You are on page 1of 248

NEW LIQUID CHROMATOGRAPHIC

METHODS FOR THE ANALYSIS OF


CATIONIC

TENSIDES

by

Ian David Bromilow

Submitted in accordance with the requirements for the degree of

Doctor of Philosophy

The University of Leeds


School of Chemistry

September 2001

The candidateconfirms that the work submittedis his own and that appropriatecredit
hasbeengiven where referencehas beenmadeto the work of others

ABSTRACT
Cationic

tensides are important

oleochemicals

that are used to prevent

being
as
adulteration of pharmaceutical preparations and personal care products, as well
the conditioning

agents in domestic fabric softeners. The widespread use of these

materials requires quantitative methods for characterising them in raw materials and
fully formulated products, whilst "down-the-drain"

release necessitates similar methods

for trace environmental

current methods used to quantify

cationic

analysis. Unfortunately,

tensides are generally

non-specific,

problematic,

or

ill-suited
are

to

environmental analysis. There is currently no generic method that can be applied to the
quantitation of these materials in all necessary matrices.

The aim of this work was to develop new liquid chromatographic(LC) methods
for the analysis of the cationic preservative and fabric conditioner actives, before
attempting to build the foundations of generic cationic tenside analysis. The
development of a new normal phase LC method is reported for the quantitation of the
in
domestic
fabric conditioners. The method yielded high
actives
present
cationic
resolution and repeatability, and allowed the quantitation of the homologuesendemic in
commercial samples.Subsequenthyphenationwith massspectrometrydemonstratedthe
for
the quantitation of thesematerials in environmental matrices.
potential
-The optimisation and validation of a reversephaseLC method for the analysisof
cationic tenside preservatives is reported. Excellent repeatability and resolution were
again attained, whilst the new method was also found to demonstrate the inherent
sensitivity required for trace environmental analysis. Subsequent hyphenation
unfortunately showed that method sensitivity was compromised by ion-suppression,
highlighting the needfor compromise in the developmentof LC/MS methods.
For both methods, stationary and mobile phaseparameterswere varied to assess
the influence on analyte resolution, and also to gauge the potential for developing a
generic liquid chromatographic method for the analysis of these materials. It was
by
held
beliefs
the
that
tensides
the
many
of
commonly
observed
on
analysisof cationic
LC
were misconceived.As a result, new insights were made into cationic
reversephase
tenside analysis, which should facilitate the development of a generic LC method
applicable to the quantitation of cationic tensides and their hydrophilic biodegradation
future.
in
the
products

Abstract

11

TABLE OF CONTENTS
Page
No.

General Introduction

CHAPTER ONE:
1.1 AMPHIPHILES,

SURFACTANTS

AND TENSIDES

1.1.1

The historical development of the detergents industry

1.1.2

"Oleochemicals"

4
7

1.2 CATIONIC TENSIDES

1.2.1

Cationic fabric conditioner actives

1.2.2

Cationic tensidepreservatives

11

1.2.3

Environmental aspectsof cationic tensides

13

1.2.3.1 Introduction to chemical risk assessment

14

1.2.3.2 "Down the drain" and "rinse-off' chemicals

15

1.2.3.3 Environmental fate, recalcitrant speciesand eco-design

17

TO CHROMATOGRAPHY

22

1.3 INTRODUCTION

AND SEPARATION

SCIENCE
1.3.1

Chromatographic theory

24

1.3.2

Introduction

32

1.4 CRITICAL

REVIEW

to High Performance Liquid Chromatography

OF CURRENT LITERATURE

METHODS

35

FOR THE ANALYSIS OF CATIONIC TENSIDES


1.4.1

Non-specific methods

36

1.4.2

Thin Layer Chromatography (TLC)

37

1.4.3

Gas Chromatography (GC)

37

1.4.4

High Performance Liquid Chromatography (HPLC)

39

1.4.5

Capillary Electrophoresis (CE)

43

1.4.6

Indirect competitive enzyme-linked immunosorbant assays

45

1.5 CURRENT FAILINGS AND FUTURE NEEDS IN THE ANALYSIS

48

OF CATIONIC TENSIDES
52

1.6 REFERENCES

CHAPTER TWO:
2.1 GENERAL

Experimental

58
59

Tableof contents

iii
2.2 CHEMICALS, SOLVENTS AND GASES

59

2.2.1

University of Leeds

59

2.2.2

Unilever Research Port Sunlight

60

60

2.3 SAMPLES
PARAMETERS

61

Normal phase liquid chromatography

61

2.4 INSTRUMENTAL
2.4.1

2.4.1.1

61

University of Leeds

62

2.4.1.2 Unilever ResearchPort Sunlight


2.4.2

Normalphase liquid chromatography /mass spectrometry

63

2.4.3

Reversephase liquid chromatography

64

2.4.4

Reverse phase liquid chromatography /mass spectrometry

65

66

2.5 REFERENCES

Development of a normal phase LC/MS method for

CHAPTER THREE:

67

the analysis of cationic fabric conditioner actives


68

3.1 INTRODUCTION
3.2 LIMITATIONS

OF STANDARD METHODOLOGY

68

3.3 EVALUATION

OF AN ALTERNATIVE

71

METHODOLOGY

3.3.1

Choice of the initial parameters

71

3.3.2

Evaluation of alternative stationary phases

73

3.3.2.1 Evaluation of the Partisil PAC phase

74

3.3.2.2 Assessment of a series of mixed-mode aminopropyl /

74

cyanopropyl bondedphases
3.3.2.3 Assessment of a series of mixed-mode alkyl-amino /
cyanopropyl bondedphases

76

3.3.2.4 Attaining improved column performance

78

3.3.3

79

Evaluation of alternative mobile phases

3.4 PEAK ELUCIDATION


3.5 OPTIMISING

THE METHODOLOGY

ENVIRONMENTAL

80

BY CO-INJECTION ANALYSIS
FOR INDUSTRIAL

AND

86

ANALYSIS

3.5.1

Analysis of Stepantex

88

3.5.2

Resolution of the components of a mixed esterquatsamples

90

3.5.3

Analysis of the cationic metabolites

92

Table of contents

iv

3.5.3.1 Gradient elution analysis for the simultaneous determination

95

of parentdiester quats and their biodegradationproducts


3.5.4

Improving

with mass

97

Evaluation of the sensitivity and robustness of the optimised

99

method sensitivity

and compatibility

spectrometry
3.5.5

liquid chromatographic

methodology

3.5.5.1 Determining the limit of detection and method linearity

100

3.5.5.2 Method validation

101

Development of the hyphenatedLC/MS methodology


3.5.6.1 Direct infusion analysis

103

3.5.6.2 Loop injection analysis

105

3.5.6

3.5.6.3

104

HyphenatedLC/MS analysis

107

3.5.6.4 Dynamic modification of the stationary phase and analyte


bleed
3.6 PROBING THE RETENTION MECHANISM

109

111

3.6.1

Assessing the influence of the stationaryphase

112

3.6.2

Assessingthe influence of the mobile phase

113

Mechanistic explanation of the experimental observations


3.7 CONCLUSIONS

115

3.8 REFERENCES

121

3.6.3

CHAPTER FOUR:

Development of a reversephase LC/MS method for

120

123

the routine analysis of alkylbenzyl quat preservatives

4.1 INTRODUCTION

124

4.2 DEVELOPMENT OF THE LC METHODOLOGY

124

4.2.1

Choice of initial parameters

124

4.2.2

Use of new silica technology

126

4.2.3

Improving sensitivity and MS compatibility

128

4.2.3.1 The peak compressioneffect


4.2.4

Determination

129

of the sensitivity, linearity and robustness of the

131

new method
4.2.4.1

Method validation

131

4.2.4.2

Method sensitivity and linearity

132

4.3 DEVELOPMENT

OF

THE

HYPHENATED

LC/MS

135

Table of contents

METHODOLOGY
4.3.1

Direct infusion and loop injection studies

4.3.2

Preliminary

4.3.3

Ion

136
139

results of LC/MS analysis

suppression,

and other problems

associated with the

141

LC/MS methodology
4.4

OPTIMISATION

OF

THE

LC/MS

HYPHENATED

142

METHODOLOGY

4.4.1

Reduction of TFA induced ion-suppression with post-column

142

modifiers
4.4.2

of the optimised

147

Increasing the sensitivity of the LC/MS methodology - "Full

149

Evaluation

of the sensitivity

and linearity

LC/MS methodology
4.4.3

scan" analysis versus "Single Ion Monitoring"

4.5 LIMITATIONS

(SIM)

151

OF THE LC METHODOLOGY

4.6 CONCLUSIONS

152

4.7 REFERENCES

153

CHAPTER FIVE:

The development of a generic chromatographic

155

methodfor cationic tenside analysis


156

5.1 INTRODUCTION
5.2 DETERMINING

A SUITABLE STARTING POINT FOR GENERIC

156

ANALYSIS
5.3 EVALUATION

FOR

157

of the new method to the analysis of the

157

OF THE NEW NP-LC/MS METHODOLOGY

GENERIC CATIONIC TENSIDE ANALYSIS


5.3.1

Application

alkylbenzyl quaternary ammonium tensides

5.3.2

161

Analysis of the alkylpyridinium tensides

5.4 EVALUATION

FOR

162

of the new RP-LC/MS methodology to the

163

OF THE NEW RP-LC/MS METHODOLOGY

GENERIC CATIONIC TENSIDE ANALYSIS


5.4.1

Applicability

analysis of alkylpyridinium

quat preservatives

5.4.1.1 Variation in separation efficiency on different cyanopropyl

164

phases
5.4.2

Implications

of

alkylpyridinium

quat

analysis

for

the

165

Table o contents

vi
development of a generic methodology

5.5 OPTIMISING

THE

RP-LC

CONDITIONS

FOR

GENERIC

166

Influence of the bonded phase on the retention characteristics

167

ANALYSIS
5.5.1

of cationic tensides
5.5.1.1 Retention of the alkylbenzyl quats on a standardODS bonded

169

stationaryphaseand a mixed-mode ODS / CN bondedphase


5.5.1.2 Retention of benzyldimethyldodecylammoniumbromide on a

170

seriesof Spherisorbcolumns
5.5.1.3 Retention of benzyldimethyldodecylammoniumbromide on a

174

bare silica column


5.5.2

Influence of mobile phase pH on the retention characteristics

175

of cationic tensides
5.5.2.1 Variation of the retention factor and the logarithm of the
retention factor of benzyldimethyldodecylammoniumbromide

176

on a series of Spherisorb columns at pH 3.0

5.5.2.2 Influence of a cyanopropyl bonded unit on the retention of

181

benzyldimethyldodecylammonium
bromide
5.5.2.3

Revisiting

the

change

in

peak

parameters

of

181

benzyldimethyldodecylammonium bromide witnessed on


SpherisorbCN and ODS / CN phases
5.5.2.4 Retention of benzyldimethyldodecylammoniumbromide on a

182

Zorbax Stable Bond cyanopropyl column

5.5.2.5 Influence of the silica substrateand mobile phase pH on the


peak

parameters

of

184

benzyldimethyldodecylammonium

bromide
5.5.3

Influence of buffer additives on the resolution of cationic

187

tensidepreservatives
5.5.3.1 The effect of different mobile phase modifiers on the

187

chromatographicparametersof cationic tensidepreservatives


5.5.3.2 Irreversible binding of cationic tensides in the absence of

191

mobile phasemodifiers
5.6 THE ROAD TO GENERIC CATIONIC TENSIDE ANALYSIS:

194

Table of contents

vii
A LESSON FROM LITERATURE
5.7 CONCLUSIONS

199

5.8

201

REFERENCES

CHAPTER SIX:

Conclusions andfuture work

210

APENDICES
APENDIX ONE:

Sample characteristics

Column characteristics

221
228

REFERENCES
APENDIX THREE:

211
220

REFERENCES
APENDIX TWO:

204

Initial mass spectrometer parameters

229

Tableof contents

viii

LIST OF FIGURES
Page
No.
Figure 1.1:

Representativestructure of a generalised tenside showing the

two distinct sub-units,which demonstratecontrasting solubility.


Figure 1.2:

Representative structures of three typical anionic (A, B and C)

tensides, one non-ionic (D) tenside, and one cationic (E)


tenside.
Figure 1.3:

Fatty acid distributions of the natural fats and oils normally used

in the production of tensides(Garrett, 1972).


Figure 1.4:

Alkyl chain lengths of typical fatty acid constituents present in

(Garrett,
1972).
tensides
commercial
Figure 1.5:

Representative structures of a generic quat tenside (A) and some

industrially important cationic tensides(B-F).


Figure 1.6:

Representative structures of the three major classes of cationic

12

preservatives.
Figure 1.7:

Figure showing the various stages of a conventional risk

16

assessment process, demonstrating how the relationship


between the PNEC and PEC affects the nature of the safety
testing required (Picture kindly donatedby Unilever Research).
Figure 1.8:

Figure showing the idealistic change in the PEC / PNEC ratio as

17

more knowledge is gained on the environmental properties of a


chemical (Picture kindly donatedby Unilever Research).
Figure 1.9:

Photographs of the River Lea in Lemsford, UK, taken in (A)

18

1960 and (B) 1964. The figure shows the environmental impact
of the replacement of soap with tetrapropylene sulphonate
(TPBS) in domestic laundry products, and the subsequent
adoption of linear alkylbenzene sulphonates in detergent
formulations (LAS) (Pictures kindly donated by Unilever
Research).
Figure 1.10:

Representation of the biodegradation pathway of Hamburg

20

diester quat (HEQ).

List ollgures

ix

Figure 1.11:

Figure showing some of the possible removal and transportation


its
by
be
that
after
release
a chemical
can
exhibited
pathways

21

into a riverine environment (Source of picture United States


Geological Surveywebsite).
Figure 1.12:

Stylised representationof a chromatographicseparation.

23

Figure 1.13:

Typical chromatogram of a non-retained species and two

26

retained components.
Figure 1.14:

Figure demonstrating how the resolution of a two-component

29

(Picture
in
be
Chromatography
manipulated
system can
reproducedfrom Start GC website).
Figure 1.15:

Individual contributions to the analyte band spreadingwitnessed

30

in conventional high performance liquid chromatography


(Reproducedfrom Snyder et al., 1979).
Figure 1.16:

Figure showing how the linear velocity of a separationaffects

31

the individual contributions to band broadening, and how these


changes subsequently affect the plate height H. The bottom
diagram shows typical van Deemter plots obtained in gas
chromatography.
Figure 1.17:

Representative structures of the active groups present on the


in
surface of silica particles used as stationary phase supports

35

HPLC.
Figure 3.1:

Chromatogram showing the analysis of the HEQ sample with

70

the current Unilever method.


Figure 3.2:

Chromatogram showing the separationof the HEQ sample that


was achieved on the Bio-Sil

73

poly-ol column with a ternary

mobile phasesystembasedon hexane


Figure 3.3:

Chromatogram showing the separation of the HEQ sample on

75

the Partisil PAC column.


Figure 3.4:

Chromatogram showing the separation of the HEQ sample on

76

the M3 column.
Figure 3.5:

Chromatogram showing the variation in resolution of species

77

in
HEQ
the
sampleon columns Ml and M3.
present
Figure 3.6:

Chromatogram showing the separation of the HEQ sample

78

M6.
column
on
achieved
List of fl ures

Figure 3.7:

Chromatogram showing the analysis of the commercial

81

DEEDMAC sampleon the M6.


Figure 3.8:

Figure

showing

the result

of

enriching

the commercial

82

DEEDMAC samplewith the C18/ C16(red trace), and the di C16


(blue trace) DEEDMAC diester samples.
Figure 3.9:

Figure showing the result of enriching the DEEDMAC sample,

83

with the C18 (red trace), and C16 (blue trace) monoester
components.
Figure 3.10:

Figure showing the resolution of the monoester positional

84

isomerspresentin the HEQ sampleon column M5.


Figure 3.11:

Chromatogram showing the analysis of the Arquad HT sample

85

on the M6 column.
Figure 3.12:

Chromatograms obtained from the analysis of the Arquad HT


and

dihexadecyldimethylammonium

column

M6.

The two

bromide

traces are overlaid

samples

86

on

to assist peak

elucidation.

Figure 3.13:

Chromatogram showing the analysis of the HEQ sample on a

89

commercial Spherisorbaminopropyl column.


Figure 3.14:

Chromatogram obtained from the analysis of the Stepantex

90

sampleon a Spherisorbaminopropyl phase.


Figure 3.15:

Figure showing the total ion chromatogram obtained from


LC/ESI-MS analysis of a mixed sample containing HEQ and

91

DEEDMAC diesters during a recent literature report (Radke et


al., 1999).
Figure 3.16:

Chromatogram

showing the separation of a mixed sample

containing the commercial HEQ and DEEDMAC

92

diester quat

samples achieved on a Spherisorb aminopropyl phase.

Figure 3.17:

Chromatogram

achieved

from

the

analysis

of

the

94

trimethylammonium propane-1,2-diol chloride sample (HEQ


diol) on a Spherisorbaminopropyl phase.
Figure 3.18:

Chromatogram obtained from the analysis of the DEEDMAC


derived diol sampleon a Spherisorbaminopropyl phase.

94

List of gures

xi

Figure 3.19

Chromatogramobtained from the gradient elution analysis of a


it's
diol
HEQ
the
and
sample
commercial
standardcontaining

96

intermediates.
Figure 3.20:

Figure showing the improvement in peak area response that was

99

diameter
internal
the
by
the
aminopropyl
of
reducing
afforded
2.0
from
4.6
to
mm.
mm
column
Figure 3.21:

Graph showing the non-linearity of the Sedex 55 evaporative

101

Figure 3.22:

light scatteringdetector to increasingHEQ concentration.


Mass spectrum obtained from the direct infusion of the HEQ

104

MS
instrument
into
Finnigan
MAT
LCQ
the
after
sample
optimisation.

Figure 3.23:

Figure showing the dimeric nature of the high molecular weight

106

ions observed during the direct infusion analysis of the


commercial HEQ sample.
Figure 3.24:

Reconstructedion chromatogramsof the major diester species

108

and two unknown resulting from the LC/MS analysis of the


commercial HEQ sample.
Figure 3.25:

Total ion chromatogram resulting from the LC/MS analysis of

109

the HEQ sample.


Figure 3.26:

Chromatogram obtained from the analysis of the HEQ sample

113

on the Spherisorbmixed mode ODS / NH2 phase.


Figure 3.27:

Chromatogram obtained from the analysis of the commercial

114

HEQ sample on column M6 with a binary mobile phasesystem


of hexaneand methanol.
Figure 3.28:

Chromatogram showing the analysis of the HEQ-derived diol

115

species on column M6 with a binary mobile phase system of


hexane and methanol.

Figure 4.1:

Chromatogramshowing the separationof four alkylbenzyl quats

126

achievedwith the new RP-LC method parameters.


Figure 4.2:

Figure showing the variation in separation efficiency afforded

128

by the Luna CN (blue trace) and Spherisorb CN (red trace)


materials.
Figure 4.3:

Figure showing the improvement in sensitivity that was

129

achievedby using a 2.0 mm W. column.


List of figures

X11

Figure 4.4:

Chromatogram

showing

the

analysis

of

100

mg/1

132

benzalkonium chloride standardperformed at the Port Sunlight


laboratory.

Figure 4.5:

Chromatogram showing the analysis of the 50 mg/1 Querton

133

KKBCL standard.
Figure 4.6:

Figure showing the linearity of the diode array detector response

134

to the C12 component in Querton KKBC1 at varying


concentration.
Figure 4.7:

Figure showing the linearity of the diode array detectorresponse


to the C14 component in Querton KKBCI

134

at varying

concentration.

Figure 4.8:

Chromatogram showing the analysis of the I mg/I Querton


KKBCL standard.

135

Figure 4.9:

Mass spectrum obtained from the direct infusion of a 100 mg/l


standardof benzalkonium chloride into the Finnigan MAT LCQ

137

instrument after MS optimisation.

Figure 4.10:

Mass spectrum of the RP-LC mobile phasesystem showing the


unknown ion at m/z = 242.3.

138

Figure 4.11:

Mass spectrum of a 100 mg/I standard of Querton KKBCL


analysedby loop injection analysis.

139

Figure 4.12:

Total ion chromatogram (TIC)

140

(top) and reconstructed ion

chromatograms (RIC's) of the major alkylbenzyl homologues


presentin the Querton KKBCL sample.
Figure 4.13:

Chromatogram showing the alkylbenzyl quat distribution in the


benzyldimethylstearyl

ammonium

chloride

141

monohydrate

sample. Peak identities of the short chain quats were confirmed

by LC/MS analysis.
Figure 4.14:

Total ion chromatogram (TIC) (top) and reconstructed ion


chromatograms (RIC's) of the quat species present in the

144

Querton sample, obtained after the addition of a fixing solution


containing IPA.
Figure 4.15:

Total ion chromatogram (TIC) (top) and reconstructed ion


chromatograms (RIC's) of the quat species present in the

145

Querton sample, obtained after the addition of a fixing solution

List ojfigures

X111

IPA.
1
3:
:
propionic acid
containing
Figure 4.16:

Plot of peak area response of the C12 quat species versus


the
KKBCI
on
the
performed
sample
querton
of
concentration

148

LCQ instrument.
Figure 4.17:

Reconstructedion chromatograms (RIC's) of the C12and C14


KKBCL
in
Querton
1
the
mg/1
present
alkylbenzyl quat species

149

in
full
The
the
scan-MS mode
analysis was performed
standard.
3:
1
fixing
propionic
the
solution containing
addition of a
after
acid : IPA.
Figure 4.18:

Reconstructed ion

Figure showing the variation in: A)

150

full
from
C12
the
the
alkylbenzyl
obtained
quat
chromatogramof
B)
0.5mg/l
Querton
KKBCL
MS
standard;
of
a
analysis
scan
Ion chromatogramof the C12alkylbenzyl quat obtained from the
SIM analysisof a 0.5mg/I Querton KKBCL standard.
Figure 5.1:

Chromatogram showing the resolution of a mixed alkylbenzyl

158

quat sampleunder normal phaseconditions.


Figure 5.2:

Figure showing the variation in the retention time of the


bromide

octadecyltrimethylammonium

159

and

benzyldimethylstearylammonium chloride dihydrate samples


under normal phaseconditions.
Figure 5.3:

Figure showing the variation


dodecylpyridinium

chloride

in

retention of

hydrate

(blue

the

trace)

1-

164

and

benzyldimethyldodecylammonium bromide (red trace) samples


on the Luna CN material.
Figure 5.4:

Figure showing the variation in retention times of a mixed

165

pyridinium quat sampleon the Luna (blue trace) and Spherisorb


materials (red trace).
Figure 5.5:

Figure showing the likelihood of co-elution of the C16

166

C12
the
and
quat
alkylbenzyl quat species,on
alkylpyridinium
A) the SpherisorbCN phaseand B) the Luna CN phase.
Figure 5.6:

Figure showing a comparison of the structuresof the quaternary

168

from
derived
three parent esterquattensidesand
amino-alcohols
the biogenic aminescholine and carnitine.
Figure 5.7:

Figure showing the retention of the 1-dodecylpyridinium

170

List ojlgures

xiv
chloride hydrate analyte on a Spherisorb CN and mixed mode
ODS / CN column.
Figure 5.8:

Figure showing the variation in the retention factor (k) and Log

172

k of the C12alkylbenzyl quat versus the carbon chain length of


the bonded phaseunit on a series of Spherisorbbasedphasesat
pH 2.0. The best-fit curve of Log k versus carbon number was
extrapolated to C1gto demonstratethe predicted retention time
of the C12componenton an ODS bondedphase.
Figure 5.9:

Figure showing the variation in the retention factor (k) and Log
k of the C12alkylbenzyl quat versus the carbon chain length of

177

the bonded phaseunit on a seriesof Spherisorbbasedphasesat


pH 3.0. The best-fit curve of Log k versus carbon number was
extrapolated to C1gto demonstratethe predicted retention time
of the C12componenton an ODS bonded phase.
Figure 5.10:

Chromatogram

resulting

from

the

analysis

of

183

benzyldimethyldodecylammonium bromide standard on a

ZorbaxStableBondcyanopropylcolumn.
Figure 5.11:

Plot showing the effect that mobile phase pH had on the


retention factor of the C12alkylbenzyl quat on the Spherisorb

185

CN phase.
Figure 5.12:

Plot showing the effect that mobile phase pH had on the


retention factor of the C12 alkylbenzyl quat on the Luna CN

186

phase.
Figure 5.13:

Figure showing how the nature of the basic and / or cationic


modifier present in the mobile phaseaffects the retention times

189

of a seriesof alkylbenzyl quats.


Figure 5.14:

Figure showing the analysis of a mixed alkylbenzyl quat sample

190

on the Luna CN column in the presenceof TBAOH.


Figure 5.15:

Figure 5.16:

Chromatogram obtained from the initial analysis of a C12


alkylbenzyl quat standardon a Spherisorb silica column with a
mobile phasecontaining no basic and / or cationic modifiers.
Sequential analysis of a 50 mg/I Benzalkonium chloride

193

193

standard on a Spherisorb silica column with a mobile phase


containing no basic and / or cationic modifiers, following the
List of rares

xv

initial analysisof the C12alkylbenzyl quat.


Figure 5.17:

Figure showing the variation in the retention time of the C12

196

alkylbenzyl quat on the Luna CN and Spherisorb CN phasesat


pH 2.0 and 3.0.
Figure 5.18:

Figure showing the effect of 5 mmol/1 ammonium acetateon the

197

retention time of the C12alkylbenzyl quat.

List of/Izires

xvi

LIST OF ABBREVIATIONS
ACN

Acetonitrile

APCI

Atmospheric pressure chemical ionisation

API

Atmospheric pressure ionisation

C18:i

Monounsaturated

CELISA

Competitive enzyme-linked immunosorbant assay

CZE

Capillary zone electrophoresis

DSBAS

Disulphine blue active substance

ELSD

Evaporative light scattering detector

ELS

Evaporative light scatterer

ESI

Electrospray ionisation

GC

Gas chromatography

HPLC

High performance liquid chromatography

W.

Internal diameter

LC/MS

Liquid chromatography/Mass spectrometry

LOD

Limit of detection

LOQ

Limit of quantitation

MeOH

Methanol

MS

Mass Spectrometry

NP/LC

Normal phase liquid chromatography

Quats

Quaternary ammonium surface active agent

RIC

Reconstructed ion chromatograms

RP/LC
SIM

Reversephase liquid chromatography


Single ion monitoring

TFA

Trifluoroacetic acid

TIC

Total ion chromatogram

TLC

Thin layer chromatography

UV/vis

Ultra-violet/visible

Cla fatty acid

List of abbreviations

xvii

ACKNOWLEDGEMENTS
I would like to take this opportunity to thank my supervisor Prof. Keith Bartle for
his
help,
for
Ph.
D.
Leeds,
this
to
the
encouragement,and
all
at
and
chance undertake
giving me
Burford
Mark
Graham
Lawrence
Thanks
Dr's
to
throughout
and
also
studies.
my
confidence
from Unilever Researchwho were involved with the project from the start and who offered
Analytical
RSC
like
I
help
the
to
throughout.
acknowledge
would also
and advice
unlimited
Division and Unilever for the provision of the joint SAC studentship,which funded this work.
I am indebted to the support staff at Leeds, who helped me throughout my four years.
Thanks to Bernard, Mike, Francis, Peter and all the workshop crew, not forgetting the two
Dave's, fellow followers of "the greatest game on earth", the friendly banter was always
beating
Saints
least
were
all-comers!
as
appreciated,not
My time at Leeds would have been much duller and less productive if not for the
"characters" around me: Thanks to Rich and Richy for takin' Robbo's flack and for looking
heavy
blood-vessel
burster
know,
To
Katherine,
I
did
I
the
than
a
night!
after
only
other
worse
Sue (who scarred me mentally), Kate P, aka the boss who always put me in my place, Krys,
Niki, Nee and Sarah, and not forgetting my personal secretary Stuart to whom I owe my first
for
favours
during writing up. To the French people who never did
the
all
three months salaries
for
Dr
Mark
Robson
Prof.
Peter
hangover
Myers
their pearls of
thanked
the
cures!
and
are
get
it
humiliation,
didn't
the
that
wisdom,
as
and
ritual
must of cured me
pure chromatographic
friends
Indeed
kill
to
all
my
and colleaguesat Leeds thanks for everything, 'cept the
me!
quite
Barbie thing, that went a little too far!
Thanks to the members of Dragonfly who welcomed me as a student and then again
haunt
back
In
Analytical
I
team
to
them!
this
to
the
thanks
motley
crew
came
special
goes
when
for all their efforts and help, to Mark, Dan and Paul for showing confidence in me (Dan I'm still
Bansai
job
in
States!
for
),
James
for
Moore
the
to
the
and
commission on your
office
waiting
banter (and sports commentary), and thanks again to Dr Paul Beers for providing realism,
laughsand a boot up the arse when it was needed...bloody biologists!
To the Gents from back home, Chris, Sid and Waz, who kept my feet on the ground
have
I
deep
in
Three
knee
finished,
haven't
I
things:
the
got
I
mire.
nope still
when was already
did
find
looking!
I
keep
Sid
to
the
third
I'll
job
type
the
secret
never
of
softness,
and
a real
Most of all I would like to thank three people in particular: my long-suffering girlfriend,
KATE, (and her family) who has been there throughout the good and bad times (and bore the
brunt of the latter), thanks hon. I love you loads! and to MY PARENTS who, as always, have
been there for me throughout, bailed me out again and again and ensured that I always knew
there were people looking out for me. To the three of you: THANK YOU VERY MUCH
THIS IS AS MUCH YOURS AS IT IS MINE!

Acknowledgements

xviii

To my grandparents,
Though my time with you was short, I am forever grateful. I hope

somedayyou'll see this.

and
To the people who lost their lives on the H"' September2001,
it puts everything into perspective.
..........

CHAPTER ONE
General Introduction

Chapter 1: General Introduction

CHAPTER ONE

General Introduction
1.1

AMPHIPHILES,

SURFACTANTS

AND TENSIDES

I planted you like a choice vinefrom the very best seed.But look what you have
become! You are like a rotten, worthless vine. Even if you washed with the strongest
soap, I would still seethe stain of your guilt.
Jeremiah 2.21-22.

"Amphiphilic molecules" or amphiphiles is the generic term applied to organic


compounds that contain at least two structural units, which demonstrate starkly
contrasting physico-chemical properties, notably solubility. At least one region of the
be
be
to
seen
will
solvophilic, or "solvent-loving", and will thus dissolve
amphiphile
least
be
at
one
other
will
solvophobic, or "solvent hating" (Garret, 1972;
whilst
readily,
Hamley, 2000) (Figure 1.1). Such ambivalence has led to them being referred to as
amphipathic, as they simultaneously show sympathy and antipathy to the solvent
(Tanford, 1980). When placed into solution such ambivalence leads to preferential
interfaces,
in
them having important advantageous
at
which
results
accumulation
been
have
by
that
exploited
man for eons (Garret, 1972).
properties
As amphiphiles are predominantely applied into aqueoussystemsthe solubility
of the solvaphilic and solvaphobic sub-units are related to water, and hence one region
is seento be hydrophilic, whilst at least one other is hydrophobic. Hydrophobic groups
long
based
hydrocarbon
on
a
chain and are thus commonly referred to as
are most often
"tails", whilst the hydrophilic group, or "head', can be chargedor uncharged(Figures
1.1 and 1.2) (Garret, 1972; Tanford, 1980; Hamley, 2000).
The tendency of amphiphiles to accumulate at surfaces has led to them being
"surfactants
"surface
". However, in the early 1960's a
to
active
agents"
or
as
referred
joint committee on nomenclatureagreed on the adoption of the term "tenside" as the
generic title for all amphiphilic molecules (Garret, 1972). Thus, tenside will be used
Chapter 1: General Introduction

throughout this work to refer to compounds, which in many other texts are referred to as
surfactants.

Hydrophilic "head group"

Figure 1.1: Representative structure of a generalised tenside showing the two distinct sub-units,
which demonstrate contrasting solubility.

1.1.1 The historical development of the detergents industry


Whilst tensides have, and still do find widespread application as foaming,
dispersing
(Garret,
1972; Hamley, 2000), history shows that they
agents
and
wetting,
have been used primarily as detergents, for several millennia. Indeed, tracing the origins
detergents
is
tenside
the
of
almost impossible. Historians believe that the
use
of
Babylonians may have been using soap (Figure 1.2), the earliest recognised tenside, to
clean their garments as early as 2600 BC. There is clear evidence that the Phoenicians
were using the product by 600 BC and the Celts had introduced the technique to the
Romans some five hundred years on, but only later did the Romans utilise soap to clean
themselves. However, whilst saponification, the change occurring in the formation of
from
Roman
it
being
from
derives
discovered
the
the
myth
of
soap,
emanating
Mount
PreSapo
Tiber,
it
is
likely
the
that
on
above
alters
sacrificial
much more
Historic Man was the first to recognise the detergent potential of tensides.

With the passing of four millennia, since the end of Babylonian civilisation,
industry.
its
detergent
has
the
the
position
as
one
of
principle
products
retained
of
soap
Soap bars and later, soap flakes formed the cornerstone of the domestic laundry market
that developed in the late nineteenth century after mass production was pioneered by the
Lever Brothers (Lever Brothers, 2000). Soap was eventually vanquished from its
detergent
laundry
in
half
the
the
principle
second
as
of the twentieth century
position
(Lever Brothers, 2000). At this point in time, alternative synthetic anionic tensides were
Chapter 1: General Introduction

4
commercialised (Garret, 1972), and later, a number of other anionic tensides were
developed for the use in domestic detergents. The use of non-ionic tensides like alcohol
has
dramatic
(AE's)
(Figure
1.2)
shown
growth, whilst the use of cationic
ethoxylates
tensides has also become widespread. From small beginnings,

and amphoteric

household detergent consumption has grown to in excess of 6800 million tonnes per
annum in Western Europe alone, of which approximately 1000 million tons comprise
2000).
(Trius
tensides
et
al.,
active

0
Oqe
,,
(A)
lJ

Sodium salt of lauric acid


Constituent of some soaps

(B)
GO

Na

SO3 -Na

Tetrapropylenebenzenesulphonate
(TPBS or ABS)
0

Me /R

H" ` ',xO

Linear alkylbenzenesulphonate
(LAS) (x + y) =6 to 9

/N\

YH

Linear alcohol ethoxylate(AE)


(D)

Me

Cl

(E)

Dialkyldimethylammoniumchloride
R is predominantlyC18and C16for
tallow derived alkylquats

Figure 1.2: Representativestructures of three typical anionic (A, B and C) tensides,one nonionic (D) tensideand one cationic (E) tenside.

1.1.2 "Oleochemicals"
With the dramatic growth in the use of detergents, and tensides in general,
increaseddemand has been placed on sourcing adequatesupplies of the raw materials
hydrophobic
As
in
the
their
unit of a tenside is normally basedon a
used
production.
long hydrocarbon chain, tensideshave traditionally been manufactured from natural or
1996).
(Dobson,
fats
oils
synthetic
and
Chanter 1: General Introduction

5
During the middle part of the twentieth century, the scale of the petroleum
industry, particularly in North America led to many of the anionic and non-ionic
tensides developed at this time, being manufactured from petroleum by-products
(Garrett, 1972). However, historically, the majority of tensideshave been manufactured
from naturally occurring edible and inedible fats and oils of animal or plant origin
(Figure 1.3). Indeed, in recent years there has been a notable decline in the use of
increase
in
the usage of renewable sourcesof raw
and
an
materials
petroleum-derived
kernel
Currently,
(Dobson,
1.3).
1996)
(Figure
oil
palm
especially
materials,
fat
90%
in
tenside manufacture, are
all
and
oil
raw
of
materials
used
approximately
found above ground (Dobson, 1996; Trius et al., 2000). As a result of their natural oil
and fat origin, tensidesare often referred to as oleochemicals(Dobson, 1996).
At this juncture it should be noted that the term fat or fatty is normally used to
refer to alkyl chains with greater than or equal to twelve carbons within their structures
(Garrett, 1972; Hamley, 2000). However, on occasions the term fatty is applied to alkyl
down
C8.
lengths
to
chain

Figure 1.3 shows the fatty acid distribution of the oils and fats commonly used
in the manufactureof tensides.It is evident that there is a distribution of chain lengths
present in the raw materials. As fractionation or purification of the oil or fat is not
generally performed prior to manufacture, commercial tenside samples contain a
distribution of alkyl chain homologues. As a result, a `pure" tenside sample will
normally contain a number of active species, which contain hydrophobic groups of
varying length (Lawrence, 1997). This can lead to some confusion as the detergents
industry often class a pure tenside sample as "that which is produced by the
does
not necessitatethat it contains only one active component.
manufacturer", which
In addition to natural fats and oils containing a distribution of alkyl chain
lengths, it is evident from Figure 1.3 that they also contain both saturatedand monofatty
acids. As a result, commercial soap samples include a
and poly-unsaturated
distribution of saturatedand unsaturatedfatty chains, which along with the chain length
distribution has a definite effect on the structural properties and solubility of the final
high
The
1972).
(Garrett,
percentageof palmitic, stearic and oleic acids present
products
in tallow fat leadsto the formation of hard, insoluble soaps,whilst the Marseilles soap,
Chapter 1: General Introduction

6
which was renowned in the nineteenth century for being "pleasant on the skin", is soft,
due to its manufacture from olive oil (Garret, 1972). As a result, blends of fats and oils
are now utilised to impart the desired structural properties to modern soaps (Garrett,
1972). Whilst many other tensides are still manufactured from a single fat or oil
(Garrett, 1972), hydrogenation of the raw material is commonly used to manipulate the
properties of the final product (Lawrence, 1997). As a consequence, the saturated to
unsaturated ratio of many tensides is very different to that of the fat or oil used during
manufacture (Lawrence, 1997).

70
60
0 50
N

CL

E 40
0
U
m 30
c
8

20
10
0

Figure 1.3: Fatty acid distributions of the natural fats and oils normally used in the production
of tensides (Garrett, 1972).

Figure

1.3 also highlights

an additional

foible

of the tenside industry:

nomenclature. The IUPAC nomenclature used to name organic compounds is often


circumvented

by the industry

in favour of trivial

nomenclature relating to the

corresponding fatty acid (Figure 1.4). As a result, lauryl and stearyl are commonly used
to refer to saturated C12 and C18 hydrocarbon chains, rather than the IUPAC defined
terms of dodecyl and octadecyl. However, further confusion arises from the continued
use of other traditional trivial names, e.g. cetyl, which is still used in place of hexadecyl
is
in
This
Appendix
problem
clearly
evident
and palmityl.

One, where a list of the

tenside samples used during this study is presented. The names of the individual
samples are given as seen on the characterisation data supplied with the sample, i. e.
Chanter 1: General Introduction

7
benzyldimethylstearylammonium chloride dihydrate may also be referred to as
benzyldimethyloctadecylammonium chloride dihydrate. For simplicity, the IUPAC
during
length
describe
this work, whenever
has
been
to
alkyl chain
used
method
possible.

FATTY ACID

ALKYL CHAIN LENGTH

Caprylic

Cs

Capric

CIO

Lauric

C12

Myristic

C14

Palmitic

C16

Stearic

C 18

Oleic

C18.1

Linoleic

C18:2

Linolenic

C18:
3

Arachidic

C20

Eicosenoic

C20:1

Figure 1.4: Alkyl chain lengths of typical fatty acid constituentspresentin commercial tensides
(Garrett, 1972).

1.2

CATIONIC TENSIDES
Cationic tensides are positively charged amphiphiles that are predominantly

based on a quaternisednitrogen atom. Figure 1.5 depicts structural representationsof


important
the
cationic tensides, along with a more generalised
most
some of
four
/
tenside
cationic
showing
alkyl
a
or alkyl-aryl groups
of
and
representation
has
that
atom
quatemised
nitrogen
an associatedanionic counter
surrounding a central
ion. The association of the counter ion leads to the formation of a quaternary
fatty
is
due
be
it
least
to
to
one
surfaceactive,
containing at
ammonium salt, which seen
tail unit. As a result, the colloquialism "quat" is commonly applied to cationic tensides,
form
(Lawrence,
1997).
is
important
that
It
demonstrates
to
this
note
generalised
which
important
have
four
distinct
industrially
tensides
cationic
groups attached
whilst many
like
1.5)
(Figure
the
to the central nitrogen atom, others
alkylpyridinium salts
(Appendix

One) encapsulate the cationic nitrogen atom within a heterocyclic ring.

Chapter 1: General Introduction

S
Nevertheless,these materials are still referred to as quats. For simplicity quat will be
fatty
to
to
this
quaternaryammonium salt.
throughout
a
refer
work
used
ooX

Me (D
/R
N

RR
N
R/

Me

Me\O/C16H35

O
Cl

/N\
Me

Me

(C)

(B)

(A)

Br0

0
30O
CH3OSO

OR

Cl
yN

-1

Me
OR

O0
Me3N

-1

HO

(D)

(E)
CIO

OR

-1
0

/P/C16H33

(F)
Figure 1.5: Representative structures of a generic quat tenside (A) and some industrially
important cationic tensides(B - F).
For speciesB, D and E, R is normally Clg and C16as they tend to be manufacturedfrom tallow
fat.

The rise to prominence of the cationic tensides came about in the early 1930's
bactericidal
Domagk
the
the
value
commercial
of
properties of these
recognised
when
first
Since
(Lawrence,
1970).
this
they
thirty
after
were
years
reported
some
materials
become
has
tensides
the
more widespread, with the materials now
time,
use of cationic
finding use in a number of different applications. Cationic tensides are applied to
1999;
fixers
dye
1987),
inhibitors
(Domagk,
(Gough
degrees
corrosion
et al.,
as
varying
Branzoi et al., 2000), oilfield extraction processes (Omar et al., 1997) and in miceller
drugs
for
(Vievsky,
delivery
therapeutic
systems
and vesicular

1997; Esposito et al.,

1999). However, whilst usage is growing in many of these areas, two traditional
domestic
in
dominate
tensides
the
market: softening agents
cationic
applications still
fabric conditioners and antibacterial agents / preservatives in personal care products,
(Jungermann,
1970;
Puchta,
1984;
Boethling
et al.,
cosmaceuticals
pharmaceuticals and
1992).

Chapter 1: General Introduction

9
1.2.1 Cationic fabric conditioner actives
The ability of cationic tensidesto condition garmentswas initially discoveredby
dyeWhen
these
applied
solutions
of
as
manufacturers
materials
clothing
accident.
fixing agents, they were seen to impart softness, often referred to as "hand", to the
in
1999).
This
(Levinson,
the mid-1950's, which
observation
made
was
garments
increased
the
timely
use of mechanical washing machines, and the
as
proved
improved
detergent
had
led
flakes
to
stain
synthetic
with
powders,
replacementof soap
improved
(Levinson,
1999).
Whilst
stain removal
removal and reduced re-deposition
led to cleaner clothes, the lack of re-deposition led to greater removal of clay and oil
fibres
built-up
had
the
that
on
over time. Unfortunately, the presenceof these
residues
improve
feel
in
found
the
the
to
of
garment
when
contact with the skin
residues was
(Levinson, 1999) and, as a result, the new detergentswhere perceived to leave garments
feeling coarse and uncomfortable making them less appealing to the consumers
(Levinson, 1999). The subsequentutilisation of rinse-addedcationic tensideswas found
to improve the feel of garments, even when washed with soap flake, and were thus
(Levinson,
1999).
by
consumers
acclaimed
Today cationic tensides have found widespread use as fabric "softening" agents
in post-wash products that are perceived to impart softness and fragrance. However, to
label these formulations as softeners does them a disservice. Research has shown that
the use of cationic tensides improves the softness and bulk of the laundry, extends fabric
lifetime, reduces electrostatic build-up, improves water and stain resistance, reduces the
burden of post-wash mechanical processes (ironing and drying), and imparts that allimportant 'freshness" (Puchta et al., 1993; Levinson, 1999).

Fabric conditioning products can be split into two main types, the liquid rinseadded formulations that are employed globally, and the sheet-basedconditioners that are
in
dryers
North
America
Whilst
in
(Levinson,
1999).
the
tumble
predominantly
used
two groups of products have very different modesof application, the mode of action and
identical
(Levinson,
1999).
liquid
Considering
the
agents
are
active
rinsestructure of
for
less
in
formulation
15%
tensides
than
the
account
cationic
added conditioners,
of
be
25%
in
the
to
can
concentration
up
whilst
concentrates(Levinson,
standardproducts,
1999). The product is added to the laundry after removal of the anionic detergent to
ensure that the cationic tensidescan effectively adsorb onto the fibres of the garments
Chapter 1: General Introduction

10
formation
(Crutzen,
1995;
from
being
through
the
complex
system
removed
without
Levinson, 1999). Whilst much of the cationic tenside is removed from the garments
during final rinsing, a discrete amount of the material is left on the garments,imparting
1997).
(Burford,
somewater and stain resistance

main

The primary compounds used in fabric conditioners can be divided into two
like
1st
traditional,
quats,
the
or
generation alkyl
classes:

dihydrogenatedtallowdimethylammoniumchloride, and the 2nd generation esterquats,


(Figure
derivative
diethylesterdimethylammonium
hydrogenated
chloride
of
such as the
1.5) (Appendix One). There is a great deal of commonality in the two groups of
fat,
from
be
fact
both
in
tallow
least
to
tend
that
the
manufactured
tensides,not
groups
1993;
(Waters
Puchta
1991;
it
fully,
et
al.,
al.,
et
partially or non-hydrogenated
whether
Lawrence, 1997; Friedli et al., 2000). In addition, the primary active agentspresent in
fabric conditioner formulations are the di-chained species,the dialkyl and diester quats
1991;
Lawrence,
1997).
due
However,
(Waters
to the method of
et
al.,
respectively
both
tensides,
to
cationic
classes
of
commercial
seen
samples
are
also
of
manufacture
(Waters
1991;
the
of
mono-chained
concentrations
species
et
al.,
contain significant
Lawrence, 1997) (Figure 1.5) (Appendix One). Indeed, in the caseof the 1stgeneration
alkylquats, and some esterquats,considerableconcentrationsof the tri-chained species
Friedli
1997;
2000)
(Appendix
(Lawrence,
One).
Both
the monoet
al.,
present
also
are
impurities,
in
that they also add to the performanceof
tri-chained
are
active
species
and
formulation.
However,
different
their
to that of the
specific
activities
are
a commercial
di-chained species.
The presence of tri-chained and / or mono-chained cationic tensides in
further
(Section
"pure"
to
adds
complication
a
so-called
sample
commercial samples
1.1.2). Therefore, some "pure" samples contain a distribution of di-chained actives
fat
in
fatty
(Figure
1.3),
tallow
the
to
acids
which may or may not show
corresponding
dependant
factor
1.1.2),
two
(Section
the
the
tallow
processing
and
of
on
a
unsaturation,
(Lawrence,
1999).
To
the
to
use
quats,
prior
worse,
matters
make
active
of
other classes
distribution
lengths,
tri-chained
species
show
a
of
and other
alkyl
chain
mono- and
impurities such as fatty acids and protonated fatty amines, which are also known to be
If
1997).
(Lawrence,
one considersthat the concentration of active agents,and
present
is
impurities
carefully controlled to ensure the efficacy of the
active and non-active
Chapter 1: General Introduction

11
is
it
formulation,
apparent that accurate characterisation data on the raw
commercial
materials, and effective quality-control protocols for fully formulated products, are of
paramount importance.

1.2.2 Cationic tenside preservatives


The secondmajor area of application for cationic tensidesare as anti-bacterial /
fungicidal agents (Lawrence, 1970). Reference has already been made to the fact that
the first reports of the beneficial properties of cationic tensides,and the first commercial
application of these materials were as anti-bacterial agents (Section 1.2). In the sixty
five years that have lapsed since the first commercialisation of these materials by
Domagk, mono-chained quats have found widespread usage as antimicrobials,
bacteriostaticsand fungicidal agents(Lawrence, 1970; Boethling et al., 1992; Lin et al.,
1991). Their potency and efficacy in these areashas led to their usageas preservatives
in decongestivenasal spraysand most ophthalmic solutions (Santoni et al., 1994; Price
et al., 1999), as well as other pharmaceutical and cosmetic products (Lin et al., 1991;
Kawakami et al., 1998). Cationic tensideshave also been applied as the active agentsin
throat sweets and lozenges (Taylor et al., 1998), and have found widespread use as
in
healthcare
industry (Lawrence, 1970;
the
agents
and
sanitisers
general anti-bacterial
Kmmerer et al., 1997). In addition, their use as sanitisers within the food and drink
industry, to prevent bacterial and microbial adulteration and fungal growth, is now
becoming more widespread (Suortti et al., 1990; Valladao et al., 1994; Zhou et al.,
1999). Finally, in recent years the application of cationic tensides to prevent fungal
in
lumber
industry,
the
sapstain
particularly in North America, has gained
and
growth
1990;
Bull
1998b;
Chen
(Suortti
al.,
et
et
al.,
et al., 1995).
pace
Figure 1.6 shows representativestructuresof the three major classesof cationic
tensidesutilised as preservatives/ sanitisersin the various applications listed above.All
three classescontain only a single fatty alkyl chain, and commercial formulations are
strictly controlled to ensure the removal of di-chained species. As a result, cationic
less
are
much
complex than the corresponding fabric conditioner
preservative samples
samples(Lawrence, 1970). An additional variation between the cationic preservatives
is
be
fabric
found
in
to
the choice of raw materials.
the
conditioner
actives
and
cationic
Whilst the conditioner actives are generally only based on tallow fat, the fatty raw
material used in the manufacture of the preservative actives is dependant on the
Chapter 1: General Introduction

12
in
healthcare
For
the
alkylbenzyl quats utilised as preservatives
example,
application.
from
One)
(Appendix
1.6)
(Figure
manufactured
are primarily
and cosmetic products
C12
distribution
demonstrate
that
on
centres
thus,
an alkyl chain
usually
coconut oil and
be
tend
to
However,
1.3).
(Figure
1972)
the
(Garrett,
C14
alkylpyridinium quats
and
industrial
being
C16
C12
homologues
the
of
species
and
with only
applied as single
relevance (Martindale,

1993; Zhou et al., 1999). In contrast, the monoalkyltrimethyl

(Lawrence,
dependant
from
C8
C18,
length
in
the
to
application
on
quats can range chain
in
the
is
difficult
it
the
to
manufacture
As
1970).
used
materials
raw
generalise
a result,

of the cationic preservativeactives.

Xe

Xe

O, R

Me
I\

N,,,,
Me

Alkylpyridinium quat
R= 12 and 16 are the only
industrially important species

Alkylbenzyldimethyl quat

R is predominantly C12and C14in


the case of coconut derived quats

0X

MR0
N
Me

\Me

Dialkyldimethylammonium chloride
R is predominantly C18and C16for
tallow derived alkylquats
Figure 1.6: Representativestructures of the three major classesof cationic preservatives.

It was mentioned above that different cationic tensides are utilised in different
the
based
Whilst
the
active
materials,
of
was
original choice
on available
applications.
be
is
individual
in
ingredient
to
dependant
is
the
the
scenario which
on
choice now
individual
homologues
different
the
tenside,
of each,
and
classes
of
applied, as
demonstrate varying efficacy (Lawrence, 1970; Prince et al., 1999). Such observed
individual
homologues
has
limits
led
defined
in
to
and
the
of
strictly
potency
variation
distributions
homologue
in
the
being
on
present
commercial samples.
placed
controls
The United States Pharmacopoeia(USP), along with the British Pharmacopoeia(BP)
(Ph.
Eur.
),
legislators,
include
defined
Pharmacopoeia
European
and
other
clearly
and
Chapter 1: General Introduction

13
boundaries on the nature and concentration of individual homologues in various
products (Martindale, 1993; Prince et al., 1999). The maximal and minimal
in
is
in
concentrationof active agentpresent a commercial sample stipulated, addition to
Prince
1993;
(Martindale,
length
the
et al.,
of
active
materials
ratio
chain
regulationson
1999). For example, the USP national formulary (USP/NF) guidelines stipulate that a
less
Chloride
(Appendix
One)
Benzalkonium
not
must contain
commercial sample of
be
less
40%
the
Of
Cis
C8
95%
than
this,
to
should
than
quats.
alkylbenzyl
not
active
C12species,and not less than 20% should be C14homologue. Additionally, the total C12
be
70%.
Any
C
than
to
greater
or
equal
material not meeting
should
and 14concentration
these criteria will be blocked from application under the pseudonym Benzalkonium
Chloride in pharmaceutical and healthcare products (Martindale, 1993). Similar
for
the alkylpyridinium quats and the monoalkyltrimethyl
restrictions are also enforced
quats(Martindale, 1993).
Whilst commercial cationic tenside preservative samples contain fewer active
in
legislation
than
conditioner
samples,
present
strict
on composition and
are
agents
formulators
implement
forces
to
and
raw
materials
suppliers
strict qualityactivity,
At
be
the
time,
the
same
regulators
must
able to test adherenceto
control protocols.
their legislation, which also requires the use of accurateanalytical methods to quantify
thesematerials as raw materials and in fully formulated products.

1.2.3 Environmental

aspects of cationic tensides

In recent years there has been a marked increasein the environmental awareness
developed
In
the
the
world especially, there is widespread
of
world's population.
industrialisation
have
that
the
potential
effects
and commercialisation
understandingof
balance
World's
Ecosystem.
Whilst major environmental
delicate
the
the
of
on
in
Exxon
Valdez
1989,
have
like
the
oil
spill
made, and always will make
catastrophes
headline news in the scientific and non-scientific press, this greater environmental
in
being
friendliness"
led
"environmental
has
to
placed
greater
emphasis
on
awareness
legislation
As
life.
environmental
a
result,
strict
now controls all of the
all areas of
influence
Man,
to
the world around us.
which are perceived
activities of

Chapter 1: General Introduction

14

1.2.3.1 Introduction to chemical risk assessment


The notion of environmental legislation is far from new in the control of
chemical usage.The inception of laws regarding the safeguardingof potable water from
sourcesof contamination can be found in the legislature of the Roman Empire, whilst
be
in
laws
to
the
of Europeancountries during the middle ages(Ahmad
seen
others are
basis
2001).
Indeed,
the
much
of
of current environmental legislation grew out of
et al.,
human health legislation developed in the late nineteenth century (Ahmad, 2001).
However, as the awareness of governments, and other authorised agencies, has
developed, with regard to the potential effects of chemical usage on the environment,
modern environmental legislature now runs in parallel to laws governing human health.
Today, the manufacture and usage of chemicals is strictly controlled by
guidelines on both perceived and documented toxicological and environmental impact
(Solbe, 2000). During the formulation of these guidelines two conceptsare taken into
hazard
assessmentand risk assessment(Solbe, 2000). A hazard is an inherent
account,
is
likelihood
the
compound,
or
whilst
action
an
risk
of
property
of that effect / property
being manifest. Death by drowning is a hazard of bathing, whilst electrocution is a
hazard of using electricity, the risk associatedwith both is lessenedif the two are not
combined! From a chemicals perspective, a hazard of nitrogen gas is that it is an
being
increased
in
the
of
asphyxiation
risk
asphyxiant;
confined areas.
Primarily, new legislation concerning chemical usage is based on risk
it
is
into
takes
quantifiable,
as
account the likelihood of a detrimental effect
assessment,
occurring, and can also be applied to new compounds as it can be used to assess
documented
as
problems. As a result, approval of the use of all new
perceived as well
in
known
to
the
use
of
addition
species for a role in which they were not
chemicals,
is
formal
by
controlled
a
risk assessmentprocedure.
originally specified,
Whilst a risk-based assessmentof chemical safety is welcomed by most, many
believe
that a hazard-basedsystem should be adopted.
groups
environmental pressure
Claims are often made that legislation is too lenient, as chemical manufacturers and
formulators are thought to hold too much influence over the decisions of the lawmakers.
Indeed, such groups often maintain that all unnaturally occurring chemicals are harmful
and hazardousto the environment and / or human health, and, as a result, their usage
Chapter 1: General Introduction

15
hazard-based
be
banned.
However,
of
adoption a
systemis often over simplified,
should
indirect
it
be
to
assess
problems associatedwith some chemical species.
as cannot used
For instance, an excess of phosphate in some confined lakes and rivers can lead to
death
body
1976).
Whilst
(Moore
the
the
of
and
ultimate
of
al.,
water
et
eutrophication
the use of phosphate-baseddetergentformulations is claimed to have greatly increased
the rapidity of eutrophication due to the occurrence of algal blooms, the process is
naturally occurring, and is not a hazard, as such of phosphate. Only risk-based
for
can
efficiently
account
such secondaryeffects.
assessments
For environmental impact assessment,the formal risk assessmentis basedon a
comparison of the lowest concentration at which a chemical will be seen to cause no
harm to the environment, the so called "predicted no-effect concentration" (PNEC),
in
"predicted
(PEC),
level
the
the
the
environmental
concentration"
with
expected
different environmental compartments (Figure 1.7). The PNEC is calculated by
base
the
the
of
compound
on
a
effect
set of sensitive organisms present in
assessing
each compartment(Figure 1.8). The conservativesafety factor is then applied to ensure
that the resulting concentrationwill have no discernible effect on the organismspresent.
Determination of the PEC is achievedby combining usage,removal, and environmental
/ geological data to determine the maximal concentration that the compound is likely to
is
It
use.
prolonged
apparent therefore that accurate risk assessmentscan
reach with
based
data,
be
details
how much compound enters the
on
accurate
which
only
from
it
where,
and
where
permeatesto ultimately.
environment,

1.2.3.2"Down the drain" and "rinse-off' chemicals


To have an effect on a specific environmental compartment, a chemical must
have the ability to reach that compartment in the first instance. There are two primary
modes by which a chemical can enter the environment, use and misuse. Of these, misuse
is probably the most recognised for causing an environmental

impact, as it is

led
that
the
to the Exxon Valdez disaster (Solbe,
one-off
pollution
event
characteristic of
2000). However, correct use of a chemical can also lead to its entry into the
environment.

