Development of Two- Dimensional Liquid Chromatographic ...88… · Development of Two- Dimensional Liquid Chromatographic Systems For the Optimised Separation and Targeted Isolation

Development of Two-

Dimensional Liquid Chromatographic

Systems

For the Optimised Separation and Targeted

Isolation of Complex Samples

by Coleen Stephanie Milroy

B.Sc. (Hons)

A thesis submitted in accord with the requisites of the degree of

Doctor of Philosophy

School of Natural Sciences University of Western Sydney

March 2010

ii

Table of Contents ────────────────────────────────────────── Table of Contents………………………………………………………………………….ii Statement of Authentication………………………………………………………….........v Acknowledgements……………………………………………………………………......vi Publications Arising From This Thesis……………………………………………..…..vii List of Abbreviations………………………………………………………………….....viii List of Symbols………………………………………………………………………….….x List of Tables………………………………………………………………………...…..xii List of Figures…………………………………………………………………………...xiv Preface………………………………………………………………………………....xxiii Chapter 1………………………………………………………………………………….........1 General Introduction…………………………………………………………….....................1

1.1 Introduction……….………………………………………………........................2 1.1.1 History of Chromatography……………………………………..……....2

1.2 One-dimensional HPLC.........................................................……………............5 1.3 Multidimensional HPLC…………………………………………………............9

1.3.1 Sample dimensionality……………………………………....................12 1.3.2 Selective and nonselective displacements……………………………...15

1.4 Two-dimensional chromatographic systems.......................................................20 1.4.1 Comprehensive and heart- cutting separations………………………....20 1.4.2 Two-dimensional system designs…………............................................23 1.4.3 Data collection..............................…………...........................................24

1.5 Applications of 2D HPLC........………………………………………………......25 1.5.1 Pharmaceutical................................................………………………....25 1.5.2 Natural products......................................................................................26 1.5.3 Traditional Chinese medicines.....…………...........................................27 1.5.4 Forensic applications...............................................................................30

1.6 Preparative chromatography........……………………………………..……......31 1.7 Objectives..................................………………………………………………......40 Chapter 2....................................................................................................................................42 General Experimental..............................................................................................................42

2.1 Chemicals .......................…………………………………………….…..............43 2.2 Chromatography columns…................................................................................43

2.2.1 Chromatography column packing conditions...………………………...44 2.3 Equipment..............................................................................................................45 2.4 Chromatographic separations..............................................................................46 Chapter 3....................................................................................................................................47 Ultra High Resolution Separations of Diastereomers on Carbon Adsorption Stationary Phases ………………………………………………………………....................................................47 3.1 Introduction……………………………………………………......……….........48 3.1.1 Selectivity of C18 and carbon clad zirconia phases.................................50

iii

3.2 Experimental…………………………………………………………..............…..53

3.2.1 Chemicals .….………………………………………….......................…53 3.2.2 Chromatographic separation.……..……………………………..........…53

3.3 Results and discussion......................................................................................…..54 3.3.1 Peak profiles for separation of oligostyrenes........................................…54 3.3.2 Ultra high resolution separations of oligostyrene isomers on CCZ................................................................................................................…58 3.4 Conclusion...............................................................................................................63

Chapter 4…………………………………………………………………………..……............64 Practical Aspects in the Optimisation of Targeted Isolations in Two-dimensional HPLC: Analytical Scale Analysis......................................................................................................…64

4.1 Introduction……………………………………………………………............…65

4.2 Experimental……………………………………………………...……..........…..66 4.2.1 Chemicals ………………………………….................………….....…...66 4.2.2 Chromatographic separation………...………………………….......…..66 4.2.3 Determination of product purity and recovery ………………….....…...66

4.3 Results and Discussion………………………….…………...…………..........….67

4.3.1 Determination of target component…….………………….........………67 4.3.2 Isolation of target component: maximising purity….…………..........…72 4.3.3 Multicomponent isolations ...........………………….…………..........…83 4.3.4 Isolation of target component: maximising recovery ...........…………...90

4.3.5 Summary………………………………………………….…….........…96 4.5 Conclusion.……………………………………………………….……............….99

Chapter 5………………………………………………………………………...…….......…101 Practical Aspects in the Optimisation of Preparative Scale Two-dimensional Isolation: Low Sample Loads...............................................................................................………………………….101 5.1 Targeting the Isolation and Purification of Specific Compounds Within the Complex

Mixture at the Preparative Level.........................................................................…102 5.2 Introduction…………………………………………………...………..........….102

5.2.1 Production rate variables ……………….................……….…........….104 5.2.1.1 Sample volume and sample concentration……..….............…..104 5.2.1.2 Product recovery yield Yi ………………..........….….........….106 5.2.1.3 Cross-sectional surface area and total porosity………........….107 5.2.1.4 Cycle time ........................………………...............….........….107 5.2.1.5 Purity................................................................………........…..108 5.2.1.6 Effective and practical production rate…................…........…..108

iv

5.3 Experimental………………………………………………..........…………...…110 5.3.1 Chemicals ………………………………….................….......….......…110 5.3.2 Chromatographic separation………...………………………......…..….110 5.3.3 Calibration....................................................... ………….....…..........…110

