Himsworth, Technometrics, 4, 441
(1962).

J.A. Nelder and R. Mead, Computer J., 7, 308 (1965).
S.N. Deming and S.L. Morgan, Anal. Chim. Acts, 150, 183(1983).
D.E. Long, Anal. Chim. Acts, 46, 193 (1969).

S.N. Deming and L.R. Parker, CRC Crit. Rev. Anal. Chem., 7, 187
(1978).

M.W. Routh, P.A. Schwarz and M.B. Denton, Anal. Chem., 49, 1422
(1977).

P.B. Ryan, R.L. Barr and H.D. Todd, Anal. Chem., 52, 1469 (1980).

E.R. Aberg and A.G.T. Gustavsson, Anal. Chim. Acta, 144, 39 (1982).

189.

190.

191.
192.

193.
194.
195.
196.

197.
198.
199.
200.

201.
202.

203.

204.

137

S.N. Deming and S.L. Morgan, Anal. Chem., 45, 278A (1973).

P.F.A. van der Wiel, R. Maassen and G. Kateman, Anal. Chim. Acta,
153, 83 (1983).

D. Betteridge, A.P. Wade and A.G. Howard, Talents, 32, 709 (1985).
D. Betteridge, A.P. Wade and A.G. Howard, Talents, 32, 723 (1985).

L. Lang, Ed., "Absorption Spectra in the Ultraviolet and Visible
Region", Vol. XV, Academic Press, Inc. New York, 1971, p. 20.

R.M. Cassidy, D.S. LeGay and R.W. Frei, J. Chromatogr. Sci., 12, 85
(1974).

S.R. Ratanathanawongs and S.R. Crouch, Anal. Chim. Acts, accepted
for publication, 1986.

W.A. Armstrong, B.W. Grant and W.G. Humphreys, Anal. Chem., 1300
(1963).

P.M. Wiegand, Ph.D. Thesis, Michigan State University, 1985.
P.M. Wiegand, FACSS XI, Philadelphia, 1984.

R.P.W. Scott and P. Kucera, J. Chromatogr. Sci., 9, 641 (1971).
L.R. Snyder, J. Chromatogr. Sci., 10, 200 (1972).

H. Janssen, U.A.Th. Brinkman and R.W. Frei, J. Chromatogr. Sci., 23,
279 (1985).

6.6. Guilbault, "Analytical Uses of Immobilized Enzymes", Marcel
Dekker, New York, 1984.

L.D. Bowers and W.D. Bostick in "Chemical Derivatization in
Analytical Chemistry", Vol. 2, R.W. Frei and J.F. Lawrence, Eds.,
Plenum Press, New York, 1982.

P.D. Wentzell, Ph.D. Thesis in progress, Michigan State University.