Cationic tensides,indeed many tensidesin general, are examples of compounds


that have the potential to be introduced into the environment through proper usage.If
Chapter 1: General Introduction

16
industrial
important
the
two
most
applications of cationic tensides, fabric
one considers
conditioners and preservatives (Section 1.2), their modes of action dictate that cationic
tensides will have the ability to be transported to the environment. Fabric conditioners,
industrial
domestic
surface cleaners are classed as "down-the-drain"
and
and

chemicals

(Burford, 1997). In the case of the developed world, the mode of action of a conditioner
dictates that it will be expelled from the washing machine into the domestic sewage
treatment system, (Versteeg, 1992). Surface cleaners are disposed of in a similar
into
the drainage system after application. If the
they
rinsed
are
normally
manner, as
local sewage treatment system is unable to completely remove the tensides, then they
into
be
the aquatic environment through wastewater or into the terrestrial
released
will
environment as sludge (Di Corcia, 1998). Cationic tenside preservatives also have a
similar

fate. Their usage to prevent adulteration of personal care, cosmetic and

"
"rinse-off
preparations, creams, pomades and ointments, also results in
pharmaceutical
them being washed into the sewage treatment system (Berger, 1997).

Step-sequence of risk assessment


ACUTE

CHRONIC

ECOSYSTEMS
Artificial
Natural

PredictedNo-EffectConcentration

limits

Overlap= possiblerisk

confidence

PredictedEnvironmentalConcentraIon

confidence
limits

MONITORING
INVESTIGATIVE
0

CONFIRMATORY
SCREENING

increasing knowledge about chemical


Figure 1.7: Figure showing the various stages of a conventional risk assessment process,
demonstrating how the relationship between the PNEC and PEC affects the nature of the safety
testing required (Picture kindly donated by Unilever Research).
As one moves towards the right hand side of the diagram, testing becomes more involved as the
test becomes increasingly realistic of environmental conditions.

Chapter 1: General Introduction

17

Exposure
concentration

Effects
concentration
phnia
algae
fish

I
The overlap (PEC >
PNEC) suggests the
possibility of a risk of
harm in the ecosystem

1
PNEC

PEC

(Al

A safety factor of 1000


is applied to the acute
toxicity value of the
most sensitive of three
species.

No-effects
concentration

Exposure
concentration

(B)

fish
algae

The lack of overlap (PEC <


PNEC)

there

that
now suggests
i s no s i gn if ican t r i s k

to the ecosystem.

4M

-"""

A safety factor of 10 is
applied to the no-effect
concentration
of the most

sensitive of three species.

PEC PNEC

Figure 1.8: Figure showing the idealistic change in the PEC / PNEC ratio as more knowledge is
gained on the environmental properties of a chemical (Picture kindly donated by Unilever
Research).
As the methods used to calculate the PEC / PNEC ratio are conservative, a value of less than
one represents a safety margin for using the chemical in it's designated role. However,
is
increase
to
the safety margin still further.
carried
out
often
additional work

1.2.3.3 Environmental fate, recalcitrant species and eco-design


The market for cationic tensides amounts to 135,000 tonnes per annum (1998
figures) in Western Europe, for detergents alone (Trius et al., 2000). It is therefore
be
deemed
be
for
to
these
they
that
species
environmentally
either
safe,
must
evident
before
in
degraded
the
be
they
to
reach
extent
environment,
great
a
or
rapidly
removed
the environmental compartment to which they are transported, and not be seen to have
factor
last
is particularly
This
to
the potential
accumulate.

important in the case of

like
high
the cationic tensides, and has been a
volume
chemicals
production
relatively
historical problem with certain species. Figure 1.9 shows two photographs taken at the
into
River
Lea
in
Lemsford, in the United
the
treatment
plant
a
sewage
outflow of
Kingdom in 1960 and 1964. The earlier picture (left hand side) coincided with the
Chanter 1: General Introduction

18
replacement

of soap flake

tetrapropylene

in domestic

laundry

benzene sulphonate (TPBS) (Figure

to be resistant to biodegradation

products

for the synthetic

1.2). Unlike soap, TPBS was found

due to the branched nature of the hydrophobic

(Beers, 2001). As a result, due to high volume

usage, the concentration

natural waters built up to the extent were it lead to the formation


rivers.

Having

biodegradable

identified

tenside:

TPBS as being responsible

in
TPBS
of

foams
on
of stable

for the foaming,

synthetic anionic tenside, linear alkylbenzene

tail unit

a new linear

sulphonate (LAS)

(Figure

1.2), was developed to replace TPBS in laundry products. The picture on the right-hand
TPBS
in
disappearance
LAS
the
the
that
of the
of
replacement
with
resulted
side shows
is
in
2001).
It
foams,
LAS
(Beers,
treatment
sewage
plants
was readily removed
as
river
important

to note that this change in ingredients

environmental

was brought about solely because of the

impact of TPBS.

Figure 1.9: Photographs of the River Lea in Lemsford, UK, taken in (A) 1960 and (B) 1964.
The figure shows the environmental impact caused by the replacement of soap with
in
laundry
domestic
(TPBS)
tetrapropylene sulphonate
products, and the subsequent adoption of
linear alkylbenzene sulphonates in detergent formulations (LAS) (Pictures kindly donated by
Unilever Research).

In addition to the problems caused by persistent

parent tensides, another

important consideration in the assessment of an environmental

risk, posed by any

form
is
its
to
recalcitrant metabolites. Whilst, the parent tenside
propensity
chemical,
initially

is
halted
biodegradation,
the
process
at an inappropriate juncture,
undergoes

due to the tendency of the metabolite to resist further breakdown (Burford, 1997). This
Chapter 1: General Introduction

19
process is characteristic of the nonylphenol ethoxylate tensides, which give rise to
nonylphenol metabolites that are recalcitrant. These metabolites show higher aquatic
toxicity compared to the parent species, and have been linked to inflicting endocrine
disturbances (Beers, 2001). Factoring these risks into the formal assessmentfor this
its
led
has
to
voluntary replacementby the more readily biodegradablealcohol
species
ethoxylates(Figure 1.2) (Beers,2001).
The propensity of many cationic tensidesto be anti-bacterial and algaecidal,and
to preferentially sorb onto negatively chargedspecieshas led to historical concernsover
the widespreaduse of these speciesin "down-the-drain" and "rinse-off' products. The
observation that the first generationfabric conditioner actives (Section 1.2.1) facilitated
low
on
sensitive
species
at
concentrations(Waters et al., 1991) did
observableeffects
fears.
help
However,
these
to
there was strong evidence that the
of
any
alleviate
not
high
tensides
removal within sewagetreatment plants, and thus only
showed
quat
alkyl
low concentrations were observed in receiving waters, even before dilution occurred
(Waters et al., 1991). Nevertheless, in light of the concerns of legislative bodies over
the environmental profile of these materials when used in fabric conditioners, the early
1990's saw formulators voluntarily replace the 1't generation alkylquats with new
1.5)
(Waters
(Figure
et al., 1991). Thesematerials demonstratedmuch
esterquatactives
improved environmental profiles, owing to the presenceof ester linkages between the
fatty
(Puchta
the
atom
and
alkyl
chain
nitrogen
cationic
et al., 1993). Theseester groups
in
the structuresof the new actives, which facilitated the
weakness
of
points
represented
fatty
from
the
the parent tensides (Figure 1.10)
of
acid
groups
cleavage
rapid
(Lawrence, 1997), greatly increasing the observed rate of primary biodegradation
(Waters et al., 1991). As with the changefrom TPBS to LAS, some twenty five to thirty
from
the use of alkyl quats to esterquats was driven by
the
switch
years earlier,
(Burford,
1997).
In
the case of the cationic tensides, these
concerns
environmental
documented
(Waters et al., 1991).
than
perceived
rather
primarily
concernswere
As risk assessmentof chemical usage must be performed separately for each
into
which the compound can ultimately pass, it is
environmental compartment
important to be able to quantify the path of a species through waste treatment and
during its progressthrough the environment. Ultimately, the transport and movement of
its
based
be
on
characteristic physico-chemical properties. In the case
any chemical will
Chapter 1: General Introduction

20
fabric
limited
the
actives,
conditioner
water solubility and high octanol /
of cationic
water partition coefficient (K),

a measure of the propensity of a compound to reside in

is
logarithmic
the
environment,
value
or
non-aqueous
which
often
quoted
as
aqueous
an
(Log P) for convenience (Hansch et al., 1979), indicates that the dialkyl

quats

As
1991).
(Waters
the
a
out
of
aqueous
phase onto solids
et al.,
preferentially partition
is
during
these
treatment
the
of
species
as a result of
of
removal
sewage
result, much
(Beers,
2001).
The
onto
sludge
solids
process of sludge amendment onto
adsorption
agricultural

land thus yields the introduction

of these species into the terrestrial

di-chained
Similarly,
into
the
residual
quats,
receiving waters,
which
pass
environment.
drop
dissolved
to
to
partition
onto
suspended
preferentially
and
solids,
which
seen
are
the riverbed to form sediments (Scott, 2000) (Figure 1.11). As a result, the di-chained
quats are introduced to the sediment compartment.

Xe
R-1

CIO

Biodegradation

Monoester intermediate
Intermediate solubility:
-t
Found on solids

Meaty

Ne3NR

OY

R-t

Fatty acid

and in aqueousphase

Parent diester quat

Biodegradation

Low aqueoussolubility:
partitionsonto solids

Fatty acid
Xe
Bacterial biomass
and
inorganic salts

o
Me3N
OH
Quaternary amino-alcohol
Readily water soluble

Figure 1.10: Representationof the biodegradationpathway of Hamburg diester quat (HEQ).


The parent di-chained quat is very hydrophobic and readily partitions onto solids. Removal of
the fatty acid groups increasesaqueoussolubility. As a result the quaternary amino-alcohol is
is
bioavailable.
thus
more
and
readily water-soluble

The partitioning behaviour of the mono-chain quats (Figures 1.6 and 1.10) is
di-chained
due
the
that
to their increasedaqueoussolubility
of
quats,
not as extensiveas
(Burford, 1997). As a result, there is increased likelihood of fording these species in
their native forms in wastewater effluents i. e. solubilised rather than associatedwith
is
for
This
1997).
true
(Burford,
the quaternaryamino-alcohol metabolites of
also
solids
Chapter 1: General Introduction

21
the esterquats (Figure 1.10) (Appendix One), which are the most readily water soluble
of the three groups of quats owing to the removal of the second hydrophobic group from
the monoester quats (Burford, 1997). As a result of their high water solubility these
found
1997).
be
in
(Burford,
to
to
associated
solids or
sediments
species are unlikely
Unfortunately, higher aqueous solubility leads to the species being more bioavailable,
for
increases
in
in
bioaccumulate
the
these
the tissues of
turn
to
potential
species
which
native

organisms

via absorption

and / or

ingestion

(Beers, 2001).

Extensive

bioaccumulation could then lead to the concentration of the compound within the
tissues of the organism approaching those at which a distinguishable effect is first
1995).
In
Leeuwen,
(van
view of recent concerns regarding metabolite
witnessed
formation during biodegradation, there has recently been an upsurge in regulatory
interest in

defining

the future

role

that metabolite

formation,

transport,

and

in
play
environmental risk assessment (Beers, 2001)
accumulation should

Figure 1.11: Figure showing some of the possible removal and transportation pathways that can
be exhibited by a chemical after its release into a riverine environment (Source of picture United
States Geological Survey website).

Chapter 1: General Introduction

22
It is apparentthat in order to carry out a formal risk assessmenton the potential
environmental impact of cationic tensides, accurate determination and quantitation is
required in a number of different environmental matrices. In order that accurate
measurement is attained, methods of analysis must be selective to minimise the
interference causedby endogenouscomponentsof complex sample matrices. Not only
does environmental assurance demand accurate quantitation of the parent cationic
tensides,but there is increasing need to be able to quantify the hydrophilic metabolites
biodegradation.
from
In addition, the ability to monitor the transport
primary
resulting
and transformation of cationic tensides through the environment is of paramount
importance if the environmental safety margins are to be maximised.

1.3

INTRODUCTION
SEPARATION

TO

CHROMATOGRAPHY

AND

SCIENCE

Chromatography is a physical method of separation in which the componentsto


be separated are distributed betweentwo phases, one of which is stationary while the
direction.
definite
in
other moves a
"Unified Nomenclature on Chromatography",,IUPAC definition of
chromatography ca. 1993 (Braithwaite et al., 1996).
Chromatography is the generic name applied to techniquesthat can be used to
separatetwo or more componentsof a mixture as a result of variations in the affinities
for
immiscible
individual
two
solutes
phases(Ruthven, 1997).
of
Natural chromatographicphenomenahave existed on Earth since the birth of the
(Braithwaite
ago
years
et al., 1996; Berezkin, 1997). However, the
planet millions of
practical relevance of chromatographic techniques was not recognised until the middyes,
the
by an early form of paper
separation
with
of
century
nineteenth
1996).
(Braithwaite
Some
ten years later Groppelsroeder
et
al.,
chromatography
reported the "capillary analysis" of pigments via separation on paper dipped into
1996).
Groppelsroeder
(Braithwaite
al.,
et
was aware that capillary action was
solvent
bringing about the migration of the various pigments. However, he was unable to justify
Chapter 1: General Introduction

23
the reasons behind the discrete coloured bands that he witnessed. Whilst these early
"Separation Scientists" had reported observations that were fundamental to modem
birth
discovery
is
the
and
of
chromatography
normally
separation,
chromatographic
Russian
Botanist
Tswett,
Twentieth
the
the
the
the
to
of
at
start
of
work
attributed
Berezkin,
demonstrated
1996;
1997;
Ruthven,
Tswett
1997).
(Braithwaite
et
al.,
century
how it was possible to separate and isolate green and yellow plant pigments, by passing
fine
(sucrose
through
or
a
glass
column
packed
a
powder
with
extracted plant material
initially
head
After
the
the
adding
plant
of the column
material
onto
calcium carbonate).
in a fine band, Tswett added additional solvent and pressure, to move the pigments
through the column (Braithwaite et al., 1996; Start GC website). The addition of the
focused
band,
facilitated
in
the
and
subsequent
addition
of
solvent,
a
small
solutes
superior component resolution,

compared to that achieved with the crude paper

fractionation of dyes and pigments (Braithwaite et al., 1996). In 1906, Tswett, during
the reporting of his findings, first coined the phrase chromatography from the Greek
"
for
"chroma
and the verb "to write" or graphein, to describe how the
colour.
words
into
discrete
bands
the
technique
plant
material
coloured
on the column.
separated
new

.r

.. n

u
i
u
C

Direction of flow

Figure 1.12: Stylised representation of a chromatographic separation.


The two components are separated over time due to their different
immiscible phases.

Some years later, the potential implications

affinities

of Tswett's

for the two,

observations were

beginnings
led
to
the
by
and
of chromatographic research. In 1941,
chemists
recognised
Martin and Synge used chromatography to separate amino acids from wool. This
(Braithwaite
chromatography
technique
partition
the
of
et al., 1996).
evolved
Chapter 1: General Introduction

24

The commercial potential of gas chromatography (GC), particularly in the


analysis of the complex samplescharacteristic of the petroleum industry, was realised
during the pioneering work of Martin and James. These researcherswere able to
separatemixtures of short chain carboxylic acids on a support material, coated with a
liquid film. By forcing the samplesthrough a glass tube with the aid of compressedgas,
it proved possible to isolate individual analyte bands. Some five years later, Golay
in
treatise
theoretical
a
on
optimising
column efficiency and peak capacity
published
GC. His theorieson the use of long, narrow columns, coated with a thin liquid film have
high
to
the
given
rise
resolving power, characteristic of modem GC
subsequently
analysis.
The development of modem day "high performance liquid chromatography"
(HPLC), came about as a result of the predictions made by Giddings in 1963. Giddings
suggested that the resolving power achieved with GC would only be forthcoming for
LC separations, if very small diameter particles were packed into narrow columns. At
the time, fabrication

technologies

interrogate Giddings'

theories, and some six years passed before the successful

could not provide

the particles

necessary to

development of 10 m silica particles proved his predictions to be true. With the


development of new bonding technologies that facilitated the attachment of functional
groups to inorganic supports, with which to bring about specific interactions with the
solute of interest, and the commercialisation

of high pressure pumping systems, the

fundamentals of modem HPLC were put in place.

1.3.1 Chromatographic theory


The operating parameters employed in a modem chromatographic separation
back
be
to chromatographic theory via a series of important fundamental
can
related
parameters: retention, efficiency, resolution and "solute band broadening". In the
following section, these parameterswill be considered with respect to their effect on
information
The
derived
separations.
contained in this section has been
experimentally
from
to
extents,
a small number of the many reference texts and
obtained,
varying
information sources on the subject of chromatographic theory. Referencesutilised at
this time include Snyder et al. (1979), Braithwaite et al. (1996), Meyer (1993), Katz et
GC
Start
(1998)
the
website.
and
al.
Chanter 1: General Introduction

25

As mentioned previously, chromatographic separation is based on the variation


in the affinity of solutes for two immiscible

phases. The rate at which an analyte

is
the
through
system therefore dependent on the length of time that the analyte
migrates
distinct
in
two
the
phases. Thus, in liquid chromatography (LC), the
spends contact with
is
dependent
intermolecular
between
forces
the
the
on
critically
acting
rate of migration
For
liquid
the
the
the
and
analyte
stationary
phase,
and
phase.
and
mobile
analyte
have
interact
high
for
thus
the stationary phase,
that
and
a
strongly,
affinity
analytes
their progress through the system is slow and retarded, in comparison to analytes that
are only weakly attracted to the stationary phase. Chromatographic

separation is

therefore dependent on a variation in the extent of which analytes distribute, or partition


between the two phases. For an analyte (solute) i, this behaviour can be described by the
distribution or partition ratio (Equation

1.1). The "partition coefficient", K, shown in

Equation 1.2 is the equilibrium constant relating to this dynamic process,

'mobile

K =

lstotionary

'"

(Equation 1.1)

(Equation 1.2)

t,m

Cl,
C1,,
in
to
the
the
the
relate
and
molar
concentration
where
of
solute
stationaryphase
m
Therefore,
the chromatogram that is achieved during
respectively.
phase
and mobile
analysis provides a pictorial representation of the history of the analytes migration
through the column over time. Analytes that are seento elute early in the chromatogram
having
be
deduced
spent the majority of their time within the mobile phase,
as
can
later
in
the trace, must have shown stronger affinity for the
that
those
elute
whilst
1.13
Figure
depicting
the separationof a twoshows
chromatogram
a
stationary phase.
baseline
The
disturbance
at the start of the trace is representativeof
component mixture.
the emergenceof the solvent front from the column, and all unretained analytes.When
it
for
is
the stationary phase and thus passes
shows
affinity
no
an analyte unretained
through the column unchecked.The time taken between the introduction of the mixture
into the system, and the elution of an unretained component is known as the "dead
time" or "void time" of the system, to. This parametercan then be used to quantify the
in
the stationary phase. The values tRI and tp
the
components
other
time
of
residence
Chapter 1: General Introduction

26
These
the
times
the
two
retained
components.
retention
of
are characteristic of the
are
individual analyte, as they are derived from the fundamental interactions between the
analyte and the stationary and mobile phases.Use of retention time data and knowledge
be
length,
L,
can subsequently
utilised to determine the average linear
of column
during
the separation,(Equation 1.3) and the averagerate at which the solute
velocity
(Equation
v
1.4).
through
the
column,
migrated

Time

Figure 1.13: Typical chromatogramof a non-retainedspeciesand two retained components.

,, =

L
(Equation

1.3)

to

v=L tR

(Equation 1.4)

As v is related to residence time in the mobile and stationary phases, the


be
from
calculated
comparing the number of moles of analyte in the
also
parametercan
in
total
the
the system. As the number of moles
number
of
moles
with
stationary phase
in
is
in
the
to
the
system
the
related
molar
of analyte present
concentration of
analyte
the stationary and mobile phases, C,,3 and Ct,m respectively, and the volume of the
V.
V3,
phases,
and
substitution and simplification results in v
mobile and stationary
being describedby Equation 1.5:

Chanter 1: General Introduction

27
v=u"1

1+ Ci'sVs / Ci,

(Equation 1.5)
mV.

Substitution of the partition coefficient (Equation 1.2) into Equation 1.5 leads
to Equation 1.6:

F=u-

(Equation 1.6)

1+K, VV/Vm

Simplification of Equation 1.6 is facilitated by utilising the equation for the


"retention factor", k, (Equation 1.7), which yields Equation 1.8. Replacementof the
terms relating to and v (Equations

1.3 and 1.4) results in an equation for which, all

from
determinable
derived
an
experimentally
chromatogram (Equation
components are
1.9).

k; =

y, V'
(Equation

V.

(Equation 1.8)

v=u .1+k

k; =

1.7)

tR - to

(Equation

1.9)

to

As k is proportional to Tp, a large increase in k is characteristic of a strongly


for
it
is
thus
and
convenience
preferable to ensurethat all k values are
retained analyte,
be
However,
if
k
is
time.
too
to
the
analysis
small prevent excessive
small
analyte may
dead
by
Control
factor
be
the
to
time.
the
to
over
retention
seen elute close
can achieved
the variation of analytical parameters.In LC separationsmodification of mobile phase
lead
to the effective control of the retention factor.
parameterscan normally
During the analysis of a multi-component sample it is often convenientto utilise
the retention factors of each of the analytes to quantify the "selectivity"
for two adjacent analyte peaks. The "separation factor",

of the method

a (Equation

1.10), is a

Chanter 1: General Introduction

28
descriptor which relates the relative retention of two components,A and B on a given
stationaryphase,under specified conditions.

(Equation 1.10)

a=KB
KA

distribution
B.
Substituting
A
KB
KA
the
coefficients
are
of
solutes
and
and
where
Equation 1.7 into Equation 1.10 highlights the relationship between selectivity and k
(Equation 1.11). It is evident that during the analysis of a multi-component mixture, if
be
be
k
to
a
within
narrow
range,
selectivity
compromised.
may
all values are seen
kB

(Equation 1.11)

kA

The ability of a column to produce narrow analytes bands is critical to high


in
liquid
Column
efficiency,
chromatography especially, is often
quality separations.
"theoretical
basis
in
lies
fractionation
in
term
the
terms
plates",
a
of
whose
quoted
towers of the petroleum industry. The higher the number of theoretical plates, the more
is
be.
is
from
This
to
the
witnessed
effect
apparent
separation
observing the
efficient
two forms of Equation 1.12. As the analyte bandwidth, w, decreases,efficiency is seen
to increase;the column is better able to resolve individual componentsof a sample into
discretenarrow bands.
2
2

N=16-

tR
WW

= 5.54 "

tR

(Equation 1.12)

V2h

It is apparentfrom Equation 1.12 that N increasesas the analytesresidencetime


is
increases,
hence
length,
N
be
by
the
also
phase
seen
affected column
stationary
within
L. It is evident that as column length increases column efficiency also increases.
Equation 1.13 demonstratesthe relationship betweenN and column length. The H term
is a descriptor relating to the efficiency of the column per unit length. It is defined as
"the height equivalent to a theoretical plate", HETP, with rearrangementof Equation
1.13 revealing that H decreasesas N increases(Equation 1.14), the distance between
the theoretical plates decreases,which when related back to petroleum distillation
Chapter 1: General Introduction

29
In
boiling
terms
isolation
in
of
the
components.
of
range
a
narrower
of
results
band,
height
lower
the
the
the
the
resulting
analyte
narrower
plate
chromatography,
better
in
higher
thus
separation.
and
efficiency
which results

N=

(Equation 1.13)

H=N

(Equation 1.14)

B
R. " als

AB

Ro- 15

Ww
rime.

1t'

mi

Figure 1.14: Figure demonstrating how the resolution of a two-component system can be
(Picture
from
Start GC website).
in
reproduced
manipulated chromatography
The terms shown are described in Equations 1.10,1.12,1.21 and 1.22. However, in the above
by
by
in
OZ,
1
2
Equations
1.21
is
and
components
and
are replaced
replaced
example a
B.
A
and
components

The processesby which multiple analytesare separatedinto discrete bands, and


by
increase
in
dwell
the
broaden
time,
bands
time
an
with
witnessed
within
w
as
analyte
derive
from
do
broadening
Instead
the
increases,
the
same
origin.
of an
not
column
has
journey
been
its
found
be
through
band
to
column
a
a result of the
on
analyte
in
Figure
1.15.
pictorially
physical effects represented

Chapter 1: General Introduction

30

(a)
INITIAL

xxxxxxxx

BAND

+txxxaxs

a1

(b)

WIDTH
JFINAL
BAND
W1D'TH

/9(

10

EDDY DIFFUSION

START
t

2
x

Li

z
C5

MOBILE PHASE
MASSTRANSFER
(C)

STAGNANT MOBILE
PHASEMASSTRANSFER

(d)

STATIONARY PHASE
MASSTRANSFER

(e)

Figure 1.15: Individual contributions to the analyte band spreading witnessed in conventional
high performanceliquid chromatography(Reproducedfrom Snyder et al., 1979).

Figure 1.15 (a) shows a cross-sectionalview through the column as the analytes
in
head
band.
In
Figure
1.15
(b),
introduced
thin
the analyte molecules
the
a
onto
are
diffusion".
"eddy
band
the
the
whereby
analyte
of
grows as a result of
process
undergo
the analyte molecules taking distinctly different paths through the support. Those
flowing through a wide channelwill be seento move faster than those moving through a
is
degree
in
The
to
time
the column.
of
spreading
seen
grow
with
spent
narrow channel.
Figure 1.15 (c) highlights band spreading by "mobile phase mass transfer" whereby
faster
band
in
than those close to the particles
the
the
move
much
the analytes
middle of
due to the viscous drag being exerted band on the liquid at this pint. Figure 1.15 (d)
"stagnant
the
contribution
of
of
representation
mobile phase mass
provides a pictorial
in
Differences
the rates to which analyte molecules
band
to
spreading.
transfer"
in
the
in
the stationary phase pores also
solvent
present
stagnant
and out of
partition
band.
Figure
1.15
final
(e)
the
the
the
the
of
shows
result of
overall size
affect
contribution

"stationary-phase mass transfer", whereby the analyte band is spread by

into
further
leads
the
to
than
stationary
phase
others,
which
permeating
analytes
some

Chapter 1: General Introduction

31
them showing additional retardation as they attempt to partition back into the bulk
solvent.
The extent of which the processes seen in Figure 1.15 affect band broadening, is
dependant on the length of time that the analytes remain in the column. As a result, it
broadening
is
inherently
band
linked
be
that
to the linear velocity at which the
seen
can
separation is performed. As a result, numerous studies have been performed to evaluate
the effect of mobile phase velocity on separation efficiency. Figure 1.16 shows the
typical form of the graph obtained from studies characterising the change in H as a
function of linear velocity. The data shown in Figure

1.16 is characteristic of that

be
during
GC
and
can
analysis,
related to Equation
obtained

1.15, which was first

reported by van Deemter.

-cw

CST

Awl

Nrr

wNew/

M-.

l`)

M ()

Ni

VMCOT

1.0

0.6

25

00

76

Figure 1.16: Figure showing how the linear velocity of a separation affects the individual
contributions to band broadening,and how these changessubsequentlyaffect the plate height H.
The bottom diagram shows typical van Deemter plots obtained in gas chromatography.

H=A+B+

Cu

(Equation 1.15)

C
band
broadening
the
to
A,
B
contributions
analyte
are
and
where
characteristic of
diffusion
longitudinal
diffusion,
and masstransfer processes,respectively.
eddy

Chanter 1: General Introduction

32
Kennedy et al. (1972) later proposed a modified version of the van Deemter
equation (Equation 1.16), which overcame the effects of particle size, column length
and solute diffusion on separation efficiency, allowing the comparison of different
disparate
being
utilised
under
experimental conditions. h is referred to as the
columns
from
being
height,
the use of Equation 1.19, where dp is the
obtained
reduced plate
linear
is
diameter.
The
from
the relationship between
reduced
velocity
obtained
particle
diffusion
dp
Dm,
the
coefficient via Equation 1.20.
solute
u, and

h= Av

Y+-+
t

Cv

(Equation

1.18)

h=d

(Equation 1.19)
P

h=

(Equation

1.20)

The final chromatographicparameter,which will be considered at this time, is


the resolution Rs,which is used to quantify the degreeof separationof adjacent analyte
bans. Resolution can be defined equally well by either Equation 1.21 or 1.22. Equation
1.22 is particularly noteworthy as it shows how modification of other chromatographic
degree
the
of resolution.
parametersaffects

tR2
R=
s

1.3.2 Introduction

12(WZ

tRl

(Equation 1.21)
+Wi

k
_Nyx(a-1)x
4a

k+1

(Equation 1.22)

to High Performance Liquid Chromatography

As with the general theory of chromatography, there are many excellent


facets
liquid
texts
on
all
of
modem
available
chromatographic theory and
reference
during
the
following
Amongst
those
the
used
preparation
of
practice.
section are:
Snyderet at, 1979; Meyer, 1993; Braithwaite et at, 1996 and Katz et at, 1998.
Chanter 1: General Introduction

33

In liquid chromatographic analysis, the resolution of a mixture of analytes is


different
for
is
their
of
affinities
a
result
a
as
stationary
phase,
achieved
which
liquid
and
a
mobile phase(Meyer, 1993). This so-called
predominantly a solid surface,
liquid-solid chromatography has in recent years made an analogous technique, liquidliquid chromatography, a technique whereby the stationary phase is also a liquid, the
two liquids being immiscible in order to create two discrete phases,almost obsolete.
The adoption of the phrase "High performance liquid chromatography" (HPLC) comes
from the fact that the stationary phase is made up of small particles, which give rise to
fact
first
in
by
1940's
Martin and
that
the
a
was
predicted
efficient separations,
early
Synge(Meyer, 1993).

Over the last thirty years there has been a surge in the interest and application of
HPLC to the separation and isolation of a host of organic and inorganic compounds.
Liquid chromatographyhas a distinct advantageover gas chromatography(GC) as it is
Gas
Chromatography
in
its
limited
application.
can only be applied successfully to
not
be
that
can
easily adaptedthrough reaction to enhancetheir
or
analytes
analytes,
volatile
is
as
a
result
only applicable to approximately 20% of the organic
volatility, and
date
been
have
to
that
characterised.In contrast liquid chromatography can
chemicals
be applied to many more compounds, as it requires only that the molecule is soluble in
the chosen mobile phase (Meyer, 1993; Katz et al., 1998). As the mobile phase
be
easily modified and adapted,they can be tailored to suit the needsof
can
parameters
an individual analyte or series of analytes,enhancing solubility and making them fit for
1993;
(Meyer,
Katz
by
HPLC
1998).
et
al.,
analysis
Modem liquid chromatographic can be performed in a number of different
from
"Size
Exclusion
Chromatography"
(SEC), to the
ranging
operation,
modes of
biologically relevant practice of "Affinity Chromatography" (Katz et al., 1998).
However only two modes of liquid chromatography will be considered at this time:
"normal phase liquid chromatography" and "reversedphase liquid chromatography".
In normal phaseliquid chromatographyseparationis performed with the aid of a
relatively non-polar mobile phaseand a polar stationary phase,which is commonly bare
1993).
Meyer,
In
1998;
this mode of operation polar analytes are
(Katz
et al.,
silica
Chanter 1: General Introduction

34
by
the stationary phase than are non-polar analytes, which pass
retained
more strongly
through the column relatively unchecked. As a result, analytes are seen to emerge from
the column in order of increasing polarity.
chromatography (NP-LC),

is normally

Retention in normal phase


liquid
-

attributed to competitive

adsorption of the

analytes and solvent molecules onto the active sites on the stationary phase support
(Katz et al., 1998; Meyer, 1993; Dorsey et al., 1994). In the case of silica, the most
different
there
phase
support
material
active
normal
are
a
number
of
used
commonly
(Figure
1.17),
the
substrate
silica
with which the solute and solvent
on
present
sites on
is
interact.
It
form
that
generally
assumed
solvent
can
molecules
a monomolecules
layer on the surface of the packing material (silica), and thus the solute molecules must
displace one or more solvent molecules, dependant on the molecular size of the species,
before they are able to interact with polar groups on stationary phase. Strong support for
this model comes from the observation of a well-defined eluotropic series of normal
1998;
Meyer,
1993),
(Katz
al.,
et
whereby non-polar solvents like
solvents
phase
hexane and and carbon tetrachloride are found at the bottom of the series, and polar
isopropanol
like
and methanol are found at the top of the series
acetonitrile,
solvents
(Katz et al., 1998). As the concentration of polar solvents within the mobile phase
increases the competitive adsorption equibrium is driven towards the analyte remaining
in the mobile phase as they are no longer able to dislodge sorbed solvent molecules.

In reversed phase liquid chromatography (RP-LC), analytes are eluted in the


"reverse" order than would be expected under normal phase conditions, i. e. the most
first,
in
This
whilst
non-polar
analytes
elute
are
strongly
polar analytes
retained.
reversal
is
due
to the employment of a non-polar stationary phase, which is
order
elution
typically

formed from bonding long hydrocarbon chains onto a silica support i. e.

(Appendix
octylsilane
octadecylsilane and

One), and a polar mobile phase, commonly

As
solvent.
a
polar
organic
and
a result, polar analytes tend to
consisting of water
favour
in
in
the
the stationary
whilst
non-polar
analytes
phase,
mobile
residence
remain
have
Two
been
1998).
(Katz
models
principal
proposed to account for
et al.,
phase
LC:
in
phase
reversed
retention

"the adsorption or solvophobic model" and "the

defining
in
Addressing
the
RP-LC is one of the
mode
and
of
retention
partition model".
in
Whilst
issues
the partition model of
chromatography.
modem
most contentious
fully
into
the
are
the chains of the
solutes
encapsulated
whereby
analyte retention,

Chanter 1: General Introduction

35
bonded phase,is generally favoured, the solvophobic model still has strong support in
from
Horvath.
noticeably
most
some areas,

Bonded silanol

Siloxane

BO, - -H, 0
j

i--,

11
/O

, -o

Metal associatedsilanol

Isolated silanol
H

'Si
/O

H,

Figure 1.17: Representativestructures of the active groups present on the surface of silica
in
HPLC.
phase
supports
particles used as stationary
The acidity of the groups increasesin moving from top left to bottom right. In normal phaseLC,
the acidic groups act as competitive adsorption targets for solutes and solvents, whilst in RP-LC
interaction with isolated and metal-associatedsilanol groups is believed to lead to the tailing of
basic analytes.

Reverse phase chromatography performed on a long alkyl-bonded stationary


liquid
is
the
most
widely
practised
mode
currently
of
chromatographic
phase support
1996).
Its
for
(Braithwaite
the analysis of small polar organic
suitability
al.,
et
analysis
drug
it
the
to
therapeutic
targets utilised
separation
applicable
of
analytesmakes readily
by the pharmaceuticalindustry. Not surprisingly this sector of industry is currently the
biggest user of liquid chromatographicproducts.

1.4

CRITICAL

REVIEW

OF

CURRENT

LITERATURE

METHODS FOR THE ANALYSIS OF CATIONIC TENSIDES


Since the discovery at the turn of the century, of the beneficial properties of
cationic

tensides and their subsequent widespread exploitation

(Section

1.2), a

Chapter 1: General Introduction

36
been
invested
into
has
optimising methods for quantifying
significant amount of work
these materials in a range of matrices. As this work has encapsulateda number of
decades, the types of methods used are numerous, and indeed, many have been
instrumental
by
techniques. However, a number of
modem
more
efficient
superseded
the older methods still remain, finding niches due to their ease of use and / or the
limitations of modem techniques.
The following sections provide a detailed and critical review of the analytical
techniques that are commonly applied to the determination of cationic tensides in a
In
historical
different
the
general,
matrices.
perspective will be limited and
range of
focus
instead
on current methods of analysis. Two recent reviews have
attention will
included in-depth coverage of the historical methods used to determine cationic tensides
(Boethling et al., 1992; Schmitt, 2001), and both of these were utilised as source
material during this review.

1.4.1 Non-specific methods


Many of the earliest methods for quantifying cationic tensides were based on
involving

non-specific

measurement, typically

fractionation

to determine the total cationic tenside active concentration (Schmitt,

the titration

of the analytes after

2001). These methodologies often only provided crude estimates of the actual cationic
tenside concentration due to cross reactivity

with fatty amines and other matrix

constituents. Of the earliest non-specific methods described for the analysis of cationic
tensides, only the colorimetric methods remain of interest today. Of particular note is
the so called "disulphine blue active substances" method that was optimised by Osburn
in 1982, after originally being reported by Waters et al. in 1976. The methodology is
based on the extraction into chloroform of the cationic tensides via complexation with
Association
dye-stuff.
of the cationic tenside analyte to the anionic dye
an anionic
in
in
the conjugation
change
a
results

system, which

allows

the quantitative

determination of the tensides by performing spectrometric analysis at a characteristic


2000).
(Burford,
wavelength

The disulphine blue active substancesmethod is applicable to all cationic


tensides, as it requires only the presenceof one fatty alkyl group and a cationic head
ionic
form
in
to
adduct with the . However, method selectivity and
a
strong
group, order
Chapter 1: General Introduction

37
dependant
investigation
the
the
tenside
on
nature
of
are
cationic
under
sensitivity
(Burford, 2000). Over years of use the method has been adapted on a number of
improve
improve
through
to
sample
put
and
analyte specificity. Indeed
occasions
recently, Unilever Research have managed to miniaturise the method described by
Osburn (1982), allowing sample throughput to be greatly increased,and solvent waste
to be minimised (Weston, 2000). Unfortunately due to the nature of the method, it is not
possible to obtain structural information on the nature of the tensidesthat are present.In
from
in
the
the
also
suffers
problems
method
with
cross-reactivity
occurring
addition,
fatty
it
be
tensides,
amines
and
amphoteric
and
cannot
of
presence
applied to the
due
back
into
to
metabolites
of
cationic
problems
analysis
with
extracting
chloroform.

1.4.2 Thin layer chromatography (TLC)


Thin layer chromatography is commonly utilised in industry to determine
cationic

tenside levels in raw material

supplies and fully

formulated

products

(Lawrence, 1997). The methodology is applicable to the determination of cationic fabric


conditioner actives and preservative actives (Paesen et al., 1994; Lawrence, 1997).
However, the developing conditions are very different. Whilst the separation of the
is
performed with the aid of a very polar solvent system, with the
preservative actives
hydrophilic

quats migrating to the greater extent, separation of the cationic fabric

conditioner actives is achieved with a non-polar elution solvent. In this scenario the
hydrophobic quat species migrate furthest down the plate (Lawrence, 2000).

Whilst both methods have been used to quantify cationic tensides in industrial
be
for
is
known
TLC
to
the analysis of environmental matrices due
unsuitable
matrices
to the fouling of the separationplates by analyte interferences(Lawrence, 1997).

1.4.3 Gas Chromatography (GC)


Cationic tensidesare non-volatile species,and thus are not directly amenableto
GC analysis.As a result, there are few reports on the direct analysis of thesecompounds
by GC. Instead, the cationic tensides are converted to their corresponding tertiary
This
improve
their
to
volatility.
process is usually achieved by the
amines
decomposition of the quaternary ammonium hydroxide species via Hoffmann
Elimination. The quat sample is reacted with an excess of alkali to form the
hydroxide
species, and only once the reaction is completed, are the
corresponding

UNIVERSITY
LEEDS
LIBRARY

Chapter 1: General Introduction

38
tertiary amines extracted from the reaction mixture with a suitable organic solvent, and
first
be
(1977a)
Takano
that
reported
et
al.
alkylbenzyl
quats
could
analysed
analysed.
in this manner when they combined a quat sample with sodium methoxide and N, Ndimethylformamide. After refluxing for one hour at 180C, the tertiary amines were
by
then
analysed packed-column GC (Takano et al., 1977a). In the same
extracted and
in
by
Hoffmann
degradation
injector
GC
this
group
performed
research
port,
a
year,
combining the quat sample with methanolic potassium hydroxide, and injecting the
injector
into
GC
290C
(Takano
1977b).
a
port,
maintained
at
et
al.,
mixture
reaction
Another alkali that has been used in a Hoffmann elimination reaction is potassium tertbutoxide, to analyse the nature and concentration of benzalkonium chlorides in
domestic sanitary wipes (Suzuki et al., 1989). More recently, Hoffmann degradation
has
been
in
tert-butoxide
used
association with GC/MS to identify
with potassium
in
Taiwanese
residues
sewage effluents and river water (Ding et al.,
quat
alkylbenzyl
2001). These researchersimplemented a combination of solid-phase extraction and
Hoffmann elimination methodologies to quantify individual

alkylbenzyl quat

homologues at 0.4 g/l levels in natural river-water. Comparison of the method with a
HPLC-UV
(Prince
method
et al., 1999) showed that the GC/MS
recently published
increase
in
ten-fold
a
sensitivity and equivalent reproducibility.
method offered
However, analysis time increasedwhilst sample preparation time doubled as a result of
the Hoffmann procedure.
Although Hoffinann Elimination is the principal method for increasing the
volatility of the cationic tensides, an early method described by Warrington, utilised
hydrogenation to form the tertiary amines (Warrington, 1961). Abidi (1980) also used
this method, but then went on to derivatise the tertiary amines to the corresponding
in
trichloroethyl
carbamate
moieties
order to facilitate analysis at low
cyanamide or
(Abidi,
1980).
temperatures
column
The need for derivatisation of long chain quaternary amines, can by
injector
by
the
of
port pyrolysis GC and GC/MS (Ng et al., 1986).Ng
use
circumvented
et al. (1986) employed this method to analyse cationic tensides samples by direct
injection into a hot injector port (>250C), with the separation of the resulting
benzyl
alkylbenzylmethyl
amines,
and
amines,
alkyldimethyl
chloride impurities being
The
DB-5
column.
capillary
responseof the alkyldimethyl amines was
performed on a
Chanter 1: General Introduction

39
then used to quantify the distribution and concentration of the alkylbenzyl quats in the
original samples.
The main drawback with the application of GC analysis to the quantitation of
in
is
difficulty
the
tensides
experienced
quantifying the di- and tri-chained quat
cationic
Few
1997).
(Lawrence,
reports are available on the analysis of cationic tensides
species
have,
have
than
those
that
one
alkyl
chain
unit,
and
many
of
more
with
shown that
formed
degradation
is
far
from
the
the
amines
origins
of
after
routine
quantifying
(Metcalfe, 1984). Indeed, quantitation of the mono-chained quats is also hampered in
the presence of other cationic tenside analytes and / or fatty amine species, which can
in
in
excess of 100 % (Metcalfe, 1984).
spiked recoveries
result

1.4.4 High Performance Liquid Chromatography (HPLC)


Today, HPLC is the preferred methodology for the analysis of quats, with
separationsbeing reported as early as 1974 (Nakae et al., 1974), for the analysis of
halides
(Nakae et al., 1974,1977). In 1980, Meyer adapted this
alkylpyridinium
300
by
i.
d.
a
x4
using
mm
methodology
p-Bondapak CN column with a 60 : 40
acetonitrile : 0.1 M sodium acetatemobile phase,adjusted to pH 5.0 with acetic acid, to
analysebenzalkonium chlorides (Meyer, 1980).
Nakamura et al. (1981) published the first of a seriesof paperson the separation
of alkylbenzyldimethylammonium chlorides using a 250 x4 mm i. d. column packed
/
ODS
ODS
TSK
5m
silica.
gel
packed columns have subsequentlybeen utilised
with
by other research groups to enable the separation of quaternary ammonium species.
However, they were found to be ill suited to the analysis of the cationic tensides, an
observation that was attributed to the strong electrostatic interactions betweenthe silica
interest
(Santoni
the
of
et al., 1993).
analytes
surfaceand
In 1988 de Schulter et al. used a novel mobile phase with two modifiers to
drugs.
The addition of an amine and a
quaternary
ammonium
of
separate a series
in
improved separationover that
to
mobile
system
phase
resulted
aqueous
an
sulphonate
However,
the
amine
present.
peak tailing was dramatically reduced,
obtained with only
when only the sulphonatewas present.

Chanter 1: General Introduction

40
The use of indirect UV absorbance was reported by Helboe (1983) who
by
the
a seriesof Nucleosil column, including silica 5, C18,
resolution offered
evaluated
C8, and CN, and Nucleosil 7 phenyl. The CN column was reported to offer the most
Larson
In
the
same
year,
et al. (1983) used a Partisil SCX column
efficient separation.
to separatea group of alkyltrimethyl quats. Huang used a Zorbax C8 column with a
5
indirect
:
water
mobile
phase
containing
mM
methanol
p-toluenesulphonic acid and
detection to identify dihydrogenatedtallowdimethylammonium chloride
UV
(DHTDMAC). However, problems of co-eluting impurities were noted. Metcalfe
reported routine analysis of the DHTDMAC compound Arquad HT on a -Bondapak
C18column with a 95 :5: 0.3 methanol : water : acetic acid mobile phasealthough no
evidenceof the attained separationwas forthcoming.
Simon et al. (1987) identified the aliphatic and polyamine intermediates of quat
manufacture using Nucleosil CN and C18 columns with an acetonitrile : water gradient
Dowle
et al. (1989) utilised a column packed with a similar
programme, whilst
polystyrene-divinylbenzene material to that used by Nakae et al. to separate amines and
60
A
38
species.
:
ammonium
:2 acetonitrile : water : acetic acid mobile
quaternary
for
being
all separations.
used
phase

In 1987 de Ruiter et al. published the first paper on the use of post-column ionin
detection
the
of quat species.After separationof the mono-, di- and
pair extraction
trihydrogenatedtallowmethylammoniumchlorides utilising a mixed mode cyano-amino
Partisil PAC 10 m column, (a gradient chloroform : methanol : acetonitrile mobile
being
employed) the analytes were mixed with methyl orange or 9,10
phase system
dimethoxyanthracene-2-sulphonate,extracted into chloroform and passed through a
into
the detection cell of a UV or fluorescence detector.
sandwich phase separator
Detection limits of 10 ng/l for fluorescencedetection were quoted. Subsequentpapers
have reported improvements in the efficiency of the system by the utilisation of a
Partisil PAC 5 column with a chloroform : methanol : acetic acid mobile phase
(Gort
1993;
Fernandez et al., 1996).
conditions
gradient
et
al.,
under
employed
Alterations and improvementsto the extraction systemhave also been reported.
Engelhardt et al. (1995) published a derivatisation method suitable for the
degradation
the
aminoalcohols,
products of the esterquats.The
analysis of quaternary
Chapter 1: General Introduction

41
in
formation
9-fluoroenylmethylformamate
the
resulted
of
addition
of a series of
by
fluorescence.
identified
Separations
were performed on a
which
were
complexes,
Nucleosil SA 5 column with a acetonitrile : water : 0.1 M sodium acetate gradient
system. The derivatisation step required 30 minutes to reach equilibrium, and in the
analysis of unknown samples,a number of reactions neededto be performed, in order to
formation.
of
complex
obtain a constantratio
Conductivity detection was first reported by Wee et al. (1982) who described the
four
quat species with a Partisil PAC 10 column and a 92 :8 chloroform :
separation of
Subsequent
by
Gerike et al. (1994) and Breen et al.
phase.
work
methanol mobile
(1996) have improved peak resolution by using a PAC 5 column and altering the mobile
phase to a 89 : 10 :1 chloroform : methanol : acetic acid system. It is interesting to note
that although the mono-, di- and tri- homlogous series were eluted as single peaks, the
is
the
three
classes
not as efficient as that obtained by de Ruiter et al.
of
separation
(1987).

Nitschke et al. (1992) used a similar system to identify a range of cationic


including
esterquats.Detection limits were reported to be in the low g /1
surfactants
range.
Alternative detection-systemshave been used previously by Spagnolo et al.
(1987). These researchersemployed a refractive index detector and a SCX column to
separate a mix of aromatic and nonaromatic quats. Although separations in a
homologous series are reported, they were only possible with the trimethylquats, as the
Caeser
as
single
eluted
peaks.
were
et al. (1989) utilised a similar detection
other series
dialkyl
in
determine
to
quats
commercial personal care formulations. Two system
Bondapak CN columns connected in series and a 55 : 45 : 0.1 water : acetonitrile
trifluoroacetic acid mobile phasewere employed.
A series of papers published by Wilkes et al. (1992,1994,1996) reported the
light
(ELS)
detector
to analyse groups the mono-, discattering
use of an evaporative
bromides
separated on a 250 x 4.6 mm i. d RSi1 polyol
tristearylammonium
and
highlighted
that the retention times obtained with the systemused
Wilkes
al.
et
column.
by Gerike et al. and Breen et al. were very much dependent on the methanol
Chapter 1: General Introduction

42
Using
the
the
and
analyte
concentration.
phase
of
mobile
a gradient elution
composition
in
hexane
based
5
5
trifluoroacetic
acid
and
on
mM
mM trifluoroacetic acid
programme
in THE : methanol 3: 1, Wilkes et al. were able to obtain stable retention times over a
injections
The
concentrations.
at
various
polyphenol column showed good
of
number
dibeing
trithe
and
constituents
mono-,
as well as
able to partially
separation of
homologous
of
a
series.
separate constituents

Mass spectrometry(MS) is well suited to the identification of minor components


first
Cotter
(1982)
the
matrix.
et
al.
papers
a
complex
published
one
of
within
present
in
the
use of mass spectrometry the analysis of cationic surfactants. Using directon
ionization
MS they were able to identify the presenceof a series of
exposurechemical
tenside molecules in a commercial sample. Lyon et al. (1984) extended this work with
the analysis of a commercial dihydrogenated-dimethylammoniumchloride. Fast atom
bombardment (FAB) mass spectrometry gave rise to abundant ions at m/z 550,522,
494,466 and 438. Collision-activated desorption spectra revealed these ions to
correspondto moleculeswith alkyl chain lengths ranging from C14to C18.
In 1988 Simms et al. (1988) published one of the first paperson the use of mass
spectrometry for the quantitation of cationic surfactants in environmental matrices.
Using isotopically labelled internal and external standards, FAB-MS and a signal
averaging technique, the group were able to analyse cationic surfactants at
concentrations approaching 100 ng/l. A subsequent paper by Simms et al. (1992)
FAB-MS
liquid
the
of
with
use
scintillation counting, in order to characterise
combined
the biodegradationpathway of a newly synthesisedester quat. Quantitation of the ester
formation
in
the
and
matrix
rate
of
of a biodegradation intermediate were
quat a sludge
in
1992,
Lawrence
(1992) used a continuous-flow fast atom
Furthermore,
reported.
bombardment interface to introduce analytes, which had previously been separatedon
W.
A
4.6
columns.
non-aqueousmobile phase system was utilised to
mm
standard
ditallow-dimethylammonium
the
of
concentration
quantify
chloride present in a
formulation.
fabric
conditioner
commercial
The use of ion-spray ionisation has also been reported in the analysis of cationic
identified
(1996)
Ogura
1996
In
the presenceof a protonated dialkyl
et
al.
surfactants.
tertiary amine in a commercial fabric conditioner, using flow injection analysis ionChapter 1: General Introduction

43
(1998)
MS.
Kawakami
utilised a similar system for the trace analysis of
et
al.
spray
benzalkonium chlorides on skin. A detection limit of 1.2 mg/1 was reported. More
ion-spray
(1998)
Di
Corcia
the
a
review
published
on
use
of
and electrospray
recently,
MS in the analysis of surfactants.Although the paper is primarily concernedwith the
it
highlights
the suitability of these techniquesin trace
analysis of non-ionic surfactants,
analysis studies.
Presently, Unilever Research, as with most major formulators of commercial
products, which utilise alkylbenzyl quats, have a standard method for the determination
forms
in
in
fully
formulated
their
these
purified
and
also
species
products. The
of
is
based
LC
a
on
reverse
phase
separation performed on a long cyanopropyl
separation
bonded stationary phase held at 35C, from which the analytes are eluted, with a mobile
phase containing a 0.2 mol/1 sodium perchlorate. Unfortunately,

this combination of

high
length,
sodium perchlorate concentration appears to lead to long
and
column
is
the
times,
as
a
result,
and,
mobile
phase
pumped at a high linear velocity,
analysis
from
the
column
achieving optimum performance.
preventing

Whilst it is unclear as to the exact influence of sodium perchlorate on the


behaviour of the analytes(this effect will be addressedin Chapter 5), what is clear, is
that such a high concentration of a non-volatile salt, hinders or even prevents the
hyphenation of the method with electrospray mass spectrometry.This presentsa major
formulators
for
draw
trying
to
when
up a risk assessmentfor the application of
problem
in
a new market, as their detection in complex matrices is
quats
alkylbenzyl
fact
by
that these speciescontain no fluorophore and show a maximum
the
compounded
in
/
UV
190-215
the
nm
vis range, the same range as many naturally and
absorbanceat
in
the environment. Therefore, in order to quantify these
ubiquitous
man-madeorganics
species in complex environmental matrices extensive sample preparation routines are
required.

1.4.5 Capillary Electrophoresis (CE)


Capillary electrophoresis (CE) would appear well suited to the analysis of
is
dependant
the
tensides,
separation
on the variation in the behaviour of ions
as
cationic
field
(Issaq,
2000). As the analytes possess a
influence
the
an
electric
of
under
in
the
the charge to size ratios of the individual
variation
charge
permanent positive
Chapter 1: General Introduction

44
homologues should allow for efficient resolution of the major species. The
by
to
suited
are
especially
analysis
capillary zone electrophoresis as
alkylbenzylquats
they possesan exploitable chromophore,which facilitates detection of these analytesby
UV analysis. As a result of this inherent chromophore there have been a number of
reports appearingon the analysis of this particular class of cationic tenside. Two recent
reviews (Vogt et al., 1998; Heinig et al., 1999) have demonstratedthat since CE was
first evaluated for the analysis of alkylbenzyl quats some nine years ago (Weiss et al.,
1992), the number of publications in this field has grown steadily to the presentdate.
Early methods for the analysis of alkylbenzyl quats utilised aqueousbuffers and
low organic solvent concentrations(Vogt et al., 1998). However, these methods were
soon replaced after the cationic surfactants were observed to favour formation of
/
micelles and or aggregatesat the walls of the fused silica capillaries (Vogt et al.,
1998). Today, most CE methods used to analysealkylbenzyl quats rely on the presence
of high concentrationsof organic solvent in the running buffer, or else the separationis
in
a completely non-aqueous environment (Heinig et al., 1999). Both of
performed
these methods of analysis have been successfully applied to the analysis of alkylbenzyl
(Heinig
in
products
commercial
et al., 1999). In a recent report, the presenceof aquats
high
-cyclodextrins,
a
and
concentration of organic solvent in a phosphatebuffer
or
system has been shown to afford excellent resolution of alkylbenzyl quat homologues
(So et al., 1999).
Unfortunately, the widespread applicability of CE to the analysis of cationic
tensides appearslimited at this moment in time for a number of reasons.Firstly there
for
detection
limited
systems
available
use with commercial CE equipment. Whilst
are
UV/ vis and fluorescence techniques are commonly applied to the analysis of other
2000),
(Issaq,
thesemethods are either completely unsuitable or can only be
compounds
applied to a small number of tenside classes.Whilst Reinig et al. (1997) have reported
the determination of four monoalkyltrimethyl quats by indirect UV / vis analysis, the
low
in
the
method
was
very
comparison with a conventional ion
repeatability of
Similar
have
observations
method.
also been made during the analysis
chromatography

Chapter 1: General Introduction

45
1.4.6 Indirect competitive enzyme-linked immunosorbant assays
Mention has already been made to the fact that the quantitation of cationic
tensides in complex matrices is far from routine (Section 1.3.2.1). Labour intensive
in
/
derivatisation
are
and
or
procedures
required
protocols,
extraction and clean-up
in
fractionate
facilitate
the
to
to
accurate
samples
order
and
pre-concentrate
order
determination of the analytes. Whilst such clean-up proceduresare critical to attaining
in
increase
in
data,
they
a
significant
analysis time, reducing sample
also result
valid
throughput. Another major limitation of all previously described methods is that they
back
in
field,
be
be
thus
that
the
samples
shipped
all
requiring
cannot readily performed
to the laboratory for analysis. For an on-going monitoring study, the necessityto ship
for
difficulty
in
lab,
large
back
the
to
the
a
and
coping
numbers
of
samples
with
samples
lead
in
In
to
sample
set.
can
problems
obtaining
a
representative
prolonged period,
have
been
hand,
in
field,
been
had
issues
to
that
the
remedied
could
results
addition,
in
delay
by
data.
be
the
receiving
study
compounded
could
The limitations described above are characteristic of those experienced by the
lumber trade, where freshly felled timber is treated with fungicidal

preparations

limit
(Chen,
to
tensides,
the
prevent
growth
cationic
of
mould,
and
sapstain
containing
1995). Due to concerns over worker health and environmental impact, and to ensure
for
fungicidal
treatment,
rapid
and
quantitative
methods
of
are
analysis
required
optimal
the determination of cationic tensides on lumber (Bull, 1998b). The scale of this task is
in
1995,
into
that
one
considers
when
cationic tensides were employed
perspective
put
in 95% of all sapstain products utilised by the massive Canadian lumber industry. In this
hundreds
to
the
set
sample
equates
per day, and where instantaneous
situation where
integrity
the
to
ensure
of the product, traditional methods of cationic
results are required
found
are
analysis

wanting.

As

immunosorbant assays (CELISA's)


of didecyldimethylammonium

a result,

indirect

competitive

linked
enzyme

have recently been developed for the determination

chloride (DDAC) (Chen et al., 1995; Bull et al., 1998b;

Chen et al., 1998) and benzyldimethyldodecylammonium

chloride (Bull et al., 1998a)

into
been
developed
field-portable
have
kits
for
Such
assays
on-site analysis,
residues.
high
have
the
criteria
aforementioned
of
sample throughput rates, cost
satisfied
and
(van
Emon
1992).
and
ease
of
use
repeatability
sensitivity,
et
al.,
effectiveness,

Chapter 1: General Introduction

46
Each of the assays that have so far been reported, involve the development of
polyclonal

is
a multi-stage process. The analyte of interest is
antibodies, which

transformed into a hapten in order to produce highly selective and specific antibodies
(Chen et al., 1995). The structure of the hapten should retain the conformation of the
impart
desired
the
those
groups
which
anti-fungal activity. In the
analyte, especially
fatty
DDAC,
the
two,
chains and the quatemised nitrogen cation are required
case of
(Chen et al., 1995), whilst the use of a "linker arm" maximises antibody production and
increases assay sensitivity (Chen et al., 1995, Bull et al., 1998a, 1998b). In order for the
hapten to be recognised as an immunogen within a mammalian system, and induce an
linked
it
be
to a large immunogenic carrier protein. As the
must
antibody response,
nature of quaternary ammonium salts prevents direct conjugation, a carboxylic acid
introduced
linker
is
the
onto
arm (Chen et al., 1995, Bull et al., 1998a,
often
group
1998b). Once a suitable immunogen has been developed, it is introduced into a
mammalian species i. e. a rabbit (Chen et al., 1995, Bull et al., 1998a, 1998b), which
then raises antibodies against the immunogen. By utilising the ELISA protocol, it is
in
titre
the
to
the rabbit's blood, and when sufficient levels
gauge
of
antibody
possible
are present, the antibodies are extracted and stored.

The sensitivity, specificity, and cross-reactivity of the antibodies produced by


the rabbit are then characterised by performing an indirect competitive ELISA
(CELISA) (Chen et al., 1995, Bull et al., 1998), in the same way as is used during
The
hapten-protein
determination.
conjugate (antigen) is coated onto
analyte
form
to
the sorbent base material. A mixture of antiserum and a
microtitration plates
known concentration of analyte is then added to each plate, and the competitive
adsorption of the analyte and the antigen allowed to reach equilibrium. After washing
the plate, a conjugated secondary antibody is added, which selectively binds to the
hapten-antibody that is attached to the antigen, sorbed onto the plate. Addition of an
initiator
causes rearrangementof the secondary antibody to facilitate the
enzymatic
formation of a chromophoric group that can be observedat 490 nm. Utilising a range of
inhibition
be
an
curve
can
set up, whereby increasing analyte
analyte concentrations
decreased
by
leads
to
the secondary antibody. The resulting
absorption
concentration
be
then
used to quantify an unknown amount of analyte in a sample of
profile can
interest. In terms of method sensitivity, the most selective hapten reported by Bull et al.

Chanter 1: General Introduction

47
(1998b) was seen to be capable of quantifying DDAC at concentrations down to 25
in
billion
(ppb)
spiked samples.
per
parts

Unfortunately, the CELISA methods do have limitations. A major disadvantage


is that the organic solvent employed during extraction has been shown to affect the
immunoreactivity of antibodies in immunoassays,with Bull et al. (1998a) reporting that
antibodies raised against benzyldimethyldodecylammonium chloride lost > 10%
reactivity in the presence of a number of organic solvents. Although not ideal,
incorporation of the solvent into the samples used to generatethe standardcurve, can
(Kramer
1994).
this
et
al.,
problem
alleviate
Another downfall of the method is that the antibodies produced for a specific
immunogen can recognise several epitopes of the antigen, with varying affinities. Thus,
"cross reactivity"
structurally

can occur whereby the polyclonal

antibodies

also recognise

similar compounds to the analytes of interest. In the case of cationic

tensides, fatty amines and other quat homologues can result in cross reactivity. As
mentioned previously, the sensitivity

and specificity

of an antibody is very much

dependent on the structure of the hapten. Bull et al. (1998b) developed three haptens for
DDAC with differing structures. It was observed that the hapten, which most resembled
the structure of DDCA,

showed a 600-fold

increase in sensitivity and selectivity

compared to that developed by Chen et al. (1995).

A final concern in the use of CELISA's for the analysis of cationic tensides
comes from the observation that the analytescan interact non-specifically with proteins,
denature
the antibody. However, Bull et al. (1998b) showed that
thus,
actually
may
and
the presenceof three other industrially important tensides had no noticeable effect on
the sensitivity of the method, and in addition, none of the three tensides were seen to
causesignificant cross-reactivity.
Though the use of CELISAs for the analysis of cationic tensides is still in its
infancy, the technique may well prove to be a rapid and effective screeningtool for the
in
tensides
environmental matrices in the future. Certainly, the
quantitation of cationic
(1998b)
by
Bull
al.
et
suggests that the method has promise.
sensitivity witnessed
However, ultimately, the rate at which ELISA's are incorporated into the suite of
Chanter 1: General Introduction

48
be
dependant
tensides
to
cationic
quantify
will
used
methods
on reducing
analytical
limiting
kits
interferences.
the
to
sensitivity
of
commercial
and
matrix
reactivity
cross

1.5

CURRENT
ANALYSIS

FAILINGS
OF CATIONIC

AND

FUTURE

NEEDS

IN

THE

TENSIDES

It was evident at the outset of this work that the suite of methods used to
in
industrial
tensides
and environmental sampleswere not sufficient to
quantify cationic
facilitate quantative and selective determination of all tenside components,of all tenside
layer
Thin
in
be
to
chromatographic
matrices.
all
methods
well
were
seen
samples,
suited to the analysis of cationic tensides in raw materials and formulated industrial
be
the
could
not
efficiently applied to environmental analysis due
method
products, yet
to plate fouling by matrix constituents.The application of gas chromatographicmethods
has been successfully applied to the environmental analysis of mono-chained quats.
However, derivatisation or degradationwas first required, and the problems experienced
during the quantitation of the di- and tri-chained quats limited the widespread
applicability of the method.
The non-selective disulphine blue active substances method and other
colourimetric methods were seen to provide data on the total concentration of active
in
As
information
sample.
structural
a
substances
on the nature of the compounds
is
it
the
with
method,
unobtainable
cannot be used for studying the fate of
present
in
In
the
tensides
environment.
addition, the cross reactivity of the method with
cationic
fatty protonated amines and amphoteric tensides results in the calculated concentration
being
above that which is actually present. Similarly, development
of active substances
linked
immunosorbant
enzyme
competitive
of
and application
assays(CELISA's) to the
is
in
its
infancy
due
tensides
to problems with crossstill
cationic
of
quantitation
reactivity and solvent effects.
Of the methods currently used to quantify cationic tensides high performance
liquid chromatography is both the most widespread and the most applicable. Yet the
be
designed
tend
to
for specific classesof tenside,
that
are
available
range of methods
from
inherent
in
light
this
many
still
suffer
of
problems. The reverse phase
and even
liquid chromatographic (RP-LC) methods that are generally applied to the analysis of
Chapter 1: General Introduction

49
the alkylbenzyl

and alkylpyridinium

hyphenation with

quats (Schmitt,

mass spectrometry, which

limits

2001)

are unsuitable

the availability

for

of structural

information on these analytes. At the same time, the methods have been found to be
instead
fabric
for
the
the
which
rely on
of
conditioner
actives,
quantitation
unsuitable
the use of normal phase LC methods for their quantitation, especially in environmental
demonstrate
Yet
2001).
(Schmitt,
these
methods
poor repeatability, are unable
matrices
to resolve the homologous series in each tenside class and are difficult to apply to the
determination

of

the hydrophilic

biodegradation

products

of

the parent quats

(Figure 1.10).

The aim of this work was to develop new liquid chromatographic methods
determination
in
for
tensides
the
quantitation
and
cationic
of
environmental
suitable
initial
The
industrial
products.
aims at the commencementof the work
matrices and
fabric
develop
LC
the
the
to
of
conditioner
resolution
actives
and
optimise
were
biodegradation
the
their
of
quantifying
parent
analytes
capable
and
methodology
hyphenation
At
ionisation
the
time,
same
with
electrospray
products simultaneously.
in
also
envisaged,
order that a method for studying the fate of
mass spectrometrywas
the parent cationic fabric conditioner actives in environmental matrices could also be
developed.The secondaim of the work was to optimise the RP-LC methods of analysis
used to quantify the alkylbenzyl quat preservative actives, and hyphenate the
in
electrospray
mass
spectrometry
with
order to study the transport of
methodology
these analytes through, and their removal from, the environment. Application of the
be
to
the
the
methodology
other
cationic
of
preservative
active
would
suitability
performed subsequentto method validation.
It was predicted that the development and subsequentoptimisation of liquid
for
the quantitation of the two main types of cationic tenside
chromatographicmethods
for
the
these components.
suite
of
analytical
methods
available
enhance
greatly
would
However, it was felt that the long-term benefits of two new methods for the specific
determination of discrete analytes would be limited. Chromatographerswould still be
left with the question "Which method do I usefor the determination of a sample in
"
Y?
matrix

Chapter 1: General Introduction

50
The plight of traditional analytical departments within the applied chemicals
industry is becoming increasingly bleak. Analytical facilities of any kind are seen in
luxury
be
by
buying-in
that
can
circumvented
expertise,
an
expensive
as
circles
many
decrease
in
funding
is
leading
the
to
a
available for many of these facilities. At
which
the same time, management is becoming increasingly wooed by the promises and
representations of contract research organisations (CRO's), who are able to use their
inherent "economies of scale" to match or even undercut the costs associated with inhouse support. In a buyers market, external competitors draw work from a broader
client base, and can therefore afford the extra investment required for new automated
demand.
laboratory
As
to
a
space,
which
are
required
meet
customer
and
equipment
have
been
by
facilities
both internal and external pressures, and
internal
squeezed
result
therefore must find and provide, the "added-value " that only long-term experience of a
fully
formulated
/
bring.
However, expertise on the
or
product
and
can
material
raw
in
dealing
in
the
the
and
experience
material,
with
matrix
of
or matrices
which
nature
they are to be found is insufficient to guarantee survival, as a potential customer can
internal
bought-in
An
facility
this
expertise.
with
achieve
must be able to deliver valid
data in a shorter time frame than can be achieved by external competitors. From a
development
implementation
this
the
requires
perspective,
and
chromatography

of

detection
for
of
extraction
and
methods
similar compounds in order to speed
generic
is
bottlenecks
development,
in the data generation process.
the
which
one
of
method
Whilst a generic methodology may not provide the perfect separation for all intended
it
will
compounds,

generate valid quantitative

data without

the need for method

development. As a result, the development of two new LC methods for the analysis of
deemed
tenside
to represent an unsuitable conclusion to
of
cationic
was
classes
specific
the work, in view of the current climate in industrial analytical facilities. Instead, a
further
knowledge
to
the
was
work
on the chromatographic behaviour
principal aim of
in
in
tensides
order that the road to a generic liquid chromatography
general,
of cationic
become
for
horizon.
these
the
the
may
of
materials
analysis
visible
on
method

Generic analysis is not a new concept in chromatography. In GC analysis it is


80%
be
that
of
all
separations
can
approximately
performed with the aid of a
reported
25m column of 0.25mm internal diameter. A stationary phase of 95% dimethyl
diphenyl
5%
film
together
polysiloxane,
thickness of 0.25 m
with
and
a
polysiloxane
2001),
(M`Nair,
whereas oven parameters may be varied and tuned
remain constant
Chapter 1: General Introduction

51
interest.
In
HPLC,
truly generic methods of analysis are
the
to
of
analyte
according
more difficult to come by. Whilst the use of an octylsilane or an octadecylsilanebonded
for
the majority of
reverse
phase
conditions,
under
accounts
stationary phase,operated
disparate
the
performed,
nature of stationary phase supports, and
currently
separations
the variation in mobile phaseparameters,makes the practice common but by no means
is
LC
For
the
temperature
the
truly
method,
only
generic
separation
a
at
which
generic.
performed and /or the solvent composition should be changed. Addition of modifiers
different
/
solvents
are
characteristics
of
of
a
method. Recently,
and or replacement
have
begun
for
have
the
to
the
which
appeared
address
requirements
reports
developmentof genericmethodsof analysis in LC (Needhamet al., 1999 and 2000).
The two reports by Needham et al. (1999 and 2000) were basedon the use of a
for
basic
drugs.
the
and
stationary
phase
mobile
parameters
set
of
analysis
of
generic
Whilst it was observed that greater efficiency was possible on cyanopropyl, phenylbased, and pentafluorophenyl-basedstationary phases, compared to that achieved on
traditional octadecylsilane bonded supports, the widespread applicability of the
impressive.
By carefully selecting mobile phaseparameters,the
most
methodology was
RP-LC methodology was seen to be applicable to a range of common basic drugs
(Needham, 2000), offering equivalent efficiency in all cases. It was subsequently
observed that the resolution and efficiency of the drug analytes was maintained during
hyphenatedLC/ESI-MS, due to the use of high organic solvent concentrations. As a
result, it was predicted that the new methodology should allow the rapid separationand
basic
drugs.
determination
of
novel
structural
It was envisagedthat by adopting many of the principles used by Needhamet al.
(2000), it could be possible to begin to develop generic methods of analysis for cationic
tensides.The ultimate aim of the developmentwould be to begin to progressdown the
for
LC
the quantitation of all cationic tenside
method
path towards a single generic
By
in
utilising electrospray mass spectrometry, the method
matrices.
all
components
benefits
for
to
LC
significant
offer
over
conventional
was envisaged
methods
in
tensides
environmental matrices. At the same, utilisation of
quantifying cationic
believed
for
to
were
offer
promise
the determination of the
profiles
elution
gradient
hydrophilic cationic metabolites formed as a result of biodegradation. It was predicted
be
also
applicable to the structural elucidation of
that such a powerful method would
Chanter 1: General Introduction

52
Initial
tensides.
unsynthesised
parent
as
yet
cationic
metabolites
and
steps on
unknown
the road to such a method, the removal of the dilemma over which analytical
burden
the
to
and
an
easing
of
of cationic tenside method
choose,
methodology
developmentwould be the ultimate goals of this work.

1.6

REFERENCES

Abian J., OosterkampA. J. and Gelpi E., J. Mass Spectrom., 1999,34,244-254.


Abidi S.L., J. Chromatogr., 1980,200,216-220.
Ahmad

R., Cartwright

Monitoring,

M. and Taylor F., Analytical

Methods for

Environmental

1stEdition, Prentice Hall, 2001.

Apffel A., Fischer S., Goldberg G., Goodley P.C. and Kuhlmann F.E., J. Chromatogr.
A, 1995,112,177-190.

.
Beers P.A., Personal Communications,2001, Unilever Research.
Berezkin V. G., Chromatography: A Separation Method or a Branch of Science?,
<http://www. richrom.com/html/bodyjocd_a. html>.
Berger H., Int. J. CosmeticSci., 1997,15.227-237.
Braithwaite

A. and Smith F.J., Chromatographic Methods, 5t' Edition, Blackie

Academic and Professional,Glasgow, 1996.


Bull J.P., SerreqiA. N., Chen T. and Breuil C., Water Research,1998a, 32,3621-3630.
Bull J.P., Serreqi AN,

Gamboa H.R. and Breuil C., J. Agric. Food Chem., 1998b, 46,

4779-4786.
Bluhm L. H. and Li T., J. Chromatogr. Sci., 1999,37,273-275.
Boethling R.S. and Lynch D.G., Quaternary Ammonium Surfactants, In: The Handbook
Volume
3,
Chemistry,
Part
F,
1st
Edition,
Environmental
Springer-Verlag, Berlin,
of
1992.
Branzoi V., Branzoi F. and Baibarac M., Materials Chem.And Physics, 2000,65,288297.
Breen D. G.P.A., Homer J.M., Bartle K. D., Clifford A. A., Waters J. and Lawrence J.G.,
Water Research,1996,30,476-480.
Burford M. D., Personal Communications,1997, Unilever Research.
Burford M. D., Personal Communications,2000, Unilever Research.
Caesar R., Weightman H. and Minty G.R., J. Chromatogr. 1989,478,191-203.
Cai B. and Li J., Anal. Chim. Acta, 1999,399,249-258.
Chanter 1: General Introduction

53
Chen T., Dwyre-Gygax C., Daniels C.R. and Breuil C., WoodScienceand Technology,
1998,32,287-296.
Chen T., Dwyre-Gygax C., Smith R.S. and Breuil C., J. Agric. and Food Chem., 1995,
43 1400-1406.
Cotter R.J. and HansenG., Anal. Chim. Acta, 1982,136,135-142.
Crutzen A. M., J Am. Oil ChemistsSoc., 1995,72,137-142.
de Schutter J.A. and de Moerloose P., J. Pharmaceutical and Biomedical. Anal., 1988,
6,879-885.
de Ruiter C., Hekens J.C.H.F., Brinkman U.A. Th., Frei R.W., Evers M., Matthijs E.
Anal.
Chem.
1987,31,325-339.
J.
Environmental
Int.
J.
A.,
Meijer
and
Di Corcia A., J. Chromatogr. A, 1998,794,165-185.
Ding W. -H. and Liao Y. -H., Anal. Chem.,2001,73 36-40.
Dobson B. C., Proceedingsof the 5" World Surfactants Congress,Florence, 2000.
Domagk L., Ullman's Encyclopedia of Industrial Chemicals, 5th Edition, Editor's:
Gerhartz W., Yamamoto Y. S., Kaudy L., Pfefferkorn R. and Rounsaville J.F., VCH,
Weinheim, 1987.
Dorsey J.G. and Cooper W.T., Anal. Chem., 1994,66,857A-867A.
Dowle C.J., Campbell W. C. and Cooksey B. G., Analyst, 1989,114,883-885.
Engelhardt H., Gotzinger W., Kramer M. and Witringer R., Acta Hydrochimica et
Hydrobiologica, 1995,23,173-179
Eshraghi J. and Chowdhury S.K., Anal. Chem., 1993,65,3528-3533.
Esposito E., Sebben S., Cortesi R., Menegatti E. and Nastruzzi C., Int. J
Pharmaceutics, 1999,189,29-41.
Fernandez P., Alder A. C., Suter M. J.-F. and Giger W., Anal. Chem., 1996,68 921929.
Friedli F.E., Koehle H.J., Fender M., Watts M., Keys R., Frank P., Toney C.J., Doerr
M. and Christakos G., Proceedings of the 5'h World Surfactants Congress, Florence,
2000.
Garrett H.E., SurfaceActive Agents, 1StEdition, PergamonPress,Oxford, 1972.
Gerike P., Klotz H., Koojman J.G.A., Matthijs E. and Waters J., Water Research,1994,
28,147-154.
Gort S.M, Hogendoorn E.A., Baumann R.A. and van Zoonen P., Int. J Environmental
Anal. Chem. 1993,53,289-296.
Gough M. A. and Langley G.J, Rapid Comm. Mass Spectrometry, 1999,13.227-236.
Chapter 1: General Introduction

54
Hamley I. W., Introduction to Soft Matter: Polymers, Colloids, Amphiphiles and Liquid
Crystals, 1StEdition, Wiley and Sons,New York, 2000.
Hansch C. and Leo A., Substituent Constantsfor Correlation Analysis in Chemistry
Ltd.,
New
York,
USA,
1979.
Sons
Wiley
John
Edition,
1"
Biology,
and
and
Helboe P., J. Chromatogr., 1983,261,117-122.
Hewlett Packard, Applications of narrow-bore columns in HPLC, 1997.
Haung C.B., J. Liq. Chromatogr. 1987,10,1103-1125.
Issaq H.J., Electrophoresis,2000,21,1921-1939.
Jungermann E., Cationic Surfactants, 1" Edition, Marcel Dekker, New York, 1970.
Katz E., Eksteen R., SchoenmakersP. and Miller N., Handbook of HPLC, Marcel
Dekker, New York, 1998.
Kawakami S., Callicott R.H. and Zhang N., Analyst, 1998,123,489-491.
Kennedy G.J. and Knox J.H., J. Chromatogr. Sci., 1972,10,549-562.
Kostiainen R. and Bruins A. P., Rapid Comm. Mass Spectrometry, 1996,0

1393-

1399.
Kramer P.M, Marco M. P. and Hammock B.D., J. Agric. and Food Chem., 1994,42,
934-943.
Kuhlmann F.E., Apffel A., Fischer S.M., Goldberg G. and Goodley P.C., J. Am. Soc.
Mass Spectrom.,1995,6 1221-1225.
Kmmerer K., Eitel A., Braun U., Hubner P.,Daschner F., Mascart G., Milandri M.,
Reinthaler F. and Verhoef J., J. Chromatogr. A, 1997,774,281-286.
Lagerwerf F.M., van Dongen W.D., SteenvoordenR.J.J.M., Honing M. and Jonkman
J.H. G., TRAC-Trendsin Analytical Chemistry, 2000,9

418-427.

Larson J.R. and Pfeiffer C.D., Anal. Chem. 1983,55.393-396.


Lawrence

C.A.,

Germicidal

Properties

of

Cationic

Surfactants,

In:

Cationic

0
Edition, Marcel Dekker, New York, 1970.
E.,
Jungermann
Surfactants, Editor:
Lawrence D. L., J ofAm. Soc.for Mass Spectrom., 1992,39 575-581.

Lawrence J.G., Personal Communications,1997, Unilever Research.


Lawrence J.G., Personal Communications,1999, Unilever Research.
Lawrence J.G., Personal Communications,2000, Unilever Research.
Lever Brothers, A hundredyears of Port Sunlight, 2000.
Levinson M. I., J. Surfactantsand Detergents, 1999,2 223-235.
Lin G.H. Y., Voss K. A. and Davidson T.J., Chemical Toxicology, 1991,29 851-854.
Lyon P.A., Crow F.W., Tomer K. B. and Gross M. L., Anal. Chem. 1984,56,2278-2284.

Chapter 1: General Introduction

55
Martin A. J.P. and SyngeR.L. M., Biochemical J., 1941,35,1358-1368.
Martindale,

The Extra Pharmacopoeia, 30`h Edition, The Pharmaceutical Press,

London, 1993.
McNair H.M., PhenomenexGC Seminar, RadissonSAS ManchesterAirport, 2001.
Metcalfe L. D., J. Am. Oil Chem.Soc., 1984,61,363-366.
Meyer V. R., Practical High-Performance Liquid Chromatography, 2d Edition, Wiley
York,
1993.
New
Sons,
and

Moore J.W. and Moore E.A., Environmental Chemistry, Academic Press,New York,
1976.
Nakae A. and Muto G., Chem.Letters 1974,6 549-550.
Nakae A., Kunihiro K. and Muto G., J. Chromatogr. 1977,134,459-466.
Nakamura K., Morikawa Y and Matsumoto I., J. Am. Oil Chemists.Soc., 1981,58,7277.
Nawrocki J., J. Chromatogr. A, 1997,779,29-71.
Needham S.R., Brown P.R. and Duff K., Rapid Comm. Mass Spectrometry, 1999,13,
2231-2236.
Needham S.R. and Brown P.R., J Pharmaceutical and Biomed. Analysis, 2000,23,
597-605.

Nitschke L., Mller R., Metzner G. and Huber L., FreseniusJ. Anal. Chem., 1992,342,
711-713.
Ng L. -K., Hupe M. and Harris A. G., J. Chrom., 1986,351,554-559.
Norberg J., Thordarson E., Mathiasson L. and JSnssonJ.A., J Chromatogr. A, 2000,
523-529.
Omar A. M. A. and Abdel-Khalek N. A., Tenside Surfactants and Detergents, 1997,34
178-182.

Ogura I., DuVal D.L., Kawakami S. and Miyajima K., J Am. Oil ChemistsSoc. 1996,
71137-142.
Osburn Q.W., J Am. Oil Chem.Soc., 1982,59 453-457.
Paesen J., Quintens I., Thoithi G., Roets E., Reybrouck G. and Hoogmartens J., J.
Chrom. A, 1994,677,377-384.
Piera E., Dominguez C., Clapes P, Erra P. and Infante M. R., J. Chromatogr. A, 1999,
852,499-506.
Prince S.J., McLaury H.J., Allen L. V. and McLaury P.J., J. Pharmaceutical and.
Biomedical Analysis, 1999,9 877-882.
Puchta It, J Am. Oil ChemistsSoc., 1984,61 367-376.
Chapter 1: General Introduction

56
Puchta R., Krings P. and SandkhlerP., TensideSurfactant and Detergents, 1993,30,
186-191.
Radke M., Behrends T., Frster J. and Herrmann R., Anal. Chem., 1999,71,53625366.
Reta M. and Carr P.W., J. Chromatogr. A, 1999,855,121-127.
Ruthven D. M., Encyclopedia of Separation Technology - Volume 1,1d Edition, Wiley
York,
1997.
New
Sons,
and

Santoni G., Tonsini A., Gratteri P., Mura P., Furlanetto S. and Pinzauti S., Int. J.
Pharmaceutics,1993,93,239-243.
Santoni G., Medica A., Gratteri P., Furlanetto S. and Pinzauti S., Il Farmaco, 1994,49,
751-754.
Schmitt T. M., Analysis of Surfactants, 2d Edition, Wiley and Sons, New York, 2001.
Scott M. J. and Jones M. N., Biochimica et Biophysica Acta, 2000,1508,235-25

1.

Simms J.R., Keough T., Ward S.R., Moore B. L. and Bandurraga M. M., Anal. Chem.
1988,60.2613-2620.

Simms J.R., Woods D.A., Walley D. R., Keough T., Schwab B. S. and Larson R.J., Anal.
Chem. 1992,64 2951-2957.
Simon P. and Lemacon C., Anal. Chem. 1987,59,480-484.
Solbe J.F., Personal Communication, 2000, Unilever Research.
Snyder L. R. and Kirkland J.J., Introduction to Modern Liquid Chromatography, Wiley
1979.
York,
New
Sons,
and
Spagnolo F., Hatcher M. T. and Faulseit B. K., J. Chromatogr. Sci., 1987,25.399-401.
Start GC website, <http: //members. kr. inter. net/guesu/gs/index. html>.

Suortti T. and Sirvi H., J. Chromatogr., 1990,507,421-425.


Suzuki S., Nakamura Y., Kaneko M., Mori K. and WatanabeY., J Chromatogr., 1989,
463,188-191.
Takano S., Takasaki C., Kunihiro K. and Yamanaka M., J. Am. Oil Chemists Soc.,
1977a, 54,139-143.
Takano S., Kuzukawa M. and Yamanaka M., J. Am. Oil Chemists Soc., 1977b, 54,
484-486.
C., The Hydrophobic Effect: Formation of Micelles and Biological
Membranes,2ndEdition, Wiley and Sons,New York, 1980.
Tanford

Taylor R.B., Toasaksiri S. and Reid R.G., J Chromatogr. A, 1998,798,335-343.

Chanter 1: General Introduction

57
Trius

A. and Brackmann B., Proceedings of the 5`h World Surfactants Congress,

Florence, 2000.

United States Geological Survey website, Contaminants in the Mississippi River


1987-1992,<http://water.usgs.gov/pubs/Circ/Circl133/organic.html>.
Valladao M. and SandineW.E., J. Dairy Sci., 1994,77,1509-1514.
W.,
Gerlach
R.
Bulletin
Environmental
Contamination
J.
M.
Emon
of
and
and
van
Toxicology, 1992,48 635-642.
An
Hermens
J.
L.
M.,
Assessment
C.
J.
Risk
Chemicals:
Leeuwen
and
of
van
Introduction, 1StEdition, Kluwer Academic Press,1995.
Versteeg D.J., Feijtel T.C.J., Cowan C.E., Ward T.E., and Rapaport R.A.,
Chemosphere,1992,245,641-662.
Vervoort

R. J.M., Debets R. J.J., Lamers R. J., Claessens H. A., Jansen J.G. M. and

Cramers C.A., J. Pharmaceutical and Biomedical Analysis, 1999,21,273-289.

Vievsky A., TensideSurfactantsand Detergents, 1997,34 18-21.


Warrington H.P., Jnr., Anal. Chem., 1961,33,1898-1900.
Waters J. and Kupfer W., Anal. Chim. Acta., 1976,85,241-249.
Waters J., How M. J., Barraft M. D., Haigh S.D., Hales S.G., Marshall S.J. and Pestell
T.C., TensideSurfactantsand Detergents, 1991,285,460-468.
Wee V. T. and Kennedy J.M., Anal. Chem., 1982,54 1631-1633.
Wehr T., LC-GC International Online, April 2000, <http://www. lcgcmag.com>.
Weston D.J., Personal Communication, 2000, Unilever Research.
Wilkes A. J., Walraven G. and Talbot J.-M., J. Am. Oil Chem.Soc., 1992,69,609-613.
Wilkes A. J., Walraven G. and Talbot J.-M., Journale Comite. Espanol Deterg., 1994,
25,209-220.
Wilkes A. J., Jacobs C. Walraven G. and Talbot J: M., Proceedings of the 41hWorld
SurfactantsCongress,Barcelona, 1996.
Yamaguchi JA., Ohmichi M., Jingu S., Ogawa N. and Higuchi S., Anal. Chem., 1999,
71,5386-5390.
Zhou X., Handie A., Salari H., Fifer E.K., Breen P.J. and Compadre C.M., J.
Chromatogr. B, 1999,728,273-277.

Chapter 1: General Introduction

58

CHAPTER TWO
Experimental

Chapter 2.- Experimental

59

CHAPTER TWO
Experimental
2.1

GENERAL
This chapter details the reagents,samples,equipment and techniquesrelevant to

the practical work described in subsequent results chapters. General descriptions of the
instrumental
and resource parameters utilised at the University
main

of Leeds and

Unilever Research Port Sunlight are detailed. Minor changes made during method
For
informative
time.
this
at
a
more
account regarding the
optimisation are omitted
is
individual
the
parameters,
reader
referred to the appropriate text and
of
optimisation
figure legends in the relevant chapters.

2.2

CHEMICALS, SOLVENTS AND GASES

2.2.1 University of Leeds


Glacial acetic acid (AcH) (AnalaR grade)- and dimethyl sulphoxide (DMSO)
(AnalaR grade) were purchased from BDH (Lutterworth, UK). Hydrochloric acid and
ammonia

solution

(Loughborough,
99+%),

formic

triethylamine

(Specified

reagents) were purchased from

Spectrophotometric

UK).

88%, American

acid (purity

(TEA)

(purity

grade trifluoroacetic

99% purity),

Chemical

ammonium

Fisher

Scientific

acid (TFA)

Society (ACS)
hydroxide

(purity
reagent),

(NH4OH)

(24%

(NH4Ac)
(purity
in
> 98%), ammonium chloride
acetate
ammonium
water),
ammonia
(NH4C1) (ACS reagent), tetramethylammonium

hydroxide

(TMAOH)

(25 % w/w

hydroxide
(TEAOH)
(35 % w/w in water)
in
tetraethylammonium
methanol),
solution
(40
hydroxide
(TBAOH)
tetrabutylammonium
weight % solution in water) were all
and
purchased from Sigma-Aldrich
N, N-dimethylammonium

Chemical Co. Ltd. (Dorset, UK). N-alkyl-N-benzyl-

chloride

(purity

> 90%)

was purchased from

Aldrich

Chemical Co. Ltd. (Dorset, UK).

All solvents were of HPLC grade or higher. Chloroform (CHCl3), methanol


(MeOH), tetrahydrofuran (THF), propan-2-ol (IPA), dichloromethane
(ACN)
diethyl
acetonitrile
and
ether
acetate,

(DCM),

ethyl

were purchased from Riedel-de-Haen


Chapter 2: Experimental

60
(Seelze, Germany). Riedel-de-Haen was also one of the suppliers of n-hexane, the
(Loughborough,
UK),
Ltd.
Philip
Scientific
Harris
Scientific
being
Fisher
and
others
(Lichfield, UK). All water used during the course of the reverse phase liquid
distilled.
double
was
chromatographywork
Oxygen-free-nitrogen (OFN) (purity >90%) was purchased from BOC gases
(Manchester, UK) and helium (purity >99%) was obtained from either BOC gasesor
Air Products Ltd. (Manchester,UK).

2.2.2 Unilever Research Port Sunlight


High purity reagentgradeTFA and AcH, Fisher Specified reagent gradeNH4Ac
from
Fisher
Scientific.
acid
propionic
were
purchased
grade
reagent
and analytical
HPLC reagentgradeNH4Ac was sourcedfrom JT Baker (Deventer, Netherlands).
All solvents were of HPLC grade or higher. McOH, hexane (hexane fraction
from petroleum), THF, IPA, CHC13and far UV grade ACN were purchasedfrom Fisher
Scientific. All water was obtained from a Milli

Q plus system (Millipore Ltd.;

Massachusetts,USA).

The nitrogen gas used for nebulisation within the Sedex 55 evaporative light
Alfortville,
France)
(See Section 2.4.1.2) was of >99%
detector
(Sedere;
scattering
by
Domnick
Hunter
NG35
(Domnick
a
nitrogen
generated
generator
and
was
purity
Hunter; Gateshead,UK). The nitrogen used as the sheath and auxiliary gas in the
Finnigan MAT LCQ benchtop ion-trap mass spectrometer (Finnigan MAT; Hemel
Hempstead, UK) (See Section 2.4.2) was provided by a Domnick Hunter Nitrox
LCMSUHPN1101 nitrogen generator(Purity > 99.5%). Helium gas (purity > 99.995%)
from
Products.
Air
was obtained

2.3

SAMPLES

It was highlighted in Chapter One that many tensides are referred to as


"oleochemicals" as they are downstream derivatives of natural fats and oils. It was
in
commercial samples containing a number of active
this
that
origin results
noted
being
dependant
distribution
the
on
the
present
number
and range of fatty
components,
Chapter 2: Experimental

61
acids present in the raw material. All of the samples studied during the course of this
work were oleochemicals,and thus some of the "pure" samplescontained in excessof
ten active components.Characterisationdata on chain length distribution, nature of the
fat / oil used during manufactureand the main industrial / researchuse of each of the
in
One.
is
Appendix
provided
samples

2.4

INSTRUMENTAL

PARAMETERS

It was observedearly on in the experimental work that it would be inefficient to


instrumental
instrumental
or
set-up,
a
generic
a
single
set
utilise
of
parameters
throughout the course of the study. As a result, the following

describe
the
sections

different
to
applicable
each
section of work.
resource
parameters
general

2.4.1 Normal phase liquid chromatography


2.4.1.1 University of Leeds
The instrumental set-up employed for normal phase HPLC work at the
University of Leeds comprised a Perkin Elmer (PE) Series Four quaternary HPLC
Research),
(ex-Unilever
and a Rheodyne Model 7125 stainless steel
system
pumping
injection valve (Phase Separations Ltd.; Clywd, UK) fitted with a 20 l external
injection loop unless otherwise stated. Two different detectors were utilised during this
work: A Perkin Elmer LC-25 detector (no suppressor column) was used for
Varex
Mk
II evaporative light scattering detector
and
a
experiments,
conductivity
(Alltech Associates; Carnforth, UK) was used for light scattering analysis. Initially,
helium (purity >99%) was chosenas the nebulising gas. However, this was replaced by
OFN due to the cost implications of running the system for long periods of time. The
55
pounds per squareinch (psi.) (379.5 kPa), and
at
regulated
nebulisation pressurewas
the evaporation temperatureat 95C. The output from the detectors was linked to a PC
(ThermoUnicam;
data
Hemel
ProGC
Hempstead, UK) for
software
capture
running
analysis and reprocessing.
Although a small number of separations were performed on a Partisil PAC
Bio-Sil
Clywd,
UK),
Separations;
(Phase
and
a
polyol column (Bio-Rad; Hemel
column
Hempstead, UK), much of the work in this section was carried out on a series of novel
based
Spherisorb
(see
Appendix Two for column
columns
available
and commercially

Chapter 2: Experimental

62
details). A series of six novel supercritically bonded mixed-mode alkylamino /
based
bonded
columns were manufacturedat the University of Leeds
silica
cyanopropyl
by Mark Robson and packed externally at Phase Separations(Carney, 1999; Dmoch,
1999; Robson, 1998). In addition to a number of commercial Spherisorb aminopropyl
formed
basis
in
the
the
the
this section.
of
much
phases
of
columns,
novel
work
However, commercial Spherisorb cyanopropyl, mixed mode octadecyl silane (ODS) /
(SCX)
exchange
and bare silica columns were also
aminopropyl, strong cation
evaluated.A guard column was not employed at any time.
All separationswere performed at ambient laboratory temperaturewith a mobile
phase flow rate of 1 ml/min. Initial mobile phase composition was identical to that
currently used by Unilever Research, namely 90: 10 CHC13 : MeOH with 10 mmol/1
AcH (Lawrence, 1997; Wee et al., 1982). However, this mobile phase was quickly
superseded by a ternary system of hexane, MeOH and THE modified with TFA.

The ratio of solvents and solvent mixing was controlled via the quaternary
pump. Two solvent reservoirs were employed, one containing

either chloroform

modified with AcH or hexane modified with TFA, the other containing either MeOH
modified with AcH or the polar constituents of hexane based solvent systems (normally
THF, MeOH modified with TFA). The desired solvent ratio was drawn from the
reservoirs, combined in the pumps' dynamic mixing chamber and then delivered to the
In
this way any variation between the specified mobile
the
rest of
analytical system.
in
the analytical system was consistent. All solvents were
that
phase composition and
degassed prior to the commencement of practical work by sparging with helium for
thirty minutes.

2.4.1.2 Unilever ResearchPort Sunlight


Separationswere driven by a PE Series 410 quaternary pump system (Perkin
Elmer; Beaconsfield, UK). Injection was performed by using either a Rheodyne Model
7125 stainless steel injection valve (Fisher Scientific) that was fitted with a PhaseSep
Event Marker (PhaseSeparations;Clywd, UK) to facilitate a synchronised start of the
data capture software, or a Kontron 360 autosampler (Bio-Tek Kontron Instruments;
Watford, UK). A Sedex 55 evaporative light scattering detector (Sedere; Alfortville,
France) was used for all separations. Nitrogen was used as the nebulisation gas
Chapter 2: Experimental

63
throughout. The output from the Sedex detector was relayed via a PE 900 Series link
box to a server running Perkin Elmer Turbochrom Client / Server software version
6.1.1. (Perkin Elmer, 1998). All method and sequence creation, along with integration,
Turbochrom
tests
the
the
were
performed
with
aid
of
suitability
and
system
reanalysis
software.

Normal phaseLC separationswere performed at ambient laboratory temperature


Leeds)
i.
d.
Spherisorb
(ex-University
3
4.6
150
of
amino
column
pm
mm
x
on either a
bonded
(Phenomenex;
3
150
Sphereclone
column
packed
with
silica
amino
m
or a
mm
Macclesfield, UK). Three different internal diameters of Sphereclone column were
i.
d.,
2.0
1.0
4.6
mm
and
mm
mm,
and all separationswere performed
evaluated,namely
For
4.6
flow
1
linear
mm
columns
a
ml/min was
velocity.
rate
of
with an equivalent
W.
flow
for
2.0
the
the
column
rate was reducedto 190 pl/min and for the 1.0
mm
used,
mm W. column the flow rate was reduced still further to 50 111/min.For the validation
fitted
SecurityGuard
cartridge
with two 4x2.0
exercisea

W.
Spherecloneamino
mm

filters (all Phenomenex;Macclesfield, UK) was introduced into the inlet end of the
column. The ternary hexane, MeOH and THE solvent system developed at the
University of Leeds was employed for all separations.The polar and non-polar solvents
were again held in separatereservoirs until needed, with the quaternary pump being
usedto control the solvent ratio.
Placement of a Jour X-Act 4 channel membrane degasser (Jour Research;
Onsala,Sweden)betweenthe solvent reservoirs and the pump inlet preventeddissolved
gasesfrom entering the analytical system.

2.4.2 Normal phase liquid chromatography I mass spectrometry


Normal phase liquid chromatography / mass spectrometry (LC/MS) experiments
were performed with the aid of a HP1100 integrated LC system (Hewlett Packard;
Waldbronn, Germany) consisting of a membrane degasser, a binary pump and a
standard auto-sampler, and a Finnigan MAT LCQ benchtop ion-trap mass spectrometer
fitted

with

an electrospray

ionisation

(ESI)

interface

(Finnigan

MAT;

Hemel

Hempstead, UK) operating in the positive ionisation mode. Full scan spectra were
obtained on the LCQ instrument across the range Wz 150 to 2000, with an ion time of
200 ms. The detector response was initially optimised with the aid of the automatic tune
Chapter 2: Experimental

64
function within the instrument software. However, additional optimisation of the Tube
Lens Offset voltage, the capillary temperature and the sheath and auxiliary gas flows
were performed manually.
During direct infusion studies and instrument tuning the integral syringe pump
was used to ensure uniform metered delivery into the instrument. For LC/MS studies,
front-end separations were performed on a 150 x 2.0 mm i. d. 3 m Sphereclone amino
column fitted with a SecurityGuard system at the column inlet (all Phenomenex;
Macclesfield, UK). A one-piece column-coupler (Anachem, Luton, UK) was used to
link the outlet of the column with the inlet of the ESI interface. The mobile phase again
flow
THE
delivered
MeOH
TFA,
hexane,
and
with
at
a
modified
and was
constituted
rate of 190 pl/min.

2.4.3 Reverse phase liquid chromatography


Reverse phase HPLC analysis was performed using a Waters Alliance HPLC
616
USA),
binary
(Waters;
Milford,
which
comprised
a
model
pump controlled
system
by a model 600s pump control module, and a model 486 tuneable wavelength detector.
Injection was performed manually with a Rheodynemodel 7125 stainlesssteel injection
fitted
)
10
injection
Ltd.
loop.
Separations
5
(Phase
The
with
a
external
or
l
valve
interface
detector
from
486
Dionex
the
was
connected
via
a
advanced
computer
output
to a PC running AI-450 chromatography automation software (Dionex Corporation;
Leeds,UK) for data capture,integration and reanalysis.
A number of silica basedLC stationary phaseswere evaluatedduring the course
Separations
this
were primarily performed on either a 150 x 4.6
practical segment.
of
(Phase
Separations;
Clywd, UK) or a 150
W.
3
Spherisorb
column
cyanopropyl
.tm
mm
(Phenomenex;
UK)).
Macclesfield,
i.
d
Luna
3
2.0
4.6
column
cyanopropyl
m
mm
or
x
However, Spherisorbsilica, methyl, hexyl, octyl, ODS2, mixed mode ODS/cyanopropyl
Zorbax
bond
SCX
evaluated,
along
with
a
also
stable
cyanopropyl
columns were
and
UK))
Clywd,
(see
Appendix
Two).
Separations;
(all
Phase
column
Separations were performed with a binary mobile phase system comprising
double
distilled
MeOH
the
ACN
solvent
strong
and
as
either
water or an
or
either
both
base,
of which were modified to the required pH with
aqueoussolution of a strong
Chapter 2: Experimental

65
TFA, as the weak solvent. It is worthy of note at this time that in the case of all reverse
phase LC and LC/MS work, modifiers were only added to the aqueous proportion of the
mobile phase, and thus pH adjustment was only ever performed on that reservoir. As a
result, the mobile phase passing through the analytical system would have demonstrated
in
higher
"apparent
than
that
the text.
specified
a
pH"

Separationsperformed on 4.6 mm i. d. columns were achieved with a mobile


phase flow rate of 1 ml/min. Flow rates were adjusted accordingly for columns with
different internal diameters in order to maintain a constant linear velocity. Solvents
were degassedthroughout the course of the practical work by continuously sparging
with helium gas.

2.4.4 Reverse phase liquid chromatography I mass spectrometry


An initial evaluation of reverse phase LC was performed at Unilever with a
Perkin Elmer Integral 4000 HPLC system,fitted with an integral autosamplerand diode
array detector (DAD). The output of the DAD was linked via a PE 900 link box to a
/
Client
Server software version 6.1.1. The Turbochrom
Turbochrom
server running
software automateddata capture, analysis and reanalysis. Chromatographic separations
were performed on either a 150 x 4.6 mm i. d. 3 pm Spherisorb cyanopropyl column
(ex-University of Leeds), or a 150 x 4.6 mm i. d. or a 150 x 2.0 mm i. d. 3 m Luna
cyano column (Phenomenex;Macclesfield, UK). Separationswere performed with a
binary mobile phase of ACN and Milli-Q water modified with a suitable base and
adjusted to pH 2.0 with TFA. The flow rate was adjusted according to the internal
diameter of the column in order to ensure separations were performed at a uniform
linear velocity.
Reverse phase LC/MS analysis was performed with the aid of a HP1100
integrated LC system and a Finnigan MAT LCQ benchtop ion-trap instrument. The
HP 1100 consisted of four modules, a membrane degas unit, a binary pump, a peltier
detector
diode
fitted with a standard flow cell. Full
array
cooled autosampler and a
Chemstation
instrument
the
the
achieved
was
via
software (Hewlett Packard;
control of
Waldbronn, Germany), the software was also used to control the output from the DAD,
instrument
LCQ
fitted
The
for
was
with an ESI interface and was
and
reanalysis.
ionisation
Both
full
in
the
mode.
scan and selected ion monitoring
positive
operated
Chapter 2: Experimental

66
(SIM) acquisition was used. In full scan mode a mass range of mz 200-500 was
acquired, whilst in SIM mode only m/z 303.5-304.5 was identified. In both modes of
operation the maximum ion time was set to 200 ms. The detector responsewas initially
function
tune
the
the
manual
within the instrument software.
aid of
optimised with
However, additional optimisation of the Tube Lens Offset voltage, the capillary
temperatureand the sheathand auxiliary gas flows were performed manually.
During direct infusion studies, and instrument tuning, analyte flow was
delivered via the integral syringepump on the LCQ instrument.
Front end LC separation was achieved on a 150 x 2.0 mm i. d. 3 m Luna
The
mobile phase composition was 50:50 ACN :5
column.
cyanopropyl

mmol/l

ammonium acetate buffer adjusted to pH 2.0 with TFA. The flow rate of the mobile
phase was set at 190 l/min. Post-column addition of modifiers was achieved with the
aid of a poly-ether-etherketone (PEEK) T-piece (Fisher Scientific) positioned between
the column outlet and the inlet of the ESI interface. This design allowed the constant
infusion of solvent and/or acidic modifiers into the flowing effluent stream, the rate of
delivery of the modifier again being controlled by the syringe pump on the LCQ.

2.5

REFERENCES

Carney R.A., Ph.D. thesis, University of Leeds,Novel Stationary Phasesfor HPLC and
CEC, 1999.
Dmoch R., Ph.D. thesis, University of Leeds, The Evaluation of Novel Supercritically
Bonded Chromatographic Stationary Phases, 1999.

Lawrence J.G., Personal Communication, 1997, University of Leeds.


Robson M. M., Personal Communication, 1998, University of Leeds.
Perkin Elmer, Turbochrom Client/Server UsersGuide, 1998.
Wee V. T. and Kennedy J.M., Anal. Chem., 1982,54,1631-1633.

Chapter 2: Experimental

67

CHAPTER THREE
Development of a new normal phase LC

for
fabric
the
analysis of cationic
method
conditioner actives

Chapter 3: A new NP-LC method or fabric conditioner actives

68

CHAPTER THREE

Development of a new normal phase LC methodfor the


fabric
conditioner actives
analysis of cationic
3.1

INTRODUCTION
With the widespreaduse of cationic tensidesas active agentsin down-the-drain

fabric conditioners there is a pressing need for methods of analysis capable of


distributions
the
alkyl-chain
of the major and minor active agents
accurately quantifying
in commercial formulations. The current methods used to quantify cationic tensidesare
non-specific, or they demonstrateother inherent problems that limit the application of
the method to industrial or environmental analysis. To date,there is no method available
in the literature that is capable of quantifying all the major cationic tenside actives
present in domestic fabric conditioners. Therefore, the aim of this section of work was
to develop a new LC method capable of quantifying cationic fabric conditioner actives
in bulk industrial formulations and environmental matrices.

3.2

LIMITATIONS

OF STANDARD

METHODOLOGY

Section 1.4 described the methods of analysis that are commonly used to
determine cationic tensidesin a range of different matrices. Though numerous methods
fabric
to
the
quantitation
of
conditioner actives, i. e. thin layer
are applicable
blue
disulphine
(TLC)
the
active substances(DSBAS) method
and
chromatography
(Section 1.4.1), normal phase liquid chromatography (NP-LC) is seen to be the most
(Schmitt,
in
literature
2001).
The
the
method
method is preferable
commonly utilised
due to its suitability for use in quality control protocols, being as it is able to distinguish
between different groups of tensides (Wee et al., 1982). The ease of hyphenating the
led
has
to the method providing unequivocal
methodology with mass spectrometry
information
on the concentration of cationic tenside residues in complex
structural
1996;
Radke
(Fernandez
et
al.,
et al., 1999).
matrices
environmental

Chapter 3: A new NP-LC method for fabric conditioner actives

69
The current standardoperating procedure (SOP) used by Unilever Researchfor
the analysis of the cationic fabric conditioner actives in environmental matrices, is
in
literature
described
(Wee et al., 1982; Nitschke et al.,
the
characteristic of methods
1992; Norberg et al., 2000). Separation is performed on a mixed-mode Partisil PAC
isocratic
(Appendix
Two),
an
mobile phase consisting of chloroform and
with
phase
improve
included
in
is
Acetic
the
to
also
mobile
phase
analyte peak
acid
methanol.
shapes. Detection is brought about with the aid of a conductivity detector, which is
Under
(Cooper,
2000).
theseconditions,
the
aid
of
a
suppresser
without
commonly used
the major active agents,the di-chained quats, are eluted first as a single peak, followed
by the mono-chainedquats,which also elute as a single peak (Section 1.4.4).
Whilst this method is capable of resolving different groups of tensides the
homologue
in
distributions
is
incapable
the
of
resolving
present
each of
methodology
these groups (Section 1.4.4). As the alkyl chain homologues show different
environmental profiles (Section 1.2.3.3), the ability to quantify the individual
componentsis a principal requirement for the effective environmental profiling of these
inability
from
Aside
to resolve cationic tenside homologues,the current
an
compounds.
Unilever SOP also suffers from poor reproducibility and the need for laborious
conditioning of the column to the cationic analytes at the commencementof a study
(Burford, 1997).

Before attempting to develop a new method of analysis for cationic fabric


Unilever
SOP
the
current
was utilised to assessthe severity of the
conditioner actives,
limitations described above. Figure 3.1 shows a typical chromatogram obtained from
the analysisof the HEQ sample(Appendix One) with the current Unilever method. The
first peak eluting after ca. 8.2 minutes corresponded to the major active agents, the
diester quats (Section 1.2.1), and the peak eluting after approximately thirteen minutes
(Appendix
One).
The two peaks were seento be
the
to
quats
monoester
corresponded
from
the
the age
of
poor
chromatography
much
whilst
stemmed
efficient,
and
poorly
detector
during
design
this work (Section 2.4.1.1),
the
employed
conductivity
and
of
(Lawrence,
is
1997).
the
methodology
of
endemic
poor peak efficiency
It should be noted at this time that the chromatogram shown in Figure 3.1 was
large
"slug"
injecting
first
a
of analyte onto the column to block the
obtained after
Chapter 3: A new NP-LC method for fabric conditioner actives

70
active sites on the silica substrate.This approach was found to be necessarybefore
irreversible
binding of the analytes to the
in
to
avoid
order
commencing a study
1997).
(Burford,
This problem was witnessed in the form of
during
use
stationaryphase
excessiveanalyte retention and low peak area response(Burford, 1997). Over time, the
active sites on the Partisil support were blocked giving rise to shortenedretention and
increased peak areas. Unfortunately, such dynamic modification was witnessed
throughout the evaluation of the Unilever SOP, and thus retention time stability was
being
by
to
the
seen
change
up to ninety secondsover the
poor, with
peak maximum
course of five sequential injections. Instability of this kind makes it difficult to
accurately define retention boundaries for the two groups of tensides in an automated
reprocessing method, which leads to additional analyst time being required for
reanalysis.

Figure 3.1: Chromatogramshowing the analysis of the HEQ sample with the current Unilever
method.
Conditions - Column: 250 x 4.6 mm i. d. 5 m Partisil PAC; Mobile phase: 85: 15 chloroform :
1
Flow
1%
rate:
ml/min; Detection method: Conductivity.
acid;
acetic
methanol modified with

The problematic nature of the Unilever SOP highlighted the need for the
development of alternative methodology that would afford improvements in analyte
resolution and retention time stability.

Chanter 3: A new NP-LC method for fabric conditioner


actives

71
3.3

EVALUATION

OF AN ALTERNATIVE

METHODOLOGY

During Section 1.4.4 referencewas paid to a series of papers that had reported
the analysis of cationic and amphoteric tensides with hexane-basedmobile phase
1996).
The
indicated
1992,1994
(Wilkes
three
that the use of
and
reports
et al.,
systems
hexane-basedmobile phasescould give rise to high efficiency separationsof cationic
tensides(Wilkes et al., 1992). Significantly, Wilkes et al. (1992) observedthat the peak
corresponding to the monoalkyl quats was split, suggesting resolution of the
homologous series (Wilkes et al., 1992). Although this group did not assess the
(Wilkes
1992,1994
1996),
the
time
of
method
et
and
previous
stability
al.,
retention
be
hexane-based
have
superior
reproducibility
suggested
may
achieved
with
reports
systems,rather than with traditional chloroform-based systems(van Damme et al, 1986;
Verzele et al., 1986). In light of these observationsthe decision was taken to evaluate a
hexane-basedNP-LC method and compareits performanceto that of the Unilever SOP.

3.3.1 Choice of the initial parameters


Wilkes et al. (1992) had utilised two polar solvents in the mobile phase system,
tetrahydrofuran (THF) and methanol (MeOH), being used in a 3: 1 ratio. As with the
Wee
(1982),
acid was added to the mobile phase to improve analyte
et
al.
of
method
Wilkes
(1992)
However,
et
al.
used trifluoroacetic acid (TFA) rather than
shapes.
peak
acetic acid at a concentration of 5 mmol/l, which improved mobile phase volatility
making the method more compatible
(Section

with

evaporative light

1.3.2.1). Separation was performed on a Bio-Sil

scattering detection

polyol stationary phase

(Appendix Two) with the aid of a gradient elution profile (Wilkes et al., 1992).

The polyol phasewas retained during the initial evaluation of the hexane-based
in
the choice of mobile phase parameters.
However,
was
apparent
variation
method.
Wilkes et al. (1992) employed a gradient elution profile of up to ninety percent polar
THE
Having
of
amounts
copious
encountered instrumental
solvent, and utilised
in
form
literature
damaged
the
the
solvent
system,
of
pump seals, the
problems with
Methanol
the
was
changed.
two
was utilised as the primary polar
polar solvents
ratio of
being
THE
MeOH
1.5:
1.
Whilst
final
it
the
of
:
ratio
was recognisedthat
with
modifier,
incompatibility
have
led
to
this solvent system could
problems due to the limited
hexane
(Phenomenex,
2001),
fears
MeOH
and
over instrument downtime
miscibility of
brought
about the change. However, contingency
and maintenance costs ultimately
Chapter 3: A new NP-LC method for fabric conditioner
actives

72
plans were put in place whereby the solvent system would have reverted back to the
literature method if problemshad beenencounteredduring use.
TFA was maintained as the acid modifier in the mobile phasesystem. However,
in another deviation from the previously reported literature method, the concentration of
the acid was raised to 25 mmol/l. This changewas not brought about consciously, but
was rather the result of an error during the initial preparation of the mobile phase that
was compoundedover time.
In the first instance,isocratic elution of the cationic tensideswas evaluated,as it
had been recognisedthat Wilkes et al. (1992) may have limited resolution by utilising a
rapid gradient elution profile. As a fundamental aim of the new methodology was to
resolve the homologous seriesof tensidesfound in commercial samples(Section 1.1.2),
it was envisaged that a reduction in the mobile phase strength might have facilitated
increased resolution. An evaporative light scattering detector was used in the method
after previously encountered problems with the use of the conductivity detector,
(Section 3.2) and in the knowledge of its compatibility with gradient elution analysis
(Wilkes et al., 1992).
Figure 3.2 shows the chromatogram achieved from the analysis of the HEQ
sample on the polyol bonded phase with an isocratic ternary solvent system based on
hexane. Two groups of peaks were apparent eluting at eight and nineteen minutes.
Whilst it was impossible to be certain of the identities of the peaks at this stage, the
characterisation data available for the HEQ sample (Appendix

One) indicated that the

early eluting peaks corresponded to the diester quats, whilst the late eluting peaks
corresponded to the monoester quats (Section 1.4.4). It was clear that both groups of
it
being
was tentatively hypothesised that resolution of the
peaks were
split, and
homologous series was therefore being witnessed.

Though a formal assessmentof the reproducibility of the methodology was not


performed at this time, the performance of a series of replicate injections showed that
the retention time stability was much higher than that witnessed during the evaluation of
the Unilever method. Additionally, it was observed that the conditioning of the new
involved
less
process, compared to that with the Partisil PAC
column was a much
Chapter 3: A new NP-LC method for fabric
conditioner actives

73
drift
Although
time
was still observedat the commencementof a study,
retention
phase.
less time elapsedbefore reproducibleretention times were obtained.

Figure 3.2: Chromatogramshowing the separationof the HEQ samplethat was achieved on the
Bio-Sil polyol column with a ternary mobile phasesystem basedon hexane.
Conditions - Column: 250 x 4.6 mm i. d. 5 m Bio-sil poly-ol; Mobile phase: 80: 12:8 hexane :
MeOH : THE modified with 25 mmol/1 TFA; Flow rate: 1 ml/min; Detection method:
Evaporative light scattering.

3.3.2 Evaluation of alternative stationary phases


Whilst the new hexane based NP-LC methodology yielded significant
improvements in peak efficiency and analyte retention time stability over the
SOP
Unilever
method, analysis time was seen to increase by over 50%.
conventional
For a method that was ultimately aimed at routine usage,increasedanalysis time would
have resulted in increasedassociatedcosts. A method was therefore sought, in which
faster elution could be achieved.
Initial attempts to reduce the analysis time by increasing the eluting strength of
the mobile phase were unsuccessful. Reducing the run-time to less than eighteen
minutes resulted in co-elution of the two groups of peaks into two discrete peaks.Whilst
increasing the mobile phaseflow rate could have yielded improvements in run time, the
LC methodology was being developed for hyphenation with electrospray ionisation
increase
in
Therefore,
(ESI-MS).
an
mobile phase flow rate would
mass spectrometry

Chapter 3: A new NP-LC method for fabric conditioner


actives

74
have adversely affected interface source dynamics and would have resulted in the
being
hyphenated
(Section
1.3.2.1).
the
compromised
methodology
of
sensitivity
An additional problem with the methodology was that the polyol column could
only be sourced from one manufacturerand would soon be obsolete due to insufficient
demand (Cooper, 1998). Whilst this initially causedinconveniencewith delivery times,
it was realised that a new SOP could not be basedon an obsoletephase.In light of these
problems, and the increasedanalysistime, alternative stationary phaseswere sought that
in
time
and
retention
resolution
stability
a shorter time period.
equivalent
would yield

3.3.2.1 Evaluation of the Partisil PAC phase


With the original Unilever SOP yielding short analysis times, the Partisil PAC
phase was assessed in association with the hexane-based mobile phase system. Figure
3.3 shows the chromatogram that was attained from the analysis of the HEQ sample on
the Partisil PAC phase with a mobile phase containing hexane, MeOH and THE
modified with 25 mmol/l TFA. It was apparent that the PAC phase afforded less
resolution than the polyol phase. However, the phase did appear to offer some benefits,
as there was evidence of four major peaks and a shoulder in the first group of peaks,
rather than the three major and two minor components seen in Figure 3.2. Having
assumed that the peaks corresponded to the diester components, the larger number of
peaks was more representative of what would be expected from the HEQ sample
(Appendix

One).

3.3.2.2 Assessment of a series of mixed-mode aminopropyl / cyanopropyl

bonded phases
Whilst it was apparentthat the commercial Partisil PAC phaseyielded reduced
increased
in
the
to
the
polyol
phase,
number of partially resolved
resolution comparison
investigate
desire
led
to
the resolution that could be achieved with
to
a
components
PAC
Whilst
the
the
exact
nature
of
phase was unknown, literature
similar phases.
reports suggestedthat aminopropyl and cyanopropyl groups were bound to the silica
(Appendix
Two). In an
secondary
amino
groups
additional
with
substrate, along
the
the
to
cationic tensides,a seriesof mixed-mode alkylof
resolution
attempt optimise
developed
(Section
2.4.1.1).
It was envisagedthat by
/
were
phases
amino cyanopropyl
bonded
the
units on the silica substrate, it would be possible to
the
varying
nature of
Chapter 3: A new NP-LC method for -fabric conditioner
actives

75
assess how resolution on the PAC phase came about, and ultimately maximise
based
As
the
on a spherical silica substrate,more uniform
were
phases
new
resolution.
leading
from
higher
the
to
the
substrate
was
envisaged,
silica
analytes
release of
increased
resolution.
efficiency and

Figure 3.3: Chromatogram showing the separation of the HEQ sample on the Partisil PAC
column.
Conditions - Column: 250 x 4.6 mm i. d. 5 m Partisil PAC; Mobile phase: 80: 12:8 hexane :
MeOH : THE modified with 25 mmol/1 TFA; Flow rate: 1 ml/min; Detection method:
Evaporative light scattering.

Initially three phaseswere manufactured by a supercritical bonding method that


had been previously shown to result in a high ligand density (Dmoch, 1999; Robson,
1998). The three columns, referred to collectively as "Type A supercritically bonded
(Appendix
columns"
mixed-mode

Two), contained varying ratios of aminopropyl and

highest
M1
Column
the
showed
percentage of cyanopropyl units,
cyanopropyl units.

Two).
lowest
(Appendix
M3
the
whereas
showed
Figure 3.4 shows the chromatogram obtained from the analysis of the HEQ
sampleon column M3. It was apparentthat the separationwas analogousto that seenon
initial
3.2)
(Figure
Bio-Sil
with
an
seriesof three major peaks,and two
the
polyol phase
late
to
two
as
a
group,
prior
eluting peaks.Comparison of
small additional peakseluting
Figures 3.2 and 3.4, revealed that the resolution obtained on the new mixed-mode
in
higher
that
than
previously,
achieved
and
addition, analysis time was seen
phasewas

Chapter 3: A new NP-LC method for fabric conditioner actives

76
to haven fallen by approximately 33%, even though the polarity of the mobile phasehad
been significantly reduced.

Figure 3.4: Chromatogramshowing the separationof the HEQ sampleon the M3 column.
Conditions - Column: 150 x 4.6 mm i.d. 5 pm M3; Mobile phase:85:9:6 hexane: MeOH : THE
I
Flow
TFA;
25
ml/min.
rate:
mmolIl
modified with

Whilst the resolution achievedwith column M3 was superior to that observedon


the poly-ol phase, the separations witnessed on columns M1 and M2 were very
different. Both columns demonstratedthe same retention characteristicsas column M3,
i.e. short analysis times with low polarity mobile phases.However, on both phasesthe
M3.
Figure
3.5
lower
that
than
column
on
achieved
shows the variation
resolution was
in the resolution of the first group of peaks, thought to be the diesters, on columns M1
the
the
It
that
M3.
number
of
aminopropyl
as
groups
on
silica
apparent
was
and
in
improvement
the
the
three
increased,
resolution
of
major components
an
substrate
M2,
found
the
This
column
trend
on
where
resolution
was
repeated
was
was witnessed.
to be intermediateof M1 and M3.

3.3.2.3 Assessmentof a series of mixed-mode alkyl-amino / cyanopropyl


bonded phases
Having witnessed a distinct effect upon changing the ratio of the aminopropyl
developed
the
silica
surface,
a
second
series
of
columns
on
were
and cyanopropyl units
to test the influence of secondary amine groups on the resolution of the cationic
Chapter 3: A new NP-LC method for fabric conditioner actives

77
tensides. The three columns in this group, referred to collectively
supercritically

bonded mixed-mode columns"

ratios of alkyl-amino

(Appendix

Two),

as "Type B

contained varying

Two) and cyanopropyl units bonded to the silica

(Appendix

lowest
M4
Column
the
showed
percentage of alkyl-amino
substrate.

groups, and

column M6 showed the highest (Appendix Two).

145

125

105

sE
85
0
a
N

65

45

25
4

4.5

5.5

6.5

Time (minutes)

Figure 3.5: Chromatogram showing the variation in resolution of species present in the HEQ
sample on columns MI and M3.
Conditions - Column dimensions: 150 x 4.6 mm i. d. 5 m packed with Spherisorb bonded
blue
Ml,
M3;
Mobile
Red
Column:
85:
9:
hexane
trace
6
trace
phase:
silica,
: MeOH : THE
modified with 25 mmol/I TFA.

Figure 3.6 shows a typical chromatogram obtained from the analysis of the
HEQ sample on column M6. Resolution was superior to that achieved with column M3.
and analysis time was reduced.

When the chromatograms achieved on the three Type B columns


increased
it
trend
that
the
of
resolution
was evident
compared,
concentration of aminopropyl

with

were

increased

units witnessed with the Type A columns, was not

Whilst
the
with
alkyl-amino
the
phases.
to
extent
same
column M4 yielded the
repeated
lowest resolution. the separation was still superior to that depicted in Figure 3.4. In
between
little
the resolution achieved on column
apparent
variation
addition, there was
M5 and column M6.
Chapter 3: A new NP-LC method for fabric conditioner
actives

78

Figure 3.6: Chromatogramshowing the separationof the HEQ sampleachievedon column M6.
Conditions - Column: 150 x 4.6 mm i.d. 5 pm M6; Mobile phase: 85:9:6 hexane: MeOH : THE
modified with 25 mmol/l TFA.

Consideration of these observations appearedto indicate that the alkyl-amino


far
in
HEQ
the
the
components
at
resolving
effective
present
sample
more
units were
than conventional aminopropyl units. With retention under normal phase conditions
interactions
being
between the analytes and the
to
electrostatic
attributed
normally
it
1994),
hypothesised
(Dorsey
that each alkyl-amino unit
et
al.,
was
stationary phase
could interact with the analytesvia the three amine groups. This would have amplified
behaviour
in
in
HEQ
the
the
the
of
electrostatic
components
sample,
of
any variation
for
increased
the
to
the
and
phases,
would
aminopropyl
account
resolution
comparison
B
These
led
belief
Type
increased
to
the
that
the
the
results
phases.
on
witnessed
in
M3,
M1,
due
increased
to
to
the
comparison
column
was
with
witnessed
resolution
in
to
functionalities
the
the
reduction
not
and
number of cyanopropyl
of
amino
number
groups.

3.3.2.4 Attaining improved column performance


An observationthat proved difficult to justify was the lack of a discernible trend
in the resolution of the three Type B phases.After examining a number of hypotheses
be
to
the complexity of the HEQ sample. It was
appeared
the most viable explanation
diester
first
if
to
the
corresponded
the
of
peaks
group
that
quats, then at least
predicted
in
120
different
were
eluted
a
secondwindow. With the variation in
twelve
components
Chapter 3: A new NP-LC method for fabric conditioner actives

79
the electrostatic behaviour of the homologues expected to be small, especially in the
it
/
C18
/
C18
C16
C20
the
components,
was envisaged that many of the
and
case of
As
hypothesised
the
tentatively
to
resolution
a
result,
was
co-elute.
components would
bonded
in
to the silica substrate.
the
of
amine
groups
presenceof an excess
plateau
Having observedthat resolution generally improved with increasing numbers of
higher
that
the
alkyl-amino
phases
yielded
resolution than the
and
amino groups,
equivalent aminopropyl phases,a pure alkyl-amino phaseappearedto provide the most
hypothesis.
by
Unfortunately, packing problems
to
test
this
which
effective method
being
in
the
effectively useless.It was predicted, without strong
new column
resulted
evidence that limited improvement in resolution would have been witnessed from a full
alkyl-amino bondedphase,comparedto that witnessedon column M6.
One other additional stationary phase was evaluated at this time, a
supercritically bonded gallic acid phase (Appendix Two). The development of this
phase came about as a result of claims that poly-phenol phases could yield higher
diol,
than
silica,
aminopropyl, cyanopropyl and poly-ol phases
resolution
conventional
(van Damme et al., 1986; Verzele et al., 1986 and 1987). In addition, the poly-ol phase
had also afforded increasedresolution over the PAC phase (Section 3.3.2.1). Though
the chromatogramrelating to the analysis of HEQ on the gallic acid phaseis not shown,
retention and resolution were equivalent to that witnessed on column M4, which was
by
Damme
the
van
et al. (1996) and Verzele et al., (1996). Due
reports
unexpectedafter
to a lack of information on the nature of the phasesused in literature reports, it was
impossible to determine whether the variation in retention was due to variations in the
bonded phaseunits that were employed. However, the increasedresolution afforded by
the alkyl-amino phasesresultedin interest in the poly-phenol phasesbeing dropped.

3.3.3 Evaluation of alternative mobile phases


Having attempted to optimise the stationary phase on which the separation of
be
based,
fabric
would
attention was turned to the mobile
cationic
conditioner actives
further
the
McOH
the
In
methodology,
to
to THE was
optimise
ratio
of
phase. order
for
impact
their
assessed
solvents
on resolution. It was
varied and alternative polar
in
MeOH:
THF
had
the
that
ratio
a significant effect on the
slight changes
evident
increased,
first
the
When
the
resolution
was
of
ratio
group of peaks suffered,
separation.
Chanter 3: A new NP-LC method for fabric conditioner
actives

80
long
increasing
THE
the
yielded
analysis times. In the same way, the
content
whilst
dichloromethane,
isopropanol
THE
and
ethyl
acetate,
with
resulted in a
replacement of
loss of resolution. It was evident that for isocratic elution purposes the initial starting
parametersafforded the optimum separation.
In addition to evaluating alternative solvent systems,the choice of acid modifier
was also reassessedin light of the well-documented problems of TFA-derived ionsuppression in hyphenated LCIESI-MS methodologies (Kuhlmann et al., 1995).
Unfortunately, direct replacementof TFA with either formic acid or acetic acid led to a
loss of resolution. As a result, it was concluded that TFA had to be maintained in the
methodology.

3.4

PEAK ELUCIDATION

BY CO-INJECTION ANALYSIS

One of the major limitations experienced when assessing the merits of the
different stationary phasesand solvent systemswas the lack of information on the peak
identities. Whilst it was tentatively hypothesised that the first group of peaks
corresponded to the diester quats and the two late eluting peaks (Figure 3.6) were
attributable to the monoesters(Section 3.3.1), confirmation of the peak identities was
urgently required. Having not yet reached the stage where a hyphenated LC/MS
for
be
structural elucidation, the commercial tenside sample
used
methodology could
individual
instead
componentsto evaluatethe effect on peak areas.
was
enriched with
Having been unable to obtain any of the single componentspresent in the HEQ
sample, a second commercial esterquat sample, diethylesterdimethylanunonium
One)
(Appendix
(DEEDMAC)
was analysed on the M6 phase (Figure 3.7).
chloride
Like HEQ, DEEDMAC is employed as an active agent in some modem fabric
Having
1.2.1).
(Section
formulations
obtained samplescorrespondingto the
conditioner
major diester components,and the two principle monoester components present in the
commercial sample, co-injection analysis was subsequentlyused to speciatethe major
peaks.
Figure

3.8 shows an overlay of the chromatograms produced when the

C18
/
DEEDMAC
the
C16diester quat sample
sample
was
enriched
with
commercial
Chanter 3: A new NP-LC methodfor fabric conditioner
actives

81
(red trace) and the C16/ C16diester quat sample (blue trace) (Appendix One). For
is
first
the
group of peaks shown, as no other change was apparent in the
clarity only
in
large
The
the
the
second
of
peak
group
area
response
showed
peak
a
chromatogram.
increaseafter enrichment with the CIS/ C16diester quat sample,whilst enrichment with
the C16/ C16diester resulted in a similar increase in the third peak. When the sample
/
diester,
increase
C18
C18
in
the
the
was
with
an
witnessed
enriched
subsequently
was
had
Support
forthcoming
first
been
in
therefore
the
that the
the
group.
of
peak
area of
first set of peakscorrespondedto the diester quats. More significantly, the methodology
had been seen to partially resolve the homologous series present in the commercial
sample.

Figure 3.7: Chromatogramshowing the analysis of the commercial DEEDMAC sample on the
M6.
Conditions - Column: 150 x 4.6 mm i. d. 5 pm M6; Mobile phase: 85:9:6 hexane: MeOH : THE
TFA.
25
mmol/1
modified with

Figure 3.9 shows a comparison of the chromatograms achieved when the


C18
DEEDMAC
the
with
sample
was
enriched
monoester quat (red trace)
commercial
figure
focuses
The
(blue
late
trace).
C16
the
two
the
quat
on
monoester
eluting peaks
and
in
It
the
further
chromatogram.
was apparent that enrichment
change occurred
as no
led
increase
in
C16
to
first
Cis
the
an
the
the
monoesters
peak
area
of
and
with
and
late
Evidently
two
the
eluting peaks correspondedto the two
secondpeaks respectively.
in
DEEDMAC
the
(Appendix
One).
present
sample
components
monoester
major
Sample enrichment provided strong evidence of the nature of the major peaks
DEEDMAC
the
during
commercial
the
of
analysis
sample, and indeed
witnessed
Chapter 3: A new NP-LC method for fabric conditioner
actives

82
from
forthcoming
data
the
characterisation
provided with the
was
evidence
additional
sample (Appendix

One). Previous analytical investigation had revealed that the diester

the
the
the
sample,
with
monoester components also
of
components
major
were
quats
having a significant presence. Assessment of the fatty acid composition showed that
approximately

60% of the 'fatty"

chains were unsaturated octadecyl chains, with

hexadecyl
being
30%
chains (Appendix
unsaturated
another
DEEDMAC
the
three
that
the
components
of
major
expected

One). It was therefore


sample were the C 18/C 18,

C18 / C16, and Cl(, / C16 diester components. With the approximate ratio of C18 to C16
being 2: 1, it was expected that the peak area of the C 18/C 18and C 18/ C16 would be
/
diester
C16
C16
twice
that
the
of
component, assuming
equivalent and approximately
no steric effects were experienced during manufacture. Subsequent evaluation of the
peak areas of the three major peaks mirrored these predictions.

Commercial DEEDMAC

800

sample

700

600
5 500
E
400
0
CL
N

300

200
100

0
3.25

3.5

3.75

4.25

4.5

4.75

Time (minutes)

Figure 3.8: Figure showing the result of enriching the commercial DEEDMAC sample with the
C1R/ C16(red trace), and the di C16(blue trace) DEEDMAC diester samples.
Conditions - Column: 150 x 4.6 mm i. d. 5 m M5: Mobile phase: 85: 9: 6 hexane : MeOH : THE
Sample:
Red
TFA:
DEEDMAC
trace
C18
/
C16
25
commercial
mmol/l
and
modified with
DEEDMAC diester. blue trace - commercial DEEDMAC and C16/ C16 DEEDMAC diester.

Chapter 3: A new NP-LC method for fabric conditioner


actives

83

45
43

Commercial

41

C 18 nionoester

DEEDMAC

sample enriched

with:

homologue

C 16 monoester homologue
39
37

35

33
N

10

31
29
27

25
10

02468

12

Time (minutes)

Figure 3.9: Figure showing the result of enriching the DEEDMAC sample, with the C18 (red
trace), and C16(blue trace) monoester components.
Conditions - Column: 150 x 4.6 mm i. d. 5 m M6; Mobile phase: 85: 9: 6 hexane : MeOH : THE
Red
Sample:
TFA;
25
trace - commercial DEEDMAC and C18
mmol/l
modified with
C16
DEEDMAC
blue
trace
and
monoester.
commercial
monoester,
-

Though single tenside components could not be sourced for the HEQ sample,
data
in
this
the
that
the
revealed
characterisation
major
of
components
close scrutiny
in
DEEDMAC
identical
the
to
those
sample were

sample, and thus peak identities

from
However,
be
the
the
observation
of
consistent.
chromatogram
achieved
would
M6
HEQ
the
on
column
showed two additional peaks eluting with
sample
of
separation
the C 18and C 16monoesters that were not witnessed with the DEEDMAC sample. Once
in
data
Although
these
two
proved
vital
speciating
components.
characterisation
again,
DEEDMAC
HEQ
the
and
structurally similar
fatty-alkyl chains. In the case of DEEDMAC
(Appendix
group
atom via an ethyl-ester

tensides do vary in the position of the


the fatty chains are linked to the nitrogen

One), and thus cleavage of the fatty-esters,

fatty
HEQ
In
the
two
the
of
case
esters are contained on the
yields an ethanolic moiety.
fatty-ester
the
Removal
ligand.
groups would therefore yield either the
of one of
same
2- or the 3-isomer. Characterisation data suggests that the 3-isomer should be formed on
3.10).
(Figure
most occasions

The four discrete monoester peaks observed in Figure 3.10 revealed that the
both
isomers,
of
resolving
NP-LC
capable
also
positional
was
method
as well as
new
Chapter 3: A new NP-LC method for fabric conditioner actives

84
It
homologous
therefore
that
the
concluded
was
new method
series.
a
of
members
fully
the
to
materials
and
ensure
quality
the
raw
of
characterise
potential
showed
formulated products. If the methodology was subsequently proved to be quantifiable
it
be
it
that
tool
used
as
a
rapid
screening
then
could
envisaged
was
reproducible,
and
for assessing ester quat concentrations in a range of samples.

Figure 3.10: Figure showing the resolution of the monoester positional isomers present in the
HEQ sampleon column M5.
Conditions - Column: 150 x 4.6 mm i.d. 5 m M6; Mobile phase: 85:9:6 hexane: MeOH : THE
TFA.
25
mmol/l
modified with

Until this point the effectiveness of the hexane-basedNP-LC method had only
been assessedfor the analysis of the second-generationesterquats(Section 1.2.1). For
for
be
the analysis of all cationic fabric conditioner
to
the methodology
proved effective
first
dialkyl
(Section
1.2.1)
the
the
of
generation
resolution
quats
of
actives, assessment
from
3.11
the
Figure
the
chromatogram
achieved
analysis of
shows
required.
was also
the Arquad HT sample (Appendix One) on column M6. As was observed with the
/
C18
C18,
C181
C16,
to
the
corresponding
three
peaks
and
principle
esterquats,a group of
C16/C16dialkylquats dominated the trace (Figure 3.12). Two late eluting peaks were
later
C18
C16
the
confirmed
as
and
monoalkyl quats
were
again evident, which
respectively.

Chapter 3: A new NP-LC method for fabric conditioner actives

85

Figure 3.11: Chromatogramshowing the analysis of the Arquad HT sampleon the M6 column.
Conditions - Column: 150 x 4.6 mm i. d. 5 pm M6; Mobile phase:90:6:4 hexane : McOH : THE
modified with 25 mmol/1TFA.

Whilst the chromatogramobtained from the analysis of the Arquad HT sample


during
the analysis of the HEQ and DEEDMAC, the
to
those
obtained
was analogous
3.9
after
minutes (Figure 3.11) had not been witnessedprior to
major peak seeneluting
this point. With the peak eluting earlier than the dialkyl quats, it was envisagedthat the
inflated
hydrophobicity compared to the dialkyl
showed
component or components
indicated
data
Characterisation
the samplecomprised approximately 15% trialkyl
quats.
quat (Appendix One), which was consistent with the observedpeak area and the early
elution time. However, only one peak was observed and the calculated number of
trialkyl components was in excess of thirty. This apparent discrepancy was
in
the eluting strength of the mobile phaseled
a
reduction
subsequentlyclarified when
to the single peak being split into a seriesof co-eluting peaks and shoulders.Ultimately,
it proved impossible to accuratelyquantify any of the individual trialkyl quats presentin
the Arquad HT sample. However, with the trialkyl quats being "active" impurities it
Similar
total
trialkyl
the
to
quat
concentration.
quantify
problems were
was sufficient
T
(Appendix
Arquad
One),
during
derived
from
the
analysis
of
a
sample
experienced
known
The
distribution
to
tallow.
was
sample
contain
a
of saturated
non-hydrogenated
fatty
in
length
from
C14to
acids
ranging
alkyl-chain
poly-unsaturated
and
and mono
C20.As a result, the chromatogramwas seento consist of three peaks corresponding to
It
dialkyl,
components.
monoalkyl
the
and
was concluded that for cationic
the trialkyl,
Chanter 3: A new NP-LC method for fabric conditioner
actives

86
tenside samples derived from non-hydrogenated fats and oils, only the total trialkyl,
dialkyl and monoalkyl concentrations could be obtained with the new methodology.

1020

Figure 3.12: Chromatograms obtained from the analysis of the Arquad HT and
dihexadecyldimethylammonium bromide samples on column M6. The two traces are overlaid to
assist peak elucidation.
Conditions - Column: 150 x 4.6 mm i. d. 5 .tm M6; Mobile phase: 90: 6: 4 hexane : MeOH : THE
Sample:
TFA;
Blue
25
trace - Arquad HT, red trace mmol/1
modified with
dihexadecyldimethylammonium bromide.

3.5

OPTIMISING

THE

METHODOLOGY

AND ENVIRONMENTAL

FOR

INDUSTRIAL

ANALYSIS

It was apparent that the new hexane-based NP-LC method offered significant
benefits over the conventional Unilever SOP. However, a number of limitations were
identified that were envisaged to limit the applicability

of the method unless further

limitations
first
The
in
TFA
these
the
of
was
presence
of
optimisation was performed.
in
is
the LC analysis of organic amines, as the
TFA
commonly used
the mobile phase.
forming
limit
the
stable
pseudo-neutral
analytes,
adducts,
which
with
acid associates
As
1.3.2.1).
(Section
is
limited,
interaction
a
result,
secondary
retention
silanol-analyte
facilitating increased resolution and reduced peak tailing. Unfortunately, whilst the use
it can lead to severe problems during LC/MS
HPLC
in
beneficial
is
TFA
methods,
of
1999)
Weston,
(Section
1.3.2.1).
1995;
It
(Kuhlmann
al.,
et
was envisaged that
analysis
be
hyphenated
ESI-MS,
in
LC
ultimately
would
with
the
methodology
order to
new
as
Chapter 3: A new NP-LC method for fabric conditioner
actives

87
in
fabric
it
levels
trace
conditioner
actives
environmental
cationic
of
matrices,
quantify
future
in
TFA
hyphenated
limit
the
to
associated
with
use
of
problems
was necessary
LGMS methodologies. The observationthat the use of acetic and formic acid resulted
in a loss of resolution in comparison to TFA (Section 3.3.3) required that the level of
TFA be modified. A damage-limitation exercise was therefore undertaken to minimise
the potential effect of TFA in a hyphenatedLC/MS methodology.
In Section 3.3.1 it was stated that the concentration of TFA utilised in the
during
had
the original
as
a
result
of
a
mix-up
occurring
come about
mobile phase
Indeed,
had
Wilkes
(1992),
the
system.
obtained
phase
et al.
mobile
preparation of
efficient resolution of a series of cationic tensides with a mobile phase containing
5 mmol/1 TFA, albeit in associationwith a gradient elution profile. In order to limit the
in
in
hyphenated
for
ion-suppression
LC/MS
the
the
method,
change
potential
resolution of the esterquatswas assessedas the concentration of TFA was reduced.
Utilisation of 10 mmoUl and 5 mmol/1 TFA yielded no apparentchangein the resolution
increase
However,
by approximately
time
to
the
analysis
was
seen
cationic analytes.
of
45% in the case of a mobile phasecontaining 5 mmoVl TFA. Further reduction in the
being
led
to
compromised and excessiveretention of the
resolution
acid concentration
monoesters. It was therefore concluded that in spite of the increased analysis time
witnessed with 5 mmol/1 TFA, the perceived reduction in ion-suppression,resulted in
the lower acid concentrationbeing maintained from here on in.
The second major limitation of the new methodology was the choice of
bonded
The
supercritically
phases had been developed at the
stationary phase.
University of Leeds to improve the resolution of the esterquatsin comparison to two
available columns (Section 3.2.2). Whilst the novel alkylfor
it
/
suitable
peak
were
elucidation
phases
studies,
was
amino cyanopropyl
impractical to assumethat they could be utilised in a method that was being groomed to

commercially

become the industry standard.One of the main reservations with the use of the Bio-Sil
from
be
it
that
sourced
only
could
one manufacturer and would soon
poly-ol phasewas
be obsolete (Cooper, 1998). This was far from ideal, as the method would need to be
became
In
the
the
redundant.
case of the novel supercritically
phase
when
redeveloped
bonded phases,the situation was far worse. Having been manufactured in-house at the
University of Leeds, the columns had not undergone any quality control protocols to
Chapter 3: A new NP-LC method for fabric conditioner
actives

88
As
the methodology would likely
and
reproducibility.
performance
satisfactory
ensure
be employed in a Good Laboratory Practice and / or Good Manufacturing Practice (GLP
lack
documentation
GMP)
of
verifying the fitness for
and
accredited environment,
purpose of a new column would prohibit the use of the methodology. As a result, work
identify
to
a commercially available stationary phase that offered
was undertaken
equivalent resolution to the novel mixed-mode phases.
Resolution of the cationic tensides had been maximised on mixed mode alkylbonded
In
/
during
the evaluation of the three
phases.
addition,
amino cyanopropyl
bonded
increase
/
to
stationary
phases,
cyanopropyl
resolution
was
seen
aminopropyl
with an increase in the concentration of aminopropyl groups present on the silica
light
In
3.3.2.2).
(Section
of these observations, the performance of a
substrate
commercial Spherisorb aminopropyl phase was assessedin the new methodology.
Figure 3.13 shows a typical chromatogram obtained from the analysis of the HEQ
diester
The
the
this
of
phase.
resolution
and monoester components was
sample on
M5,
to
that
on
column
observed
with the monoestercomponents still being
equivalent
into
3-isomers
3.4).
In
light
2(Section
the
and
of the guaranteedperformance
resolved
and reproducibility of commercial stationary phases, and the ability to source a
from
different
a
number
of
column
replacement
suppliers, the Spherisorb aminopropyl
for
the
the new NP-LC methodology.
as
column
of
choice
phasewas adopted

3.5.1 Analysis ofStepantex


The new normal phase LC methodology had been found to be suitable for the
DEEDMAC
HEQ
the
and
esterquats.In order to test the applicability of the
analysis of
fabric
to
conditioner actives, a commercial Stepantex
commercial
other
methodology
sample (Appendix One) was analysedwith the optimised methodology. The structure
but
DEEDMAC,
is
to
Stepantex
that
of
with one of the methyl groups
related
closely
of
by
being
to
the
replaced
an ethanolic group (Appendix One).
nitrogen atom,
attached
As a result, Stepantex components are more hydrophilic than corresponding
HEQ
DEEDMAC.
in
In
found
or
addition, the ethanolic group is
either
components
known to undergo esterification, which results in the commercial sample containing a
high percentage of triester quats (Lawrence, 1999). With the commercial Stepantex
from
hydrogenated
being
tallow, the commercial
partially
manufactured
material

Chanter 3: A new NP-LC method for fabric conditioner


actives

89
Stepantexsample was also known to have a more complex fatty acid distribution than
One).
(Appendix
DEEDMAC
HEQ
or
either

Figure 3.13: Chromatogram showing the analysis of the HEQ sample on a commercial
Spherisorbaminopropyl column.
Conditions - Column: 150 x 4.6 mm i.d. 3 m Spherisorb aminopropyl column; Mobile phase:
85:9: 6 hexane : McOH : THE modified with 5 mmoUl TFA.

Figure 3.14 shows the chromatogramthat was achieved from the analysis of the
Spherisorb
large
The
Stepantex
on
a
aminopropyl
sample
phase.
stationary
commercial
large
five
to
the
the
total
triester
corresponded
minutes
quat species,
peak eluting after
number of components present preventing resolution. The group of peaks eluting
between eight and ten minutes were identified as the diester components.Again, due to
the sample being only partially hydrogenated, it proved impossible to quantify
individual components of this series, even after reducing the strength of the mobile
be
diester
to
the
total
The
only
utilised
calculate
quat
could
method
phase.
concentration.
During the initial analysis of the Stepantex sample, no evidence of the
from
injection.
However,
the
twenty
minutes
after
point
of
monoesterswas apparent
during subsequentanalyses(Figure 3.14), a broad peak was seen to elute prior to the
had
for
C16
to
front,
that
the
peak
area
response
similar
a
expected
which
solvent
due
hypothesised
It
that
to
the
was
presenceof the ethanolic
monoester components.
limits
for
the
the
outside
set
eluting
were
analysis, and thus were
the
monoesters
group,
injection.
Further support for this theory was
the
the
next
being carried-over to
start of
Chapter 3: A new NP-LC method for fabric conditioner actives

90
C18
be
the
trace,
to
monoester
components
the
where
were
seen
of
at
end
apparent
late
these
Confirmation
the
of
components. was subsequently
elution
of
eluting.
forthcoming when the analysis time was increasedto thirty minutes. Two broad peaks
27.5
25.5
and
minutes, which correspondedto the
minutes
ca.
at
were witnessedeluting
Ci8:
C16
C18:
Cls:
the
two
C18
and
and
monoester
three
i,
2,
monoester components,
0,
C16:
C16:
and
components,
i, respectively.
0

Triesters

Diesters

'

Carry- over

Monoester

22

Figure 3.14: Chromatogram obtained from the analysis of the Stepantex sample on a
Spherisorbaminopropyl phase.
Conditions - Column: 150 x 4.6 mm i.d. 3 pm Spherisorb aminopropyl column; Mobile phase:
85:9:6 hexane : McOH : THE modified with 5 mmol/I TFA.

3.5.2 Resolution of the components of mixed esterquat samples


The late elution of the diester and monoester components of the Stepantex
from
facilitate
these
known
the
those
to
of
components
resolution
of either
was
samples
HEQ or DEEDMAC under the Unilever SOP method (Lawrence, 1999). Five peaks
Stepantex
HEQ
DEEDMAC
triesters,
the
to
the
or
were witnessed corresponding
diesters, the Stepantexdiesters, the HEQ or DEEDMAC monoesters, and finally the
Stepantex
(Lawrence,
1999;
Cooper
2000).
the
monoesters
peak,
strongest retained
However, utilisation of the SOP method could not facilitate the resolution of the
Instead,
DEEDMAC
diester
HEQ
samples.
the
and
a
single
peak and a
componentsof
(Cooper,
2000).
A
NP-LC
modified
the
witnessed
are
version
peak
of
single monoester
has
(1982),
been
Wee
developed,
by
in
recently
et
al.
which
results
published
method
Chapter 3: A new NP-LC method for fabric conditioner actives

91
diester
total
the
and monoester components of the two samples
partial resolution of
(Radke et al., 1999), but with no resolution of the four separate homologous series
(Figure 3.15).

100 Total DEEDMAC

Total HEQ diesters


C
diesters 12.22
TIC
3.33e5
A

X0.32

Internal

2.50

5.00

71.

standard

10.00

15.00

Tim

Figure 3.15: Figure showing the total ion chromatogramobtained from LC/ESI-MS analysis of
HEQ
DEEDMAC
diesters
during a recent literature report
and
containing
a mixed sample
(Radke et al., 1999).
Conditions - Column: 150 x 1.0 mm i. d. 5 m Partisil PAC column; Mobile phase: Gradient
elution with chloroform, acetonitrile and methanol modified with 0.1% glacial acetic acid.

Comparisonof the chromatogramsachieved from the analysis of the commercial


HEQ and DEEDMAC samples(Figure 3.6 and 3.7) have shown that the retention times
diester
the
and monoester components of the two samples vary. It was
of
major
therefore envisagedthat it should be possible, during the analysis of a mixed sample,to
different
from
the
samples and partially resolve the homologous
components
resolve
Figure
3.16
tensides.
cationic
of
shows the chromatogramthat was
group
seriesof each
achieved from the analysis of a mixed sample containing both the commercial
DEEDMAC and HEQ samples.The resolution was found to be much higher than that
(1999),
DEEDMAC
HEQ
by
Radke
the
as
and
al.
et
componentswere not only
reported
but
homologous
from
the
resolution
of
partial
other,
each
series was also
resolved
Stepantex
HEQ
Subsequent
the
to
the
of
commercial
addition
sample
mixed
witnessed.
/ DEEDMAC sample, studied in Figure 3.16, revealed that the new NP-LC
be
to
in
the
three
components
quantify
of
used
esterquat
could
samples
a
methodology
Stepantex
triesters
DEEDMAC
The
to
diesters, which
the
eluted
prior
analysis.
single
diesters.
HEQ
The
Stepantex
diesters
the
with
co-eluted
partially
were found to be
again
Chapter 3: A new NP-LC method for fabric conditioner
actives

92
DEEDMAC
the
though
from
the
of
sample,
and
components
and
all other
well resolved
HEQ monoesters demonstrated partial co-elution, the Stepantex monoester eluted as
in
from
distinct
all others the trace.
two
peaks. separate

UI: ED1C dcriNed


components
225

DI! EDMAC C,
And
111':Q C

II E.Q deriN cd
components
ti

175
E

a,
N

N
CD

C,

DFEDMAC

FIEQ
C,
6

125

ct

75

14

16

22

20

18

25

024

10

12

14

16

18

20

22

Time (minutes)

3.16: Chromatogram showing the separation of a mixed sample containing the


diester
DEEDMAC
IIEQ
quat samples achieved on a Spherisorb aminopropyl
and
commercial
Figure

phase.
Conditions - Column: 150 x 4.6 mm i. d. 3 m Spherisorb aminopropyl column; Mobile phase:
85: 9: 6 hexane : MeOH : THE modified with 5 mmol/l TFA.

The observations made during the analysis of the mixed diester quat samples led
to the realisation that the new methodology may have had the potential to determine
individual

in
tenside
residues
rivers, wastewater, and sediments. It was
cationic

in
Figure
3.16, was maintained during the
if
the
witnessed
that
resolution
predicted
study

of

environmental

matrices,

individual
of
concentrations

it

would

be possible

to

quantify

residual

components, derived from a known sample origin. In

in
be
to
assist
studies evaluating the removal rates of
used
addition, the method could
the different diester quats, in order to build a more accurate environmental profile of
these analytes.

3.5.3 Analysis of the cationic metabolites


Studying the fate and biotransformation of the esterquats after their release into
biodegradation
to
the
the
quantify
major
ability
the environment requires
products that
breakdown
In
Section
the
during
1.2.3.3 an outline
formed
of
parent
the
species.
are
Chapter 3: A new NP-LC method for fabric conditioner
actives

93
in
diester
Removal
the
breakdown
the
quats
the
parent
environment.
of
was presentedof
the
fatty
the
subsequent
species,
whilst
removal
of
monoester
yields
group
acid
one
of
formation
in
the
fatty
the
correspondingquaternaryamino-alcohol
of
results
acid
second
(Section 1.2.3.3 and Appendix One). In the caseof HEQ and DEEDMAC the resulting
from
breakdown
diols,
the
triol
to
of
whilst a
usually results
speciesare often referred as
the Stepantexesters(Appendix One).
To test the effectiveness of the new NP-LC methodology in the qualitative
diol
DEEDMAC
HEQ
diol
triol
the
and
and
metabolites,
analysis of the cationic
diethanoldimethyl
chloride, and

species, trimethylammonium propane-1,2-diol


One),
In
iodide
(Appendix
were
analysed.
view of the greater
ammonium
hydrophilicity of the quaternaryamino-alcohols, it was necessaryto increasethe eluting
strength of the mobile phaseat the commencementof the study.
Figure 3.17 shows the chromatogram obtained from the analysis of the HEQderived diol specieson a Spherisorbaminopropyl stationary phase.The severefronting
demonstrated by this component was very unusual and was initially believed to be
/
incompatibility.
Subsequent
overload,
or
and
sample-solvent
of
sample
representative
investigation revealedthat sampleconcentrationhad little effect on peak shape,and thus
the poor chromatographywas attributed to an incompatibility between the analyte and
the eluting solvent. As a consequenceof the poor peak shape,the observed efficiency
low,
it
height
difficult
to assessthe point at which
were
which
made
response
and peak
the HEQ diol started to elute. Whilst this was a minor irritation with high analyte
low
3.17),
height
(Figure
led
belief
the
that
to
the
peak
response
concentration
differentiating betweenthe analyte and the baselinenoise would compromise sensitivity
during trace analysis.
Subsequentanalysis of the DEEDMAC-derived diol species gave rise to an
during
HEQ-diol
the
(Figure
3.18).
the
that
analysis
to
of
witnessed
peak
analogous
The component again demonstratedsignificant fronting, which could not be attributed
to sample overload. In addition, a small secondarypeak was also apparenteluting at the
tail of the first peak. Unfortunately the origin of this peak could not be determined.

Chanter 3: A new NP-LC method for fabric conditioner actives

94

Figure 3.17: Chromatogram achieved from the analysis of the trimethylammonium propane1, 2-diol chloride sample(HEQ diol) on a Spherisorbaminopropyl phase.
Conditions - Column: 150 x 4.6 mm i. d. 3 m Spherisorb aminopropyl column; Mobile phase:
70: 18:12 hexane: MeOH : THE modified with 5 mmol/1TFA.

Figure 3.18: Chromatogramobtained from the analysis of the DEEDMAC derived diol sample
on a Spherisorbaminopropyl phase.
Conditions - Column: 150 x 4.6 mm i.d. 3 pm Spherisorb aminopropyl column; Mobile phase:
70: 18:12 hexane : MeOH : THE modified with 5 mmol/l TFA.

The strange peaks associatedwith the analysis of the two quaternary aminobelieved
initially
to
provide experimental proof of the limited solubility
alcohols were
Chapter 3: A new NP-LC method for fabric
conditioner actives

95
in
hexane-based
However,
hydrophilic
the
mobile
phase
system.
a
analytes
additional
of
during
development
NP-LC
the
the
of
points
methodology,
observations made at other
been
have
Indeed,
the
that
more
complex.
might
possible explanations
origin
suggested
to account for these, and other observations will be presented later in the chapter.
Nonetheless,it was apparentthat a methodology that had been found to offer excellent
benefits over conventional methods for the analysis of the parent cationic fabric
found
been
during
had
the analysis of their hydrophilic
wanting
conditioner actives,
metabolites.

3.5.3.1 Gradient elution analysis for the simultaneous determination of


biodegradation
diester
their
quats and
products
parent
Utilising the Unilever SOP method for the quantitation of the esterquats and
their cationic metabolites is known to be labour-intensive, as each sample has to be
analysedunder two different solvent systemsin order to quantify the parent speciesand
(Burford,
diol
1997). At the outset of this work,
triol
the
or
metabolites
subsequently
in
developing
the
new methodology was that it should be capable
aims
one of
primary
of quantifying parent analytesand metabolites simultaneously, as it was envisagedthat
this capability would greatly assist biotransformation and bioaccumulation studies
performed on these species(Section 1.2.3.3). At the same time, it was predicted that a
method capable of simultaneously quantifying parent analytes and their metabolites
may ultimately lead to a generic method of cationic tenside analysis(Section 1.5).
The poor chromatography witnessed during the analysis of the quaternary
amino-alcohols gave rise to scepticism that a single methodology capableof quantifying
the esterquatsand quaternaryamino-alcohols could be developed from the new NP-LC
increase
it
However,
the
that
to
the eluting strength of the
recognised
need
was
method.
during
(Section
the
3.5.2)
the
analysis
quaternary
of
amino-alcohols
phase
mobile
for
the simultaneous analysis of
the
of
a
gradient
elution
profile
use
would necessitate
the cationic analytes.As gradient elution profiles are known to increasepeak efficiency
it
hoped
1.3.2.1),
(Section
that the chromatography of the
was
and method sensitivity
be
improved.
Gradient
could
elution was thus evaluated for
quaternary amino-alcohols
the simultaneous analysis of the esterquatsand their corresponding quaternary aminoalcohol metabolites.

Chapter 3: A new NP-LC method for fabric conditioner


actives

96
Figure 3.19 shows a chromatogramachieved from the gradient elution analysis
its
diol
HEQ
the
and
metabolite. The chromatogram shows the
sample
of
commercial
M6
the
column with the mobile phase being modified with
on
separation performed
25 mmol/l TFA. Due to the nature of the gradient elution profile that was applied during
the course of the analysis,the diester and monoestercomponentswere eluted under the
isocratic conditions reported in Section 3.3.1. Initiating the onset of the gradient prior to
the elution of the monoestersresulted in co-elution of the two principle components.
Whilst it is clear from Figure 3.19 that the gradient profile was not optimised for the
improvement
diol
the
component, some
was witnessed in peak shape and
elution of
peak height responseof the diol analyte. Further evaluation revealed that the analysis
time could be reduced to twenty three minutes without observing a significant increase
in baselinenoise as a result of incomplete evaporation of the polar solvent.

270
35

220

30

170

20

15

W
c

0
CL

w5

120-

10

is

zo

25

30

35

45

45

Mm(n*u
s)
70

20

05

10

15
Time (minutes)

20

25

Figure 3.19: Chromatogram obtained from the gradient elution analysis of a standard
its
diol
intermediate.
HEQ
the
sample
and
commercial
containing
Conditions - Column: 150 x 4.6 mm W. 5 pm M6; Solvents: A- hexane modified with
25 mmol/l TFA, B -1.5: 1: MeOH : THE modified with 25 mmoll TFA.

Considerationof the above results showed that the methodology had potential to
be adapted to quantify parent diesters and their biointermediates in a single analysis.
However, it was clear that accuratequantitation of the diol specieswould be difficult

Chapter 3: A new NP-LC methodfor fabric conditioner


actives

97
during trace environmental analysis due to the severefronting that was still evident for
the analyte.

3.5.4 Improving

method

sensitivity

and

compatibility

with

mass

spectrometry
It was mentioned in Section 1.3.2.1 that detection limits are improved in liquid
internal
diameter
by
(i. d.) of the column on which the
the
reducing
chromatography
bands
in
The
takes
analyte
elute
reduced peak volumes, giving rise to
place.
separation
increased
bands
and
peak height responses(Hewlett Packard, 1997),
more concentrated
is
linear
the
consistent, peak area responseis also increased(Hewlett
velocity
and as
Packard, 1997). An additional benefit of performing separations on narrow-bore
columns is that the reduced mobile phase flow rate improves the compatibility of the
LC method with electrospray mass spectrometry (ESI-MS). The response of the
detector is increasedand interface dynamics are improved (Abian et al., 1999) (Section
1.3.2.1).

In order to improve the limit of detection (LOD) of the new NP-LC


its
hyphenation with ESI-MS, a study was undertaken to
facilitate
methodology and
assessthe effect of reducing column i. d. on the peak parameters of the diester and
monoester quats. Three aminopropyl bonded stationary phases were evaluated with
internal diameters of 4.6,2.0 and 1.0 mm. Due to the cost implications of purchasing
three Spherisorb columns an alternative stationary phase was sought that could be
obtained at reduced expense. A review of commercial literature revealed that the
Sphereclonematerial was marketed by Phenomenex(Macclesfield, UK) as a suitable
developed
for
methods
and validated on Spherisorb. Indeed the
replacement column
far
as to guaranteeequivalent performance from this material,
as
went
manufacturers
(Phenomenex,
2001).
As
identical
the Sphereclone columns were
conditions
under
found to retail for less than equivalent Spherisorb phases,three Sphereclonecolumns
W.
for
the
the
of
of
column
effect
assessment
were used
Whilst manufacturers' information may have shown that the Spherisorb and
Sphereclone materials yielded identical chromatography, it was soon apparent that
did
in
these
the case of the new NP-LC
not
support
claims
experimental observations
longer
that
It
a
equilibration period was required for the
was noted
methodology.
Chapter 3: A new NP-LC method for fabric
conditioner actives

98
Sphereclone columns in comparison to the Spherisorb columns. In addition, the
Sphereclone
led
Spherisorb
the
the
to a reduction
with
material
material
replacementof
in resolution and retention. As a result, the eluting strength of the mobile phase was
reduced to bring the resolution and elution times of the tenside components in line with
those witnessed with the Spherisorb material. Nevertheless, analogous separations of
the three parent diester quat samples, and the two quaternary amino-alcohol samples
(Appendix One) were ultimately achieved on the new Sphereclone aminopropyl
phases.
Figure 3.20 shows a comparison of the chromatogramsobtained on the 4.6 mm
i. d. and 2.0 mm i.d. Spherecloneaminopropyl phases.It was evident that the peak areas
of the three main diester componentswere significantly increasedon the narrow bore
column. Subsequentcomparison of the peak area responsesrevealed that the use of the
2.0 mm i. d. column gave rise to an approximate five-fold increase in sensitivity with
little associatedloss in resolution.
The successfulevaluation of the 2.0 mm W. column led to the hope that the use
improvements
W.
further
in both sensitivity and ESI-MS
1.0
the
would yield
mm
of
compatibility. Unfortunately, subsequentanalysis of the commercial HEQ sample on
this phaserevealedthat the diester componentseluted as a broad inefficient peak, whilst
the monoesterswere barely visible above the baseline. It was subsequentlydiscovered
that the problems experiencedwith the 1.0 mm W. column could be traced to the use of
instrumentation designed for industry-standard 4.6 nun W. columns. Whilst the
(Section
2.4.1.2)
HPLC
pump
was capableof delivering a reproducible flow
quaternary
of 50 l/min, problems were experienced with the performance of the commercial
light
detector.
the
scattering
evaporative
autosamplerand
The problems experiencedwith the autosamplercame from an observedinability
delivery
injection
instrument
to
uptake
and
the
meter
accurately
of a1 l
plug. The
of
led
injection
to
volume overload occurring on the column
volume
l
a5
of
utilisation
(Meyer, 1993), and excessivebroadening of the analyte band within the injection port.
However, the most serious problems experienced during the evaluation of the 1.0 mm
W. column were traced to the evaporative light scattering detector. Although
instrument
be
the
literature
that
could
suggested
utilised with mobile flow
promotional
Chapter 3: A new NP-LC method or fabric
conditioner actives

99
100
less
to
than
l/min,
equal
or
rates of

an inconsistent sputtering aerosol was

from
Advice
instrument
from
the
the
obtained
nebuliser.
vendor
witnessed emerging
detector
known
fall-off
to
the
the
that
of
was
response
suggested

when used in

below
150
(Anachem,
2000).
flow
l/min
of
a
collaboration with

4.6 mm i. d. column

1)

0123456789

11 12 13 9

'5

VSB

Z) 21 2B

2.0 mm i. d. column

Figure 3.20: Figure showing the improvement in peak area response that was afforded by
reducing the internal diameter of the aminopropyl column from 4.6 mm to 2.0 mm.
Conditions - Column: Top - 150 x 4.6 mm i. d. 3 m Sphereclone aminopropyl, bottom - 150 x
2.0 mm i. d. 3 m Sphereclone aminopropyl; Mobile phase: 90: 6: 4 hexane : MeOH : THE
modified with 5 mmol/I TFA: Flow rate: 4.6 mm i. d. column - 1000 gl/min, 2.0 mm W.
column - 190 gl/min.

In view of the problems experienced during the evaluation of the 1.0 mm i. d.


Sphereclone aminopropyl column. and the increased sensitivity obtained in comparison
i.
d.
2.0
W.
4.6
mm
columns.
mm
was adopted as the column
with conventional
dimension in the optimised method.

3.5.5 Evaluation of the sensitivity and robustness of the optimised liquid


chromatographic methodology
The new normal phase LC methodology had been seen to offer significant
benefits over both the Unilever

SOP methodology

and the updated methodology

described by Radke et al. (1999). In order to test the validity of the methodology for
the
control
protocols,
quality
and
robustness, reproducibility,
analysis
environmental
and sensitivity of the method were subsequently evaluated.
Chapter 3: A new NP-LC method for fabric
conditioner actives

100

3.5.5.1 Determining the limit of detection and method linearity


Evaporative light scatteringdetectors are known to suffer from an inherent lack
in
linearity
UV
/
comparison
with
more
conventional
and
of sensitivity
vis and
fluorescent detectors. As a result, a study was undertaken to evaluate the limit of
detection (LOD), limit of quantitation (LOQ), and linearity of the optimised NP-LC
methodology. A series of eight HEQ calibration standardswere prepared ranging in
from
0
100
The
in
to
mg/l.
standards
were analysed triplicate, and
active concentration
the mean peak areaof the diester speciessubsequentlycalculated. At the sametime the
chromatographic data system (Section 2.4.1.2) was utilised to obtain an approximation
of the signal to noise ratio (S/N) correspondingto the total diester concentrationpresent
in each standard.The signal to noise ratios were then used to calculate the LOD and
LOQ of the method, defined as S/N =3 and S/N = 10 respectively in this case.
Figure 3.21 shows the plot of average peak area response of the total diester
for
HEQ
the
eight
observed
concentration
calibration standards. It was
quats versus
linear
limited,
the
that
the
range
of
plot
was
evident

and instead showed a good

approximation to a quadratic. Whilst the non-linear nature of the plot was not ideal, it
was expected in light of previous work performed on the detector (Burford, 2000).

Observation of the signal to noise ratios of the eight calibration standards


revealed that both the LOD and LOQ lay between 5 and 10 mg/l, further
LOD
be
6
be
It
the
LOQ
be
9
to
the
to
revealing
mg/I
experimentation
and
mg/l. must
noted that these values can only be considered as estimates, as the splitting of the
diesters into three component peaks (Section 3.4) made accurate determination of the
it
difficult.
Nonetheless,
to
was apparent that in its current form, the
signal noise ratio
inherent
for
lacked
it
the
to
sensitivity
necessary
make
useful
new method
is
1
A
LOD
deemed
for
mg/1
usually
of
necessary
an
analysis.
environmental
facilitates
(Burford,
2000),
in
LOD
low
this
the
as
a
parts
methodology
environmental
in
billion
combination with a suitable extraction and preused
range when
per
level
This
for
the expected loss in
of
sensitivity
also
accounts
concentration method.
during
the analysis of complex environmental matrices
witnessed
method sensitivity
(Burford, 2000).

Chapter 3: A new NP-LC method for fabric conditioner


actives

101

4.5E+07

y= -22.003x3 +3967.4 X2 + 230540x


R2 = 0.9999

4.0E+07

8852.3
-

3.5E+07
3.0E+07
U)
2.5E+07
_
2.0E+07
l0

CL

1.5E+07
1.0E+07
5.0E+06
0.0E+00

--

10

20

30

40

50

60

80

70

90

100

Concentration (mg/I)

Figure 3.21: Graph showing the non-linearity


detector to increasing HEQ concentration.

of the Sedex 55 evaporative light scattering

3.5.5.2 Method validation


Assessment of

the robustness and reproducibility

of

the new

NP-LC

formal
by
its
However,
to
means
of
a
validation
was
made
protocol.
prior
methodology
it
was predicted that the separation was reproducible and the method
commencement
robust in view of the "informal"

validation, which had been on going throughout the

development of the method. In Section 3.3.1 it was mentioned that during the original
be
had
been
the
to
time
stationary
phase,
poly-ol
retention
of
seen
evaluation
stability
Similar
of
weeks.
number
reproducibility
a
excellent over

was subsequently witnessed

(Section
3.3.2.2)
the
phases
and with the commercial aminopropyl
mixed-mode
with
be
Indeed
time
the
to
the
stability
of
aminopropyl
retention
phase
was
seen
phase.
dedicated
that
the
to method optimisation.
throughout
six-month
period
was
excellent
During this time a number of Spherisorb aminopropyl columns were utilised, all of
which gave rise to equivalent separations, thus showing that the methodology was not
based solely on the merits of a single "deactivated" column.

The second observation made during the informal validation process was that
University
Leeds
the
of
at
was recreated at Unilever Research
the separation achieved
The
being
to
the
had
methodology.
made
therefore been
methodology
without changes
At
transfer
test.
inter-laboratory
the
instrumentation
time,
the
same
to
an
seen pass
used
Chapter 3: A new NP-LC method for fabric
conditioner actives

102
different,
Unilever
Leeds
the method could apparently be performed on
very
was
at
and
injection
LC
with
performed either manually or via an
equipment,
standard
autosampler. In the same way, reagent and solvent suppliers were different at the two
be
therefore
to
not
need
specified when a new SOP was
would
sites; solvent suppliers
eventually created.
The final observationsmade during method development,which were critical to
assessing the robustness of the methodology, were made on the sensitivity of the
separation to changesin mobile phase composition and temperature. Variation in the
retention time had been witnessed when the mobile phase composition had been
changed (Section 3.3.3). However, by ensuring that the mixed polar solvent was
prepared accurately, and that the two solvents were dosed correctly with TFA,
In
the sameway, the methodology was never carried out at a
consistent.
was
resolution
fixed column temperature,and was insteadperformed in ambient laboratory conditions.
As a changein retention time was only witnessedonce during method development,and
then only under extreme conditions in which laboratory temperaturehad fallen to less
than 10C, it was concluded that the methodology was insensitive to small changesin
temperature.

Having observed excellent reproducibility and robustness during method


development, the formal validation study aimed to mathematically quantify the degree
in
the retention times and peak area responsesof the major
of variation witnessed
components. A series of six HEQ calibration standards were prepared ranging in
from
100
Over
10
to
three consecutive days a total of fifty two
mg/I.
concentration
HEQ
the
standards were performed. The chromatographic data
various
analyses of
into
then
transferred
to
the System Suitability
analysis
was
each
corresponding
data
Turbochrom
the
chromatographic
system(Section 2.4.1.2) to assess
on
programme
the variation in the retention times and peak arearesponsesof each component.
The percentagerelative standarddeviation (%RSD) of the retention times of the
three principle diester and two monoestercomponentsduring the fifty two analyseswas
highlighted
This
1.10.
less
the excellent reproducibility of the
be
than
value
to
seen
be
to
Retention
seen
consistent over time and showed no variation
was
methodology.
fact
in
that
a
was
stark contrast to the changing retention
with analyte concentration,
Chapter 3: A new NP-LC methodfor fabric conditioner
actives

103
times witnessed with varying analyte concentration in the Unilever SOP method
(Lawrence, 1999). When the reproducibility of the peak arearesponsewas subsequently
%RSD
for
found
it
the total diester concentration was less than
that
the
evaluated, was
1.49, whilst the %RSD's for the three individual componentswere less than 2.40. It was
apparentthat the new methodologydemonstratedreproducible peak arearesponses.
Though the new methodology facilitated high retention time stability and high
inherent
lack
the
of sensitivity prohibited the use of the methodology for
resolution,
in
its
form.
However, no such problems were envisaged
current
environmental analysis
during the evaluation of raw materials and the analysis of industrial samples.In both of
these scenariosanalyte concentrationwould be above trace levels and thus combination
of the NP-LC methodology and a suitable extraction procedurewas predicted to yield a
method suitable for use in quality control protocols, assuming calibration was carefully
controlled in light of the observednon-linearity.

3.5.6 Development of the hyphenated LCIMS methodology


The lack of sensitivity witnessed during the evaluation of the NP-LC-ELSD
its
in
limited
the quantitation of cationic fabric conditioner actives in
use
methodology
environmental matrices. As a result, an alternative detection method was sourced that
could offer greater sensitivity. In Section 1.4.4 it was described how a number of
improved
had
the sensitivity and selectivity of NP-LC methods for the
groups
determination of cationic fabric conditioner actives by hyphenating the methods with
ionisation
mass spectrometry (ESI-MS) (Fernandez et al., 1996; Radke et
electrospray
hyphenated
Utilisation
1999).
the
of
methodology facilitated structural confirmation
al.,
in
the
componentspresent each of the samplesstudied, and also yielded quantitation
of
billion
tenside
the
at
sub
per
parts
analytes
concentrations (Radke et al., 1999).
of
Therefore, in an attempt to improve the selectivity, specificity and sensitivity of the
for
environmental analysis, attempts were made to hyphenate
methodology, particularly
the new LC methodologywith electrospraymass spectrometry.

Chapter 3: A new NP-LC method for fabric


conditioner actives

104

3.5.6.1 Direct infusion analysis


The initial

mass spectral assessment of the cationic tenside samples was

infusion
2.4.2).
direct
(Section
Figure 3.22 shows
the
analysis
aid
of
performed with
the mass spectrum obtained from the direct infusion of the commercial HEQ sample
into the ESI interface at a flow rate of 3 pl/min. The mass spectral parameters used
during the acquisition were obtained with the automatic tune facility

in the LCQ

instrument control software (Section 2.4.2).

538.7 C

loo

18

/C

16

es'
80

c
18/C

85

18

75i

Diesters

70
ss=
60

45-.

1-9 4o-_
35

16/C 16

Thouht

7-5

to be relics

61 Q8

3eft
0
9

Monoesters

25
20

1339.4
1311.

570.6

15

CIh

CIK

101

128$. 5

1.5

1$7.3

55.6

372'(

400-5

624.6
I L. I
z, 6
Bo0

200

860

101

1073.3

1000

125

.313

1200

155.8
14bC

1804.6

1400,1800

rti

Figure 3.22: Mass spectrum obtained from the direct infusion of the HEQ sample into the
Finnigan MAT LCQ instrument after MS optimisation.
Infusion rate: 3 I/min.

The two principle

ions witnessed in the mass spectra, the base peak at

ion
638.7
percent
abundant
ninety
at m/z = 666.8 were seen to correspond to
and a
m/z =
the molecular ions (M+) of the C18 / C16 and C18 / C18 diester components respectively,
from
been
had
previous work carried out on the mass spectral analysis
expected
which
of HEQ (Burford,
Similarly,

1999; Radke et al., 1999; Waters et al., 2000; Weston, 2000).

the ions witnessed at m/z = 610.6,400.5,

and 372.5 were known to

C16
diester,
C18
/
C16
the
C16
the
the
to
monoester,
and
monoester
correspond
However,
1999).
(Radke
what was unexpected was the group of
al.,
et
respectively
1395.
These
high
1255
between
and
m/z
=
molecular weight ions had
peaks witnessed
during
HEQ
the
by
been
analysis
of
previously
mass spectrometry
observed
not
(Weston, 2000), and indeed, characterisation data supplied by Unilever Research
Chapter 3: A new NP-LC method for fabric conditioner
actives

105
to
did
this
that
molecular
corresponding
weight
components
range
suggested
not exist in
the commercial sample (Appendix One). It was therefore hypothesised that these high
ions
being
the
of
electrospray
process,
were
relics
generated in-situ,
molecular weight
(Weston,
2000). Further
the
therefore
constituents
of
sample
native
were not
and
ions
distribution
to
the
the
mass
and
charge
ratios
of
revealed that they may
scrutiny of
have been dimeric adducts of the diester quats. However, the variation in the mass to
ions
between
doubly
that
they
showed
adjacent
were
not
charged species.
ratios
charge
Tandem MS studies performed on the high molecular weight ions revealed that they
ions
high
fragment
"product"
to
two
collisions
corresponding
energy
emit
under
would
to the M+ ions of the principle diester quat components (Figure 3.23). However, the
fragmentation
bring
to
about
energy required

was significantly

higher than that

normally required to fragment small organic analytes (Weston, 2000), adding further
belief
that they were relics of the electrospray process. A second
to
the
support
important observation made during tandem MS experiments was that the high molecular
weight components did not correspond to pure dimeric species, and instead appeared to
ions
diester
M+
two
to
quat
co-ordinated with a single chloride anion, which
correspond
for
ions
demonstrating
believed
the
to
account
an apparent one plus charge.
was

Direct infusion studieswere also performed on many of the other cationic fabric
conditioner active samplesthat were utilised during the development of the new NP-LC
identities
during
Peak
Section 3.4 were subsequentlyconfirmed
assigned
methodology.
by massspectral identification.

3.5.6.2 Loop injection analysis


Prior to the assessmentof the fully hyphenatedLGMS system, loop injection
analysis of a number of cationic tenside sampleswas performed to assessthe effect of
the LC mobile phase system on the ionisation efficiency of the analytes and the
electrosprayinterface.

Chapter 3: A new NP-LC method for fabric conditioner


actives

106
('

100

l'

l6

638 5

6.5

('

/C

19

9P

Product

85-

JI
45 =

ions

ffi7 5

40

ion

Parent

30

1342.3

25

2O
.3

1510

31

5
477-3

O400

579.5
1L1

5=7

787.1

887.7

M_

860

1088-3
10o

1254.0

1153.8
120

1307-3

1453.9

1400

rn/z

Figure 3.23: Figure showing the dimeric nature of the high molecular weight ions observed
during the direct infusion analysis of the commercial HEQ sample.
Infusion rate: 3 l/min; Trapped ions: m/z = 1340 - 1343; Relative collision energy: 100%.

For tuning purposes, a poly (ether (ether) ketone) (PEEK) T-piece was inserted
into the instrumental set-up between the injector on the LC instrument and the inlet to
the ESI interface. In this way, the integral syringe pump on the LCQ instrument
(Section 2.4.2) could be used to constantly infuse the cationic tenside sample into the
flowing LC mobile phase to efficiently optimise the MS response. The optimised MS
loop
injection
Three.
in
Appendix
the
throughout
studies
are
shown
utilised
parameters

Loop injection analysis was performed with the same instrumental set-up that
for
be
hyphenated
LC/MS
the
the
evaluation
used
of
would subsequently

system

(Section 2.4.2). However, the HPLC column was omitted from the set-up and thus the
directly
inlet
interface
injector
to
the
the
connected
of
was
electrospray
via
port
outlet of
Observation
during
loop
injection
the
PEEK
length
tubing.
of
results
obtained
of
a
during
to
those
were
produced
mass
spectra
witnessed
analysis revealed analogous
direct infusion studies. However, it was evident that there was no longer any evidence
further
detailed
for
ions
the
theory that
high
above,
support
weight
the
molecular
of
It
been
the
had
ions
process.
electrospray
of
was therefore concluded that
relics
these
in
leading
to
the electrospray process.
subtle
changes
the LC mobile phase was
However, these changes did not appear to adversely affect the ionisation efficiency of

the analytes.

Chapter 3: A new NP-LC method for fabric conditioner


actives

107

3.5.6.3 HyphenatedLC/MS analysis


Full hyphenation of the LC method with ESI-MS was simple after the successful
further
injection
loop
as
analysis,
no
changes were made to the MS
evaluation of
instrumental
A
Three).
(Appendix
to
the
change
was
made
set-up, whereby
parameters
the column outlet was connected directly to the ESI interface via a one-piece column
brought
by
This
2.4.2).
(Section
the observation that the
change
was
about
coupler
injection
during
loop
broadened
band
analysis, as a result of the adsorption of
analyte
the cationic tenside analytes onto the interior walls of the PEEK tubing linking the
injector to the interface. During initial

LC/MS

analysis it was observed that the

in
diester
lost
HEQ
the
the
components
present
as a result of
of
sample
was
resolution
extra-column band broadening. Replacement of the PEEK tubing with the columncoupler was seen to resolve this problem.

Figure 3.24 shows the reconstructed ion chromatograms (RIC's) of the M+ ions
/
/
diester
/
C13
C16,
C16
C16
C1
CIS,
the
and
components (m/z = 666.5,638.5
of

and

610.5) obtained from the LC/MS analysis of the HEQ sample. When the RIC's were
compared with the total ion chromatogram (TIC) obtained from the same analysis
(Figure 3.25 shows the TIC from a later analysis when retention times had stabilised), it
diester
the
three
that
main
peaks separated by the front-end LC system,
was apparent
had been correctly identified in Section 3.4. In addition, the RIC's showed that the
front-end LC separation was efficiently

resolving the individual

tenside components

into discrete peaks, i. e. analyte carry-over and late elution were minimised. It was
therefore concluded that the new NP-LC method represented an efficient means of
diester
individual
the
and monoester components of the HEQ sample into
resolving
discrete peaks. At the same time, it was envisaged that the use of the RIC's would
individual
the
of
quantifying
means
cationic tenside components
provide an effective
if
in
sample,
and when the sensitivity provided by ELS detection
a commercial
present
found
wanting.
was

Comparison of the RIC's of m/z = 694 and 582 (not shown), the M+ ions of the
C20 / C18 and C16 / C14 diester quats with the TIC, revealed that these two species
two
the
to
small peaks eluting prior to, and subsequent to the three
corresponded
3.25).
(Figure
Mass
diester
spectral evaluation of the two late
components
principle

Chapter 3: A new NP-LC method for fabric conditioner


actives

108
NP-LC
by
C18
that
these
the
the
also
method
confirmed
were
eluting peaks separated
400.5
372.5)
(m/z
C16
=
and
respectively.
monoester quat components
and
--

401=

IiT-

si0

m/z = 694-695

pi

/C
2,

.`

iaai,

, oo

a
7

---

--'--

"o

--_

ao

'

t a4

,isos
/

/C
18

UP

m/z = 638-639
C

I,

m
v

18

8os.

`$

"oe

-0

18/C

',

I"

m/z = 666-667

/ "l`;

moo
U

eo

Is

16

m/z = 610-611
C

16/'

16

m/z = 579-580
o

q
M-

a14i0

m/z = 551-552

1
"0
'\
Time

(minutes)

Figure 3.24: Reconstructed ion chromatograms of the major diester species and two unknown
HEQ
from
LC/MS
the
the
commercial
of
sample.
analysis
resulting
Conditions - Column: 150 x 2.0 mm i. d. 3m Sphereclone aminopropyl; Mobile phase: 90: 6: 4
hexane : MeOH : THE modified with 5 mmol/I TFA; Flow rate: 190 l/min.

The hyphenatedLC/MS methodology was subsequentlyutilised to study many


listed
in
Appendix
tenside
samples
the
cationic
other
of

One. In each case, LC/MS

3.4
identities
(Section
the
the
and
validity of
pre-assigned peak
analysis confirmed
3.5.1). It was observed that the hyphenated method was also well suited to the analysis
of the DEEDMAC.

Arquad and Stepantex samples, as well as the individual ester and

alkyl quats (Appendix

One). The quaternary amino-alcohols were not studied with the

method.

The observations made during the evaluation of the new hyphenated NPLC/ESI-MS

it
be
indicated
that
might
well suited to the analysis of
methodology

due
Unfortunately
it
to
time
fabric
actives.
constraints
was not
conditioner
cationic
linearity
indeed
the
to
and
of
the
methodology,
sensitivity
or
to
evaluate
possible
It
formal
that
formal
the
envisaged
was
a
validation
of
validation.
perform a
the
the
to
the
of
the
methodology
suitability
of
analysis
of
assessment
and
methodology,

Chapter 3: A new NP-LC method for fabric conditioner actives

109
be
in
future
the
to more fully assess its
required
near
quaternary amino-alcohols, would
applicability to environmental analysis.

6.36

100-

NL:
9.39E7

F:

IA" IM1

95

3.03

90

eih6,97
\

1J

20.02000.. 00]

85
60

75y

Ii

5.05

70-'

55

: o/C IaIII

60
L

Ihi(
970

16

55

18

50

23.63

53

45_

(16(1

40-

1 63
,

Z'

7.42_

30

13.68

' 1226

25

21

16.51
v.

22 60

ti,

'
_

20

! r

18:
10.3

01

2.12

1 117
,.

2466

10

12

14

16

18

20

22

24

Figure 3.25: Total ion chromatogram resulting from the LC/MS analysis of the HEQ sample.
Conditions - Column: 150 x 2.0 mm i. d. 34m Sphereclone aminopropyl; Mobile phase: 90: 6: 4
hexane : MeOH : THE modified with 5 mmol/I TFA: Flow rate: 190 l/min.

3.5.6.4 Dynamic modification of the stationary phase and analyte bleed


The two additional RIC's shown in Figure 3.24 corresponded to two unknown
ions witnessed during direct infusion studies (Figure 3.22), m/z = 579 and 551. As
literature reports had not previously noted the existence of these two species in the
HEQ
sample (Radke et al., 1999; Waters et al., 2000), it was originally
commercial
ions
that
the
were additional relics of the electrospray process (Section
assumed
3.5.6.1). However, observation of these ions during loop injection and LC/MS analysis,
ions
(Section
3.5.6.2)
high
longer
the
weight
molecular
were
no
where
evident, brought
this hypothesis into doubt.

Having been unsuccessful in attributing the two ions with known impurities of
the HEQ sample. attention turned to the origin of the ions being the presence of other
It
in
the
sample.
was soon realised that m/z = 579 and 551
tenside
analytes
cationic
/
ions
C20
C18
C18
/
M+
C18
dialkyl
the
the
of
and
to
quats (Section 3.4).
corresponded
Scrutiny of the characterisation data supplied with the HEQ sample suggested there was
little evidence of the existence of these tensides in the HEQ sample (Appendix One),
Chanter 3: A new NP-LC method for fabric
conditioner actives

110
had
HEQ
become
it
the
that
thus
sample
of
assumed
adulterated and / or the
was
and
interior
the
tensides
the
onto
cationic
of the ESI interface was leading to
adsorption of
fouling

and subsequent analyte carry-over (Cooper, 2000). However,

subsequent

comparison of the RIC's of m/z = 579 and 551 with the corresponding TIC revealed that
the two components were eluting at two points in the trace; after the solvent front and
between the three principle diester components (Figures 3.22 and 3.24).

The observation of the two components eluting close to the solvent front was
attributed to the short elution times of these speciesunder the mobile phaseconditions
utilised during LGMS analysis (Section 3.4 and Figures 3.11 and 3.25). However,
examination of the chromatogramsobtained from the NP-LC analysis of HEQ (Figure
3.13) revealed the same impurity had been witnessed in the HEQ trace throughout the
development.
It
was apparent that the existence of the two components could
method
instrument
be
MS
to
the
either
attributed
or to sample adulteration. Instead, it was
not
hypothesised that the cationic tensides were dynamically modifying the stationary
in
phase,and addition, a number of the analyteswere liberating other bound analytesvia
an exchangeprocess.
Whilst it proved impossible to ascertain whether this hypothesis was correct, it
was believed that support for the theory was forthcoming from two areas.Firstly, it was
observed that at the commencementof a study, a number of replicate injections were
required before consistent retention was attained (Section 3.5.4). In the case of the
Spherisorb material, it was observedthat this processwas only required during the use
of a new column, whilst the Sphereclone material required this procedure to be
performed whenever the system was halted. It was predicted that the shortening of
dynamic
the
times
about
as
a
of
came
result
modification of the active sites on
retention
the stationary phase (Nawrocki, 1997), stable retention being achieved when the most
difference
in
The
blocked.
the equilibration time of the Sphereclone
acidic sites were
believed
to provide evidence that the two silica supports
Spherisorb
was
materials
and
behaveddifferently (Section 3.5.4).

Support for the dynamic exchangeof the cationic tensides on the silica support
bound
"bleed"
of
previously
analytes was provided by the LC/MS
the
continuous
and
dialkyl
in
the
trace
The
quats
of
a
corresponding to a commercial
observation
work.
Chanter 3: A new NP-LC method for fabric
conditioner actives

111
by
have
the
come
about
removal of a number of the dialkyl
only
could
esterquatsample
during
having
been confident that
bound
to
the
a
previous
analysis,
silica substrate
quats
the samplehad not beenadulterated.More significantly the dialkyl quats were also seen
hydrophobicity
diester
The
in
the
the
quats.
greater
of
middle
of the dialkyl
eluting
demonstrating
in
have
them
a much shorter retention time if they
resulted
quats should
had been present in the analyte band at the point of injection. Their elution with the
diester quats could therefore only be justified by the fact that the dialkyl quats were
being desorbedfrom the silica substratein-situ.

Validation of the hypotheseson dynamic modification of the stationary phase,


and subsequent analyte bleed, was not forthcoming before work on the new NP-LC
in
However,
ceased.
view of the fact that analyte bleed was probably
methodology
occurring continuously, caution must be observed if and when the methodology is used
for quantitation purposes. It is important that the extent to which the stationary phase is
fully
be
in-situ
assessedin the future.
modified

3.6

PROBING THE RETENTION MECHANISM


The new NP-LC methodology developed for the analysis of cationic fabric

conditioner

been
had
shown to offer
actives

reproducibility,

superior resolution

(Section 3.4),

and robustness (Sections 3.5.5.1 and 3.5.5.2) to that achieved with the

Unilever SOP method (Section 3.2), the commonly used methodology of Wee et al.
(1982), and a recent method developed by Radke et al. (1999). However, problems were
experienced when the new methodology was applied to the analysis of the quaternary
3.53).
Co-injection
(Section
analysis and subsequent LC/MS analysis
amino-alcohols
revealed that the new methodology could resolve the different groups of cationic tenside
in a commercial sample (Figure 3.14), partially resolve the homologous series present
in each group (Figure 3.8), resolve components from different tenside samples (Figure
3.16), and separate isomeric species (Figure 3.10). However, in spite of the efficient
"How
tensides
the
questions
still
remained:
cationic
of
and why was the
resolution
be
"
"Could
improved
further?
the
" and finally
resolution
still
occurring?
resolution
"Could the methodology be optimised to yield efficient analysis of the quaternary
It
that
"
to
the
answers
envisaged
was
above questions would be
amino-alcohols?

Chanter 3: A new NP-LC method for fabric


conditioner actives

112
forthcoming from an understandingof the mechanism by which the cationic tenside
analyteswere being retainedon column.

3.6.1 Assessing the influence of the stationary phase


During the development of the hexane-based NP-LC methodology a series of
different stationary phases were evaluated. Whilst many of them were seen to give rise
to analogous separations, it was realised that the subtle differences in resolution could
be used to shed light on the retention mechanism. For example, during the evaluation of
the supercritically

bonded mixed-mode

phases, it became apparent that superior

resolution was achieved on the aminopropyl phases as the percentage of amino groups
increased (Section 3.3.2.2). At the same time, resolution, on the alkyl-amino phases
(Appendix

Two) was higher than that achieved on the corresponding aminopropyl

phases (Section 3.3.2.3). It appeared that as the number of amine groups attached to the
silica surface increased, there was a general increase in the resolution of the mono and
di-chained cationic tensides. This apparent trend was supported by the observation that
efficient

resolution of the cationic tensides could be achieved on a commercial

aminopropyl bonded phase (Figure 3.13), yet no evidence of the cationic tensides could
be seen when the same sample was subsequently analysed on a commercial cyanopropyl
phase. A reduction in resolution was subsequently observed when the HEQ sample was
analysed on a mixed-mode aminopropyl / octadecylsilane (ODS) bonded phase (Figure
3.26). The diester quats were evidently eluting as a single peak.

Comparison of the observations made on all of the stationary phasesrevealed


that the presenceof hydroxyl groups or amine groups bonded to the silica substrate,
facilitated the resolution of the homologous series present in each group of cationic
tensides. As the number of amino and hydroxyl groups increased, a general increase
in
was witnessed resolution, yet consistent retention times were witnessed on most of
the different phases.A change in the nature of the stationary phase led to negligible
large
but
times
a
effect on resolution. Indeed, increased
effect on analyte retention
Bio-Sil
Partisil
the
PAC phases.Both of
on
poly-ol
witnessed
and
only
was
retention
these phaseswere basedon irregular silica supports (Appendix Two) and thus it was
feasible that the increasedretention came about as a result of the nature of the support
bonded
hypothesised
the
that
the
It
nature
of
phase was only affecting
material. was
resolution of the analytes,and thus the mode of retention was equivalent on each.
Chapter 3: A new NP-LC method for fabric
conditioner actives

113

Figure 3.26: Chromatogramobtained from the analysis of the HEQ sample on the Spherisorb
mixed mode ODS / NH2 phase.
Conditions - Column: 250 x 4.6 mm i. d. 5 m Spherisorb mixed-mode aminopropyl /
6
hexane
85:
9:
Mobile
phase:
: MeOH : THE modified with 25 mmol/1 TFA;
octadecylsilane;
Flow rate: 1 ml/min; Detection method: Evaporative light scattering.

3.6.2 Assessing the influence of the mobile phase


It was stated in Section 333

that an evaluation into the effect of varying the

direct
had
THE
THE
the
and
replacement
ratio
solvents
of
:
with other organic
methanol
led to changes in the resolution of the analytes and / or changes in analyte retention
time. It was apparent that the replacement of THE with other organic solvents higher in
the eluotropic series of normal phase solvents (Meyer, 1993) (Section 1.3.2) had a
instead
led
in
time,
to
the resolution of
the
and
retention
a reduction
negligible effect on
the analytes. This observation appeared to provide strong evidence that the mode of
different
LC
to
the
to
method
adsorption
usually
attributed
normal
phase
retention was
(Section 1.3.2). In contrast, variation in the methanol : THE ratio led to very significant
Increasing
both
in
the proportion of methanol
and
resolution.
retention
analyte
changes
in the mixed polar solvent led to a reduction in analysis time and resolution, as was
However,
from
the
the
solvent.
a
stronger
of
utilisation
when
proportion of
expected
THE was increased, a disproportionate increase in retention and resolution was
indicate
both
these
for
When
to
observations
the
appeared
combined,
need
witnessed.
facilitate
in
to
THE
the
phase
mobile
efficient resolution of the cationic
methanol and
tensides. The presence of methanol appeared to prevent excessive analysis time, whilst

Chapter 3: A new NP-LC method for fabric conditioner


actives

114
the THE appearedto facilitate resolution of the tenside homologues. In order to test this
hypothesis the HEQ and HEQ diol sampleswere analysedwith a mobile phasesystem
hexane
and methanol.
only
consisting of
Figure 3.27 shows the chromatogram obtained from the analysis of the HEQ
binary
hexane
methanol
solvent system. The diester quats were seen
and
sample with a
to elute as a single peak with an unusual band profile similar to those witnessedduring
the analysis of lycopene by NP-LC on a Partisil PAC phase (Piretti et al., 1996).
However, the monoestercomponentswere actually seento show increasedresolution in
comparison to that witnessed with the ternary mobile phase system (Figure 3.10).
Subsequentanalysis of the commercial DEEDMAC sample resulted in an analogous
separation.

Figure 3.27: Chromatogram obtained from the analysis of the commercial HEQ sample on
hexane
binary
M6
mobile
phase
system
of
and methanol.
a
with
column
Conditions - Column: 150 x 4.6 mm i.d. 5m M6; Mobile phase: 90: 10 hexane : MeOH
I
TFA;
Flow
Detection
25
rate:
ml/min;
mmol/I
method: Evaporative light
modified with
scattering.

Figure 3.28 shows the chromatogram obtained from the analysis of the HEQderived quaternaryamino-alcohol component (Appendix One) on the M6 column with
hexane
The
binary
and
methanol.
of
chromatography of the analyte
system
solvent
a
different
be
to
that
to
witnessed with the ternary solvent system.
very
seen
was again

Chapter 3: A new NP-LC method for fabric conditioner


actives

115
From showing severefronting under the ternary mobile phase system, the analyte was
binary
in
the
the
tailing
to
presence
of
solvent system.
to
peak
seen give rise a

57
52

47 -!

m42c
0

37
I
32
27
22
10

0123456789
Time (minutes)

Figure 3.28: Chromatogramshowing the analysis of the HEQ-derived diol specieson column
M6 with a binary mobile phasesystemof hexaneand methanol.
Conditions - Column: 150 x 4.6 mm W. 5 m M6; Mobile phase: 75:25 hexane : McOH
light
1
Evaporative
TFA;
Flow
Detection
5
rate:
ml/min;
method:
mmol/1
modified with
scattering.

It was evident from the results obtained with the binary solvent systemthat the
diesters,monoesters,and diol componentswere affected differently by the changein the
mobile phasesystem.

3.6.3 Mechanistic explanation of the experimental observations


Many observationswere made during the development and optimisation of the
the
LC
the
that
the
regarding
especially
of
effect
nature
methodology,
phase
normal
had
the
tenside
the
chromatographic
on
of
cationic
profiles
phases
mobile
stationary and
it
it
the
that
Initially,
of
many
observations were contradictory and
appeared
analytes.
justifiable
develop
hypothesis
for
impossible
to
to
a
all of
account
thus
prove
would
during
hexane
the
the
However,
the
evaluation
obtained
of
and methanol
results
them.
implicated
by
described
a
retention
mechanism
previously
mobile phase system,
Kaaoka and co-workers (Kazoka et al., 1992 and 1996; Kazoka, 2000).

Chapter 3: A new NP-LC method for fabric conditioner actives

116
The above authors studied the effect that changesto the mobile system had on
the resolution of a series of purine and pyrimidine analytes separatedon a bare silica
(Ka2oka
1992
conditions
et
al.,
phase
and 1996). The group
normal
phase under
improved
immiscible
that
was
achieved
when
separation
mutually
observed
solvent
systems were employed, attributing these observationsto the formation of a stationary
solvent in the pores of the silica particles. Retention was thus believed to occur via
adsorption onto the surface silanol groups of the silica and partitioning into and out of
the stationary liquid (Kaloka, 2000). Importantly, as the ratio of the solvents varied,
beyond
its
limit,
the extent of partitioning
that
the
was
pushed
solubility
system
such
increased,facilitating resolution of the analytesof interest.
In view of the immiscibility of methanol and hexane at the levels used in the
new NP-LC methodology (Meyer, 1993) it was hypothesisedthat a stationary liquid of
THE and MeOH had accumulatedin the pores of the silica during column equilibration.
Whilst this proposal initially appearedto be a little too convenient, it was soon realised
that many of the observations made during the development of the methodology
hypothesis.
this
supported
Assuming the presenceof stationary-liquid phase,the analyteswould have first
had to partition into this phase before interacting with the stationary phase. This
from
fact
is
the
that the majority of the bonded phaseresideswithin
evident
prediction
the pores of the silica (Nawrocki, 1997). As a result, the rate at which the analytes
into
liquid,
the
out
of
and
stationary
partitioned
and the degreeto which they interacted
have
the
would
phase
governedretention.
stationary
with
Having predicted that retention of the analytes relied on both adsorption and
in
it
became
a
change
apparent
why
stationary phase was seen to have a
partitioning,
but
the
minimal effect on the retention of the di-chained
resolution,
major effect on
in
have
led
the
The
the
to the
solvent
pores
of
polar
of
silica
would
accumulation
quats.
being
in
bulk,
than
that
the
the
more
polar
pores
and thus a concentration
within
solvent
been
have
With
the
the analytespartitioning into
set-up
within
column
gradient would
hydrophobicity
have
been seento have a
the
phase,
would
stationary-liquid
and out of
times
influence
the
residence
of the analytes within this phase,which in
on
significant
bonded
the
phase. Under stationary-liquid conditions,
turn would affect contact with
Chapter 3: A new NP-LC method for fabric
conditioner actives

117
hydrophobic analytessuch as the di-chained quats would spend less time in the pores of
the silica than more hydrophilic species.Having minimal residence time in the pores,
the di-chained quats would also have little potential to interact with the bonded phase,
bonded
have
in
had
little
the
the
thus
of
phase
nature
would
and
changes
effect on the
retention times of these analytes, which was evident throughout method development
(Figure 3.5).

In light of the mixed adsorption - partition model, the increased resolution of the
di-chained analytes witnessed in the presence of increasing numbers of amino groups
(Section 3.3.2.2), was believed to derive from adsorption onto the bonded phase.
Having shown that resolution of the ester and alkyl quats was maximised in the
presence of large numbers of amine or hydroxyl groups (Section 3.6.1), it was predicted
that the quaternised nitrogen group would undergo electrostatic interaction with these
it
in
Thus,
that
appeared
variation
electrostatic interaction was giving
respective groups.
rise to homologue resolution, whilst partitioning was bringing about class separation.
This hypothesis was in direct contrast to normal theories on liquid chromatography
(Dorsey, 1994), which assume that partitioning behaviour generally leads to homologue
(Dorsey,
1994). Nevertheless, it was believed
phase
reverse
conditions
resolution under
that the lack of resolution witnessed during the analysis of the diester quats, with the
hexane - methanol solvent system (Section 3.6.2) provided support for this theory.

It was hypothesised that in the presence of the binary solvent system a


stationary-liquid phase would still exist within the pores of the solvent. However, as
in
the
the pores, residence times of the dionly
solvent
now
present
methanol was
decrease.
In
to
expected
were
quats
addition, the bulk mobile phase was
chained
increased
demonstrate
due
higher
to
to
the
eluotropic
strength
expected
methanol
concentration. These two predictions would then account for the shorter analyte
di-chained
in
the
times
the
times,
the
residence
reduced
whilst
of
quats
retention
have
limited
interaction
in
the
analyte-bonded
would
phase
resulting
stationary-liquid
3.27).
(Figure
loss
of resolution
observed
Additional support for the homologue resolution being derived from electrostatic
interactions with the bonded phase, rather than partitioning, was believed to be
forthcoming from the observed behaviour of the monoester quats in the two solvent
Chapter 3: A new NP-LC method for fabric conditioner
actives

118
hydrophilicity
have
The
these
of
analytes
would
greater
systems.
resulted in increased
in
di-chained
in
to
the
the
phase,
comparison
stationary-liquid
residence
quats, in both
in
the
the electrostatic behaviour of these
variation
solvent systems, accentuating
in
in
greater
resolution
comparison with the diesters (Figure
components, resulting
3.10). In addition, contrary to what was predicted for the diesters, it was envisagedthat
have
been affected when the
the
times
components
monoester
would
not
of
residence
hexane-methanol
due
to
to the observedsolubility of
switched
was
system
mobile phase
these analytesin methanol. As a result, resolution would have been maintained, as was
(Figure
3.27).
by
experimentation
witnessed
Whilst it was not possible to conclusively justify how the resolution of the
monoesters increased upon removal of THF, it was proposed that additional stationary
/
interactions
or solvent-induced retention might have been responsible. It
and
phase
hydrogen
bonding with the stationary
that
the
would
monoesters
under-go
was predicted
phase and the methanol in the stationary-liquid phase. With the methanol content of the
being
far higher in the binary solvent system, hydrogen bonding
stationary-liquid phase
further
increased,
have
in
the behaviour of the
accentuating
any
variations
would
hypothesised
increased
It
hydrogen bonding facilitated
therefore
that
was
analytes.
increased resolution in the binary solvent system.

The mixed retention model and the observations made with the binary solvent
3.28)
be
(Figure
fronting
to
the
could
also
used
system
explain
severe peak
demonstratedby the quaternary amino-alcohols in the ternary mobile phase system. It
fronting
demonstrated
by
derived
from
diols
that
the
the
assumed
originally
was
incompatibility of the analyte with the mobile phase. However, in light of the
it
instead
hypothesis
was
proposed that the diol analytes partitioned
stationary-liquid
into the stationary liquid and resisted partitioning back out. It was envisaged that
hydrogen bonding between the analytes, and between the analytes and the stationary
long-term
their
the
stabilise
solvent,
would
methanol
residence in the
phase and
in
bulk
limited
hexane
Having
the
liquid.
solubility
solvent, partitioning of a
stationary
diol analyte out the pores of the silica would only come about as a result of hydrogen
bonding with a methanol molecule in the bulk liquid phase. As the analytes were
it
hypothesised
interact
loss
that
other,
was
the
each
to
with
of one analyte
predicted
into the bulk would promote the loss of others i. e. partitioning back into the bulk mobile
Chapter 3: A new NP-LC method for fabric
conditioner actives

119
back
into
Assuming
bulk
transfer
that
the
was a co-operative
phase was co-operative.
leave
the stationary-liquid phasesimultaneously.
process,most analytemoleculeswould
Though transfer would be slow at first, the rate would increasewith time until all of the
been
liberated
bulk.
had
This release pattern would be
to
the
analyte molecules
increased
band
before
falling back to the
by
that
exponentially
characterised an analyte
baseline as all of the molecules in the pores were exhausted. Such a band profile was
during
fronting
diol
the
the
the
to
witnessed
peak
analysis
of
with the ternary
analogous
solvent system.
Utilisation of the binary solvent system would have resulted in a significant
increase in the concentration of methanol in the bulk liquid phase, which would have
greatly increasedthe solubility of the diol component. As a result, a conventional band
profile would be expected,where peak tailing would be likely to be severeas a result of
from
the
the stationary phaseand the stationary-liquid phase.
analytes
of
uneven release
Experimental observations for the diol species under the two solvent systems could
therefore bejustified.

Whilst experimental evidence is available to support the theory of a mixed


adsorption - partition model of cationic tenside retention, a number of potentially
important experiments need to be carried out in the future to resolve a number of
Firstly,
questions.
a bare silica phase needs to be evaluated with both
unanswered
determine
to
whether any resolution of the tenside homologues is
solvent systems
forthcoming from the silanol groups on the surface of the support. Similarly, a strong
(SAX)
column should also be evaluated, as this phase contains a
anion exchange
be
quaternary
ammonium
charged
group,
which could
used to determine
permanently
bonded
facilitate
homologue
in
the
on
phase
actually
groups
charged
whether
resolution
the new NP-LC methodology. Finally, having predicted that the peak shape of the diol
dependant
in
the
the solvent system,
concentration
on
of
was
methanol
component
THE
decrease
in
fronting, and
to
the
allow
should
one
of
study
gradual replacement
in
increase
peak tailing.
subsequent

Until such a time occurs when these experiments are undertaken, it will be
difficult to justify with any certainty the actual mode of retention in the new
it
is
Nonetheless,
that
apparent
whatever the retention mechanism, it
methodology.
Chapter 3: A new NP-LC method for fabric
conditioner actives

120
facilitates efficient reproducibleresolution of first and second-generationcationic fabric
conditioner actives.

3.7

CONCLUSIONS

This chapter has described the development of a new isocratic normal phase
liquid chromatographic method for the analysis of cationic fabric conditioner actives,
the method being developed from a previous literature report after problems were
Unilever
Evaluation
the
the
current
use
of
standard
procedure.
operating
witnessedwith
of a range of stationary and mobile phase systemsrevealed that optimum resolution of
the tensides was achieved with a ternary mobile phase system on an amino-bonded
stationaryphase.
Utilisation of co-injection analysis showed that the method was capable of
separatingthe major classesof cationic tensidespresentin commercial samples,as well
homologous
the
series present in each class. Subsequentanalysis
as partially resolving
of mixed surfactant samples revealed that the new methodology could also resolve
individual components from a specified sample origin, a feat that had not previously
beenpossible with literature methods.
The capability of the new method to partially resolve the homologous series
in
present commercial sampleswas unusual for a normal phaseLC method. As a result,
literal
investigation
led to the proposal of a mixed
and
experimental
extensive
being
for
high
the
model
responsible
partition
resolving power of the new
adsorption method. The resolution was therefore critically dependant on the formation of a
in
liquid
the pores of the silica substrate,yet all of the observed separations
stationary
demonstratedexcellent long-term reproducibility, and the method itself was found to be
inherently robust.
Whilst the method was found to be well suited to the analysis of the parent
ill
it
tensides,
suited to the analysis of the quaternary amino-alcohol
was
cationic
degradationproducts. Some improvement was gained in the chromatographyof the diol
However,
from
it was ultimately
of
a
gradient
elution
profile.
utilisation
species

Chapter 3: A new NP-LC method for fabric conditioner


actives

121
from
the analysis of these analytes
that
the
resulting
chromatography
poor
concluded
with the new methodology,would prohibit their efficient quantitation at trace levels.
Unfortunately, the new method also suffered from a number of other inherent
limitations. Method sensitivity was low, even after inclusion of aa narrow-bore column,
light
detector
the
the
scattering
evaporative
was seento be non-linear. It
and
responseof
lack
in
limit
the
the applicability of
that
of
sensitivity
particular
would
was envisaged
the methodology to environmental analysis. However, the potential to increasemethod
detection
with
mass
spectrometric
was evident after the
selectivity
sensitivity and
hyphenated
of
a
methodology.
successfulevaluation
In its current form the new optimised NP-LC methodology appears to offer
in
for
routine use the quality control assessmentof the conditioner-active raw
potential
fabric
in
domestic
be
However,
further
conditioners.
optimisation will
materials used
required before the methodology can be routinely applied to the analysis of industrial
formulations and environmentalmatrices.

3.8

REFERENCES

Abian J., OosterkampA. J. and Gelpi E., J. Mass Spectrom.,1999,34,244-254.


Anachem, Personal Communications,2000, Unilever Research.
Burford

M. D., Personal Communications, 1997, Unilever Research.

Burford

M. D., Personal Communications, 1999, Unilever Research.

Burford

M. D., Personal Communications, 2000, Unilever Research.

Cooper D., Personal Communications, 1998, Unilever Research.


Cooper D., Personal Communications, 2000, Unilever Research.

Dmoch R., The Evaluation of Novel Supercritically Bonded Chromatographic


Stationary Phases,Ph.D. Thesis,University of Leeds, 1999.
Dorsey J.G. and Cooper W.T., Anal. Chem., 1994,66,857A-867A.
Fernandez P., Alder A. C., Suter M. J.-F. and Giger W., Anal. Chem., 1996,68,921929.
Hewlett Packard, Applications of narrow-bore columns in HPLC, 1997.
Kaloka H.A. and ShatzV. D., J. Chromatogr., 1992,626,181-186.
Kazoka H. and ShatzV., J. Chromatogr. A, 1996,732,231-238.
Chapter 3: A new NP-LC method for fabric conditioner
actives

122
Kazoka H., J. Chromatogr.A, 2000,874,45-53.
Kuhlmann F.E., Apffel A., Fischer S.M., Goldberg G. and Goodley P.C., J. Am. Soc.
Mass Spectrom.,1995,6 1221-1225.
Lawrence J.G., Personal Communications,1997, Unilever Research.
Lawrence J.G., Personal Communications,1999, Unilever Research.
Meyer V. R., Practical High-Performance Liquid Chromatography, 2d Edition, Wiley
and Sons,New York, 1993.
Nawrocki J., J. Chromatogr. A, 1997,779,29-71.
Nitschke L., Mller R., Metzner G. and Huber L., FreseniusJ. Anal. Chem., 1992,342,
711-713.
Norberg J., Thordarson E., Mathiasson L. and JnssonJ.A., J Chromatogr. A, 2000,
523-529.
Phenomenex, Catalogue for Separation Science, 2001.

Piretti

M. V.,

Diamante M.,

Bombardelli E. and Morazzoni P., Biomedical

Chromatography, 1996,10.43-45.
Radke M., Behrends T., Frster J. and Herrmann R., Anal. Chem., 1999,71,53625366.
Robson M. M., Personal Communication, 1998, University of Leeds.
Schmitt T.M., Analysis of Surfactants, 2dEdition, Wiley and Sons,New York, 2001.
van Damme F. and Verzele M., J. Chromatogr., 1986,351,506-511.
Verzele M. and van Damme F., J. Chromatogr., 1986,362,23-31.
Verzele M. and van Damme F., J. Chromatogr., 1987,393,25-37.
Waters J., Lee K. S., PerchardV., FlanaganM. and Clarke P., TensideSurfactants and
Detergents,2000,37,161-17 1.
Wee V. T. and Kennedy J.M., Anal. Chem., 1982,54,1631-1633.
Wehr T., LC-GC International Online, April 2000, <http://www. lcgcmag.com>.
Weston D.J., Personal Communications,1999, Unilever Research.
Weston D.J., Personal Communications,2000, Unilever Research.
Wilkes A. J., Walraven G. and Talbot J: M., J. Am. Oil Chem.Soc., 1992,69 609-613.
Wilkes A. J., Wairaven G. and Talbot J.-M., Journale Comite. Espanol Deterg., 1994,
25.209-220.
Wilkes A. J., Jacobs C. Walraven G. and Talbot J.-M., Proceedings of the 4`h World
Surfactants Congress, Barcelona, 1996.

Chapter 3: A new NP-LC method for fabric


conditioner actives

123

CHAPTER FOUR
Development of a reverse phase LC/MS
for
the
routine analysis of alkylbenzyl
method

quat preservatives

Chapter4: RP-LC/MSof alkvlbenzylauats

124

CHAPTER FOUR

Development of a reverse phase LC/MS method for the


routine analysis of alkylbenzyl quat preservatives
4.1

INTRODUCTION
With the widespread application of alkylbenzyl quats as bactericides and

is
for
in
household
there
and
personal
care
products
a
requirement
preservatives
methods of analysis to quantify these compounds in complex industrial and
Currently,
liquid
(RP)
(LC)
reverse
phase
matrices.
chromatography
environmental
separations performed on long cyanopropyl (CN) bonded columns are the primary
Separations
industry.
in
utilise mobile phases containing high
methods used
ion-pairing
high
linear velocities to facilitate
reagents
of
non-volatile
and
concentrations
homologue
While
distributions present in commercial samples
the
times.
short analysis
by
be
such methods, the mobile phase conditions prevent hyphenation
resolved
may
in
quantitation
making
mass
spectrometry
environmental matrices troublesome.
with
The aim of this section of work was to develop a hyphenatedLC/MS method capableof
in
fully
formulated
quats
commercial products and environmental
analysing alkylbenzyl
wastewaters,rivers and sediments.

4.2

DEVELOPMENT OF THE LC METHODOLOGY

4.2.1 Choice of initial parameters


For convenience, the method described in the current Unilever Research
(Section
(SOP)
1.4.4)
base
procedure
was
chosen
operating
as
standard
an appropriate
from which to develop the new methodology. With the use of a CN column being
literature
in
the use of octadecylsilane
and
methods,
reported
problems
of
characteristic
(ODS) bonded stationaryphasesfor the analysis of quaternaryammonium preservatives
(Valladao et al., 1994; Zhou et al., 1999) (see also Section 1.4.4), a CN phase was
basis
However,
length
from
the
the
of
new
method.
as
column
maintained
was reduced
250 mm to 150 mm to promote rapid analysis at reduced mobile phase velocities. (See
Section 1.3.1).

Chapter 4: RP-LC/MS of al

lbenzyl quats

125
ACN was retained as the strong solvent constituent of the mobile phaseafter the
use of MeOH was found to result in lower efficiency, reduced resolution and longer
improve
To
time.
compatibility of the LC methodology with electrospraymass
analysis
spectrometry (ESI-MS) the non-volatile sodium perchlorate (NaClO4) used in the
Unilever SOP was replaced with volatile alternatives that would allow for efficient
resolution and short analysis times of the quat species (Vervoort et al., 1999; de
Schutter et al., 1988). In Section 1.3.2.1 the advantagesand disadvantagesof using
various organic modifiers were discussedin relation to the analysis of organic basesand
quaternaryammonium species.It was noted how efficient resolution and short analysis
times could be achievedwhen both an organic acid and basewere presentin the mobile
phase. In light of such observations, two modifiers were incorporated into the new
methodology, a competitor quaternary ammonium salt and a strong organic acid.
Ammonium acetate(NH4Ac) was chosenas the competitor species,due to its suitability
for ESI-MS and its efficacy in the analysis of a number of other quaternaryammonium
1991;
(Barcelli
Moyano et al., 1996; van der Hoeven et al., 1996; Evans et
et
al,
species
Trifluoroacetic
(TFA)
2000).
acid
was chosen as the organic acid due to the higher
al.,
it
had
in
the NP-LC method in comparison with acetic acid and
that
afforded
resolution
formic acid (Section 33.3), and its higher volatility and hence greater ESI-MS
compatibility comparedto octanesulphonicacid (de Schutteret al., 1988). The choice of
mobile phasepH was dictated by the durability of the Spherisorb CN stationary phase
that was initially used in the methodology. Manufacturers' guidelines recommendedthe
column be limited to a working range of pH 3-7 (Waters, 1999). However, the mobile
2.0
at
pH
as previous work had shown only a small reduction in
operated
phase was
(Myers,
2000),
lifetime
this
pH
at
column
which was compensatedby a reduction in
secondaryretention effects on surface silanol groups (Section 1.3.2.1). The strong acid
TFA
also allowed pH adjustment to be achieved with lower volumes of
of
properties
likelihood
interface
fouling and blockage (Lagerwerf et al.,
the
of
reagent, reducing
2000).

Of the instrumental parametersused in the Unilever SOP, only two were altered
initial
development
during
the
of the new methodology. The linear velocity
significantly
flow
by
the
third,
phase
mobile
a
rate was reduced from 1.5 mumin to
was reduced
1.0 mumin, whilst the UV / vis detection wavelength was modified from 254 nm to 214

Chapter 4: RP-LC/MS o alkvlhenzyl quats

126
increase
following
in
that
ten-fold
the
a
observation
sensitivity would be afforded at
nm,
this new wavelength.
A chromatogramobtained from the analysis of a mixed alkylbenzyl quat sample
with the modified methodology is shown in Figure 4.1. The samplecomprised 250 mg/1
C12,C14,C16and C18alkylbenzyl quat (Appendix One), and from reversephasetheory
it was predicted that the first abundantpeak (ca. 11.8 minutes) correspondedto the C12
homologue, whilst with the last correspondedto the C18component. Evidence that the
had
been
identities
correctly assigned came via sequential enrichment of the
peak
individual
homologues.
The figure also shows that the new
the
sample with
methodology afforded efficient resolution of the four quats, and that the homologous
series actually appearedto commenceat Cg rather than C12,indicating one or more of
the individual quat samplescontained short-chain impurities.

Figure 4.1: Chromatogramshowing the separationof four alkylbenzyl quats achieved with the
new RP-LC method parameters.
Conditions - Column: 150 x 4.6 nun i.d. Spherisorb 5 m CN; Mobile phase: 50:50
ACN: 5 mmol/1NH4Ac (pH 2.0); Flow rate: I ml/min; Detection wavelength: 214 nm.

4.2.2 Use of new silica technology


Whilst the new methodology facilitated quantitation of the individual alkyl
benzyl quats, a number of limitations were evident. Firstly, analysis time had stretched
to 21 minutes, an increase of over 260% on the Unilever method, and subsequent
by
increasing
time
the
to
analysis
the eluting strength of the mobile
attempts reduce
Chapter 4: RP-LC/MS Ofalkylbenzyl

quats

127
lost.
Secondly,
the method resulted in the
was
as
resolution
unsuccessful
were
phase
quats demonstrating severe peak tailing as a result of uneven release from the silica
surface brought about by additional interactions with surface silanol groups and / or
in
the Spherisorb material (Myers, 1999).
the
analytes within cusps
retardation of
Whilst peak tailing is an aesthetic problem in all LC separations,in trace analysis it
leads to difficulties in differentiating between background noise and the tail of the
in
being
results
method
sensitivity
which
compromised. During the
analyte peak,
analysis of complex matrices this problem is compounded by co-eluting interferences.
The broader the peak the greater the number of possible interferences,and the greater
the likelihood of further analyst and instrument time being taken up by reanalysis.
To overcome the problems of peak tailing witnessed in the analysis of basic
based
are
silica
phases
on high purity supports that are smoother, more
analytesmodem
distribution
have
even
a
more
of surface silanol groups (Section 1.3.2.1).
and
spherical
Such improvements give rise to even release of basic and / or cationic analytes from
their surfaces(Myers, 1999; Phenomenex,2000), and, whilst they do not eliminate the
interactions
for
improve
they
secondary
often
potential
efficiency and peak symmetry.
The separationof the mixed alkylbenzyl quat was thus repeatedon a secondCN
based
that
was
on a modem high purity silica support (Luna from
material
Phenomenex).Figure 4.2 shows the variation in separation that was achieved on the
two phases.The blue trace corresponding to the Luna phase clearly demonstratesthat
this material provided a reduction in peak tailing, analysis time and analyte peak
volumes.
Manufacturers' guaranteeson working pH range (1.5 to 7.0), column lifetime
(guaranteedfor > 1000 hours continuous running at pH 1.5) and the provision of a full
in
(Phenomenex,
2000),
analysis
addition to independentobservationson
certificate of
the batch-to-batch reproducibility and robustness of the silica support (Kele et al.,
2000), as well as the observations detailed above, led to the Luna CN phase being
in
the new methodology.
the
of
choice
column
adoptedas

Chanter 4: RP-LC/MS o alkvlbenz l guats

128

4.2.3 Improving sensitivity and MS compatibility


During the optimisation of the NP-LC method (Chapter Three) the use of a 2.0
five-fold
increase
had
in method sensitivity, compared to an
i.
d.
a
provided
mm
column
improve
To
i.
d.
4.6
the sensitivity of the RP-LC method, and
column.
analogous
mm
improve MS response and interface dynamics (Abian et al., 1999), the

ultimately

i.
d.
2.0
Luna
CN
mm
of
a
column was evaluated. Figure 4.3 shows an
performance
overlay of the chromatograms resulting from the analysis of a 50 mg/1 benzalkonium
chloride standard on a 4.6 mm and 2.0 mm i. d. Luna 3 m CN column. The two peaks
corresponded to the C 12 and C 14 alkylbenzyl quats, with the ratio of the two species
being characteristic of a sample derived from coconut oil. It is important to note that the
linear velocity was constant in both analyses, and thus the observed variation in
retention time was as a result of the influence of instrument and extra-column void
volumes on narrow bore separations (Wehr, 2000; Hewlett Packard, 1997).

The

chromatogram

obtained

on

the

2.0

mm

i. d.

showed

significant

improvements in analyte peak area response compared to the 4.6 mm i. d. alternative,


and thus the narrow-bore column was adopted for all further work.

120

100

80
E
m

60

0
0.
N

40

20
AKU--ZX
0
02468

10

12
Time

14

16

18

20

22

24

26

(minutes)

Figure 4.2: Figure showing the variation in separation efficiency afforded by the Luna CN
(blue trace) and Spherisorb CN (red trace) materials.
Conditions - Column dimensions: 150 x 4.6 mm i. d. packed with 3 m silica particles; Mobile
NH4Ac
2.0);
Flow
5
(pH
50
ACN:
50:
mmol/I
rate: 1 ml/min.
phase:

Chapter 4: RP-LC/MS of alkylbenzylguars

129

4.2.3.1 The peak compression effect


"Analyze peak compression " is a recognised LC phenomenon in which the
in
lower
is
dissolved
interest
a
solvent
of
eluting strength than the LC mobile
sample of
into
is
injected
When
the eluant stream the analyte plug is compacted,
the
sample
phase.
band
is
in
interior.
the
the
than
that
the
the
rear
of
at
strength
stronger
eluting
as
Although the analytes undergo the same degree of band broadening during their passage
through the column. having derived from a narrower band, peak volumes are reduced
leading to an increase in efficiency and method sensitivity.

Peak compressionwas overlooked during the analysis of the fabric conditioner


actives due to the limited solubility of the analytes in pure hexane. However, the effect
is referenced here as the chromatograms in Figure 4.3 took advantage of the effect, as
the benzalkonium chloride standard that was used for the two analyses was prepared in
80: 20 water: ACN, rather than 50: 50 water: ACN.

25

20

15

U)
C
co
0

CL

Od)

10

0
0123456789

10
Time (minutes)

Figure 43: Figure showing the improvement in sensitivity that was achieved by using a 2.0 mm
W. column.
Conditions - Column: 150 x 4.6 mm W. Luna 3 m CN column (blue trace), 150 x 2.0 mm i. d.
Luna 3 m CN column (red trace); Mobile phase: 50: 50 ACN: 5 mmol/l NH4Ac (pH 2.0);
Sample: 50 mg/I benzalkonium chloride in 80: 20 water: ACN; Flow rate: 1000 Rumin (blue
trace), 190 l/min (red trace).

Chapter 4: RP-LC/MS of alk

'1benzylquats

130
Peak compressionwas observedin both of the separationsshown in Figure 4.3.
However, the effect was more pronounced on the narrow-bore column due to the
injector,
flow
In
HPLC
it
be a manually actuated
an
rate.
whether
phase
reducedmobile
valve such as a Rheodynemodel 7125 or a mechanically actuatedvalve like those found
in commercial autosamplers,solute injection follows the switching of the valve from the
load to the inject position. Eluant flow is diverted through the injection loop washing
the solute band onto the analytical system. When a filled 10 l loop is flushed with
loop
is
in
1000
600 ms. However, when the flow
the
l/min,
at
emptied
mobile phase
loop
filled
for
3000
is
190
the
to
l/min
remains
partially
with
sample
over
reduced
rate
is
formed,
broader
A
analyte plug
which results in larger peak volumes and
ms.
The
lower
severity of the band broadening occurring in the
efficiency.
ultimately
is
compounded when the sample diluent is of a higher eluting
column
narrow-bore
strength than the mobile phaseeluant. Here, the front of band moves more rapidly than
the tail leading to further expansion of the plug. This effect was witnessed with the
ACN
RP-LC
methodology
when
was used as the sample diluent. Instead of
optimised
two discrete peaks, a single poorly efficient peak was observed on the 2.0 mm i.d.
demonstrated
baseline
a
width in excess of four minutes. When the
column, which
i.
4.6
d.
the
on
mm
was
performed
column, little variation was apparent
analysis
betweenthe compressedand non-compressedbands.
Peak compression is commonly used in the analysis of therapeutic drugs in
biological matrices like blood and plasma. The technique allows for improvements in
is
limited,
size
sample
as it facilitates large volume injection (50sensitivity when
100 l) onto narrow or micro-bore columns without deterioration in efficiency (Mills et
4.3
Figure
1997).
appearedto provide encouragingresults in the use of narrow bore
al.,
for
One
the
of
alkylbenzyl
quats.
analysis
was led to believe that the increase
columns
in sensitivity was obtained at the expense of an insignificant reduction in efficiency.
However, problems were soon encountered when consideration was not paid to the
Major
in
sample
a
quat
was
prepared.
which
problems were also envisagedin
manner
for
the environmental analysis after discussions with Unilever
technique
the
the use of
Research (Sparham, 2000). Sample preparation protocols used to extract cationic
tensidesfrom environmental matrices include a 100 to a 1000 fold pre-concentrationof
in
final
the
step the processbeing the resolvation of the analytesin the
the analytes,with
Whilst
for
straightforward
solvent.
of
most analytes,the propensity of
minimum volume
Chapter 4: RP-LC/MS of
alkylbenMI guats

131
has
been
lead
inefficient
the
to
to
to
tensides
to
walls
of
a
vessel
seen
sorb
cationic
is
if
not carried out with extreme care.
resolvation
recovery

Work performed at Unilever Researchhas shown that 80:20 water:ACN is an


inefficient resolvation system for most cationic tensides(Sparham, 2000), and that pure
ACN is required to guaranteeefficient recoveries of the alkylbenzyl quats. When used
in conjunction with the new methodology, efficient recoverieswould be achievedat the
expenseof poor chromatography.As a result, although the 2.0 mm W. column and the
diluent
for
RP-LC
the
the
were
work,
used
of
sample
remainder
use of a weak
Unilever
Research
it
be
that
suggest
at
made
may
more effective to use a
observations
4.6 mm i. d. column for environmental analysis.

4.2.4 Determination

of the sensitivity, linearity

and robustness of the

new method
4.2.4.1 Method validation
A formal validation of the RP-LC-UV method was never performed due to time
during
this segmentof work. Instead, experiencegained during
restrictions experienced
the validation of the NP-LC method was applied to this RP-LC work:
During NP-LC method development an aged column could be replaced with a new
in
loss
performance.A similar observation was made during the RP-LC
column with no
work with agedCN columns.
Transfer of the NP-LC method to Unilever Research yielded equivalent separation
University
the
to
that
witnessed
at
of Leeds. Variations in instrumentation
efficiency
had
apparently
no effect on the separationefficiency. Theseresults
and reagentsupplier
RP-LC
later
the
transferred(Figure 4.4).
methodology
was
when
were echoed
The above parameters form the basis of an inter-lab validation study, and are thus
inherent
in
determining
the
robustnessof new chromatographic methodology.
critical
Although the observationsdid not representa rigorous validation of the method, it was
it
be
to
was
sufficiently
the
to
transferred
that
methodology
robust
new
allow
predicted
between laboratories, and utilised by adequatelyqualified personnel, without resolution
being compromised.

Chapter 4: RP-LC/MS ofalkvlbenzvl quats

132

4.2.4.2 Method sensitivity and linearity


The sensitivity

of an analytical

method is ultimately

dependant on the

is
It
is
impossible
define
detector
limit
to
that
the
used.
an
exact
of
of
performance
detection for a new methodology without specifying that the same model of detector be
Even
be
further
if
for
then,
equivalent
performance
work.
will
only
achieved
all
used
both detectors have been maintained to the same degree. For linearity, the same is also
true.

To determine the linearity of the optimised RP-LC-UV methodology, a series of


ten Querton

KKBCL

calibration

standards were

prepared,

ranging

in

active

The
from
0
100
in
to
triplicate, and the
mg/l.
standards
were
analysed
concentration
average peak areas of the two principle components, the C12 and C14 homologues,
determined (for an indication of the separation see Figure 4.5). Calibration curves for
the two components were created, and a first order polynomial regression fit applied to
4.7
4.6
both
Figures
data
that
and
show
curves demonstrated excellent
set.
each
linearity, with the R2 values being in excess of 0.995, and approaching 1.0 in the case of
the C12component.

Figure 4.4: Chromatogram showing the analysis of a 100 mg/I benzalkonium chloride standard
laboratory.
Sunlight
Port
the
performed at
Conditions - Column: 150 x 2.0 mm i. d. 3 m Luna CN; Mobile phase: 50: 50 ACN: 5 mmol/I
NH4Ac (pH 2.0); Sample diluent: 80:20 water: ACN; Flow rate: 190l/min.

Chanter 4: RP-LGMS of alkylbenzyl auats

133
It was evident after the initial transfer of the RP-LC-UV

method to Unilever

Research that the diode array detector (DAD) present in the HP 1100 system, would
detection
detector
lower
limit
(LOD)
than
the
of
a
used at Leeds. However,
provide
during the linearity study it was apparent that a number of standards were close to, or
beyond both the limit of quantitation (LOQ), defined here as a signal to noise ratio
(S/N) ratio of 10, and the LOD, defined as a S/N ratio of 3. When the 1 mg/I Querton
KKBCL standard was analysed the peak area responses of the C12 and C14 components
LOQ
be
(Figure 4.8). However, when the 500 g/1 standard was
the
to
above
were seen
C14
longer
the
component
analysed
was
no
visible, and though the peak
subsequently
relating to the C12 component was present, the peak area response was below the LOD.
Further work revealed that the actual limit of detection of the C12 component was 850
g/l Querton (x 650 parts per billion (ppb) C12), for a 10 l injection. In Section 3.5.5.1
it was stated that a limit of detection of 1 mg/l is required for an LC method to be
acceptable for environmental analysis. It was therefore evident that the new LC-DAD
for
the quantitation of alkylbenzyl quats in environmental
suitable
was
methodology
matrices.

Ihm

120

C'12

,ao.

C14

M.
C10

('16

C18

sk
1

Figure 4.5: Chromatogram showing the analysis of the 50 mg/I Querton KKBCL standard.
Conditions - Column: 150 x 2.0 mm i. d. 3 gm Luna CN; Mobile phase: 50: 50 ACN: 5 mmol/l
NH4Ac (pH 2.0); Sample diluent: 80: 20 water: ACN; Flow rate: 190l/min.

Chapter 4: RP-LC/MS ofalkvlbenzvl

guats

134

3000

y= 27.963x
R2 = 0.9999

2500
2000
E
1500
IY

1000

500

0
0

10

20

30

40

50

60

Concentration

(mgiL)

70

80

90

100

4.6: Figure showing the linearity of the diode array detector response to the C12
KKBCI
in
Querton
at varying concentration.
component
Conditions - Column: 150 x 2.0 mm W. Luna CN column; Mobile phase: 50: 50 ACN: 5 mmol/I
NH4Ac (pH 2.0); Sample diluent: 80: 20 water: ACN; Flow rate: 190 1/min.
Figure

900
800

y=7.6766x

5.2522
-

700
F

600

E 500
400
300
200
100
0
0

10

20

30

40

50

Concentration

60

70

80

90

100

(mg/1)

Figure 4.7: Figure showing the linearity of the diode array detector response to the C14
KKBCI
Querton
in
at varying concentration.
component
Conditions - Column: 150 x 2.0 mm W. Luna CN column; Mobile phase: 50: 50 ACN: 5 mmol/l
NH4Ac (pH 2.0); Sample diluent: 80: 20 water: ACN; Flow rate: 190 l/min.

Chapter 4: RP-LC/MS ofalkylbenzyl quats

135

DAD1 B Sig=214,4

Ref=360.100

(IBROM\MAY23D08

D)

Norm

35

C12
CD
rn
25

2
II

I
15

Impurity

C10and
Impurity

14

05
0

-0.5

T--

02468

10

12

rr

Figure 4.8: Chromatogram showing the analysis of the I mg/I Querton KKBCL standard.
Conditions - Column: 150 x 2.0 mm W. Luna CN column; Mobile phase: 50: 50 ACN: 5 mmol/l
NH4Ac (pH 2.0).

The lack of a formal validation and application of the methodology to the


ideal,
fortified
samples
environmental
was
not
especially considering the
analysis of
developed
for
in
laboratory
that complied with the
was
routine
use
a
methodology
new
protocols

of

reproducibility

Good Laboratory

Practice (GLP).

of the separation and suitability

Uncertainty

over the long-term

of the method for environmental

being
led
have
favour
in
to
the
method
shelved
of the SOP. However,
analysis could
Unilever chose to adopt the new methodology, and have subsequently validated its use
in the analysis of alkylbenzyl quats in a number of environmental matrices (Sparham,
2000).

4.3

DEVELOPMENT

OF

THE

HYPHENATED

LC/MS

METHODOLOGY
The unknown impurity

that was seen to co-elute with the C10 component

its
interfere
4.8).
thus
with
quantitation, provides an example of the main
(Figure
and
methodology; i. e. an absence of structural specificity to
identities.
The
UV
peaks
spectra of the alkylbenzyl quats are
unambiguously confirm
limitation

of the LC-DAD

Chapter 4: RP-LC/MS o alkylbenzyl


quats

136
have
(190-260
low
few
if
nm),
and
at
wavelengths
any characteristic
unremarkable
bands at higher wavelengths.As a result, the spectral information provided by the DAD
in
be
to
used assist peak purity assessmentwhen co-eluting impurities had
could only
characteristicabsorbancespectra.
In order to improve the selectivity and specificity of the methodology, especially
in the analysis of alkylbenzyl quats in complex matrices, work was undertaken to
hyphenatethe LC-DAD methodology with electrospraymass spectrometry(ESI-MS). It
information
in
LC/MS
to
the
that
addition
structural
new
methodology
was envisaged
lower
LOD
had
been
for
LC-DAD
than
that
the
a
which
also
yield
reported
would
method.

4.3.1 Direct infusion and loop injection studies


Initial mass spectral assessmentof the alkylbenzyl quat sampleswas performed
infusion
loop
injection
direct
the
and
analysis(Sections 3.5.6.1 and 3.5.6.2).
aid of
with
In direct infusion mode a 100 mg/1benzalkonium chloride standard,preparedin
80:20 water:ACN, was used to tune the MS instrument (Appendix Three). Observation
4.9),
(Figure
the
spectrum
showed two principle ions, a base peak at
of
resulting
ion
50%
304.3
abundant
at m/z = 332.2, correspondingto the molecular ions
and a
m/z =
(M), of the C12and C14alkylbenzyl quats. These ions had been expected after results
fabric
the
conditioner actives and previous RP-LC/MS analysis of
observed with
(Hau
2000;
Evans
2000).
Importantly, the
species
et
al.,
ammonium
et
al.,
quaternary
ions
two
the
correlated with the peak areaswitnessed during the LC-UV work
ratio of
data
(See
Appendix
One). In a repetition of observations
the
manufacturers
and with
3.5.6.1,
high
Section
during
a
series
of
molecular weight ions were again
made
data
be
derived
from
fragmentation
them
to
the presenceof dimeric
revealing
apparent,
it
impossible
Whilst
/
C14
C12
to be certain of the origin of
quats.
was
and or
adductsof
hypothesised
it
that they derived from the electrospray process
ions,
these
was again
(Section 3.5.6.1).

During loop injection analysis tuning of the MS instrument was performed with
KKBCL
Querton
100
standard. A poly (ether (ether) ketone) (PEEK)
mg/1
the aid of a
T-piece was inserted into the instrumental set-up (see Section 3.5.6.2) to facilitate

Chanter 4: RP-LC/MS ofalkvlbenzyl puats

137
into
(infusion
infusion
the
the
eluant
stream
rate =3
analyte
of
continuous

l/min).

Initial MS parameters were based on a tune file obtained previously (Appendix

Three).

The capillary temperature was increased from 220C to 250C, and the sheath gas being
for
higher
in
35
the
to
to
compensate
volume of eluant
order
units
arbitrary
regulated
entering the source.

Figure 4.9: Mass spectrum obtained from the direct infusion of a 100 mg/1 standard of
benzalkonium chloride into the Finnigan MAT LCQ instrument after MS optimisation.
Infusion rate: 3Wmin.

When the analytes were initially infused into the flowing eluant, the molecular
ion of the C12component(m/z = 304.3) was dwarfed by an unknown ion at m/z = 242.3.
Variation of the Tube Lens Offset voltage led to a variation in the abundanceof the C12
/
fragmentation
less
but
ion
more
occurred,
as
negligible effect was witnessed
molecular
infusion
direct
ion.
Previous
MS
had
the
studies
shown low abundanceof
unknown
on
data
(NMR)
242.3,
nuclear
magnetic
resonance
and
provided by Unilever
m/z =
Research,indicated no major impurities were present in the sample (Appendix One). It
ion
hypothesised
that
the
the
origin
of
therefore
was the LC mobile phase, the
was
hypothesis being subsequentlyconfirmed when mobile phase was pumped directly into
the ESI interface in the absenceof any quat samples(Figure 4.10).
Having identified the origin of the ion, attempts were made to reduce its
flow
The
temperature
and
sheath
gas
capillary
rate were increased and
abundance.
facilitate
droplet
to
added
evaporation. Unfortunately, the
auxiliary gas was
Chanter 4: RP-LC/MS of alkvlbenzyl quats

138
did
batches
base
little
had
too
the
use
of
new
solvent
of
and
as
effect,
and
modifications
the replacementof TFA with AcH.

Figure 4.10: Mass spectrum of the RP-LC mobile phase system showing the unknown ion at
m/z = 242.3.
Conditions - Mobile phase:50:50 ACN: 5 mmol/l NH4Ac (pH 2.0); Flow rate: 190 l/min.

High abundancyimpuritieshad beenwitnessedin a numberof earlier LC/MS


formed
basis
In
(Weston,
2000).
ACN
in
these
the
the
of
mobile
phase
studies, which
ions
formed
from
to
the co-ordination
the
were
attributed
salt
clusters
unknown
studies
in
ions
HPLC
Na
In
the
ACN
present
solvent.
an attempt to confirm the origin
with
of
ion,
MS
the
tandem
the
composition
structural
of
experiments were
elucidate
and
fragmentation
Unfortunately,
the
work
was
wholly
unsuccessful,as
of the
performed.
ion resulted in it being obliterated, liberating a host of product ions that were uselessfor
information,
Without
it
impossible
to categorisethe
structural
was
elucidation purposes.
hypothesis.
However,
ion.
the
origin
a
salt
cluster
remained
most
viable
unknown

Onemethodwas identifiedto preventthe impurity ion from swampingthe total


ion chromatograms (TIC's) of the respective quat samples. Prior to each set of analyses
from
the
subsequently
which
was
stored,
subtracted
of
was
each
spectrum
a solvent
242.3
ion,
the
the
TIC's,
of
m1z
=
presence
and yielding clean analyte
removing
analyte
from
injection
4.11
that
loop
the
Figure
spectrum
the
mass
was
shows
obtained
spectra.
Querton
KKBCI,
been
had
100
the
of
after
standard
mg/l
solvent
spectrum
analysis of a
The
C10,
C12
the
C14
from
trace.
the
abundances
of
and
components were
subtracted

Chapter 4: RP-LC/MS of al kylbenzylguats

139
data
for
indicating
in
this
the
available
sample,
characterisation
a
again agreementwith
lack of MS detector bias. This problem had been witnessed at Unilever Researchduring
the analysis of non-ionic alcohol ethoxylate surfactants (Weston, 2000), where
found
be
flawed
data
detector
to
the
as
was
struggled to
quantative
qualitative and
identify the low molecular weight ethoxymers.

Figure 4.11: Mass spectrum of a 100 mg/I standard of Querton KKBCL analysed by loop
injection analysis.
Conditions - Mobile phase: 50:50 ACN: 5 mmol/I NH4Ac (pH 2.0); Flow rate: 190 l/min;
Acquisition time: 2 minutes.

4.3.2 Preliminary results of LC/MS analysis


After overcoming a number of problems during loop injection analysis, the
transfer to a fully hyphenatedLC/MS methodology was relatively simple. Preliminary
from
loop
be
tune
injection
the
that
parameters
optimised
obtained
could
work showed
hyphenated
in
4.12
for
Figure
the
system
with
minimal
modification.
use
adapted
(TIC)
ion
from
LC/MS
the
total
the
obtained
chromatogram
analysis of a
shows
100 mg/1 Querton KKBCL standard. The reconstructed ion chromatograms (RIC's),
M}
ions
the
the
to
of
major quat species are also shown, providing
corresponding
data
interferences.
the
Information
species,
of
quat
without
each
external
on
quantative
TIC,
RIC's
by
that
in
the
LC
the
three
the
confirmed
and
the
principle
peaks
provided
Cio
C12,
C14
the
to
and
alkylbenzyl quats respectively. At the same
trace corresponded
Chanter 4: RP-LC/MS ofalkvlbenzyl
guats

140
Figure
4.12
C18
RIC's
that
the
to
also
made.
was
shows
time, confirmation of aCio
demonstrated
ion
the
only
to
accumulation at
quats
alkylbenzyl
each of
corresponding
ion
i.
276.3
(M+
C10
the
discrete
trace,
the
=
of
component)
e.
m/z
points within
single
be
The
LC
6.5
6
to
between
methodology
was
seen
minutes.
and
was only observed
length,
little
basis
the
chain
with
sign of
of alkyl
resolving the alkylbenzyl quats on
long
latter
into
the
trace,
the
or
early
eluting
part
of
carry over of short chain species
chain species.

Dal

- 13 to

IfT.

Tu

R7: 6.69
1m

RT* 6.21

11T: 10.74

IR

12.69

7n"n
1: ". s

w
0

RT. s, 9

11 -

NL: 1.74 E6

cio
J
. 71
s165

NL: 1.96 E7

12
I

"
..,
.

..

9o. s

NL: 9.04 E6

10.26
1rte-.
nn.
86i!

C18

Im

,
T.

A.,

0 b)

Figure 4.12: Total ion chromatogram (TIC) (top) and reconstructed ion chromatograms (RIC's)
homologues
in
Querton
KKBCL
the
present
sample.
the
alkylbenzyl
major
of

Conditions - Column: 150 x 2.0 mm i.d. Luna 3 m CN; Mobile phase: 50:50 ACN: 5 mmol/l
NH4Ac (pH 2.0).
Analysis of the remaining quat samples by LC/MS allowed confirmation of the
in
Figure
4.13
distributions
elsewhere.
present
shows that the analysis of a
alkyl chain
100 mg/1 benzyldimethylstearyl ammonium chloride monohydrate standard (Appendix
One) revealed the presence of a quat distribution

ranging from C8 to C1g. It was

in
Figure
the
that
chain
components
short
witnessed
therefore predicted

4.1 were

derived from the presence of this sample within the mixture.

Chapter 4: RP-LC/MS of al lbenzyl guats

141

4.13: Chromatogram
Figure
in the
the alkylbenzyl
showing
quat distribution
benzyldimethylstearyl ammonium chloride monohydrate sample. Peak identities of the short
chain quats were confirmed by LC/MS analysis.
Conditions - Column: 150 x 2.0 mm i. d. Luna 3 m CN; Mobile phase: 50: 50 ACN: 5 mmol/1
NH4Ac (pH 2.0).

4.3.3 Ion suppression, and other problems associated with the LC/MS
methodology
Whilst information

generated by LC/MS analysis was used to confirm peak

identities and characterise the minor quat impurities present in some samples, a number
in
flaws
inherent
the methodology. The propensity of quats to sorb to
observed
were
of
led
to problems when the analytes were transferred from the outlet of the DAD
surfaces
to the inlet of the ESI interface. During previous LC/MS work (Section 3.5.6.3) column
into
interface
ESI
flowed
had
the
via a one piece column-coupler. Unfortunately,
eluant
this design was not viable during the RP-LCIMS work, and as a result, a long length (z
50 cm) of 0.12 mm i. d. PEEK tubing was used as a transfer line between the DAD and
the ESI inlet capillary. When the TIC's and RIC's were observed it was apparent that
they showed severe band broadening, peak tailing and asymmetry (Figure 4.12) when
(Figure
LC-UV
4.5).
seen
earlier
to
the
peak
profiles
compared

Problems were also witnessed with cross contamination;

analyte carry-over

being due to the fouling of the interface by sorbed quat species.

Chanter 4: RP-LC/MS o alkvlbenzyl


quats

142
While the aforementioned problems were derived from the inherent physicochemical propertiesof the quat analytes,an apparentinherent lack of method sensitivity
in
directly
be
to
the
the new methodology.
to
parameters
used
related
appeared
Figure 4.12 had initially causedconcern as the abundanceof M' ion of the C12
injection
below
loop
the
abundances
witnessed
with
analysis as well as
quat was well
instrument.
is
Whilst
LC/MS
the
to
analysis
performed
on
same
some
variation
previous
be expected in the ionisation efficiency

and ease of ion expulsion of structurally

disparate analytes, the degree to which the abundancies had decreased, in addition to the
knowledge that the analytes were charged in solution, indicated the influence of an
external factor. When a series of calibration standards were subsequently analysed it
was clear the hyphenated methodology demonstrated an approximate four-fold loss in
LC-DAD
to
the
compared
sensitivity

method, the LOD of the C12 component having

3
Querton.
to
mg/l
approximately
risen

4.4

OPTIMISATION

OF

THE

HYPHENATED

LCIMS

METHODOLOGY
The limitations of using TFA as an ion-pair reagent in the LC/MS analysis of
basic analyteswere describedin Section 1.3.2.1. A number of articles are referencedin
TFA
led
to significant ion-suppressionof the organic bases,the
the
of
presence
which
inefficient
desolvation
being
to
attributed
effect
and ion expulsion occurring, as a result
TFA-base
the
nature
of
recalcitrant
adducts (Apffel et al., 1995). Severe ionof
been
has
in
to
reported
occur
negative ESI-MS mode in the presenceof
suppression
100 mmol/1 acetateions, whilst the presenceof TFA led to the effect being witnessedat
much lower concentrations (Yamaguchi et al., 1999; Kuhlmann et al. 1995). Closer
inspection of the LC mobile phaserevealed that approximately 0.5% TFA was present
in the new methodology.

4.4.1 Reduction

induced
ion-suppression
TFA
of

with post-column

modifiers
Having previously observed that the use of AcH and formic acid resulted in
increased
in
tailing
LC
the
peak
and
system in comparison to TFA, it
reduced efficiency
the
that
of
methodology would be laborious and time
redevelopment
was apparent
Chanter 4: RP-LC/MS of alkvlbenzyl

quats

143
linearity
LC-UV
Having
the
the
sensitivity
and
assessed
of
already
consuming.
method,
in addition to building a profile of robustnessand reproducibility, it was inconceivable
to re-designthe LC/MS methodology from scratch. Instead, a technique was sought that
fixed"
be
interim
':
TFA
to
the
as
an
measure.A literature search
problem
would allow
/
liquid,
infused
into
the
that
the
and
an
organic
or
acidic
use
of
sheath
of
revealed
lead
in
ion
to
the
could
column
outlet
significant
after
reductions
eluant stream
suppression(Kuhlmann et al., 1995; see also Section 1.3.2.1). As a result, this method
in
was subsequentlyevaluated the LC/MS work.
Sheath liquid was infused into the column effluent via a PEEK T-piece in the
infused
into the eluant stream during the tuning of
by
analyte
was
same manner which
the MS detector in loop injection mode. Initial attempts to reduce ion-suppression
HPLC
100
sheath
of
grade propan-2-ol (IPA), and whilst most MS
l/min
a
utilised
kept
ESI
the
constant,
capillary temperaturewas increasedto 285C,
parameterswere
flows
the
gas
and
auxiliary
were regulated to 45 and 10 arbitrary units
sheath
and
respectively to assistsolvent removal.
Figure 4.14 shows the chromatogram that resulted from the analysis of a
100 mg/1 Querton standard in the presenceof a sheath of IPA `fixing solution". The
abundanciesof the major quat species were significantly higher than those witnessed
above, with the calculated peak areas increasing three-fold on average. These
improvements were higher than those reported previously in the literature (Apffel et at,
1995).

It was predicted that the combined action of two factors was leading to the
in
Increases
in
improvements
sensitivity.
conductivity and surface tension of
observed
had
been
TFA
thought to be the root cause of a number of
containing
solvent systems
during
ESI-MS
the
of solvent systems containing this reagent.
problems experienced
High droplet conductivity was thought to lead to interface shorting and an intermittent
high
droplet
At
time,
the
same
surface tension was thought to limit
spray current.
1993).
(Eshraghi
Literature
ion
have
et
al.,
therefore attributed
expulsion
reports
analyte
improvements in analyte ion abundancy in the presenceof IPA, to a reduction in the
liquid
droplets
formed
in
the
the ESI interface. A significant
tension
of
surface
droplet
in
tension
ion
would
surface
greatly
assist
analyte
reduction
expulsion and
Chanter 4: RP-LC/MS o alkylbenzyl quats

144
TFA,
formation
to
the
of
which
would
greater
volatilisation
allow
would also yield
of a
likelihood
the
of shorting, and also reduce the amount of
reducing
current,
stable spray
reagent available in the interface to mop-up the generated charge.

It was predicted that the larger than expected improvements in analyte ion
during
this work, represented a more significant
abundancy witnessed

reduction in

droplet surface tension as a result of excessive use of TFA in the new methodology,
compared to previous literature reports.

RT 091

-II

Ri

u:

i1e

RT,BM

r. _
MUND

RT: t0.36

NL 6 63 E6
.

NL 5.17 E7

l IrT

C
s
t

NL3.62E7

mo-ms

+n
ins

33M

Trr'

0)

Figure 4.14: Total ion chromatogram (TIC) (top) and reconstructed ion chromatograms (RIC's)
in
Querton
the
the
present
sample, obtained after the addition of a fixing solution
species
quat
of
containing IPA.
Conditions - Column: 150 x 2.0 mm i. d. Luna 3 m CN; Mobile phase: 50: 50 ACN: 5 mmol/I
NH4Ac (pH 2.0): Fixing solution: IPA: Fixing solution flow rate: 100 l/min.

One major drawback with the use of the sheath of IPA was that a change was
C12
C14
It
ion
in
M+
the
the
and
of
components.
that
the
ratio
appeared
witnessed
favoured
being
C14
C12:
C14
the
the
to
quat
was
as
ratio fell from 3: 1 to
corresponding
2.2: 1. It was hypothesised that the change in the C12:C14 ratio reflected a variation in the
different
had
demonstrated
the
analytes,
as
previous
that the
of
reports
stability
adduct
formation of TFA-base adducts was governed by the mutual accessibility of the ions,
between
interactions
(Cai
them
1999).
the
et
al.,
non-columbic
and

Chanter 4: RP-LC/MS ofalkylbenzyl

quats

145
To overcome the discrepancies witnessed with the C12:C14 ratio, the composition
3:
1
IPA.
This
fixing
to
the
propionic
adjusted
acid
:
was
solution
solvent system had
of
been reported to yield a 30 fold increase in sensitivity in the abundances of small
peptides in the presence of TFA compared to a system in which no fixing solution was
used (Kulmann et al., 1995). The presence of an excess of propionic acid was thought to
split the TFA-base adducts, leading to the volatilisation

of TFA and the formation of

demonstrated
adducts,
which
a reversible equilibrium
weak propionate-base

allowing

them to be broken down in the ESI interface yielding higher abundancies of analyte ions
increase
in
for
(Section
detector
the
thus
the
accounting
observed
sensitivity
and
within
1.3.2.1).

Figure 4.15 shows that on average an eight-fold increase in the peak areas of the
M` RIC's was observed when a 100 mg/I standard of Querton KKBCL was analysed in
the presence of a 100 1/min sheath of 3: 1 propionic acid : IPA. In addition to yielding
did
lead
the
to observable analyte MS bias as the
new
modifier
sensitivity
not
additional
original C12:C14ratio was maintained.

rr uni -1395

IL
Is..

rrtz
Rr. 5.6

Rr 10.13

Dim=

n; oa

NL9.7E6;

2ms

IL

NL 8.7 E7

a,.

D, DLI.

aL
aim
tl.

NL 6.0 E7

3A
I..

uoas
i.

ome

L
tEti
M-

the'

0)

II

Figure 4.15: Total ion chromatogram (TIC) (top) and reconstructed ion chromatograms (RIC's)
in
Querton
the
sample, obtained after the addition of a fixing solution
present
the
quat species
of
IPA.
3:
1
acid
:
propionic
containing
Conditions - Column: 150 x 2.0 mm i. d. Luna 3 .tm CN; Mobile phase: 50: 50 ACN: 5 mmol/I
NH4Ac (pH 2.0); Fixing solution: 3: 1 propionic acid : IPA; Fixing solution flow rate:
100 I/min.

Chapter 4: RP-LC/MS of alkylbenzyl

guats

146
Agreement has yet to be reachedamong mass spectrometristsas to the reasons
/
IPA
leads
in
to
the amount
the
acid
of
propionic
excess
a
reduction
of
an
presence
why
basic
/
by
ion-suppression,
in
the
and
or
positively
charged
exhibited
analytes,
of
is
in
The
TFA.
technique
conflict with normal recommendationson the
presence of
development of LC/ESI-MS methodologies, in that the concentration of modifying
interface
is
increased
ESI
(Lagerwerf
the
than
rather
reduced
et al.,
entering
reagents
2000). There is normally a general fear that the excessiveuse of reagentswill increase
the likelihood of interface fouling and / or that the excess reagent will mop-up the
high
background
ionisation
leading
interface,
in
to
the
and
reduced
analyte
noise,
charge
ionisation
decrease
background
With
to
efficiency
expected
as
noise
efficiency.
increases,a low S/N ratio is normally predicted.
Such a sequenceof eventswas thought to have accountedfor the low analyte ion
by
in
LC/MS
the
the
when
alkylbenzyl
quats
witnessed
were
analysed
abundances
fixing
Although
ionisation
solution.
analyte
of
a
efficiency was predicted to
absence
have had little influence on the system,the predicted increasein surface tension and the
presenceof stable TFA-analyte adducts (Kulmann et al., 1995) was thought to lead to
limited ion expulsion, and account for the inherent lack of sensitivity in the method. At
the same time, excess TFA was swamping the detector, causing an increase in
backgroundnoise and leading to the lower than expectedS/N ratio.
Support for the theories of Kuhlmann et al. (1995) and Apffel et al. (1995), on
the reduction of ion-suppressionby modifying liquids, appearsto have come out of this
In
in
13.2
far
the
of
a
modifier
a
spray
absence
current
of
A
was
work.
recorded,
be
from
that
would
normally
which
expected
of
a 3.7 kV source voltage. It was
excess
high
derived
from
ions
the
that
current
was
charged
originating from the
predicted
intermittent
that
the
current
explained
and
observed
problems
phase,
with
spray
mobile
in
build-up
droplet
ESI-interface.
liquid
the
profiles and
In the presenceof a sheath of IPA, the spray current fell to 9.2 A. With IPA
droplets,
tension
the
the
fall
in spray current
to
the
surface
of
spray
affect
only
predicted
decreased,
theory
that
tension
the
TFA was driven out of
to
as
surface
support
appeared
into
the gas phase.As a result, fewer TFA and TFA-induced
droplets
the
and volatilised

Chapter 4: RP-LC/MS ofalkvlbenzvl gnats

147
interface
in
leading
to
the
to the observed
generate
excess
charge
were
present
species
decreasein current.
When a modifying liquid of propionic acid - IPA (3: 1 v/v) was used in the
LC/MS methodology the spraycurrent was seento fall still further to 6.8 A. Due to the
respective strong and weak acid nature of TFA and propionic acid, the combination of
the two reagents in an aqueous system would normally yield dissociated TFA and
protonatedpropionic acid. However, Apffel et al. (1995) predicted the reverseto be true
interface.
dissociation
As
ESI
the
acid
was occurring at the sametime as droplet
within
it
decreasing,
was predicted that the higher volatility of TFA would lead to the
size was
driven
dissociation
being
towards
of the propionic acid, as the protonated
equilibria
TFA evaporatedout of the droplet into the gas phase.As droplet evaporation was being
facilitated by surface tension reducing IPA, it was thought that the effect would occur
leading
to the removal of much of the TFA from the liquid droplets, and
rapidly,
TFA-positively
the
of
strong
many
of
cleavage
charged analyte adducts.Observation of
in
ion
increase
abundancyand a two-fold decreasein spray current in this
an eight-fold
work appearsto suggestthat the predictions regarding the acid equilibria were correct.
TFA was being driven out of the source, leaving behind fewer charged species and
fewer strong TFA-analyte adducts.
Whilst ion abundancy was seen to increase in the presence of a modifier
/
IPA,
improvements
the
acid
containing propionic
witnessed during the course of this
'those
did
in
the literature (Kuhlmann et al., 1995).
approach
reported
not
work
However, it was thought that due to the excessiveuse of TFA in the LC methodology,
far more TFA remained in the ESI interface than in previous work. Although the spray
6.8
fell
levels
this
to
the
A,
was
still
well
above
current
witnessed during previous
hypothesised
instrument.
It
that an excessof charged specieswere still
the
was
work on
interface.
in
the
present

4.4.2 Evaluation of the sensitivity and linearity of the optimised LC/MS


methodology
Having observed significant improvements in analyte ion abundancy during
LGMS analysesperformed in the presenceof a 3: 1 propionic acid : IPA fixing solution,

Chanter 4: RP-LC/MS of

lbenzvl guats

148
this modifier was incorporated into the optimised LC/MS methodology and was thus
used in all further RP-LC/MS work.

To determine the sensitivity and linearity of the optimised methodology a series


in
from
0 to 50
Querton
standards
was
prepared
ranging
concentration
calibration
of
ions
formed
between
Having
that
the
observed
no
adduct
previously
were
mg/l.
in
the solvent system, only the M+ ions
other
species
present
alkylbenzyl quats and
for
during
be
to
quantative work. Figure 4.16 shows a plot of the RIC
accounted
needed
instrument
ion
C12
LCQ
M+
to
the
the
the
of
of
alkylbenzyl quat
peak area response
304.5),
304.0
(m/z
to
versus concentration of Querton. The detector
=
component
showed good linearity over the specified calibration

range, which was not fully

in
light
ion
instruments
trap
previous
quantitation
of
problems
witnessed
with
expected
(Weston, 2000).

1.80E+09

y= 3E+07x + 4E+06

1.60E+09
1.40E+09
1.20E+09
m
T 1.00E+09
m

uni 8.00E+08
CL
6.00E+08
4.00E+08
2.00E+08
0.00E+00

4
05

10

15

20

25

30

Concentration

(mg/L)

35

40

45

50

Figure 4.16: Plot of peak area response of the C12 quat species versus concentration of the
Querton KKBCI sample performed on the LCQ instrument.
Conditions - Column: 150 x 2.0 mm W. Luna 3 m CN; Mobile phase: 50: 50 ACN: 5 mmol/l
NH4Ac (pH 2.0); Fixing solution: 3: 1 propionic acid : IPA; Fixing solution flow rate:
100 l/min.

In addition

to good linearity,

the optimised

LC/MS

demonstrated improved sensitivity compared to the LC-DAD

methodology

also

method reported above.

fell
650
Querton,
C12
to
LOD
The
the
g/l
component
approximately 450 ppb of the
of

Chanter 4: RP-LC/MS of al lbenzyquats

149
C12 component (see Figure 4.17 for the RIC's of the C12 and C14 components obtained
from the analysis of aI mg/I Querton standard).

4.4.3 Increasing the sensitivity of the LC/MS methodology - "Full scan"


(SIM)
Ion
Monitoring"
"Single
analysis versus
The optimised LC/MS methodology had afforded an improvement in sensitivity
over

the previously

reported

LC-DAD

method.

However,

application

of

the

biodegradation
in
the
the
to
of
of
quats
environmental compartments
study
methodology
further
improvements
in
Prior
be
to this time all
achieved
without
sensitivity.
not
could
MS analyses had been performed in the `full-scan

MS mode " (Section

1.3.2.1).

However, having observed that the M+ ions were the only abundant species present, a
preliminary evaluation of "single ion monitoring MS mode " (SIM-MS) was performed.

3l:
1686

NL 3.46 E6

C1
2

s,:.

tL
ESSE

C14

NL 8.1 E5

:o-

35,

o.,

-n.. aN

Figure 4.17: Reconstructed ion chromatograms (RIC's) of the C12 and C14 alkylbenzyl quat
Querton
I
KKBCL
in
The
the
mg/I
standard.
analysis was performed in the full
present
species
fixing
the
addition
of
a
solution containing 3: 1 propionic acid : IPA.
after
scan-MS mode
Conditions - Column: 150 x 2.0 mm W. Luna 3 m CN; Mobile phase: 50: 50 ACN: 5 mmol/1
NH4Ac (pH 2.0); Fixing solution: 3: 1 propionic acid : IPA; Fixing solution flow rate:
100 1/min.

Figure 4.18 shows the response of the MS instrument to the C12 alkylbenzyl
in
full
Querton
A)
500
in
standard
g/1
a
scan MS mode and B) SIM-MS
quat present
from
SIM-MS
ion
ion
The
higher
obtained
chromatograms
showed
significantly
mode.
RIC
during
the
full
More
to
MS.
corresponding
obtained
compared
scan
abundancy
detailed evaluation of SIM-MS revealed that the LOD for the C12component was below
300 pg/I, and thus it was predicted that a methodology utilising solid phase extraction
Chapter 4: RP-LC/MS ofalkylbenzyl quats

150
(SPE) and LC/SIM-MS should allow quantitation of alkylbenzyl quats at ng/1 levels
(parts per trillion).

A) Full scan MS

B) SIM MS

Figure 4.18: Figure showing the variation in: A) Reconstructed ion chromatogram of the C12
from
full
MS
KKBCL
0.5mg/I
Querton
the
scan
analysis
of
a
obtained
quat
alkylbenzyl
C12
from
SIM
Ion
B)
the
the
of
alkylbenzyl
obtained
analysis of a
quat
chromatogram
standard.
0.5mg/I Querton KKBCL standard.
Conditions - Column: 150 x 2.0 mm i. d. Luna 3 p.m CN; Mobile phase: 50: 50 ACN: 5 mmol/I
NH4Ac (pH 2.0); Fixing solution: 3: 1 propionic acid : IPA; Fixing solution flow rate:
100 l/min: Scan range for full scan MS: m/z = 200 to 500; Scan range for SIM-MS:
ml.- = 303.5-304.5

In the above example only one component of the Querton sample was studied by
SIM-MS. However. within the LCQ Navigator control software, as with many other
instruments,
MS
benchtop
the
to
option
was
available
create
multiple-segment
modern
SIM-MS
SIM-MS
/
two
at
or
more
whereby
mass
ranges
or
and
acquisition methods.
discrete
be
LC/MS
Although
MS
full
at
points
performed
within
could
an
run.
scan
and
MS
during
the
this
the
of
work,
use
course
of
analysis
multiple-segment
not utilised
by
SIM-MS,
/or
detect,
to
a
number
and
of
the
active components
capability
provides
it
the
Having
to
analysis
suited
of
well
quat
samples.
previously
making
metabolites.
LC
SIM-MS
time
the
stability
of
retention
the
methodology,
excellent
observed
for
individual
been
have
the
improving
the
each
of
created
components,
could
segments
LOD for each active species present in the sample. Unfortunately, due to time restraints
breakdown,
the
instrumental
MS
application
of
multiple-segment
methods
major
a
and
into
SIM-MS
integration
hyphenated
the
of
methodology were not
and the subsequent
Chapter 4: RP-LC/MS of alkylbenzyquats

151
before
However,
SIM-MS
during
be
this
the
work.
of
course
can
adopted for
evaluated
linearity
importantly
the
of
reproducibility,
sensitivity,
evaluation
and
most
routine use
be
must
performed.
robustness

4.5

LIMITATIONS

OF THE LC/MS METHODOLOGY

Whilst the optimised LC/MS methodology, incorporating a sheath of propionic


improvements
LC-UV
SIM-MS
IPA,
the
over
current
of
methods,
and
use
:
offers
acid
further
LOD,
due
the
the
to the
reduce
methodology
still
suffers
could potentially
TFA
in
high
Whilst
the
the addition of
of
concentrations
mobile
of
phase.
presence
is
if
is
LC
to
the
simple
an
additional
was
modifiers
pump
required
method
post-column
be left unattended.Unfortunately, the use of 75% propionic acid leads to a number of
limit
will
which
routine use. The modifier liquid produced a
aesthetic problems,
lab,
through
the
that
perpetrated
smell
making the analyst unpopular with copungent
fixing
In
the
the
propensity
of
solution to attack the sealspresent in a
workers. addition,
led
belief
HPLC
to
the
that the method (and analyst) would quickly
pump
a
syringe or
become unpopular with management,due to the prohibitive costs of purchasing new
frequency
instrument
increased
down-time
the
of
associatedwith repairs and
parts, and
replacements.

Ultimately, the presenceof high concentrationsof TFA in the LC mobile phase


LC/MS
limit
the
the
of
reported
applicability
methodology, and thus redevelopment
will
is
LC
front-end
the
separation
paramount. An important question that needs to be
of
be
is
"Does
2.0?
the
to
" If a higher pH can
to
mobile
phase
need
adjusted
pH
addressed
be utilised without compromising the resolution of the alkylbenzyl quats, less acid will
be needed in the system. As a result, the extent of ion suppression is expected to be
is
be
that
in
This
Chapter Five.
questions
a
of
of
number
will
one
considered
reduced.
Since the completion of this work a report has appeared on the use of SPELGMS and SPE-LC/MS/MS for the analysis of alkylbenzyl quats in wastewatersand
2001).
levels
(Ferrer
literature
The
has
et
al.,
g/l
method
sub
at
a number of
rivers
inherent benefits over the method described above, as it does not require the use of a
high
fixing
to
ions
ion
a
obtain
solution
abundancy
the
of
analyte
post-column
within
Chapter 4: RP-LC/MS of alkylbenzIL11
quats

152
trap. However, the literature method relies on the use of a gradient elution profile, as an
ODS material was chosen as the stationary phase, yet the C16 alkylbenzyl quat still
five
If
to
twenty
elute.
one also considers that a post time of ten
minutes
requires
between
be
the
to
column
required
re-equilibrate
successive
minutes would also
longer
depending
(potentially
on the measuresadopted to avoid void volumes
analyses
in the HPLC system),then the literature method shows an approximate 3.5 fold increase
in analysis time compared to the method reported above. The advantages and
disadvantagesof the individual parameters used in the two methodologies will be
consideredin Chapter Five.

4.6

CONCLUSIONS
This chapter has described the development of a new hyphenated RP-LC/MS

based
for
The
LC
the
analysis
of
alkylbenzyl
routine
quats.
methodology
was
method
on a Unilever Researchstandardoperating procedure that is representativeof methods
Improvements
in
literature.
in
LC
the
the
separationwere achievedby utilising
available
based
high
on
a
phase
purity silica support. In comparison with
a short cyanopropyl
traditional sol-gel silica-based stationary phases the high purity Luna phase yielded
more symmetrical, faster eluting analyte peaksthat showed reducedpeak volumes.
Utilisation of a narrow-bore (2.0 mm i. d.) column improved analyte peak areas
by a factor of five, and in addition to the replacementof non-volatile sodium perchlorate
improved the compatibility of the method with electrospray mass spectrometry.
However, problems were experiencedwith maintaining the resolution of the alkylbenzyl
forcing
the
the use of analyte peak compression.
narrower
column,
on
quat species
Recent problems at Unilever Research in the extraction of cationic tensides from
bore
in
that
the
suggest
use
of
a
narrow
matrices
column
associationwith
environmental
the peak compression effect will not be suitable for the application to environmental
d.
i.
to
4.6
thus
to
the
was
a
recommendation
made
revert
use of a
mm
analysis, and
column when problems are experienced.
The optimised LC-DAD

methodology showed excellent linearity and


demonstratedthe necessarysensitivity to make it applicable to environmental analysis.
formally
the
not
was
In addition, whilst
method
validated, long-term observations on
Chapter 4: RP-LC/MS of al

lbenzyl quats

153
the reproducibility of the separation,its easeof transfer to an alternative laboratory, and
subsequentobservationsmade by Unilever Research,suggestthe method is inherently
robust.
Hyphenation of the LC methodology with massspectrometrywas more difficult.
Problems were experienced with unknown ions swamping the detector, whilst
hyphenation of the LC method with ESI-MS revealed a lack'of method sensitivity, due
to trifluoroacetic acid (TFA) induced ion suppression.The subsequentuse of an acidic
fixing
influence
increased
ion
/
leading
TFA
the
to
solution
reduced
or
organic
of
and
abundancyand sensitivity. Observationsmade during this section of work also appeared
to support previous theories on the improvements in signal to noise ratio that can be
achievedwith the use of post-column modifiers.
Evaluation of the optimised LC/MS methodology revealed that good linearity
in
full
scan mode, and that an approximate 25% reduction in the limit of
was shown
detection could be achieved by using LC/MS rather than LC-UV. In addition, a
preliminary investigation of the use of single ion monitoring MS revealed that it should
be possible to develop a hyphenated LC/SIM-MS method that could be used in
associationwith suitable selective extraction and pre-concentrationtechniquesto study
the fate of alkylbenzyl quats in wastewatersand rivers.
Whilst the LC-UV

method reported here does provide benefits over other

by
industry,
the hyphenated LC/MS methodology ultimately
used
methods currently
suffers from the presence of an excess of TFA within the system. As a result, use and
applicability

of the new methodology

will

be limited.

Of greater use are the

difficulties
the
that can be experienced in the development
observations made regarding
for
basic
/
LC/MS
and
or cationic analytes.
methods
of

4.7

REFERENCES

Abian J., OosterkampA. J. and Gelpi E., J Mass Spectrom., 1999,34,244-254.


Apffel A., Fischer S., Goldberg G., Goodley P.C. and Kuhlmann F.E., J. Chromatogr.
A, 1995,712,177-190.

Chapter 4: RP-LC/MS ofalkvlbenMl

154
Barcel D., Durand G., Vreeken R.J., de Jong G.J., Lingman H. and Brinkman U. A. T.,
J. Chromatogr., 1991,553,311-328.
Cai B. and Li J., Anal. Chim. Acta, 1999,399,249-258.
de Schutter J.A. and de Moerloose P., J Pharm. & Biomed. Anal., 1988,6,879-885.
Eshraghi J. and Chowdhury S.K., Anal. Chem., 1993,65 3528-3533.
Evans C.E., Startin J.R., Goodhall D. M. and Keely B. J., Rapid Comm. Mass
Spectrometry,2000,14,112-117.
Ferrer I. And Furlong E.T., Environ. Sci. Technol., 2001,35,2583-2588.
Hau J., Riediker S., Varga N. and Stadler R.H., J. Chromatogr. A, 2000,878,77-86.
Hewlett Packard, Applications of narrow-bore columns in HPLC, 1997.
Kele M. and Guiochon G., J. Chromatogr. A, 2000,869,181-209.
Kuhlmann F.E., Apfel A., Fischer S.M., Goldberg G. and Goodley P.C., J Am. Soc.
Mass Spectrom.,1995,6 1221-1225.
Lagerwerf F.M., van Dongen W.D., SteenvoordenR.J.J.M., Honing M. and Jonkman
J.H. G., TRAC-Trendsin Analytical Chemistry, 2000,9 418-427.
Mills M. J., Maltas J. and Lough W.J., J Chromatogr. A, 1997,759,1-11.
Moyano E., GamesD.E. and Galceran M. T., Rapid Comm. Mass Spectrometry, 1996,
10,1379-1385.
Myers P., Personal Communications,1999, University of Leeds.
Phenomenex, Explore Luna, Promotional Literature, 2000.
Sparham C., Personal Communications,2000, Unilever Research.
Valladao M. and SandineW.E., J. Dairy S&L,1994,77,1509-1514.
Reeuwijk
A.
M.,
H.
J.
R.
E.
M.,
Tjaden
U.
R.
der
Greef
J.,
J.
Hoeven
der
and
van
van
Chromatogr. A, 1996,741,75-84.
Vervoort R.J.M., Debets R.J.J., Lamers R.J., Claessens H.A., Jansen J.G.M. and
CramersC.A., J. Pharm. & Biomed. Analysis, 1999,21,273-289.
Waters, Chromatography Columnsand Supplies Catalogue, 1999.
Wehr T., LC-GC International Online, April 2000, <http://www. lcgcmag.com>.
Weston D.J., Personal Communications,2000, Unilever Research.
Yamaguchi JA., Ohmichi M., Jingu S., Ogawa N. and Higuchi S., Anal. Chem., 1999,
71 5386-5390.
Zhou X., Handie A., Salad H., Fifer E.K., Breen P.J. and Compardre C.M., J
Chromatogr. B, 1999,728,273-277.

Chanter 4: RP-LC/MS of alkvlbenzvl quats

155

CHAPTER FIVE

The development of a generic

for
tenside
cationic
chromatographic method
analysis

Chapter5: Genericanalysisof cationictensides

156

CHAPTER FIVE

The developmentof a generic chromatographic methodfor


cationic tenside analysis
5.1

INTRODUCTION
With the need to deliver analytical solutions on increasing numbers of parent

is
their
there
tensides
specific
metabolites,
an urgent need to reduce the
and
cationic
time taken for analytical method development. Such improvements in speed can be
developing
for
by
generic
methods
of
analysis
structurally analogous
achieved
for
be
However,
to
analysis
generic
successfully applied to environmental
compounds.
identification
/
data
is
the
of
metabolites,
or
unambiguous
structural
and
matrices,
be
by
hyphenated
/
only
provided
can
which
chromatographic
mass
required,
The
aim of this segment of work was to addressthe question
methods.
spectrometric
"Can generic methods of analysis be developedfor the determination of cationic
tensidesin industrial and environmental matrices?" This is achievedby building on the
findings of previous chapters, with the experimental work setting out to provide a
be
the
that
of
some
of
problems
understanding
will
encounteredon the path
preliminary
to a single generic method of analysis for cationic tensides.

5.2

DETERMINING

A SUITABLE

STARTING

POINT

FOR

GENERIC ANALYSIS
Literature currently available on both separation science in general and the
in
indicates
four
tensides
that
there
particular,
cationic
are
of
prime starting
analysis
for
develop
from
to
the analysis of cationic tensides:
a
generic
method
which
points
1.) The techniqueof capillary zone electrophoresis(CZE).
2.) The disulphine blue active substancesmethodology (DBAS).
3.) Normal phaseliquid chromatography/ mass spectrometry(NP-LC/MS)
4.) Reversephaseliquid chromatography/ mass spectrometry(RP-LC/MS)
In Section 1.4 a critical review was presented of the common methods used to
included
details
inherent
tensides,
the
which
of
advantages and
analyse cationic
disadvantagesof each of the methods. The review illustrated the major limitations that
Chapter 5: Generic analysis ofcationic tensides

157
the CZE and the DSBAS methods harbour, which prohibit their application to generic
analysis (Sections 1.4.5 and 1.4.1). As a result, it was clear that a liquid
chromatographicmethod representedthe most appropriate starting point from which to
develop a genericmethod of analysis for cationic tensides.
In Chapters Three and Four the development of two new hyphenatedLC/MS
for
described
the analysis of specific cationic tensides. Both of these
methods are
inherent
literature
of
advantages
over
current
methods, and
a
number
offered
methods
thus it was concluded that the new methods should be evaluated for widespread
applicability to cationic tensides.

53

EVALUATION

OF THE NEW NP-LC/MS METHODOLOGY

FOR GENERIC

CATIONIC

TENSIDE ANALYSIS

The new NP-LC/MS methodology described in Chapter Three was developed


to improve the resolution of the cationic actives utilised in fabric conditioner
formulations (Section 1.2.1). The LC methodology was shown to be suitable for the
diester,
dialkyl,
of
monoester,and monoalkyl quats, that are, or have
analysis of a range
been the primary active agentspresent in domestic fabric softeners(Section 1.2.1). The
method was also successfully applied to the identification of the triester and trialkyl
in
impurities
a number of the samples(Appendix One). However, limited
present
quat
in
the quantitation of the quaternary amino-alcohols, the
success was witnessed
degradation products of the esterquat tensides (Section 1.2.3.3). For the new NPLC/MS methodology to be adopted for the generic analysis of cationic tensides, its
be
tensides
to
to
cationic
needed
other
assessed,and the peak parametersof
applicability
the quaternaryamino-alcoholsimproved.

5.3.1 Application of the new method to the analysis of the alkylbenzyl


tensides
ammonium
quaternary
Figure 5.1 shows the chromatogram obtained from the analysis of a mixed
Spherisorb
amino column with the ternary mobile phase
on
a
sample
alkylbenzyl quat
3.5.
Baseline
in
Section
described
resolution of the four homologues was
system
C18
the
in
thirty
with
alkylbenzyl quat eluting first and the C12
minutes,
achieved
3.4).
last
(Section
componenteluting
Chapter 5: Generic analysis of
cationic tensides

158

Figure 5.1: Chromatogram showing the resolution of a mixed alkylbenzyl quat sample under
normal phaseconditions.
Conditions - Column: 150 x 4.6 mm i. d. Spherisorb 3 pm amino; Mobile phase: 88:7.2:4.8
Hexane : MeOH : THE modified with 5 mmol/l TFA; Detection: Evaporative light scattering.

It was apparentafter comparing the chromatogramsobtained from the analysis


of

the

octadecyltrimethylammonium

bromide

sample

and

the

benzyldimethylstearylammonium chloride dihydrate sample under normal phase


5.2)
(Figure
that the retention times of the alkylbenzyl quats were much
conditions
increase
in
Indeed,
the
the
than
alkyltrimethyl
retention
corresponding
quats.
greater
time was such that the calculated selectivity (see Section 1.3.1 and Equation 1.11) of
be
C18
to
was
two
species
seen
the
approximately 1.6. This result was
respective
in
data
for
the
the
two
physico-chemical
as
available
species,
somewhat unexpected
/
Log
log
P)
(log
K0
the
octanol
water
partition
coefficients
or
predicted
particular
(Section 1.2.3.3), suggested that the alkylbenzyl quat should have been more
hydrophobic than the corresponding alkyltrimethyl quat (Hansch et al., 1979; Syracuse
ResearchCorporation).
It is usually accepted that analyte retention is brought about by electrostatic
interactions between the analyte and the stationary phase under normal phase LC
After consideration of the mutual incompatibility of the
1.3.1).
(Section
conditions
data
the
chromatographic
and
collected during the analysis of
ternary solvent system,
Chanter 5: Generic analysis of cationic tensides

159
the fabric conditioner actives (Chapter Three), a hypothesis was proposed whereby the
resolution of the cationic tensides was derived from the formation

of a band of

stationary polar solvent within the pores of the silica (Section 3.6.3). Under these
conditions partitioning into the stationary liquid is thought to be the principal driver of
2000).
(Kazoka.
rather than conventional electrostatic interactions. It
analyte retention
hydrophobicity
that
the
therefore
envisaged
was
of the analytes should have influenced
their retention characteristics. as hydrophobicity

increased, retention was predicted to

decrease due to preferential partitioning into the bulk mobile phase. Consideration of
the two chromatograms shown in Figure 5.2 revealed that experimentally derived data
did not agree with these theoretical predictions.

320

270

E 220
m
C

170

120

70

20
05

10

15

20

Time (minutes)

Figure

in the retention
showing the variation
time between the
bromide
benzyldimethylstearylammonium
and
octadecyltrimethylammonium
chloride dihydrate
samples under normal phase conditions.
Conditions - Column: 150 x 4.6 mm i. d. Spherisorb 3 gm amino; Mobile phase: 88: 7.2: 4.8
Hexane : MeOH : THE modified with 5 mmol/I TFA; Detection: Evaporative light scattering.
Figure

5.2:

Although

it proved impossible to develop a definitive

theory for the long

it
the
times
quats,
alkylbenzyl
of
was predicted that one or both of the
retention
following was partly or wholly responsible for the long retention times:

1.) Stability of the TFA-analyte adducts

Chapter S: Generic analysis

of cationic tensides

160
2.) Interactions between the stationary liquid and the phenyl unit present in

the alkylbenzyl quats

1.)

Referencehas already been made to the fact that the association of TFA with

basic and / or cationic analytesleads to the formation of pseudo-neutraladducts, which


demonstrate greater hydrophobicity than the free analytes (Section 1.3.2.1). The
formation of such adductsis governedby the mutual accessibility of the two ions (Cai et
hinder
bulk
in
1999),
to
the
thus
close
charge
will
any
steric
proximity
cationic
and
al.,
the formation and overall stability of the adducts. It was hypothesisedthat the increased
have
indicated
instability
the
quats
could
alkylbenzyl
of the adducts
greater
retention of
formed by these quats compared to those of the corresponding monoalkyltrimethyl
decrease
in TFA concentration resulted in
Having
that
observed
a
already
quats.
increasedretention under normal phase conditions (Section 3.5), it was envisagedthat
forming
demonstrate
increased
adducts
would
unstable
also
analytes
retention under
such conditions.
2.)

The addition of THE as a ternary solvent to a methanol-water mobile phase,has

been shown to assist the resolution of straight chain alkenes and cyclic aromatic
behaviour
1993).
This
indicate
(Meyer,
type
that THE can undergo n-ic
of
may
analytes
interactions with analytescontaining a delocalised electron cloud. During the resolution
of an aliphatic and an aromatic analyte, preferential interaction could be witnessed
between the solvent and the aromatic analytes. Having proposed that polar solvent had
accumulated within the pores of the silica, under the NP-LC conditions described in
Section 3.6.3, the silica would have been enriched in THE in relation to the bulk mobile
in
THE
feasible
liquid
is
the
the
It
that
stationary
would interact with the phenyl
phase.
leading
to an additional solvent-induced retention mode,
the
quats
alkylbenzyl
groups of
been
for
have
the alkyltrimethyl quats. Whilst this theory
evident
which would not
be
found
in
tenuous
the original work of
perhaps
contrived,
support
could
and
appeared
Wilkes et al. (1992), on the analysis of cationic tensides with hexane-basedmobile
Contrary
Wilkes
to
the
observations
reported
above,
systems.
et al. witnessedthe
phase
C16alkylbenzyl quat eluting prior to the C18and C16alkyltrimethyl quats, which was
log
I(
for
the
in
these tensides (Hansch et al.,
of
estimates
with
agreement
more
1979). However, in the literature report, THE and MeOH had been employed in a 3: 1
during
1:
1.5
that
the
this
Under
was
used
than
ratio
theseconditions,
work.
rather
ratio,
Chapter 5: Generic analysis
of cationic tensides

161
the THE concentration in the bulk mobile phase and the stationary liquid would have
been equal. As a result, the alkylbenzyl quats would not have undergone additional
have
been
liquid,
due
in
through
the
the
even
sped
and
may
column
stationary
retention
to the excessTHE in the mobile phase,accounting for the earlier elution and the switch
in elution order.

5.3.2 Analysis of the alkylpyridinium tensides


When a 500 mg/I standardof cetylpyridinium chloride monohydrate(Appendix
One) was analysedunder normal phase conditions a broad, poorly efficient peak was
in
The
31
the
tailing,
to
peak
ca.
minutes.
showed
severe
which
resulted
seen elute after
baseline width being in excessof seven minutes. Whilst increasing the eluting strength
led
in
to
the retention time of the analyte, the
the
a
reduction
phase
mobile
of
in
far
less. When the 1-dodecylpyridinium
symmetry
peak
was
comparative change
chloride hydrate samplewas subsequentlyanalysed,peak symmetry and efficiency were
for
C16
homologue,
be
that
than
the
to
witnessed
worse
even
showing that the
seen
normal phasemethodology was not well suited to the analysisof thesespecies.
Reliable physico-chemical data on the pyridinium quats was sparse.However, a
for
homologue
K.,,
C16
log
(Hansch
it
1979)
the
that
value
et
suggested
al.,
predicted
had similar hydrophilicity to the C16 alkyltrimethyl quat, with the variation being
insufficient to account for the increasein the retention time of the alkylpyridinium quat.
In view of the analytes partitioning into a stationary liquid, it was envisaged that
hydrophobicity / hydrophilicity should have played some part in the retention
i.
have
been expected to have eluted
the
alkylpyridinium
quats
e.
would
mechanism,
it
throughout
Indeed,
than
the
what
was
observed
was
experiment.
much earlier
if
hydrophobicity
factor
the
that
were
only
causing retention then the
predicted
have
would
eluted close to the alkyltrimethyl quats.
alkylpyridinium quats
In light of the predicted log Koy values (Hansch et al., 1979), the excessive
retention

and

alkylpyridinium

significant

peak

tailing

observed

during

the

analysis

of

the

quats, provided further support for the theories regarding stability of

interaction
between
the
THE and quats containing
TFA-analyte
strong
and
adducts
the
5.3.1.1). With the quatemized
(Section
functionalities
being
nitrogen
aromatic
heterocyclic
it
in
ring
was plausible that the approach of a
six-membered
a
constrained
Chanter S: Generic analem
of cationic tensides

162
limited,
hindering
be
formation.
A reduction in
thus
severely
adduct
would
counter-ion
the rate of formation and / or the stability of the adducts was predicted to lead to lower
increase
in
hydrophobicity
and yield an
retention (Section 3.5). At the same
apparent
time, free analyteswould have more opportunity to interact with the silanol and bonded
in
increase
in
the
silica
surface
resulting
an
electrostatic interaction.
phase groups on
The analytes therefore undergo partitioning and electrostatic interaction with the silica
is
analogous to the mechanism ascribed to peak tailing under reverse
support, which
1.3.2.1).
(Section
phaseconditions
The low peak efficiency, and problematic quantitation witnessed during the
analysis of the alkylpyridinium quats and the quaternaryamino-alcohols (Section 3.5.3)
NP-LC
the
the
of
application
methodology to thesetensides.It appearedthat
prohibited
/
instability
hydrophilicity
and
or
of the TFA-analyte adducts increased,the
as analyte
NP-LC
the
of

decreased.
Unfortunately, the focus of
method
applicability
has
in
shifted
recent years to the study of hydrophilic analytesas
environmental analysis
their aqueous solubility increasesthe likelihood of them being found unassociatedin
leading
to high bioavailability (Section 1.2.3.3).
wastewaters,
rivers and environmental
An inability to deal effectively with these important analytes resulted in the
being
abandonedat this point.
methodology

5.4

EVALUATION

OF THE NEW RP-LC/MS METHODOLOGY

FOR GENERIC CATIONIC TENSIDE ANALYSIS


In Section 3.6.3 it was predicted that the unusual peak shapeswitnessed during
the analysis of the quaternary amino-alcohols indicated a non-linear adsorption /
desorption isotherm (Section 1.3.2.1) for these analytes. A retention mechanism was
desorption
the
of one analyte molecule from the surface of the silica
predicted whereby
further
from
desorption
i.
the
molecules
of
silica
e. desorption appearedto be
promoted
for
force
driving
The
this unusual retention mechanism was predicted to
co-operative.
be the low solubility of the diol and triol speciesin the bulk hexane-basedmobile phase,
favour
in
to
the stationary liquid that was envisaged
residence
as the analytesappeared
(Section
3.6.3).
In
the
the
silica
of
order to obtain efficient peak shapes
pores
within
for
these
times
/
analytes,
a
water
and
retention
or polar organic solvent-based
short
and
It
was predicted that reverse phase LC (RP-LC)
was
required.
mobile phase system
Chanter 5: Generic analysis
sis of cationic tensides

163
in
which to obtain efficient resolution of the
presented a more suitable environment
quaternaryamino-alcohols.

5.4.1 Applicability

of the new RP-LC/MS methodology to the analysis of

alkylpyridinium

quat preservatives

Whilst the optimised RP-LC methodology described in Section 4.2.3 had been
well suited to the rapid and efficient

analysis of the alkylbenzyl

its
quats,
wider

had
tensides
to
cationic
other
yet to be evaluated. The second major group
applicability
of cationic tenside preservatives, the alkylpyridinium

quats (Appendix

One) appeared

to represent a suitable starting point from which to test the generic nature of the method,
been
had
these
previously
analysed under similar RP-LC conditions to those
analytes
as
described in Section 4.2.1 (Taylor et al., 1997; Zhou et al., 1999).

Figure 5.3 shows a comparison of the chromatograms achieved during the


analysis of 150 mg/1 standards of 1-dodecylpyridinium chloride hydrate and
benzyldimethyldodecylammonium bromide on the Luna CN phase (Appendix Two).
The peak corresponding to the C12pyridinium quat was seen to elute earlier than the
equivalent alkylbenzyl quat species,which supportedprevious theories on the instability
(Section
TFA-alkylpyridinium
5.3.1.1).
the
quat
adducts
of
Aside from a shorter retention time, the C12alkylpyridinium quat also showed a
290% increase in peak area responseover that of the corresponding C12 alkylbenzyl
in
to
the extinction coefficients of the two species
attributed
a
variation
quat, which was
increased
It
that
the
214
envisaged
was
responsewould facilitate a lower limit of
nm.
at
detection for the alkylpyridinium quats than that reported for the alkylbenzyl quats in
Section 4.2.4.2.

Chapter 5: Generic analysis


of cationic tensides

164

450
400
350
300
E

250

4)
c
Q 200
150
100

50
0 -
02

Figure 5.3: Figure showing the variation in retention of the 1-dodecylpyridinium chloride
hydrate (blue trace) and benzyldimethyldodecyl ammonium bromide (red trace) samples on the
Luna CN material.
Conditions - Column: 150 x 4.6 mm W. 3 m Luna CN; Mobile phase: 50: 50 ACN: 5 mmoll
NH4Ac (pH 2.0).

5.4.1.1 Variation

in separation efficiency

on different cyanopropyl

phases

It was previously reported that the retention times of the alkylbenzyl quats were
different on the Spherisorb CN and Luna CN phases (See Section 4.2.2 and Figure
4.2). Similar observations were also made during the analysis of the alkylpyridinium
quats (Figure

5.4). The Luna CN material again afforded greater peak symmetry,

in
Spherisorb
times
to
the
tailing
shorter
and
resulted
analysis
compared
reduced peak
material. However, when the retention times of the two groups of cationic tensides were
C
became
it
the
that
apparent
16pyridinium quat eluted at a similar retention
compared,
time to the C12 alkylbenzyl quat on the Luna material. Figure 5.5 shows an overlay of
from
the analyses of a mixed alkylpyridinium
resulting
the chromatograms

quat sample

(blue trace) and a mixed alkylbenzyl quat sample (red trace) on A) the Spherisorb CN
first
For
CN
Luna
Spherisorb
the
B)
time
the
the
phase.
material offered
and
phase
benefits over the high purity Luna material, as there was less overlap between the
C12
C16
It
the
this
and
alkylpyridinium
alkylbenzyl
times
phase.
quats
of
on
retention
degree
during
increase
the
two
the
the
that
overlap
of
of
species
would
was predicted
industrial
due
(Section
to
the
or
samples,
environmental
matrix
effects
of
analysis

Chapter 5: Generic analysis of cationic tensides

165
1.3.2.1). As a result, the quantitation of the two species would be problematic without
adjustments being made to the chromatographic parameters.

300

250

200
E

150

0
C2.
N
m

100

50

0
2468

10

12

14

Time (minutes)

Figure 5.4: Figure showing the variation in retention times of a mixed pyridinium quat sample
Spherisorb
(red
Luna
(blue
trace).
trace)
materials
the
and
on
Conditions - Column dimensions : 150 x 4.6 mm i. d. packed with 3 m CN bonded silica
50
ACN:
5
NH4Ac
2.0).
50:
(pH
Mobile
mmol/I
phase:
particles:

5.4.2 Implications

of alkylpyridinium

for
development
the
quas analysis

of a generic methodology
The overlap of the quat species witnessed in Figure 5.5 caused concerns over
the analysis of the quaternary amino alcohols under RP-LC conditions. With their
increased
it
NP-LC
hydrophilicity
and
retention
under
conditions,
was predicted
greater
little
show
retention under RP-LC conditions. With the C 12
that these analytes would
alkylpyridinium

quat having demonstrated a retention factor (k) of approximately 1.0 on

the Luna CN phase, it was envisaged that the diol and triol species (Appendix

One)

CN
Luna
to
the
be
the
column,
on
unless
made
a
reduction
was
unretained
would
/
different
the
the
and
phase
or
separation
mobile
on
a
was
performed
of
eluting strength
stationary phase.

Chapter 5: Generic analysis of cationic tensides

166

200
180
160
140

E
120
100

80
60
40
20
0
0

10

Time

25

20

15
(minutes)

250

200
E

150
100
50

pnn
02

468

10
Time

12

14

(minutes)

Figure 5.5: Figure showing the likelihood of co-elution of the C1, alkylpyridinium quat and the
Cie alkylbenzyl quat species. on A) the Spherisorb CN phase and B) the Luna CN phase.
Conditions - Column dimensions: 150 x 4.6 mm W. packed with 3 m CN bonded silica
50
5
50:
ACN:
Mobile
mmol/l NH4Ac (pH 2.0); Sample: Blue traces - mixed
phase:
particles;
alkylpyridinium quat sample. red traces - mixed alkylbenzyl quat sample.

5.5

OPTIMISING

THE

RP-LC

CONDITIONS

FOR

GENERIC

ANALYSIS
Previous observations regarding the effect of the eluting strength of the mobile
have
decreased
the
that
alkylbenzyl
of
quats
the
elution
revealed
resolution
phase on
increased
lower
in
literature
In
tailing
peak
and
efficiency.
addition,
strength resulted
in
increasing
LC
the
indicated
the
that
aqueous
content
of
mobile
resulted
phase,
reports
ionisation
(Kostiainen
increased
1996),
due
efficiency
to
the
et
al.,
reduced electrospray
in
lower
of
water,
volatility
comparison to common reverse phase
tension
and
surface
1993)
(Section
1.3.2.1).
(Eshraghi
et
al..
organic solvents
Chanter 5: Generic analysis of cationic tensides

167
Whilst the problems associatedwith high water content mobile phases were
equivalent for all quat analytes, additional problems were identified in the analysis of
the dialkyl and diester quats. The low water solubility of these species has already been
/
dissolved
in
to
their
onto
suspended
partitioning
solids in rivers
relation
considered
and environmental wastewaters (Section

1.2.3.3). As a result, their low aqueous

solubility was predicted to lead to one or more of the following problems: excessive
binding
irreversible
of the quat analytes to the silica support, poor peak shape,
retention,
by
have
been
Whilst
/
low
these
some
of
could
overcome
efficiency.
problems
and or
utilising a gradient elution profile, a more retentive stationary phase was sought, that
for
hold
the
the
amino-alcohols
quaternary
on
column
sufficient time as to allow
would
efficient

increasing
By
the retention of the quats, an increased organic
resolution.

solvent content would be required in the mobile phase to facilitate short analysis times.
Improvements

in electrospray efficiency

and a reduction

in TFA-induced

ion-

in
to
predicted
accompany
were
a
suppression
reduction
water content, resulting in
higher sensitivity and the potential to obviate the need for the post-column modifiers
during MS analysis (Section 4.4.1).

5.5.1 Influence of the bonded phase on the retention characteristics of


cationic tensides
In order to maximise the retention of the quaternary amino-alcohols on a silica
based stationary phase support, an octadecylsilane (ODS) (C18) bonded phase was
evaluated. However, when a 500 mg/l standard of benzyldimethyldodecylammonium
bromide was analysed on a 250 x 4.6 mm i. d. 5 m Spherisorb ODS2 column
(Appendix Two), no peak was observedwithin one hundred and fifty minutes from the
in
This
line with previous studies on the analysis of
injection.
observation
was
point of
long
alkyl-bonded stationary phases,in which it was reported that
on
alkylbenzyl quats
for
1989;
this
(van
Liderkerke
challenging
unsuitable
were
analysis
al.,
et
such columns
Santoni et al., 1993; Zhou et al., 1999). Most authors predicted that the elution
interactions
between the cationic
the
of
secondary
electrostatic
result
were
problems
deprotonated
silanol groups on the silica surface (Dowle et al., 1989).
charge and
Indeed, some groups went further, suggesting that the adoption of the cyano phase as
for
the analysis of the alkylbenzyl quats was due to the cyano
the column of choice
functionality interfering with silanol-analyte interactions, resulting in shorter elution
times (Santoni et al., 1993).
Chanter 5: Generic analysis of cationic tensides

168
If electrostaticinteractions were solely responsible for the excessiveretention of
the alkylbenzyl quats, then it should have proved impossible to elute other quaternary
literature
bonded
Having
have
from
ODS
that
seen
phases.
reports
ammonium species
like
biogenic
choline,
that
ammonium
compounds
quaternary
small
shown
in
be
from
ODS
(Figure
5.6)
could
eluted
an
column under
acetylcholine and carnitine
five minutes (Liberato et al., 1986; Tsai, 2000; Zhu et al. 2000), this was evidently not
the case.

CIO
Me

CIS

OH
,

0
CH3OSO3

Me

H,

Me
N

<Me

Me"'

OH
H

OH

DEEDMAC diol

HEQ diol

Mew
me-'

Stepantex triol

CIS

CP
Me
OH

Choline

''OH

MeN Me

OH

0
OH

Carnitine

Figure 5.6: Figure showing a comparison of the structures of the quaternary amino-alcohols
derived from three parent esterquattensidesand the biogenic aminescholine and carnitine.

It was soon realised that the chromatographic parametersbeing utilised in the


RP-LC methodology were limiting the potential for silanol-analyte interactions.
Adjustment of the mobile phasepH to 2.0 was done deliberately to limit the numbersof
charged silanol groups on the silica surface, as characterisation of the silica supports
2.0,
but
had
in
HPLC,
that
the most acidic silanol groups are
at
pH
all
revealed
utilised
Nawrocki,
for
1997
At
1990;
(Section
1.3.2.1).
(Hill,
the
see
or
a
review)
protonated
believed
further
limit
to
the potential
the
modifier
acetate
was
time,
ammonium
same
for secondaryelectrostatic interactions (de Schutter et al., 1988; Nawrocki, 1997), by
for
forming
/
the
the
active
sites
on
analytes
a
silica
substrate,and or
competing with
to
the
close
silica surface, which repelled the approachof
charge
cationic
of
mono-layer
1988;
Vervoort
Schutter
(de
et
al.,
et al., 1992; Nawrocki, 1997). In
the analytes
Chapter 5: Generic analysis of cationic tensides

169
during
ion-suppression
RP-LC/MS
the
the
witnessed
addition,
analysis of the
43.3)
(Section
confirmed the existence of stable TFA-analyte
alkylbenzyl quats
Under
the
reverse phase conditions, association of
system.
analytical
adducts within
increased
leads
TFA
to
tailing
reduced
peak
and
retention, as the
with
cationic analytes
TFA effectively shields the cationic charge and results in the formation of pseudoneutral adducts(Cai et al., 1999).
Consideration of the above information indicated that the likelihood of silanoltenside interactions was low. It was hypothesised that silanol-analyte interaction played
a minor role in the retention of the alkylbenzyl quats on reverse phase supports at
for
2.0.
Evidently
retention
unidentified
mechanism
was
responsible
an
excessive
pH
ODS
bonded
the
on
cationic
preservatives
phases.
of
retention

5.5.1.1 Retention of the alkylbenzyl quats on a standard ODS bonded


stationary phaseand a mixed-mode ODS / CN bonded phase
Having seenthat elution of the C12benzyl quat could not be achieved from an
ODS bonded phase,yet it was achieved on a cyanopropyl bonded phase,a mixed-mode
both
ODS
and cyanopropyl moieties was assessedto evaluate what
containing
phase
both
functional
if
have
the
of
presence
groups
would
on the resolution of the
effect, any,
quat species.Figure 5.7 shows that the retention time of the 1-dodecylpyridinium quat
increased by approximately 150% on the mixed-mode phase in comparison to the
conventional Spherisorb CN phase, whilst the peak width increased by in excess of
400%. It was apparentthat the presenceof the cyano groups was reducing the retention
in
alkylpyridinium
the
comparison to the standard ODS bonded phase.
quat
of
However, it was far from clear as to how or why the presenceof the cyanopropyl groups
having
such a significant effect on retention.
was
The work carried out on the ODS, cyano and mixed-mode ODS/CN phaseshad
bonded
had
two
the
that
of
units
presence
a significant effect on the retention
shown
in
However,
due
tenside
the
to
the
cationic
preservatives.
of
characteristics
variation the
bonding
density
(Appendix
ligands
Two),
it
the
the
and
was unclear as to
of
nature
interfering
the
was
with silanol-analyte interactions, or
cyanopropyl group
whether
in
density,
bonding
length,
the
alkyl
chain
variation
or some other factor was
whether
for
time.
the
shorter
elution
responsible
Chanter 5: Generic analysis
sis of cationic tensides

170

190
170
150
130
E

110
90
70
50
30
10
0

10

20

30

40

50

Time (minutes)

Figure 5.7: Figure showing the retention of the 1-dodecylpyridinium chloride hydrate analyte
on a Spherisorb CN and mixed mode ODS / CN column.
Conditions - Column: Blue trace - 250 x 4.6 mm W. 5 m Spherisorb CN, red trace - 250 x 4.6
mm W. 5 .im Spherisorb ODS/CN. Mobile phase: 50: 50 ACN: 5 mmol/I NH4Ac (pH 2.5).

5.5.1.2 Retention of benzyldimethyldodecylammonium

bromide on a series

Spherisorb
columns
of
There is still a significant

controversy surrounding the actual method of

LC
in
(Section
1.3.2). However, it is now generally accepted
phase
reverse
retention
that retention in RP-LC is brought about by the partitioning of analytes into the alkyl
bonded
(Dorsey
1994).
have
A
literature
the
phase
et
al.,
of
number
chains
of
reports
studied the retention characteristics of alkyl-bonded phases, which varied in carbon
ligand
density,
/
length
through the analysis of homologous series of
or
and
chain
Lork
1987;
1988;
Martin
(Sander
et
al.,
et
al.,
et al., 1988). In general, as alkyl
analytes
ligand
density
increased,
/
length
increase
in
or
a
and
general
retention was
chain
1988;
Dorsey
1994).
(Lork
et
al.,
al..
et
witnessed

If partitioning was the major retention mechanism affecting the cationic tensides
increase
in
supports,
phase
a
general
reverse
retention should have
on alkyl-bonded
been witnessed in moving from a trimethylsilane (C1) bonded phase to an octylsilane
(Cg) bonded phase. In order to determine the influence if any, of the nature of the
Chapter 5: Generic analysis of cationic tensides

171
bonded phase on the retention of the alkylbenzyl quats, a 100 mg/1 standard of
benzyldimethyldodecylammonium bromide was analysed on five Spherisorb based
stationary phases:silica, trimethylsilane (Spherisorb methyl), cyanopropyl (Spherisorb
CN), hexylsilane (Spherisorb C6) and octylsilane (Spherisorb C8). The five phaseswere
bonded onto the same batch of Spherisorb substrate,and thus variations in retentivity,
acidity, metal content, and shape selectivity were kept to a minimum. In addition, the
column dimensions were kept constant, and mobile phase conditions and injection
volumes were uniform. As a result, only the length of the bonded phaseunit attachedto
the silica substrateand the ligand density varied during the evaluation (Appendix Two).
The tenside standard was analysed in triplicate on each column, with the average
retention time being used to calculate the retention factor (k) (Section 1.3.1) (Equation
1.9) of the C12componentand subsequentlythe logarithm of the retention factor (log k).
Figure 5.8 shows the graphs obtained when A) the retention factor (k) and B)
log k of the C12 alkylbenzyl quat were plotted against the number of carbons in the
bonded unit on the silica support. For both graphsthe three data points correspondingto
the C1, C6 and Cg phaseswere plotted first, and a suitable regression fit applied to the
data
The
points correspondingto the silica and cyanopropyl phaseswere added
series.
later. It was immediately apparent that the plot of k versus carbon number was nonlinear and instead showedgood approximation to an exponential, whilst the plot of log k
showedexcellent linearity, with a correlation factor approaching 1.0.
The exponential increase in k and the linear increase of log k with increasing
bonded unit length on the silica support indicated that under reverse phase conditions
the main retention mechanismfor the cationic tensideswas partitioning. When the bestfit line of log k plot was subsequentlyextrapolated, it was found that the predicted
C12
factor
the
alkylbenzyl quat on a 250 mm ODS bonded column was
of
retention
824.5, corresponding to a retention time of approximately 28 hours. This revelation
had
been
it
the
why
analyte
not
seento elute in Section 5.5.1.1.
made very apparent
A linear correlation between log k and bonded phase chain length had been
homologues
during
the
analysis
of
groups
of
on different alkyl-bonded phases
witnessed
(Tchapla et al., 1994). Amongst the earliest of these reports was the work performed by
Henion et al. (1978) who observeda linear increasein log k versus length of the bonded
Chapter 5: Generic analysis of cationic tensides

172
hydrocarbons.
Having
for
aromatic
maintained a consistent
poly-cyclic
of
a series
phase
ligand density on all of the columns, the authors interpreted
confirmation

these results as

that the whole of the bonded phase interacted with the solute. It was

hypothesised that the analytes were intercalating with the alkyl chains of the bonded
be
for
by
that
a partitioning
could only
accounted
phase, a process

model of solute

retention (Section 1.3.2).

25

y=1.0735e0
R2 = 0.9994

20

369x

OCiyl

(C8)

16

18

15

!O

110
13,

Ix

Methyl

(C)
Cyanopropyl (CN)

Silic
0
012345678

Carbon chain length on silica support

3.5

+ 0.0308
R2 = 0.9994

gy=0.1603x

K' = 824.5

25

YpO

TR= 1668 minutes

2
1.5
1

0.5

10

02a68
Carbon

chain

length

on silica

12

14

support

Figure 5.8: Figure showing the variation in the retention factor (k) and log k of the C12
length
bonded
the
the
chain
carbon
of
versus
phase unit on a series of
quat
alkylbenzyl
Spherisorb based phases at pH 2.0. The best-fit curve of log k versus carbon number was
demonstrate
C12
ODS
C18
the
time
to
predicted
the
retention
to
of
an
component
on
extrapolated
bonded phase.
Conditions - Column dimensions: 250 x 4.6 mm i. d. packed with 5 m Spherisorb silica
5 mmol/I NH4Ac (pH 2.0).
70:
30
ACN:
Mobile
phase:
particles;

Chapter 5: Generic analysis of cationic tensides

173
Though Henion et al. (1978) interpreted the linear correlation between log k and
chain length of the bonded phase as confirmation of a partitioning mechanism, most
subsequentauthors have refuted these claims, suggestingthe trend actually confirms the
solvophobic theory of retention (Section 1.3.2) (Tchapla et al., 1994; Barrett et al.,
1996a). However, supporters of the partitioning model of retention have shown that
is
k
linear
whilst a
correlation of
more likely to satisfy the partition theory (Barrett et
linear
log
k
bonded
length
1996a),
plots
of
so-called
versus
phase
chain
actually
al.,
show characteristicdiscontinuations at points in the graph where the analyte can begin
to or can no longer intercalate into the chains of the bonded phase (Tchapla et al.,
1994). Indeed three separatescenariosare often recognisedas the length of the bonded
phasechangesand / or the size of the analyte is varied:
1.) The analyte is much smaller than the alkyl unit bonded to the silica support and
therefore intercalatesinto the chains.
2.) The analyte has approximately the samemolecular size as the bonded phaseand
though it will intercalate, chain orientations are changedsuch that interaction is
between
the adjacentchains.
much weaker
3.) The analyte is too large to fully intercalate into the bonded phaseas part of the
from
bonded
the
structureemerges
phasechains.
The observationsmade by Tchapla et al. (1994) and Barrett et al. (1996a) on the
discontinuations present in log k plots provided additional support for the partition
model of analyte retention, and may also provide a viable explanation for the equivalent
(Katz
selectivity
methylene
et al., 1998) of a new commercial C12bonded
retention and
in
comparison to conventional ODS bonded phases for the
reverse phase column,
(Phenomenex,
2001
analytes
organic
of
small
a). In terms of the retention of the
analysis
C12 alkylbenzyl quat on the allcyl bonded Spherisorb columns, the observations of
Tchapla et al. (1994), brought into doubt the validity of extending the log k plot to a
bonded phase chain length of C18,as it becameapparent that the gradient of the graph
length
bonded
C12.
At C18it was envisagedthat the
the
phase
reached
as
would change
into
fully
intercalate
bonded
be
to
the
able
may
phase and thus the predicted
analyte
have
time
actually representedan under estimation of the actual elution
may
retention
time. This realisation provided additional justification for believing that an ODS bonded
for
the analysis of alkylbenzyl quat analytes.
phase
poor
stationary
phase representsa
Chapter 5: Generic analysis of cationic tensides

174
With the fabric conditioner actives being significantly more hydrophobic than the
it
monoalkyl quats, was predicted that elution of these analytes from an ODS phase
would not be achieved without utilising non-aqueousreverse-phaseconditions, with
dichloromethane,
chloroform,
or a similar elution strength solvent.
of
copious amounts

5.5.1.3 Retention of benzyldimethyldodecylammonium bromide on a bare


silica column
Mention has been made to the fact that many authors who have witnessed the
bonded
for
ODS
the analysis of cationic tenside preservatives
phases
of
unsuitability
have hypothesised that problems result from electrostatic interactions between the
analyte and surface silanol groups on the silica support (Santoni et al., 1993). These
theories had already been brought into doubt earlier in this chapter when the degree of
be
ionisation
low
to
was
predicted
at pH 2.0 (Section 5.5.1.1). The additional
silanol
observation that the retention factor of the C12alkylbenzyl quat increasedexponentially
length
increasing
chain
alkyl
on the silica support, also implicated partitioning as
with
the major retention mechanism of the quats, which added further weight to the theory
that electrostatic interactions played only a minor role in the retention of cationic
tensides.However, to better understandthe degreeto which electrostatic interactions at
pH 2.0 influenced the retention of cationic tensides,elution of the C12alkylbenzyl quat
from the silica phasewas probed.
With the degree of silanol-analyte interaction being directly related to the
deprotonated
silanol groups present on the silica, it should follow that
of
proportion
interaction
is
witnessed when all of the original silanol groups
maximum electrostatic
bare
If
interactions
i.
had
intact
silica.
electrostatic
on
e.
played a major role in the
are
have
increased
then
the
tensides,
the
silica
cationic
column
should
of
shown
retention
for
Co
by
that
column
a
predicted
extrapolation of the log k versus
retention over
bonded chain length plot generatedin Figure 5.8. It was apparent after comparing the
for
Co
a
column to the experimentally derived value for the silica phase
value
predicted
that increasedretention was occurring. However, whilst the silica column demonstrated
in
it
increase
time
16%
over
what
was
predicted,
retention
was envisagedthat much of
a
the additional retention was due to interactions between the analytes and the bulk
factor
1997)
has
(Nawrocki,
been
that
a
well documented in literature
siloxane surface
(Bij
1981;
Cox
et
al.,
occasions
of
et al., 1987; Cox, 1993). In the presence
on a number
Chapter 5: Generic analysis of cationic tensides

175
into
limiting
bonded
thought
to
the
the
are
partition
chains,
analytes
phase
of an alkyl
for
interaction.
As
the
the
influence
a
result,
extrapolated
value
the
of siloxane-analyte
Co column would not account for this mode of retention and would be lower than the
actual experimentalvalue.
Whilst this hypothesis may seem a little contrived, and could be viewed as an
interactions
between
the analytes and
the
to
existence of electrostatic
attempt obviate
deprotonatedsilanol groups, strong support for this theory can be witnessed from the
k
log
in
Whilst
5.8.
Figure
the
best-fit
versus
regression plot shown
equation of the
linearity,
length
excellent
and a steeppositive gradient, which
shows
plot
carbon chain
is
increases
hence
k
that
time
unit
added
that
with
each
methylene
retention
and
shows
to the bonded phase,the intercept value of the line is almost negligible. As the intercept
is
derived
from
it
be
that
the
not
partitioning,
can
seen
retention
provides an account of
in
interactions
that electrostatic
play a small part the retention of the cationic tensidesat
interactions
between
bonded
dispersive
Strong
2.0.
the
phase and the cationic
pH
1.3.2)
drive
(Section
1993)
(Carr
to
appear
retention under reversephase
et al.,
analytes
interactions
between
deprotonated
Electrostatic
low
the
and
analytes
pH.
conditions at
have
influence
little
to
the
support
seem
silica
under such mobile
silanol groups on
phaseconditions.

5.5.2 Influence of mobile phase pH on the retention characteristics of


cationic tensides
One of the main limitations of the RP-LC/MS methodology described in
Chapter Four was the severe ion-suppression and hence lack of sensitivity observed
improve
Whilst
detection.
MS
to
of
addition
post-column
modifiers
was
seen
with
ionisation efficiency and thus reduce the limit of detection (Section 4.4.2), it would
have been preferable if its use could have been avoided. With the ion-suppression
believed to derive from the presenceof TFA (Kuhlmann et al., 1995), a reduction in the
been
by
have
ion-suppression
thus
accompanied
and
reduced
should
acid concentration
increasedsensitivity in LGMS.
Figure 5.8 had provided a strong indication that retention of the C12alkylbenzyl
due
to the partitioning of the analyte
conditions
was
phase
reverse
quat speciesunder
into the bondedphase,with electrostatic interaction appearingto play a very minor role.
Chapter 5: Generic analysis of cationic tensides

176
However, the proportion of deprotonatedsilanols at pH 2.0 is known to be low, with
impurities
(Section
those
the
associated
with
potentially
metal
groups,
most acidic
only
13.2.1) remaining uncharged (Nawrocki, 1997). As a result, electrostatic interactions
between the analytesand the silica support were likely to have been prohibited by the
mobile phasepH.
Being acidic in nature (Section 1.3.2.1), the deprotonation of silanol groups is
(Nawrocki,
increased
mobile
phase
pH
at
promoted

1997), which can result in an

increased electrostatic interaction between the analytes and the silica support. Even so,
having observed that this type of retention had little effect at pH 2.0, the possibility of
dramatically
increasing
TFA
the
without
analyte retention time
concentration
reducing
increased
The
mobile phase pH on the peak parameters of the
effect of
was envisaged.
cationic tenside preservatives was subsequently assessed. Whilst the average pKa of the
is
be
to
substrate
predicted
a
silica
approximately 7.1 (Nawrocki,
on
silanol groups
1997), a cautious approach was adopted during this exercise, and thus peak parameters
first
instance.
in
3.0
the
were assessedat pH

5.5.2.1 Variation of the retention factor and the logarithm of the retention
factor of benzyldimethyldodecylammonium bromide on a series of
Spherisorbcolumns at pH 3.0
Figure 5.9 shows the graphs of retention factor (k) and log k versus bonded
length
for
five
Spherisorb columns utilised in Section 5.5.1.2 at
the
phasecarbon chain
from
The
3.0.
the averagedretention time
curves
were
again
generated
regression
pH
data of the C1, C6 and C8 bonded phases,with the silica and cyanopropyl data points
being added later. It was apparentthat although the mobile phasepH had increased,the
identical
k
forms
2.0;
to
the
those
the
were
of
graphs
plot
of
observed
pH
at
general
log
k
the
to
a
showed
good
approximation
number
an exponential, whilst
versus carbon
linearity.
demonstrated
good
plot again
With k still reflecting an exponential increasewith increasing carbon number in
it
initially
bonded
appearedthat partitioning was still the only major retention
the
phase,
3.0.
However,
for
tensides
the
at
pH
cationic
comparison of the equationsof
mechanism
log
k
for
(Figures
5.8 and 5.9) showed that significant
the
two
best-fit
plots
the
curves
increase
in
had
the
pH.
accompanied
changes

Chapter 5: Generic anal sy


is of cationic tensides

177

50
45
40

35
U

30
25
20
15
10

5
0

"
12345678

Carbon
4

chain

length

on silica

support

3.5

K'

2.5

+ 0.1799
y=0.1848x
R2 = 0.9981

3208.5

TR = 6483 minutes

2
J

1.5
1
0.5

10

02468
Carbon

chain

length

on silica

12

14

16

18

support

5.9: Figure showing the variation in the retention factor (k) and log k of the C12
length
bonded
the
the
chain
carbon
of
phase unit on a series of
versus
quat
alkylbenzyl
Spherisorb based phases at pH 3.0. The best-fit curve of log k versus carbon number was
ODS
demonstrate
C12
C18
the
time
the
to
predicted
to
retention
on
an
of
component
extrapolated
bonded phase.
Conditions - Column dimensions: 250 x 4.6 mm i. d. packed with 5 .tm Spherisorb silica
NH4Ac
3.0).
5
(pH
70:
30
ACN:
Mobile
mmol/I
phase:
particles;
Figure

An increase in the intercept value of the regression plot was immediately


been
having
intercept
With
3.0.
the
recognised as providing an estimate
apparent at pH
induced
increased
(Section
5.5.1.3),
degree
the
retention
non-partition
of
of the
intercept reflected an increased contribution from a second retention mechanism. At pH
2.0 the intercept value of log k was seen to be negligible,

reflecting the minor

However,
had
3.0
intercept
this
the
at
pH.
retention
at
pH
contribution of secondary
increased by approximately 480%, indicating that an additional retention mechanism
influence
the
higher
this
on
cationic
having
analyte
at
pH.
significant
a
was
Chapter 5: Generic analysis of cationic tensides

178

Having increasedthe pH, it was predicted that more silanol groups would have
been ionised, leading to increasedelectrostatic interaction betweenthe silica surface and
the analytes. When the theoretical elution time of the C12 alkylbenzyl quat was
it
bonded
3.0,
increased
for
Co
had
that
at
was
a
phase
pH
observed
calculated
retention
from 4.2 minutes at pH 2.0 to 5.5 minutes. Having assumedthat the degreeof siloxanehave
increasing
interaction
changed
with
significantly
pH, any
would not
analyte
increased
interaction.
in
the
contribution of silanol-analyte
change retention reflected
Thus it was hypothesisedthat the increasein the intercept value of the log k regression
indicative
3.0
of silanol-analyte interaction having a significant role in
was
plot at pH
the retention of the cationic tensidesat the higher pH.
Figures 5.8 and 5.9 also show that a 15% increase in the gradient of the log k
increase
in
to
the
accompany
was
graph
seen
mobile phasepH. As the gradient of the
log k graph was related to the partitioning of the analytes into the stationary phase, the
indicate
3.0
to
that either the analyte was showing
appeared
steeper gradient at pH
for
favouritism
the stationary phase at higher pH, or that as the length of the
greater
alkyl-bonded phase increased, co-operation was witnessed between partitioning and
electrostaticretention.
In order to test the validity of the two above hypotheses,it was necessaryto
for
the C12 alkylbenzyl quat on each of the
the
parameters
obtained
peak
examine
bonded phases,as it was recognisedthat for the increasedresidencetheory to have been
proved correct only the retention time of the analyte should have been seen to vary
between the different phasesat the two pH values. When the peak parameters were
in
found
in
it
that
to
addition
a
change
retention time, peak tailing
compared was
3.0,
increase
increased
in
whilst
a
more
significant
at
pH
peak tailing was
generally
bonded
from
Cg
C1
in
These
to
the
the
phase.
going
witnessed
observations seemedto
instead
increased
lent
theory,
the
to
residence
and
support to the co-operative
paid
put
retention theory.
In Section 5.5.1.2 it was mentioned that a number of literature reports have
four
last
in
the
in
the
the peak parametersof organic
variation
years,
regarding
appeared
amines witnessed on a range of reverse phase materials (Barrett et al., 1996a and b;
Chapter 5: Generic analysis
sis of cationic tensides

179
M`Calley, 1999). In each of these reports it was observed that as the length of the
bonded phase increased,there was also a general increase in the retention time of the
basic
At
time,
the
the
all
analytes studied, showed strong secondary
same
analytes.
form
bonded
in
tailing
the
the
on
of
analyte
peak
phases(Barrett et al., 1996a
retention,
and b). Of more significance to the current discussion was the observation that the peak
tailing was much more pronouncedon the longer alkyl bondedphasesi.e. Cg and Clg, in
i.
to
the
alkyl
short
phases e. C1 and C4 (Barrett et al., 1996a).Barrett et al.
comparison
(1996) concluded that a Cg bonded phase provided superior performance than the
conventional C13bonded phasesfor the selected analytes, due to reduced peak tailing
and increasedpeak efficiency.
Whilst the above reports demonstrated that increased peak tailing had been
witnessed with increasedbonded phase length during the analysis of organic amines,
none of the reports could provide a suitable explanation to account for the experimental
hypothesis
Instead
a
sound
was found in a much earlier report on the
observations.
analysis of a group of neutral, acidic, and basic organic analytes on two C18bonded
phasesand a silica phaseof the same support material (Bidlingmeyer et al., 1982). The
first important experimental observation reported in this paper was that benzocaine,
2.8,
has
of
showed consistent retention acrossthe pH range 3.0 to 9.0. This
a
pKa
which
result provided the final proof that the increasedresidencetheory was not responsible
for the changein the gradient of the log k plot witnessedduring this work. Bidlingmeyer
four
(1982)
that
showed
when
also
anaestheticswere analysed on a C18bonded
et al.
half
bonded
C18
the ligand density of the conventional phase,and a
phase
with
phase,a
bare silica column, the peak tailing of the four basic analyteswas seento decreasein the
i.
Cis
50%
C18
>
>
silica,
e. maximum peak tailing was seenon the conventional
order
C18phase and maximum peak symmetry was witnessed on the bare silica phase. By
for
the
silica
support
all three columns, variation in retentivity, acidity,
same
utilising
distribution
the
of silanol groups was kept to a minimum. Thus, the
and
number
and
in
peak tailing reflected a variation in the retention mechanism on
observed variation
the three columns. These observations led Bidlingmeyer et al. to conclude that the
is
due
held
tailing
that
to the interaction of the analytes
view
peak
solely
commonly
(Daldrup
1984),
groups
et
silanol
al.,
and hence the notion of silanol
with surface
"bad",
The
being
was
misconceived.
authors instead proposed a mechanism
as
groups
is
brought
by
inaccessibility
the
tailing
about
whereby peak
of surface silanol groups,
Chapter 5: Generic analysis of
cationic tensides

180
by
difficulties
in
the
encountered
analytes as their attempts to move away
and
particular
from the surface are hindered by the bonded phase. It was envisaged that the reduced
by
C18
50%
demonstrated
the
tailing
with
phase
coverage showed the analytes
peak
bonded
journey
into
back
less
likely
their
to
with
a
phase
unit
on
come
contact
were
into the bulk from the silica surface. Peak symmetry was therefore maximised on the
lack
hindrance
from
bonded
due
to
the
of
phase units as the analytes passed
silica phase,
back into the bulk mobile phase. As a result, non-uniform peak shapes on the silica
in
the pores of the
the
the
of
of
analytes
representative
retardation
phase were probably
/
in
1.3.1)
(Section
the acidity of the silanol groups themselves
and
or
variation
silica
(Section 1.3.2.1).

Support for the predictions and observationsmade by Bidlingmeyer et al. (1982)


(1987),
later
by
Cox
five
et
al.
who characterisedthe retention on silica and
years
came
conventional reversephasesupportsunder pseudo-reversephaseconditions. Once again
peak symmetry and efficiency was observed to be higher on silica than on the alkyl
bonded phases.
It appearedthat a suitable hypothesis had been found that could account for all
during
based
Spherisorb
the
the
the
made
evaluation
of
observations
columns under
of
during
this work. As the bonded phase length increased,peak
reverse phaseconditions
tailing also increasedas a result of retardation of the tenside analytes by the chains of
the bonded phase.This observation could not simply have resulted from the variation in
the ligand densities of the C1, C6 and Cg phases,as the peak tailing observedon the Cg
phasewas much more severethan that witnessed on the C6 phase,yet the C8 phasewas
ligand
density;
lower
3.12
have
to
a
mol/m2 versus 3.36 mol/m2 (Appendix
reported
Two) (Phenomenex,2001b).

Whilst peak tailing was in general, more severeat pH 3.0, it was also witnessed
longer
bonded
In
further
the
2.0,
to
on
phases.
addition
especially
substantiating
at pH
the predictions made by Bidlingmeyer et al. (1982), these results showed that though
few surface silanol groups were believed to be deprotonatedat pH 2.0, their existence
by
tensides.
be
the
cationic
of
analysis
revealed
could

Chanter 5: Generic anal sY


is of cationic tensides

181

5.5.2.2 Influence of a cyanopropyl bonded unit on the retention of


benzyldimethyldodecylammoniumbromide
Having observed that at low pH, partitioning of the analytes is primarily
long
bonded
for
retention
on
excessive
phases,and that secondaryretention
responsible
is
dependant
length
it
bonded
the
tailing
the
also
on
of
was
or peak
phase unit,
hypothesis
head
the
that
the
regarding
action
of
cyano-propyl
groups should
concluded
be re-evaluated (Section 5.5.1). If the cyano groups were affecting retention to any
log
k
for
degree,
have
been
then
the
this
values
column
significantly
should
significant
lower than those predicted for a C4 phase, especially at pH 3.0, where this type of
interaction was more prevalent. Inspection of Figure 5.9 showed that the data point
corresponding to the cyanopropyl column lay very close to the best-fit line obtained
from the alkyl bonded phasedata. When the predicted retention time for the C4 column
for
it
CN
to
the
the
experimental
value
compared
phase
was
was found that there was
less than 3% variation in the two results. Indeed, when the log k versus carbon chain
include
CN
data
to
the
the correlation factor of the regressionfit was
repeated
plot was
unchanged.
It appearedthat under the reverse phaseconditions developed in Chapter Four
the cyanopropyl phasewas actually behaving as a C4 alkyl bonded phase,with the CN
head group acting as a conventional terminal methyl group. It was apparent that the
had
the
cyanopropyl
of
adoption
phase
come about due to false assumptions
widespread
had
fears
the
unknown, which
of
no strong foundation in experimental observation
and
For
theory.
generic cationic tenside analysis the choice of stationary
or chromatographic
longer
be
restricted to one phase.
phaseshould no

5.5.2.3

Revisiting

the

change

in

benzyldimethyldodecylammonium

peak
bromide

parameters

of

witnessed on

SpherisorbCN and ODS / CN phases


In addition to helping to explain the variation in the gradients of the log k plots
at the differing pH values, the observationsand subsequenthypothesesof Bidlingmeyer
5.5.2.1),
be
(Section
(1982)
could
also
used in association with the results
et al.
in
to
the
the peak parametersof the C12alkylbenzyl
variation
explain
presentedabove
CN
Spherisorb
the
and mixed-mode ODS / CN phases(Figure 5.7
on
quat witnessed
5.5.1.1).
Section
and
Chapter S: Generic analysis of cationic tensides

182

The alkylbenzyl quat analyte demonstrated a significant increase in both


retention time and peak tailing on the mixed-mode phase in comparison with the
increase
in
The
phase.
retention time can be explained in
conventional cyanopropyl
terms of increased residence of the analyte in the stationary phase in the presence of the
ODS bonded units, in the same way that increased retention was witnessed on the C8
Cl
bonded
increased
in
The
the
the
with
phase.
explanation
peak
comparison
of
phase
tailing had been much harder to justify.
interfere
did
with
not
group
preferentially

However, having recognised that the cyano

silanol-analyte

interactions,

did
it
interact
neither

with the cationic analytes, as had been witnessed previously

analysis of organic amines (Barrett et al., 1996a; MCalley,

in the

1999; Needham et al.,

2000), the increase in peak tailing was attributed to the increased retardation of the
in
leaving
the
the presence of ODS bonded units. Of greater
silica
surface
analyte
significance than the actual peak tailing was the degree to which peak tailing had
increased on the mixed-mode phase. The 400% increase had come about in spite of the
ligand density of the column being half that of a conventional cyanopropyl phase
(Appendix

Two). Although

retardation by the ODS bonded moieties would have

increased peak tailing (Section 5.5.2.1), the poor surface coverage of the ligands
in
the
the rates of release of the analytes from the
to
variation
accentuate
appeared
surface, greatly enhancing peak tailing. The conclusion was reached that a mixed-mode
be
this
nature
should
not
chosen as the stationary phase for the analysis of
phase of
cationic tensides.

5.5.2.4 Retention of benzyldimethyldodecylammonium bromide on a


Zorbax Stable Bond cyanopropyl column
Having witnessed that the logarithm of the retention factor of the C12
increased
linearly
bonded
on
short
alkyl
quat
stationary phasesat low pH
alkylbenzyl
(Figures 5.8 and 5.9), it was predicted that a stationary phase which had two or more
bonded
(<
C4)
demonstrate
discrete
to
the
silica
units
support
may
a seriesof
short alkyl
for
In
tenside
the Zorbax Stable Bond CN phase
cationic
component.
a
single
peaks
(Appendix Two), a suitable column was found to evaluate this theory, as the column
has iso-butyl and iso-propyl groups at the base of the silica, which are designed to
from
hydrolysis
the
acid
and thus extend column lifetime, in addition to
surface
protect
the standard cyanopropyl unit (Phenomenex, 2001b). With a Spherisorb CN phase
Chapter 5: Generic analysis of cationic tensides

183
having been seen to behave like a C4 phase, it was envisagedthat the Zorbax SB CN
phasewould act as a mixed-mode phaseunder the conditions describedabove.
Figure 5.10 shows the chromatogram obtained from the analysis of a 100 mg/l
benzyldimethyldodecylammonium
bromide
standard of
on a 150 x 4.6 mm i.d. 5 m
Zorbax SB CN column. The two peaks seen eluting at ca. 4.8 minutes and 6.4 minutes
were both seento correspondto the C12species,as peak arearesponsevaried according
to the concentration of the standard and the injection volume. It was proposed that the
into
the
the
to
the
analyte
partitioning of
early eluting peak probably corresponded
cyanopropyl chains, whilst the later peak was probably brought about by additional
ligands
from
Bond
dissociate
Stable
the
to
the
the
as
analytes attempted
retention on
silica substrate.

Figure 5.10: Chromatogramresulting from the analysis of a benzyldimethyldodecylammonium


bromide standardon a Zorbax Stable Bond cyanopropyl column.
Conditions - Column: 150 x 4.6 mm i.d. 5 m Zorbax SB CN; Mobile phase: 50:50
ACN: 5 mmoUl NH4Ac (pH 2.0).

Recently, the use of a 50 x 4.6 nun 5 m Zorbax SB ODS phasewas also seen
during
the analysisof single non-ionic tensidesat Unilever
to
to give rise multiple peaks
Research(Cooper, 2000). At the same time, long chain tensideswere observedto elute
in
disparate
Whilst
the
two
points
chromatogram.
these results demonstratedthat
at
Chapter 5: Generic analysis of cationic tensides

184
Figure 5.10 was not an isolated event, it was still unclear as to why the two retention
in
been
light
had
more
witnessed
often
not
of the methyl groups being
mechanisms
present on other reversephase supports. On the Zorbax SB CN column at least, there
was little difference in alkyl chain length between the main bonded unit and the steric
groups at the silica support. In the case of the ODS bonded phase, it was thought that
the high flow rate (2.0 ml/min) and short column length had combined to split the
it
is
is
band.
Nonetheless
that
work
clear
considerable
required to evaluate
analyte
SB
CN
Zorbax
freak
the
two
these
observationswere
whether
occurrences,or whether
be
important
fundamental
information
in
to
the
able
on
shed
phase
particular may
method of retention under reversephaseconditions.

5.5.2.5 Influence of the silica substrate and mobile phase pH on the peak
benzyldimethyldodecylammonium
bromide
of
parameters
The RP-LC/MS methodology described in Chapter Four suffered from an
inherent lack of sensitivity resulting from the ion-suppressionwitnessed in the presence
in
increase
TFA.
Therefore
mobile phasepH would reduce the concentration of
an
of
TFA and hence begin to limit ion-suppression. A brief evaluation of the effect of
increasedpH had already been performed on a series of Spherisorb columns (Section
5.5.2.1). However, to gain a greater understandingof the influence of pH on the peak
parameters

of

the

cationic

tensides,

100

mg/I

standard

of

benzyldimethyldodecylammoniumbromide was analysedin triplicate on the 150 x 4.6


3
i.
Luna
CN
d.
Spherisorb
and
m
columns at varying pH.
mm
Figure 5.11 shows the variation in the retention factor (k) of the C12alkylbenzyl
Spherisorb
CN
factor
The
the
pH
on
phase.
changing
was seen to
quat with
retention
increasemarkedly over the pH range 2.0 to 3.75, with a slower rate of growth between
Having
5.0.
3.75
maintained the ratio of acetonitrile to water at 1:1 throughout
and
pH
the duration of the experiment, changes in analyte retention must have been derived
from increased electrostatic interaction between the analytes and the silica support,
deprotonated
in
increase
it
the
As
number
of
silanol
an
groups.
a
result,
reflecting
became apparent that the Spherisorb material contained numerous silanol groups with
low pK8's and hence high acidity (Nawrocki, 1997). These observations went against
by
Vervoort
(1992),
findings
made
et
al.
who reported that peak asymmetry
previous
between
3
6
fairly
constant
pH
and
on a -Bondapak column, and that no
remained
Chapter 5: Generic analysis of cationic tensides

185
between
pH 2.5 and 3.5. The authors of this report hypothesised
variation was apparent
that these observations showed that silanol groups deprotonated at pH 3.5 would remain
deprotonated at lower pH. A major limitation of the work performed by Vervoort et al.
(1992) was that there was a lack of consistency in the mobile phase conditions
employed at the different pH's, and thus the lack of variation in peak asymmetry, and
the predictions regarding the nature of the silanols were probably attributable to mobile
Indeed numerous reports are available in the literature, which

phase conditions.

demonstrate that silanols have been identified with pKa values ranging from 1.0 to 10.0
(Nawrocki, 1997).

35
30
25
t5 20
t9
O

15
10
4
5
p.
1.5

Figure 5.11: Plot showing the effect that mobile phase pH had on the retention factor of the C12
CN
Spherisorb
the
phase.
on
quat
alkylbenzyl
Conditions - Column: 150 x 4.6 mm i. d. 3 p.m Spherisorb CN; Mobile phase: 50: 50
ACN: 5 mmol/l NH4Ac adjusted to pH with TFA.

When the variation in the retention factor was subsequently plotted against
CN
for
Luna
(Figure
5.12),
it
the
phase
pH
phase
was observed that the silanol
mobile
different
this
to those on the Spherisorb material.
material
were
on
very
groups present
Little increase in k was witnessed in going from pH 2.0 to 3.0 compared to what had
been witnessed in Figure 5.11. It was therefore predicted that the Luna phase contained
fewer silanol groups with pKa values in the range 2.0 to 3.0 than its Spherisorb
Luna
did
big
However,
increase
the
in k in going from pH 3.0
phase
show
a
counterpart.
been
had
Spherisorb
the
3.75.
not
observed
with
to
which
material.
Chapter 5. Generic analysis
of cationic tensides

186

25

20

l0

15

10

0
1.5

2.5

3.5

4.5

Adjusted pH of mobile phase

Figure 5.12: Plot showing the effect that mobile phase pH had on the retention factor of the C12
alkylbenzyl quat on the Luna CN phase.
Conditions - Column: 150 x 4.6 mm i. d. 3 m Luna CN; Mobile phase: 50: 50 ACN: 5 mmol/l
NH4Ac adjusted to pH with TFA.

Taken together the observations made on the Spherisorb and Luna CN columns
led to the prediction that the silanol groups on the Luna material were less acidic than
those on Spherisorb. It was also noted that the silanols on the Luna material showed far
less variation in pKa than those on the Spherisorb silica as demonstrated by the increase
in k between pH 3.0 and 3.75. Such observations are in line with the reports from
high
new
purity silica supports, of which Luna is one,
regarding
column manufacturers
have
distribution
these
that
materials
an
even
claim
of silanol groups that show
which
less variation in their acidity (Phenomenex, 2000). The observation that the silanol
Spherisorb
high
Luna
than
those
the
the
more
were
acidic
material
on
purity
on
groups
further
to
the
theory that silanol groups associated with
support
provided
silica also
high
(Section
demonstrate
1.3.2.1),
is
impurities
Luna
the
acidity
as
material
metallic
lower
its
Spherisorb
than
to
concentration
of
a
residual
contain
metals
reported
b).
2001
(Phenomenex,
counterpart

A major limitation of the study on the change in retention factor with mobile
limited
the
phase pH was

range over which

the work

was performed.

Useful

information regarding the numbers, nature, and distribution of the silanol groups on the
have
been
two
the
supports
could
gained from evaluating the change in
surface of
Chapter 5. Generic analysis of cationic tensides

187
including
7.5,
limit
factor
the
to
pH
and
of the working range of
up
retention
between
in
1.5
2.0
the case of the Luna column
pH
and
cyanopropyl columns, and also
(Phenomenex,2000). Nonetheless,the observations described in this section show that
important
be
fundamental
to
tensides
could
used
provide
cationic
short chain monoalkyl
information on the nature of silica substrates.As these analytes possessa permanent
cationic charge across the whole of the pH range they will only undergo secondary
in
Thus,
deprotonated
any change analyte peak parameters
silanol groups.
retention on
in
be
to
the nature of the surfacesilanol groups.
changes
can relatedunequivocally

5.5.3 Influence of buffer additives on the resolution of cationic tenside


preservatives
It is well documentedthat the analysis of organic amines is problematic in liquid
interactions
due
between
to
the analytes and surface silanol
secondary
chromatography
Indeed,
basis
for
1996a).
described
in
(Barrett
Sections
the
the
et
al.,
work
groups
5.5.2.1 and 5.5.2.5 was to understandhow the peak parametersof the cationic tensides
in
order to minimise analysis time and
change with chromatographic conditions,
One
limiting
interactions
method
sensitivity.
of
and
secondary
efficiency
maximise
between organic amines and the silica support, which had not been fully evaluated up to
this point, was the addition of a basic and / or cationic modifier to the mobile phase (de
Schulter et al., 1988).

5.5.3.1 The effect of

different

mobile phase modifiers on the

chromatographicparametersof cationic tenside preservatives


Whilst ammonium acetate had been used throughout the course of the reverse
had
been
limit
based
literature
LC
to
secondary
retention,
usage
on
previous
work
phase
(Section
4.2.1),
ESI-MS
compatibility
rather than experimental
reports and
brief
As
the
evaluation
of
result,
a
a
suitability of a selected group of
observations.
influence
to
their
undertaken
assess
was
on the peak parametersof
cationic modifiers
hence
their ability to limit silanol-analyte interaction.
tensides,
and
the alkylbenzyl quat
Four cationic modifiers were evaluated during the test, ammonium hydroxide
(NH4011), ammonium acetate (NH4Ac), tetramethylammonium hydroxide (TMAOH),
hydroxide
(TBAOH).
Each
tetrabutylammonium
modifier was addedto the aqueous
and
to
5
before
the
concentration
a
phase
mobile
the pH was
of
mmol/l,
of
proportion
Chanter 5: Generic analysis of cationic tensides

188
regulated to 2.0 with TFA. In addition to the cationic modifiers, an assessmentwas
made on the effect of excluding a modifier by regulating the pH of double-distilled
water (Section 2.2.1) to 2.0 with TFA. Each of the mobile phaseswere then utilised for
the analysis of a mixed alkylbenzyl quat sample on the Spherisorb CN phase (Section
4.2.1), to evaluatethe resulting changesin peak shapeand resolution.
Figure 5.13 shows an overlay of the chromatogramsthat were achieved with
each of the five different mobile phases.It was apparentthat as the strength and size of
the quaternaryammonium cation increased,there was an associateddecreasein analyte
retention time and peak tailing. It was envisaged that these results indicated that the
larger and more basic the cation, the more effective it was at blocking and / or
competing with the analyte for the surface silanol groups i. e. the apparenttrend was that
NH3+ < N(Me)4+ < N(Bu)4+ at interfering with silanol-analyte interactions (Nawrocki,
1997).

In regard to which of the bases is best suited to the analysis of cationic tensides,
TBAOH was seen to give rise to the shortest analysis time and the largest peak height
response (Figure 5.13). However, when the analysis of the mixed alkylbenzyl quat
sample was repeated with the Luna CN column, the use of TBAOH

caused the C8

alkylbenzyl quat to be eluted close to the solvent front (Figure 5.14). Whilst it was
envisaged that increased retention time could probably be attained through the use of a
reduced modifier

concentration, the lower volatility

of TBAOH

in comparison to

TMAOH and NH4Ac led to the TBAOH being declared unsuitable for use in the generic
analysis of cationic tensides.

Of the other modifiers that were evaluated TMAOH and NH4Ac appearedto
benefits
for
the analysis of cationic tensides,whilst being sufficiently
the
optimum
offer
ESI-MS.
In
light
be
its
to
compatible
with
of
volatile
widespread use in ESI-MS
(Section 4.2.1), NH4Ac was deemed to be the most suitable modifier of the four
in
for
evaluated, use a generic methodology.

Chanter S." Generic analysis of'cationic tensides

189

No base
-Ammonium

hydroxide

--Ammonium

acetate

Tetramethylammonium
hydroxide
--Tetrabutylammonium hydroxide

02468

10

12

14

16

18

20

22

24

Time (minutes)

Figure 5.13: Figure showing how the nature of the basic and / or cationic modifier present in
the mobile phase affects the retention times of a series of alkylbenzyl quats.
Conditions - Column: 150 x 4.6 mm i. d. 3 m Spherisorb CN; Mobile phase: 50: 50
ACN: 5 mmol/l modifier (pH 2.0).

Whilst

the size and basicity

of the cations could explain

some of the

in
Figure
5.13,
the variation in the chromatograms achieved in the
shown
observations
NH4Ac
be
NH4OH
from
the nature of the cation
and
could
not
explained
presence of
implied
influenced
These
that
the
the
the retention of
nature
of
results
anion
also
alone.
the analytes, the presence of the much larger organic acetate anion leading to a
in
hydroxide
in
Whilst
time
the
the
comparison
with
retention
smaller
anion.
reduction
been
had
the
theory
to
that many of the analyte molecules
support
obtained
evidence
in
TFA
(Section
it
4.4),
adducts
pseudo-neutral
with
was observed that
were associated
silanol-analyte

interactions were still evident at pH 2.0 (Section 5.5.1.3). It was

hypothesised that the counter-ion may have associated with the free analytes in the
system.

Chapter 5: Generic analysis of cationic tensides

190

Figure 5.14: Figure showing the analysis of a mixed alkylbenzyl quat sample on the Luna CN
column in the presenceof TBAOH.
Conditions - Column: 150 x 4.6 mm i. d. Luna 3 m CN; Mobile phase: 50:50 ACN: 5 mmol/1
TBAOH (pH 2.0); Flow rate: 1 ml/min.
Although it proved impossible to determine the influence of the anion on the
peak parameters of the cationic tensides, a subsequent search through the available
literature revealed that this phenomenon had been witnessed before during the analysis
of quaternary ammonium analytes (Bluhm et al., 1999). Two quaternary ammonium
analytes, benzyltrimethylammonium
triethiodide,

bromide

and a tri-quaternary

drug gallamine

had been analysed on a silica column with a series of mobile phases

containing different competitor reagents that showed variation in the nature of the cation
It
in
had
that
the
the
the
the
was
reported
a
anion.
change
of
cation
nature
nature of
and
a significant effect on the retention time and peak shapes of the analytes. However, it
was also observed that when tetramethylanimonium chloride and tetramethylammonium
bromide were evaluated as mobile phase modifiers, the peak parameters of the two
distinguishable
a
analytes showed

variation.

Bluhm

et al. (1999) presented no

for
However,
their
to
observations.
account
when taken together with the
explanations
during
this work, two possible theories would appear to account for the
results observed
during
Either
two
the
studies.
a reduction in the molecular size of the
made
observations
likelihood
it
being
interact
increases
the
to
of
able
with free analytes, due to
anion
leading
formation
inhibition,
to
the
of pseudo-neutral analytes, which
reduced steric

Chanter 5: Generic analysis of cationic tensides

191
therefore show less interaction with the silica substrate,or it is the variation in Log P of
the adducts, which leads to a change in the degree of partitioning of the analytes
(Hanscheta!., 1979).
The above observations made it very apparent that the nature of the mobile
has
influence
a
significant
phase additives
on the peak parameters of the cationic
tensides.As a result, work is neededto examine the influence of high concentrationsof
other cations and anions on the resolution of the tenside analytes.This work could yield
important information on optimising mobile phase parameters,and at the same time
begin to predict what changes are likely to be experienced when industrial,
pharmacologicaland / or environmental samplesare analysedwith this methodology.

5.5.3.2 Irreversible binding of cationic tensides in the absenceof mobile


phasemodifiers
In view of previous theories regarding the analysis of cationic tensides(Santoni
et al., 1993), the observation that good resolution of four alkylbenzyl quats could be
achieved in the absenceof a basic and /or cationic modifier (Figure 5.13) should have
been a startling revelation. However, having observed that silanol-analyte interaction
played only a minor role in the retention of cationic tensidesat pH 2.0 (Section 5.5.1.3),
this result was far less unexpected. Nevertheless, it was recognised that if efficient
resolution of the tensides could be achieved in the absence of a modifier, LC/MS
be
improved
could
significantly.
compatibility
The effectiveness of a mobile phase utilising no modifying reagents was
therefore evaluatedduring the analysis of the alkylbenzyl quat preservativeson each of
the bonded phase columns tested in Section 5.5.1.2. It was observed that although the
from
did
each of the phases,with a high elution strength mobile
elute
quats
alkylbenzyl
phase, analysis time was excessive, peak shapes were poor, and co-elution of the
homologues was often witnessed. However, of greater significance was the low
retention time stability that was witnessedon eachof the columns.
Numerous reports are available in the literature on the employment of cationic
tensides to dynamically modify silica supports, in order to provide novel
be
that
cannot
always
achieved with commercially available phases
chromatography
Chanter 5: Generic analysis of cationic tensides

192
(Hansen et al., 1987; Cafias-Montalvo et al., 1994; Lavine et al., 1997). Each of the
reports relies, to some extent, on the binding of the cationic tensides to the silicasupport via strong electrostatic-interactionswith surface silanol groups (Hansen et al.,
1987). Having read a number of reports on the dynamic modification of silica by
cationic tensides, and having witnessed problems with the fabric conditioner actives
during the normal phase work (Section 3.5.6.4), it was hypothesised that the
inconsistent retention times witnessed in the absenceof a modifier were derived from
the irreversible binding of the analytesto the substrate.In order to test this hypothesis,a
seriesof alkylbenzyl quats were analysedin successionon a Spherisorbsilica column in
the absenceof mobile phasemodifiers. The pH of the mobile phasewas regulated to 3.0
deprotonation
TFA
to
promote
with
of the surface silanol groups, prior to the mobile
flushed
being
through the silica column for three hours to ensure complete
phase
removal of residual ammonium acetate.
Figure 5.15 shows the chromatogramachieved from the first analysisperformed
in
the
on
silica phase the absenceof a modifier. The peak area of the C12alkylbenzyl
quat was approximately 25% of that witnessed in the presenceof NH4Ac, and though
had
increased
by
over 450% (Figure 5.9), efficiency had fallen in excess of
retention
25%. It was apparentfrom the retention time shift that interactions betweenthe analytes
and the silica substratewere much stronger in the absenceof ammonium acetate.
When a Benzalkonium Chloride (Appendix One) standard was subsequently
analysed in triplicate on the same phase, a decreasein the retention time of the C12
from
approximately 33 minutes during the first analysis to 17
observed
species was
(Figure
fourth
during
5.16).
At the same time, an increasein the
the
analysis
minutes
peak area responseof the two cationic tensides was also observed between sequential
injections. It was predicted that the observations made on shifting retention time and
increasing peak area indicated that some of the analyte molecules were irreversibly
binding to the silica support during each analysis, and henceblocking active sites on the
support. As a result, analytes introduced to the system during subsequentinjections
faster
due
interaction
to
the
reduced
eluted
were
with the substrate.At the same time,
fewer analyteswere irreversibly binding to the surface during later injections due to the
reducednumbersof strongly acidic sites.

Chanter5: Genericanal sisof cationictensides

193

14

12
10

>
E
c
0
a
6

2
0
0

15

10

20

30

25

35

40

Time (minutes)

5.15: Chromatogram obtained from the initial analysis of a C12 alkylbenzyl quat
basic
/
Spherisorb
column
with
a
mobile
phase
containing
silica
no
and
or cationic
on
a
standard
modifiers.
Conditions - Column: 250 x 4.6 mm W. 5 m Spherisorb silica; Mobile phase: 70: 30
ACN: H, O (pH 3.0).
Figure

20
18
16
14
E
12
CD
c 10
0
CL
8
In
6
4
2
0

15

10
Time

20

25

30

(minutes)

Figure 5.16: Sequential analysis of a 50 mg/I Benzalkonium chloride standard on a Spherisorb


basic
/
following
the
containing
mobile
phase
no
a
and
or
cationic
with
modifiers,
silica column
initial analysis of the C12alkylbenzyl quat.
Conditions - Column: 250 x 4.6 mm i. d. 5 m Spherisorb silica; Mobile phase: 70: 30
ACN: H, O (pH 3.0): Traces: Black - run one, Red - run two, Blue - run three.

Chapter 5: Generic analysis of cationic tensides

194
It was hypothesisedthat the short retention time witnessed in the presence of
brought
by
5.5.2.1)
(Section
was
about
ammonium cations binding
ammonium acetate
to the most acidic silanol groups on the silica substrate. The remaining ammonium
for
the remaining silanol groups. This
the
then
with
analytes
compete
cations would
theory appearedto be confirmed when the mobile phasewas switched to one containing
5 mmol/l ammonium chloride. After flushing the column for five minutes with the new
just
fell
C12
three
to
the
time
minutes,
the
over
quat
alkylbenzyl
of
elution
mobile phase,
in keeping with that witnessedduring the work describedSection 5.5.2.1.

5.6

THE ROAD TO GENERIC CATIONIC TENSIDE ANALYSIS:


A LESSON FROM LITERATURE

A literature report has appeared very recently in which the analysis of


in
described
environmental wastewatersand rivers (Ferrer et al.,
alkylbenzyl quats was
2001). By utilising solid phaseextraction followed by reversephaseLC/MS the authors
billion
individual
homologues
low
determine
to
alkylbenzyl
per
quat
at
part
were able
levels in complex environmental matrices. To evaluate whether this report could
developed
in
indeed,
to
theories
this
chapter, or
whether
provide any additional support
be
learned
development
for
lessons
to
the
could
ease
method
of a generic
any additional
by
Ferrer et al. (2001),
the
tenside
chromatographic
analysis
parametersutilised
cationic
in
described
Section 4.2.3.
to
those
were compared
In general many of the chromatographic parametersused in the recent literature
in
in
developed
2001)
(Ferrer
the
were
accordance
with
et al.,
methodology
report
Section 4.2.3. The mobile phase consisted of a mixture of acetonitrile and an aqueous
in
10
formate
this
case
mmol/I
ammonium
was used to
modifier;
a
cationic
solution of
formic
interaction
(Section
5.5.3),
being
acid
addedto the mobile
reduce silanol-analyte
The
3.5.
the
to
to
separation was also performed on a narrow-bore
pH
phase regulate
(Section
ESI-MS
in
4.2.3),
3.0
improve
250
this
to
compatibility
mm
x
case
a
column
W. mm long ODS bonded phase was chosen for the separation. At this point the
literature
in
between
By
Section
4.2.3
the
that
method
and
reported
cease.
similarities
instead
TFA,
Ferrer
formic
of
acid
et al. avoided the problems associatedwith
utilising
ion-suppression in the electrospray interface, and thus obtained high MS sensitivity
for
(Section
distinct
4.4.1).
This
the
modifiers
post-column
need
a
represented
without
Chanter 5: Generic analysis ofcationic tensides

195
LC/MS
in
Section 4.4.2, as the method
the
optimised
method
reported
advantageover
was simpler and avoided the problems described in Section 4.5 that were observed
during the use of post-column modifiers.
One of the main drawbacks of the literature methodology (Ferrer et al., 2001)
was the choice of the long ODS stationary phase. A gradient elution profile was
increased
from
100%
in
50%
to
the
over
acetonitrile concentration
which
required
fifteen minutes, with the system being maintained at this composition for a further ten
flow
In
addition,
a
mobile
phase
rate of 600 l/min was applied, equivalent to
minutes.
a flow rate of 1.4 ml/min on a conventional 4.6 mm i. d. column. In spite of these
in
C16
demonstrate
increase
150%
the
to
alkylbenzyl quat was still seen
modifications,
a
in
in
Section 4.5. It was apparentthat the method
to
that
time
comparison
seen
retention
suffered from excessive retention, with the authors being unable to elute the C18
in
alkylbenzyl quat the specified time frame. Having witnessed the excessiveretention
Spherisorb
C12
C8 at pH 3.0 (Figure 5.8), the long
the
the
alkylbenzyl
quat
on
of
retention of the alkylbenzyl quats on the ODS bonded phase was not unexpected.
However, examination of the retention time of the C12 component in the literature
in
there
that
the retentivity of the silica supports
was
a
significant
variation
showed
utilised in the two reports.
Having witnessed that the retention time of the C12 alkylbenzyl quat was in
five
Spherisorb
Cg phase, and having been unable to
the
on
minutes
of
eighty
excess
from
this
a Spherisorb ODS phase (Section 5.5.1), elution of the C12
component
elute
component in under eleven minutes, as reported by Ferrer et al. (2001) was surprising.
Even after accounting for the application of a gradient elution profile, and the higher
flow rate used in the literature report, it was concluded that the Phenomenexcolumn
less
by
Ferrer
electrostatic retention than did the Spherisorb
showed
et
al.
employed
in
/
increase
led
have
10
the
the
to
modifier
concentration
or
and
material,
mmol/l must
to a major reduction in silanol-analyte interaction, and hence the retention factor of the
C12alkylbenzyl quat.
It has been reported how the high purity Luna phase was seen to give rise to
3.0
times
than did the Spherisorb material (Figures 5.11,5.12
at
pH
shorter retention
Although
5.17).
the
nature of the silica used by Ferrer et al. was not specified
and

Chapter S: Generic analysis of cationic tensides

196
it
based
that
the
the
predicted
stationary
was
phase
was
report.
on a high purity
within
less
acidic silanol groups than the Spherisorb material,
contained
which
silica support.
However,
5.5.2.5).
it
(Section
thus
was predicted that the
retention
analste
reduced
and
had
influence
to
the
a
much
phase
more
significant
on
mobile
reagents
changes made
the shift in analvte retention.

140
120

100
80
a,
c
0

60

40

20

Figure 5.17: Figure showing the variation in the retention time of the C12 alkylbenzyl quat on
2.0
CN
Spherisorb
CN
Luna
phases
at
pH
and 3.0.
the
and
Conditions - Column dimensions: 150 x 4.6 mm W. packed with 3 .tm silica particles; Column:
Red trace and blue trace - Luna CN. black trace and green trace - Spherisorb CN; Mobile
Blue
NH4Ac
TFA;
5
trace
50
ACN:
50:
to
adjusted
pH
mobile
phase:
mmol/l
with
pH
of
phase:
3.0;
Flow
2.0.
I
trace
black
trace
green
trace
and
red
rate:
ml/min.
and
-

Having observed that the use of ammonium acetate and ammonium hydroxide
it
four
5.13),
in
(Figure
the
alkylbenzyl quats
resolution of
gave rise to some variation
have
formate
2001)
(Ferrer
that
the
ammonium
would
et al.,
use of
was postulated
justification
in
However,
parameters.
as
sound
to
peak
analyte
a slight change
given rise
for a variation in the behaviour of the formate and acetate anions was not forthcoming,
it was concluded that the increase in modifier concentration was responsible for the shift
in analyte retention time. Whilst
parameters of the alkylbenzyl

a full

assessment of the variation

quats with

changing modifier

in the peak

concentration,

was

in
during
the
this
the
work.
comparison
of
chromatograms
achieved
overlooked
5
begin
(Figure
5.18)
to
mmol/l
of
ammonium
absence
and
acetate
can
presence

Chapter 5: Generic analysis of cationic tensides

197
provide some indication of the likely result of increasing the modifier concentration.
Figure 5.18 shows that a significant reduction in the retention of the C12 alkylbenzyl
quat accompanied the introduction of the cationic modifier to the system. Whilst an
increase in concentration from 5 to 10 mmol/l may not have yielded the same shift in
retention as that witnessed in Figure 5.18, due to the fewer numbers of silanol groups
left unassociated on the silica surface, it was predicted that a significant reduction in the
analvte retention times would still be witnessed. Support for this hypothesis was found
in a short communication by Bluhm et al. (1999), which assessed the influence of
varying modifier concentration on the retention of benzyltrimethylammonium

bromide.

It was reported that as the concentration of the cationic modifier, tetramethylammonium


hydroxide. was increased from 125 mmol/l to 250 mmol/I and subsequently to 500
mmol/l the retention factor of the analyte fell to 75% and then to 60% of its original
value.

100
90
80
70
E
m

0
CL

60
50

40
30
20
10

0
Time (minutes)

Figure 5.18: Figure showing the effect of 5 mmol/l ammonium acetate on the retention time of
the C, _ alkylbenzyl quat.
Conditions - Column: 250 x 4.6 mm i. d. Spherisorb 5 m CN; Mobile phase: Blue trace 70: 30
ACN: 5 mmol/I NH4Ac (pH 3.0). Red trace - 70: 30 ACN: H, O (pH 3.0).

A second critical inference that could be deduced from the successful utilisation
formate
by
Ferrer et al. (2001), was that selected mobile phase
10
ammonium
mmol/l
of
be
utilised at concentrations in excess of 5 mmol/1 in a hyphenated
could
modifiers
LC/MS methodology without affecting ESI-MS source dynamics and sensitivity. The
Chapter 5: Generic analysis
of cationic tensides

198
significance of this result was increasedby the fact that the high mobile phaseflow rate
would have given rise to sub-maximal detector response and would have led to
problems with source dynamics in comparison to the flow rate used in this work
(Section 13.2.1).

Whilst a change to the stationary phase and an increase in modifier


concentration could together account for the shift in retention time betweenthe work of
Ferrer et al. (2001), and that presented in Section 4.2.3, the variation in the choice of
acid used to regulatepH was also predicted to have some influence on analyte retention.
The use of TFA in the methodology reported in this work was found to result in less
in
tailing
peak
comparison to acetic and formic acid (Section 4.4.1). However the
effectiveness of TFA in reducing peak tailing may have also inadvertently led to an
increase in analyte retention time, as the pseudo-neutral adducts formed from the
association of TFA with cationic tensides, were believed to be more hydrophobic than
the free analytes.This effect has previously been seento lead to increasedpartitioning,
and hence retention under reverse phase conditions in the analysis of proteins and
1999).
It was therefore hypothesisedthat the addition of an excess
(Cai
et
al.,
peptides
LC
in
longer
TFA
to
the
mobile
phase
resulted
of
retention in comparison to what
would have beenwitnessedin the presenceof either acetic or formic acid.
The literature report of Ferrer et al. (2001) revealed a number of important
insights into the how generic cationic tenside analysis will progress in the future, not
least of which was the additional support for earlier theories on the choice of a suitable
due
However,
in
to
the nature and concentration of the
variations
phase.
stationary
in
literature
the
that
employed
were
reagents
report, compared to those
mobile phase
it
difficult
develop
in
justifiable
to
this
theories on optimal
was
chapter,
used earlier
it
became
Nonetheless,
very apparent that further studies on
mobile phase parameters.
the choice and concentration of mobile phase modifiers would be needed in order to
interaction
hence
and
maximise peak efficiency. The variation
minimise silanol-analyte
in chromatographic peak parameters and MS response should be assessedas the
concentrationof ammonium acetateand ammonium formate is varied. At the sametime,
be
should
performed with triethylamine as literature reports have
study
an analogous
recently revealed the greater efficacy of tertiary amines at blocking silanol-analyte

Chapter 5: Generic analysis of


cationic tensides

199
interaction than quaternary ammonium salts (Hill, 1990; Nawrocki, 1997; MCalley,
1999; Reta et al., 1999).

5.7

CONCLUSIONS
Although this chapter may appear to consist of the discussion of a series of

unconnected experiments, the theme and main challenge of all the work was "the quest
to develop a generic method of analysis for cationic tensides ". Having recognised that
liquid chromatographic methods offered the greatest potential for realising generic
broader
tenside
the
analysis,
cationic
applicability of the two new LC methods reported
in Chapters Three and Four were assessed. The observation of poor chromatography
during the analysis of the alkylpyridinium

quats, in addition to previous problems with

the analysis of the quaternary amino-alcohol

metabolites showed that the inherent

limitations of the new normal phase methodology would prohibit its use as a generic
methodology.

Application of the new reverse phase LC method to the alkylpyridinium quats


facilitated efficient resolution of two homologues of this quat series. However,
comparison of the retention times of the alkylbenzyl quats and the alkylpyridinium
quats revealed that co-elution was a problem on both of the cyanopropyl phasesthat
interpreted
which
was
were evaluated,
as an indication that the quaternary aminobe
unretainedon thesetypes of support.
would
alcohols
The search for a stationary phase that would provide stronger retention of the
hydrophilic cationic tensidesled to the evaluation of a series of commercially available
based
Spherisorb
the
on
columns
silica. Elution of a short-chain
reverse phase
be
achievedon a conventional octadecylsilanebonded silica,
not
quat
could
alkylbenzyl
yet the correspondingalkylpyridinium quat was seento elute from a mixed-mode ODS /
bonded
phase, albeit retention time and peak tailing were considerably
cyanopropyl
increasedin comparisonto a conventional cyanopropyl bonded phase.
Further investigation into the retention of the cationic tensideson reversephase
low
that
at
revealed
pH, contrary to popular belief, silanol-analyte interactions
supports
had little effect on retention, and instead conventional reverse phase partitioning was
Chanter 5: Generic analysis ofcationic tensides

200
for
responsible retaining the analyteson column. At higher pH, electrostatic interactions
were seen to be much more prevalent in the retention mechanism, giving rise to long
Interestingly,
tailing.
times
and
peak
peak tailing was seen to be much more
elution
long-alkyl
bonded
the
supports which are predominantly chosen for the basis
severeon
be
This
to
thought
separations.
was
as a result of the retardation of the
phase
of reverse
analytesin the bonded phase as they desorbedfrom deprotonatedsilanol groups on the
silica surface. Such observations revealed that the increasedpeak tailing witnessed on
the mixed-mode ODS / CN column was due to variations in the rates of analyte transfer
betweenthe silica surface and the bulk mobile phasedue to the presenceof cyanopropyl
bonded
units on the silica surface.
and octadecyl silane
Having assessedthe influence of a series of quaternary ammonium salts on the
resolution of four cationic tenside preservatives, it was apparent that silanol-analyte
interaction fell as the basicity of the cation increased. An effect on the analyte peak
parameters was also witnessed when the nature of the anion in the quaternary
The
however,
less
change
salt
was
varied.
was
ammonium
much
pronounced than that
interesting
An
alternative
cations.
with
observation of the modifier work was
observed
the recognition that efficient resolution of the analytes occurred at low pH in the
basic
/
a
and
or cationic modifier. Unfortunately, dynamic modification of a
of
absence
silica column soon showed why the analysis of cationic tensides should be avoided
under reversephaseconditions if a suitable basic modifier is unavailable.
Comparison of the results produced in this chapter with those documentedin a
has
high
literature
that
shown
report,
purity silica supports can yield significant
recent
in
in
time
tailing,
and
retention
peak
analyte
comparison with older silica
reductions
high
Luna
far
fewer
have
The
the
that
to
purity
observation
support appeared
materials.
highly acidic groups on its surface than the corresponding Spherisorbmaterial may go a
long way to explaining these observations. Having witnessed a large increase in
Luna
increase
time
the
than
material
over
a
short
on
pH
range,
rather
even
retention
an
in retention over a wider pH range, as was the casewith the Spherisorbmaterial, it was
high
the
that
the
silanol
groups
on
purity phase demonstratedmuch more
predicted
uniform pKa's than those on the Spherisorbmaterial.

Chapter 5 Generic anal sy


is of cationic tensides

201
Ultimately the results presentedin this chapter show that more efficient analysis
of monoalkyl cationic tensides can be achieved on short alkyl bonded or bare silica
columns, operated under reverse or pseudo-reverse phase conditions, than can be
achieved with normal phase LC and reverse phase LC, performed on conventional
octadecylsilane-bondedstationary phases.
The most important realisation of this chapter is that many of the preconceived
theories on the analysis of cationic tensides under reverse phase conditions are not
basedon either sound chromatographic theory or experimental fact. The path to generic
cationic tenside analysis will be much simpler having removed many of the mist clouds
surrounding theseanalytes.

5.8

REFERENCES

Barrett D.A., Brown V. A., Shaw P.N., Davies M. C., Ritchie H. and Ross P., J.
Chromatogr. Sci., 1996a, 34.146-156.
Barrett D.A., Brown V. A., Davies M. C. and Shaw P.N., Anal. Chem., 1996b, 68,21702178.
Bidlingmeyer B.A., Del Rios J.K. and Korpi J., Anal. Chem., 1982,54,442-447.
Bij K. E., Horvath C., Melander W. R. and Nahum A., J. Chromatogr., 1981,203,65-84.
Bluhm L. H. and Li T., J. Chromatogr. Sci., 1999,37,273-275.
Cai B. and Li J., Anal. Chim. Acta, 1999,399,249-258.
Caflas-Montalvo B. and Izquierdo-Hornillos R.C., J. Liq. Chromatogr., 1994,17,
1461-1478.
Carr P.W., Li J., Dallas A. J., Eikens D. I. and Tan L. C., J Chromatogr., 1993,656,
113-133.
Cooper D., Personal Communication, 2000, Unilever Research.
Cox G.B., J. Chromatogr. A, 1993,656,353-367.
Cox G.B. and Stout R.W., J. Chromatogr., 1987,348,315-336.
Daldrup T. and Kardel B., Chromatographia, 1984,18,81-83.
de Schutter J.A. and de Moerloose P., J. Pharm. & Biomed. Anal., 1988,6,879-885.
Dorsey J.G. and Cooper W. T., Anal. Chem., 1994,66,857A-867A.

Dowle C.J., Campbell W.C. and Cooksey B. G., Analyst, 1989,114,883-885.


Eshraghi J. and Chowdhury S.K., Anal. Chem., 1993,65 3528-3533.
Chapter 5: Generic analem of cationic tensides

202
Ferrer I. and Furlong E.T., Environ. Sci. Technol.,2001,35,2583-2588.
Hansch C. and Leo A., Substituent Constantsfor Correlation Analysis in Chemistry
and Biology, In Edition, John Wiley and SonsLtd., New York, USA, 1979.
Hansen S.H., Helboe P. and Thomsen M., J Chromatogr., 1987,409,71-80.
Hill D.W., J. Liq. Chromatogr., 1990,3

3147-3175.

Katz E., Eksteen R., SchoenmakersP. and Miller N., Handbook of HPLC, Marcel
Dekker, New York, 1998.
Kaioka H., J. Chromatogr. A, 2000,874,45-53.
Kostiainen R. and Bruins A. P., Rapid Comm. Mass Spectrometry, 1996,10,13931399.
Kuhlmann F.E., Apfel A., Fischer S.M., Goldberg G. and Goodley P.C., J. Am. Soc.
Mass Spectrom.,1995,6 1221-1225.
Lavine B. K., HendayanaS., Cooper W.T. and He Y., J. Liq. Chromatogr., 1997,20,
351-376.
Liberato

D.J., Yergey A. L. and Weintraub S.T., Biomed & Environ. Mass

Spectrometry,1986,13.171-174.

Lork K.D. andUngerK.K., Chromatographia,1988,26,115-119.


Martin M., Thevenon G. and Tchapla A., J. Chromatogr., 1988,452,157-173.
M`Calley D.V., J. Chromatogr. A, 1999,844,23-3 8.
Meyer V. R., Practical High Performance Liquid Chromatography, 2ndEdition, John
Wiley and SonsLtd., New York, USA, 1993.
Nawrocki J., J. Chromatogr. A, 1997,779,29-71.
Needham S.R. and Brown P.R., J. Pharm. & Biomed. Anal., 2000,23 597-605.
Phenomenex,Explore Luna, Promotional Literature, 2000.
Phenomenex, Synergi: simplifying HPLC column selection and method development,
Promotional Literature, 2001.
Phenomenex, Cataloguefor Separation Science,2001.
Piera E., Dominguez C., Clapes P, Erra P. and Infante M. R., I. Chromatogr. A, 1999,
.
852,499-506.
Reta M. and Carr P.W., J Chromatogr. A, 1999,855,121-127.
Sander L. C. and Wise S.A., Anal. Chem., 1987,59 2309-2313.
Santoni G., Tonsini A., Gratteri P., Mura P., Furlanetto S. and Pinzauti S., Int. J
Pharmaceutics,1993,93,239-243.

Chanter5 Genericanal sis of cationictensides

203
Syracuse

Research

Corporation,

LogKow

KowWin

Program,

<httpJ/esc.syrres.com/interkow/kowdemo.htm>.
Tchapla A. and Heron S., J. Chromatogr. A, 1994,684,175-188.
Tsai T. -H., J. Chromatogr. B, 2000,747,111-122.
van Liedekerke B.M., Nelis H.J., Lambert W.E. and de LeenheerA. P., Anal. Chem.,
1989,61,728-732.
Vervoort R.J.M., Mans F.A. and Hindriks H., J. Chromatogr., 1992,6239207-220.
Wilkes A. J., Walraven G. and Talbot J.-M., J. Am. Oil ChemistsSoc., 1992,69,609613.
Zhou X., Handie A., Salari H., Fifer E.K., Breen P.J. and Compardre C.M., J.
Chromatogr. B, 1999,728,273-277.
Zhu Y., Wong P.S.H., Cregor M., Gitzen J.F., Coury L. A. and Kissinger P.T., Rapid
Comm.Mass Spectrometry,2000,14,1695-1700.

Chanter S" Generic analysis of cationic tensides

204

CHAPTER SIX
Conclusions and future work

Chapter6," Conclusionsand futurework

205

CHAPTER

SIX

Conclusions and future work


The widespreadapplication of cationic tensides in pharmaceutical preparations,
and home and personal care products, requires the capability to accurately quantify
these analytes in a number of different classesof sample. As the nature of the matrix
formulated
from
"pure"
raw
material
or
can range
product, to natural water, sediment
or soil, methods of analysis must be adaptable,unaffected by matrix interferences,and
selectively speciatethe analytesof interest at trace levels.
To date, most methods used to quantify cationic tensides demonstrate
insufficient reliability, sensitivity, and / or selectivity for widespread use. In addition,
they can often only be effectively applied to distinct matrices; no method is capable of
speciating all of the cationic actives present in domestic fabric conditioners and
pharmaceutical preparations, prior to use, and after "down-the-drain release". As a
fill
to
this
the
work
was
some of the voids apparentin the current suite of
aim of
result,
important
to
these
quantify
oleochemicals.Furthermore, it was envisaged
methods used
that the foundations for generic cationic tenside analysiscould also be built.
In Chapter Three a new normal phase liquid chromatographic method was
developed to quantify the cationic actives present in domestic fabric softeners.The new
method was seen to offer superior resolution and retention time stability compared to
first
fabric
to
and
second
used
quantify
generation
methods
current
conditioner actives.
It was observedthat the new method could resolve different classesof cationic tenside,
for
LC
phase
method, partial resolution of the endemic
a normal
whilst unexpectedly
homologues was also attained. In addition, the subsequentanalysis of a tenside sample
fabricated from a number of commercial samples, revealed that the new method also
individual
homologue
from a specified sample
tenside
the
of
an
quantitation
allowed
for
the
mass spectrometry. This representedthe first time such
need
without
origin,
been
liquid
had
achievable
with
a
chromatographicmethodology.
resolution

Chapter 6: Conclusions and future work

206
Although the new method was thought to be suitable for the quality assuranceof
industrial
formulated
products, an observed lack of sensitivity limited
raw materials and
its applicability to the environmental analysis of cationic tensides. Furthermore,
ill
the
that
method
was
suited to the quantitation of hydrophilic cationic
observation
tensides (Chapter Five) and cationic biodegradation products (Chapter Three), led to
the conclusion that the method could not be used for generic analysis.
In Chapter Four, experimentation centred on the development of a new
analytical method capableof quantifying trace levels of cationic tenside preservativesin
environmental matrices. Building

on literature reports and current Unilever

methodology, a reverse phase liquid chromatographic method was optimised for use
with electrospraymass spectrometry. Utilisation of narrow-bore columns was seen to
improve method sensitivity, whilst the employment of volatile organic modifiers
improved the compatibility of the method with mass spectrometry. Although the new
method was subsequentlyvalidated and found to demonstrate excellent linearity and
reproducibility, the necessityto utilise peak compression was thought likely to lead to
during
environmental analysis.
resolvation problems
Hyphenation of the LC method with mass spectrometry was found to be
inefficient as sensitivity was compromised by ion-suppression, resulting from the
presenceof a high concentration of trifluoroacetic acid in the LC effluent. Whilst the
use of a post-column modifier reduced ion-suppression, the optimised methodology
proved unwieldy, contrived, and would result in increased instrument downtime.
Experience highlighted the need for careful consideration of liquid chromatographic
hyphenation
to
with mass spectrometry. Compromise is evidently the
prior
parameters
key to the developmentof an efficient LC/MS methodology, and whilst bench-top MS
instruments may have introduced LC/MS into many new laboratories,new lessonsneed
to be learnedbefore the technique is applied effectively and intuitively.
Following the observation that the new normal phase liquid chromatographic
little
potential for the generic analysis of cationic tensides,efforts
methodology showed
focused on optimising the reverse phase LC/MS method for its application to other
groups of cationic tensides. Whilst success was forthcoming in the analysis of the
alkylpyridinium preservatives,problems were anticipated for the determination of fabric
Chanter 6: Conclusions and future work

207
in
light
especially
of the well-documented problems of silanolactives,
conditioner
interaction.
Preliminary
tenside
results showed that the commonly used reverse
cationic
inefficient
and
octadecylsilane
octylsilane,
provided
phase supports,
resolution of the
cationic tenside preservatives,due to excessiveretention and peak tailing. These results
appearedto support previous theories on reversephaseLC analysis of cationic tensides,
whereby retention was brought about by electroststic interactions between the analytes
and surface silanol groups of the silica support. However, the observation that peak
tailing was minimised on a bare silica phase, where such interactions should have been
maximised, led to the hypothesisthat strong partitioning and not electrostaticinteraction
for
fundamental
the
excessive analyte retention i. e. original theories were
reason
was
later
found
It
that the severity of peak tailing was highest on long
misconceived. was
alkyl-bonded stationary phasesas the analyteswere strongly retardedon their path from
the silica surfaceto the bulk mobile phase.
The use of competitive modifiers was found to be important in the analysis of
bare
tensides,
on
silica, where their presenceprevented dynamic
particularly
cationic
by
the tenside analytes. Similarly, regulation of mobile
the
surface
of
modification
be
important
found
in
to
also
achieving efficient chromatographic
phase pH was
low
Regulation
the
to
of
mobile
phase
pH conditions was found to limit the
separations.
influence of silanol-analyte interaction on analyte retention. However, it was observed
that under conditions normally perceived to protonate all the silanol groups on the silica
interactions
were still evident between the analytes and the
electrostatic
substrate,
interaction
degree
The
this
to
which
occurred was subsequentlyfound to be
surface.
high
purity silica supports, which justified claims that these
reduced on modem
less
highly
acidic adsorption sites, and demonstrategreater uniformity
materials contain
in the nature of the silanol groups.
The work described herein was focussed on the development of new
methodology for the analysis of cationic tensides.However, many of the lessonslearned
during Chapters Four and Five regarding peak tailing and LC/MS analysiswill also be
applicable to the analysis of small proteins and peptides, and to the amphoteric tensides,
which also contain a quaternisednitrogen group within their structure.

Chanter 6: Conclusionsand future work

208
Looking ahead, evaluation of the suitability of the new normal phase LC/MS
and reverse phase LCIMS methods for the analysis of "real-world" samples, will be
critical in determining the future potential of each of these methods. The normal phase
methodology seemswell suited to industrial applications due to the superior resolution
that is offered in comparison to literature methods. Additional work is required on
understandingthe retention mechanism and the influences that adduct stability have on
analyte retention times, as the findings should allow resolution to be maximised and the
method to be made more applicable to the analysis of hydrophilic tensides. Whilst the
sensitivity achievedwith the evaporative light scatteringdetector was low, improvement
could well be forthcoming from the use of conductivity detection, particularly when
used with automatedsuppressersystems.However, the effect of gradient elution should
be thoroughly assessedto ensurethat baselinedrift is not a major problem in the future.
The normal phase LC/MS method appears to demonstrate potential for the
analysis of the parent ester and alkyl quats in environmental matrices. A formal
have
that
the
constituents
matrix
of
effect
evaluation
on the resolution of the individual
tenside homologues is required, and method sensitivity and linearity should
flow
be
The
use
of
evaluated.
splitting, prior to the electrospray interface
subsequently
ion
improvements
in detector response and
also
should
yield
monitoring
and single
method sensitivity, which potentially could lead to the application of the methodology
to environmental fate studies. Utilisation of MS instruments with an orthogonal
interface design will be beneficial during the analysis of "dirty" samplesthat would be
encounteredat this point.
For the new reverse phase LGMS method, the future would appear bleak. It is
front-end
LC
be
improve
its
the
to
that
to
separation
needs
evident
redeveloped
compatibility with electrospray mass spectrometry. It was apparent that the use of
trifluoroacetic acid should be avoided or severely limited in the future to ensure the
by
LCIMS. Nonetheless, the new reverse
tensides
of
cationic
sensitive quantitation
has
in
LC-DAD
the routine analysis of cationic
method
already
shown
worth
phase
tenside preservativesin environmental matrices at Unilever Research.
From the perspective of the generic analysis of cationic tensides, the work
insufficient
is
for
herein
a generic method to be immediately evident. However,
reported
Chanter 6: Conclusions and future work

209
it is clear that for rapid and efficient quantitation of these analytes, chromatographers
should focus on developing methods that utilise short (C4 to C8) alkyl bonded stationary
high
based
The
that
purity
silica
on
supports.
phases
are
use of low concentrations of
organic amines such as triethylamine and ammonium acetate will also facilitate short
analysis times and speed sample throughput, whilst their inherent volatility

will aid

hyphenation with mass spectrometry.

The very different physico-chemical properties of the cationic metabolites and


the dialkyl parent quats will require generic cationic tenside analysis to be performed
with the aid of a gradient elution method. It is envisagedthat a suitable mobile phase
from
an aqueous-organicmixture similar to that used in Chapters
system would start
Four and Five, and change to a fully non-aqueous system, based on a strong solvent
such as dichloromethaneover time. This systemshould facilitate elution of parent quats
in
frame
time
a
and cationic metabolites
suitable for routine analysis.
Utilisation of the new monolithic stationary phase technology should also be
in
the development of a generic methodology. The
the
convenience
earliest
evaluatedat
novel pore structure of this support material allows the utilisation of high linear
high
backpressure. The ability to vary
the
problems
normal
of
velocities without
separation velocity representsa powerful tool that could be used in conjunction with
fast gradient elution analysis and elevated temperature,to speed analytical throughput
and lift someof the pressureon entrenchedanalytical facilities.
Although generic cationic tenside analysis may still be some way off, many of
the previous misconceptionsthat have led to irrational fears of irreversible binding and
been
dispelled,
have
is
based
the
tailing
now
and
road
on the
peak
ahead
severe
foundation of chromatographictheory rather than preconceivedideas.

Chanter b." Conclusions and future work

210

APPENDICES

Apaendices

211

APPENDIX ONE
Sample characteristics
Arauad HT
General description: Hydrogenatedtallow derived dialkyl quat. Used as the principal
active agent in a number of fabric conditioners formulations prior to the mid 1990's,
when it was replacedin Europe by the esterquat surfactants.Arquad HT and analogous
products are still utilised in some fabric conditioners being sold outside of Europe.
Principal active: Dialkyl dimethyl ammonium chloride
Me0/R
Me

N
R

0
CI

Chain length distribution

Impurities present:

(%): R= C205 2, C18= 60-80, C16= 25-35, C14= 1-5

15% monoalkyl quat; x 10% trialkyl quat

Source: Unilever Research

Arguad T
General description: Tallow derived dialkyl quat. Has been used as an active agent in
formulations.
fabric
conditioner
some
Principal active: Dialkyl dimethyl ammonium chloride
Me/R
Me

N
\R

O
Cl

Chain length distribution (%): Unknown but principally C16and C18saturated and
unsaturated
Impurities present:

17.5% monoalkyl quat;

15% trialkyl quat

Source: Unilever Research

Hamburg Ester Ouat (HEO)


General description: Hydrogenatedtallow derived diester quat. Used as the principal
fabric
in
some
modem
conditioners formulations.
active agent

Appendices

212
Principal active: Diester quat

R -1

Chain length distribution (%): R= C20=1.4, C18=63.2, C17-'1,C16=31.1, C14=2


Impurities present: ft 5% monoesterquat; x 3% fatty amine; x 10% free fatty acid
Source: Unilever Research

Diethylesterdimethylammonium

chloride (DEEDMAC)

General description: Hydrogenatedtallow derived diester quat. Used as the principal


fabric
formulations.
in
conditioners
modem
some
active agent
Principal active: Diester quat
(D
C1
Me

0N,

R -1Y

Me

N`O

R -1

Chain length distribution (%): R= C18= 64.2, C17= 2.1, C16= 28.8, C14= 3.26
Impurities present: P,3% monoester quat; x 1.6% free fatty acid

Source: Unilever Research

Dimethyldioctadecylammonium

bromide

General description: Hydrogenated dialkyl quat with two C18H37alkyl chains.


Representativeof one of the main actives presentin Arquad 2ht-75.
Principal active:
Mel+/C18H37
/N\

Me

Br

C18H37

Chain length distribution (%): C18/ C18-- 97


Impurities

present:

C18/ C16m 3%

Source: Sigma-Aldrich

Appendices

213
Dihexadecyldimethylammonium

bromide

General description: Hydrogenated dialkyl quat with two C16H35alkyl chains.


Representativeof a major active component presentin Arquad HT.
Principal

active:

Me\0/C16H33
O

Me

/N\

Br
C16 H 33

Chain length distribution

(%): C16I C16> 98

Impurities present: CIS/C16<2%


Source: Sigma-Aldrich

Octadecyltrimethylammonium

bromide

General description: Hydrogenated monoalkyl quat with a C1gH37alkyl chain.


Representativeof a minor impurity presentin Arquad HT.
Principal active:
Mel D C18H37
/N\
Me

BOO
Me

Chain length distribution (%): C18> 90


Impurities present: Mainly C16but C14and C12are also present
Source: Sigma-Aldrich

Cetyltrimethylammonium

bromide

(Cetrimonium

bromide

or CTAB)

General description: Hydrogenatedmonoalkyl quat with a C16H37


alkyl chain.
Principal active:

Me\/Cl6H35
/N\
Me

Boo
Me

Chain length distribution (%): C16> 95


Impurities

present: Mainly Clg

Source: Sigma-Aldrich

Appendices

214
DEEDMAC

C18/ CIg diester guat

General description: Hydrogenatedtallow diester quat with two C18H350tail units. A


principal active componentof the commercial DEEDMAC sample.
Principal active:

10
O,,

C17H35

\NiMe

",,

Me

C17H35

Chain length distribution (%): C18/ C15> 95


Impurities present: Remainderis C18/ C16
Source: Unilever Research

DEEDMAC

C1

C16diester guat

General description: Hydrogenatedtallow derived diester quat with one C18H350tail


A
C16H310
principal component of the commercial DEEDMAC
unit.
unit and one
sample.
Principal

active:

Ie
O

0-

C15H31
Y

O
Me
, _,,
C17H35

Mew

O
Chain length distribution (%): C18/ C10'95%
Impurities present: Remainderis C18/ C18or C16/ C16
Source: Unilever Research

DEEDMAC

C16 C16diester Quat

General description: Hydrogenated tallow derived diester quat with two C16H310
DEEDMAC
the
A
of
commercial
active
component
principal
sample.
units.
Principal active:
I0

O,,NiMe

C15H31Y

Me
0

OCH

15 31

Appendices

215
Chain length distribution (%): C16/ C16> 95%
Impurities present: Remainderis C18/ C16
Source: Unilever Research

CL monoester guat

DEEDMAC

General description: Hydrogenatedtallow derived monoesterquat with a C1gH350tail.


Active impurity present in the commercial DEEDMAC sample, and a biodegradation
diester
the
parent
species.
of
product
Principal active:
I0
HO,,
'.,

O
Me

"'O

MeN

C17H35

Chain length distribution (%): C18> 98%


Impurities present: Remainder is C16

Source: Unilever Research

C16monoester guat

DEEDMAC

General description: Hydrogenatedtallow derived monoesterquat with a C16H310tail


unit. Active impurity presentin the commercial DEEDMAC sample.
Principal active:
I0
HO"_,

Me

N'--'0
Mew

Ct5H3t

Chain length distribution (%): C10'98%


Impurities present: Remainderis Cis
Source: Unilever Research

Trimethylammonium

propane-1.2-diol

chloride (HEQ diol)

General description: Quaternary amino alcohol formed during the biodegradation of


the HEQ derived monoesterquat.

Appendices

216
Principal active:
OH

CI0
O
Me3N
OH

Impurities present: Unknown


Source: Unilever Research

iodide (DEEDMAC

Diethanoldimethylammonium

diol

General description: Quaternary amino alcohol formed from the breakdown of the
DEEDMAC derived monoesterquat.
Principal active:
I
HMe

Mew

OH

Impurities present: No information


Source: Unilever Research

Ste=
General description: Tallow derived diester quat. Used as the principal active agent in
formulations.
fabric
conditioners
somemodem
Principal active:

R -1

O
CH3OSO3
Me
O"O

Ir

,,,

O
0R

-1

HO
Chain length distribution
C14--2

(%): C18

19, C18:i 39, C18:2


--

5, C16 st;26, C16:1

4,

Impurities present: Triester, monoester, triol and free fatty acids are known to be
levels
but
the
are unknown
present
Source: Unilever Research

Appendices

217
Benzalkonium chloride
General description: Commercial alkyl benzyl quat derived from coconut oil. Samples
in
kind
this
are
used
as
preservatives
personal care products and pharmaceutical
of
preparations.
Principal active:
CI0
(D Me
N,,,,
Me
Chain length distribution (%): C12 67, C14-- 32
Impurities present: Tracesof C16and Clo are also present.
Source: Sigma-Aldrich

Querton KKBCL
General description: Commercial alkyl benzyl quat derived from coconut oil. Often
in
used as a preservative personalcare products and pharmaceuticalpreparations.
Principal active:

C'0
Me
,._,
R
Me
Chain length distribution

(%): C16m 1, C14se24, C12 71, Clo; zr3

Impurities present: Tracesof C18are also present.


Source: Unilever Research

Benzvldimethylstearylammonium
General description:

chloride dihydrate

Alkyl benzyl quat with a C18H37alkyl chain.

Principal active:

CIS
-,, Me
C18H37
Me

ppendices

218
Chain length distribution (%): Cig

90

Impurities present: Remainder is C167C14and C12


Source: Sigma-Aldrich

Benzvlcetvldimethvlammonium

chloride monohvdrate

General description: Alkyl benzyl quat with a C16H33


alkyl chain.
Principal active:

CI

N Me
N"I
\C16H35

Me
Chain length distribution (%): C16> 95
Impurities present: Remainderis Clg
Source: Sigma-Aldrich

Benzvldimethvltetradecvlammonium
General description:

chloride dihvdrate

Alkyl benzyl quat with a C14H29alkyl chain.

Principal active:

CI

O
Me

'*'

\C14H29

IIN Me

Chain length distribution

(%): C14> 99

Impurities present: Remainderis C12


Source: Sigma-Aldrich

Benzyldimethyldodecylammonium

bromide

General description: Alkyl benzyl quat with a CI2H25alkyl chain. Analogous to the
in
benzalkonium
the
commercial
chloride sample
principal species

Appendices

Appendices

219
Principal active:

0
Br
Me
_,
N `C12H25
I

Me
Chain length distribution (%): C12 97
Impurities present: Remainderis C14
Source: Sigma-Aldrich

Cetylpyridinium

chloride monohydrate

General description: Alkyl pyridinium quat with a C16H29alkyl chain. Pyridinium


in
and
as
preservatives
antibacterial
utilised
agents
some personal care
quats are
products.
Principal active:
CIO
(D/C16H33
N

Chain length distribution (%): C16-298


Impurities

present: Unknown

Source: Sigma-Aldrich

1-Dodecylpyridinium

hydrate
chloride

General description: Alkyl pyridinium quat with a C12H29alkyl chain.


Principal active:
CIE)
N/C12H25

Chain length distribution

(%): C16= 98%

Impurities present: Unknown


Source: Sigma-Aldrich

220

REFERENCES
EcoLIMS,

SEAC Environment Ecotoxicology

LIMS, 1998, Unilever Research Port

Sunlight.

Lawrence J.G., Personal communication, 1999, University of Leeds.


Sigma-Aldrich,

Online Product

Reference Catalogue and MSDS Database,

<http://www. sigma-aldrich.com>.

Appendices

221

APPENDIX TWO
Column characteristics
Partisil PAC
General:

Mixed-mode alkylamino/cyanopropyl bonded irregular shaped silica

(Maidstone,
UK).
by
Whatman
produced
Designedas a normal phasematerial, and acceptablefor use as a L18 column following
USP designation.Most current methods for analysing cationic fabric conditioner actives
(Wee
1982).
this
phase
et
al.,
stationary
use
Bonded phase units:
Me

e
O

Me
S',...

\Me
Sj,,,,

NH2

CN

O'

e
,`Me

NH2
n

Column dimensions: 250 x 4.6 mm W.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 85 / 350
Carbon loading (%) / ligand coverage (mol/m2): 4.0 / 2.04
Source: Phase Separations

Bio-sil polyol
General: Poly-hydoxyl normal phasematerial that is claimed to offer higher selectivity
for lipids than other NP materials.
This stationary phase has been used previously in the analysis of cationic fabric
1992).
(Wilkes
al.,
et
conditioner actives
Bonded phase unit: Unknown but thought to resemblea standarddiol phasei. e.
OH
Me
\Me

OH
Column dimensions: 250 x 4.6 mm i. d.

Appendices

222
Particle size (pm) / pore size (A) / surface area (m2/g): 5/ 90 / unknown
Carbon loading (%): unknown
Source: Bio-Rad

Supercritically

bonded mixed mode columns


Type
A
-

General: Mixed-mode aminopropyl/cyanopropyl bonded Spherisorbsilica.


Manufactured at the University of Leeds by M. M. Robson to study the influence of the
bonded phaseunits presenton the Partisil PAC material on the resolution of the cationic
fabric conditioner actives (Robson, 1998). The ratio of the amino/cyano functionalities
was varied during the manufacturing processgiving rise to columns M1, M2 and M3.
Bonded phase units:
Me
l,,,..
\Me
5
O'

Me
NH2

r\

%\Me
Si.,.,,
0 ,

M1:

1.5

M2:

M3:

1.5

CN

Column dimensions: 150 x 4.6 mm i. d.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%): M1 = 2.8; M2 = 2.5; M3 = 2.1
Source: University of Leeds

Sunercritically

bonded mixed mode columns Tvae B


-

General: Mixed-mode aminopropyl / cyanopropyl bonded Spherisorbsilica.


The ratio of the amino/cyano functionalities was varied during the manufacturing
M4,
M5
M6.
to
rise
columns
and
giving
process
Bonded phase units:
Me
``Me
SIH
..,,

m
NH2

Me
g

Appendices

223
Bottom: B

Top: A
M4:

M5:

11

M6:

1.5

1.5

Column dimensions: 150 x 4.6 mm i. d.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
M4 - 5.9
M5 - 5.7

Carbon loading (%):

M6 - 5.3
Source: University of Leeds

bonded 2allic acid phase

Supercritically

General: Gallic acid-derived phase developed to test the resolution offered by a polyhydroxyl / poly-phenol phase
Bonded phase units:
Me
gl.
OO

0
OH

"Me

I
OH
OH

Column dimensions: 150 x 4.6 mm i. d.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%): 4.85
Source: University of Leeds

Spherisorb aminopropyl
General: Monofunctional aminopropyl bonded Spherisorb silica that meets the criteria
for a USP L8 column. This column was employed as a normal phasesupport during this
work.

Appendices

224
Bonded phase unit:
Me

Column dimensions: 250 x 4.6 mm i. d. and 150 x 4.6 mm i. d.


Particle size (m) / pore size (A) / surface area (m2/g): 5,3 / 80 / 200
Carbon loading (%) / ligand coverage (pmol/m2): 1.9 / 2.64
Source: PhaseSeparations

Suhereclone aminopropyl
General: Direct alternative to Spherisorb aminopropyl material. Manufacturer claims
the phaseoffers equivalent resolution to the Spherisorbmaterial.
Bonded phase unit:
Me
Me

f\I,,,.,,
O

NH2

Column dimensions: 150 x 4.6 mm i. d., 150 x 2.0 mm i. d. and 150 x 1.0 mm i. d.
Particle size (m) / pore size (A) / surface area (m2/g): 3/ 80 / 200
Carbon loading (%): 2
Source: Phenomenex

Spherisorb

silane (ODS) / aminopropyl

mixed mode octadecyl

General: Mixed-mode ODS/aminopropyl bonded Spherisorbsilica.


Designed primarily as a reversephasematerial the material was actually evaluated as a
during
this work.
support
normal phase
Bonded phase unit:
Me

Me
r

S1_%F**(CH2)17Me
-1

SI

NH2

Column dimensions: 250 x 4.6 mm W.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%): unknown

Source: PhaseSeparations

Appendices

225

Spherisorb cyanopropyl
General: Monofunctional cyanopropyl bonded Spherisorb silica that meets the criteria
for a USP L10 column. This type of column was employed in both normal and reverse
during
the courseof this work.
modes
phase

Bondedphaseunit:
Me
-,, Me
,
r NI,
Si-,

CN

Column dimensions: 250 x 4.6 mm i. d. and 150 x 4.6 mm W.


Particle size (m) / pore size (A) / surface area (m2/g): 5,3 / 80 / 200
Carbon loading (%) / ligand coverage (mol/m2): 3.1 / 3.29
Source: PhaseSeparations

Luna cyanopropyl
General description: Cyanopropyl chemically bonded onto high purity Luna silica.
Column meetsthe criteria for a USP L10 column.
This type of column was employed in the reversephasemode during this work.

Bonded phase unit:

Me
r

SI

CN

Column dimensions used: 150 x 4.6 mm i. d. and 150 x 2.0 mm i. d.


Particle size (m) / pore size (A) / surface area (m2/g): 3/ 100/ 400
Carbon loading (%) / ligand coverage (pmol/m2): 7/3.8
Source: Phenomenex

Spherisorb ODS 2
General description: Octadecylsilane(ODS) chemically bonded to porous Spherisorb
is
An
group
present to reduce the number of free surface silanol
end-capping
silica.
This
type of reverse phase material adheresto the USP
the
support.
silica
on
groups
L1
is
for
the most common LC stationaryphasecurrently
stationary
phase,
a
and
criteria
in use.

Appendices

226
Bonded phase unit:

Me
Me
r

S""*(CH2)17Me

Column dimensions used: 250 x 4.6 mm Ld.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%) / ligand coverage (mol/m2): 11.5/2.98
Source: PhaseSeparations

Spherisorb mixed mode ODS / cyanopropyl


General: Mixed-mode ODS/cyanopropyl bonded Spherisorbsilica with an end-capping
free
the
that
number
of
surface silanol groups on the silica support.
reduces
group
Reverse phase material that offers increased polarity compared to a standard ODS
bondedphase.
Bonded phase unit:
Me

O"S

,.
\Me
,X,
*(CH2)17Me

Me
I
\Me
,,
"'S

CN

Column dimensions used: 250 x 4.6 min i. d.


Particle size (gm) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%) / ligand coverage (mol/m2): 5/1.15
Source: Phase Separations

Spherisorb silica
General: Porous Spherisorb silica. This material meets the requirementsof the USP L3
stationaryphasenotation.
Silica is primarily used as a normal phase material. However, it was evaluated as a
during
this work.
reversephasesubstrate
Bonded phase unit: None
Column dimensions: 250 x 4.6 nun i. d. and 150 x 4.6 mm i. d.
Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Source: Phase Separations

Appendices

227

Spherisorb methyl
General description: Trimethylsilane chemically bonded to porous Spherisorbsilica.
This reversephasematerial adheresto the USP criteria for a L13 stationaryphase.
Bonded phase unit:

Me
%o`Me
,
S)
Me
O.
Column dimensions used: 250 x 4.6 mm i. d.
Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%) / ligand coverage (mol/m2): 2.15 / 2.97

Source: PhaseSeparations

Spherisorb hexyl
General description: Hexylsilane chemically bonded to porous Spherisorb silica. An
free
bonded
is
the
to
the
onto
silica
support
number
reduce
of
end-capping group also
for
USP
L15
This
to
the
criteria
a
phase
material
adheres
reverse
surface silanol groups.
stationaryphase.

Bonded phase unit:


Me

r O'

\Me
SI ,,..
(CH2)5Me

Column dimensions used: 250 x 4.6 mm W.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%) / ligand coverage (pmol/m2): 4.7 / 3.36
Source: PhaseSeparations

Spherisorb octvl
General description: Octylsilane chemically bonded to porous Spherisorb silica. An
free
is
the
to
the
number of
support reduce
additional end-cappinggroup also presenton
for
This
L7
USP
to
the
reverse
a
groups.
phase
material
adheres
criteria
silanol
surface
stationaryphase.

Appendices

228
Bonded phase unit:
Me

\Me
%,,,
SI.(CH2)7Me

Column dimensions used: 250 x 4.6 mm i. d.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 200
Carbon loading (%) / ligand coverage (pmol/m2): 5.75 / 3.12
Source: PhaseSeparations

Zorbax stable bond cyanonroyyl


General description: Monofunctional cyanopropyl bonded Zorbax silica manufactured
by Agilent Technologies.This column was used in the reversephasemode.
Bonded phase unit:

1-Bu

-Bu
`,
rmo0 SI,.,

crv

Column dimensions used: 150 x 4.6 mm W.


Particle size (m) / pore size (A) / surface area (m2/g): 5/ 80 / 180
Carbon loading (%) / ligand coverage (mol/m2): 4/3.04

Source: PhaseSeparations

REFERENCES
Bio-Rad, HPLC Columns,Methods and Applications Catalogue, 1998.
Fisher Scientific UK, TheFisher Chromatography Catalogue, 1999.
Myers P., Personal Communication, 1998, University of Leeds.
Phenomenex, Cataloguefor Separation Science,2001.
Robson M. M., Personal Communication, 1998, University of Leeds.
Waters, Chromatography Columns and Supplies Catalogue, 1999.

Wilkes A. J., Walraven G. and Talbot J.M., J. Am. Oil Chem.Soc., 1992,69,609-613.
Wee V. T. and Kennedy J.M., Anal. Chem., 1982,54.1631-1633.

Appendices

229

APPENDIX

THREE

Initial mass spectrometerparameters


Capillary Temperature(C)
SourceVoltage (kV)
SheathGas Flow

220.00
3.7
25.00

Auxiliary Gas Flow

0.00

Capillary Voltage (V)

3.00

Tube Lens Offset (V)

7.5

Octapole RF Amplifier (Vp-p)

400.00

Octapole 1 Offset (V)

4.00
-

Octapole2 Offset (V)

8.00
-

InteroctapoleLens Voltage (V)

16.00
-

Trap DC Offset Voltage (V)

10.00
-

Maximum Ion Time (ms)

200.00

Ion Time (ms)


SourceType
Polarity
Zoom Micro Scans

5.00
Electrospray
Positive
10

Full Micro Scans

SIM Micro Scans

Apendices