5.4 Results and Discussion…………………………………………....….….........…111 5.4.1 Recovery yield of target component...................…………..…..........….116 5.4.1.1 Product recovery of the target component from central area of band in 1D and 2D…………..……………………………………..….............116 5.4.1.2 Off-centre in 1D...........……………………………..…........…...119

5.4.1.3 Multicomponent..............………………………….…….............125 5.4.1.4 Maximising recovery.....................................................................128

5.4.2 Summary………………………………………………………………...129 5.5 Conclusion………………………………………....…………...………........…...131

Chapter 6……………………………………………………..………………………........…..133 Practical Aspects in the Optimisation of Preparative Scale Two-dimensional Isolations: High SampleLoads...............................................................…………………………….….........…133

6.1 Introduction…………………………......…………………………...........……..134

6.2 Experimental……………………………...….........…………………….........…134 6.2.1 Chemicals .....................….……………………………..……..........….134 6.2.2 Chromatographic separations……………………...…….…….........….134 6.2.3 Determination of product purity and recover.………………............….135 6.2.4 Calibration.............................………....……………………...........…...135

6.3 Results and Discussion…………………………..…………………............……135

6.3.1 Sample load limitations on 1st dimension and 2nd dimension columns........................................................................………….…..........…..135 6.3.2 Increase in N in the second dimension….…………………..…........….139 6.3.2.1 Production................................................................................….148

6.3.3 Increasing the Peak Capacity in the First Dimension......…................….150 6.3.3.1 Recovery, Purity and Production Rate as a Function of Sample Injection

Volume.....................................................................................................…….151 6.3.3.1.1 Injection volume: 50 µL.....................................................….151 6.3.3.1.2 Injection volume: 100 µL...................................................….155 6.3.4 Summary………………………………………………………………. 159

6.4 Conclusion........................................................................................................……165

Chapter 7..............................................................................................................................….166 General Conclusion............................................................................................................….166 7.1 General Conclusion.........................................................................................….167 References.......................................................................................................................….....174

v

STATEMENT OF AUTHENTICATION The work presented in this thesis is, to the best of my knowledge, original unless specifically

acknowledged. I hereby declare that I have not submitted this material, either in whole or in part,

for a degree at this or any other institution unless specifically acknowledged.

……………………………… Signed: Coleen S Milroy

vi

ACKNOWLEDGEMENTS

I would like to express my gratitude and thanks to my principal supervisor, Associate Professor

R. Andrew Shalliker for his patient guidance, encouragement and continual support throughout

my project and for his contribution to my understanding of chromatography.

I would also like to thank Dr Gary Dennis, my secondary supervisor also for his encouragement

and helpful discussions that have made this work successful.

To my colleagues Michael Gray, Sindy Kayillo, Heather Catchpoole, Larissa Borysko, Paul

Stevenson, Arianne Soliven, Kirsty Mayfield and Mariam Mnatasakyan I am grateful for their

thoughtful advice and the pleasant working environment that they have created.

I would like to acknowledge the assistance of a University of Western Sydney Postgraduate

Award.

Finally I would like to acknowledge my family for their unreserved belief in me and for always

being my source of inspiration and purpose, my heartfelt thanks to my husband Jim and my

beautiful children Megan, Daniel and Liam. Thanks to my sisters Fiona, Kristina, Carole and

their families for all their support and also to Ian and Beth. And ultimately my gratitude to my

dad Richard Cooke who showed me that it is never too late in life to learn and that anything is

possible if you have faith in your choices.

vii

PUBLICATIONS Book Chapter C.S Milroy, P.G Stevenson,M. Mnatasakyan and R.A Shalliker, „Hyphenated and Alternative

Methods of Detection in Chromatography‟ in Multidimensional High Performance Liquid

Chromatography (Chromatographic Science). Ed. R.A Shalliker CRC Press Inc. (2010) (in

press).

Journal Publications Arising From This Work

C.S Milroy, G.R Dennis and R.A Shalliker, Ultra High Resolution Separations of Diastereomers

on Carbon Adsorption Stationary Phases. J. Liq. Chrom. Rel.Tech. A, 2007, 30(8) 991-999.

viii

LIST OF ABBREVIATIONS

1D: One-Dimension or Dimensional

2D: Two-Dimension or Dimensional

ABA: Abscisic Acid

ACN: Acetonitrile

APCI: Atmospheric pressure chemical ionization

CCC: Counter Current Chromatography

CCZ: Carbon Clad Zirconia

CF: Clerodendrum floribundum

CN: Cyano

COMET: Comprehensive Orthogonal Method Evaluation Technology

CVD: Chemical Vapour Deposition

DAD: Diode Array Detection

DCM: dichloromethane

GC: Gas Chromatography

GPC: Gel-permeation Chromatography

HPLC: High Performance Liquid Chromatography

Hz: Hertz (Unit of)

IAA: Auxin Acid

IR: Infrared

LC: Liquid Chromatography

LCC: Liquid Liquid Chromatography

LSC: Liquid Solid Chromatography

ix

MeOH: Methanol

MS: Mass Spectrometry

NMR: Nuclear Magnetic Resonance

NPLC: Normal Phase Liquid Chromatography

PAH: Polycyclic Aromatic Hydrocarbons

PHPLC: Preparative High Performance Liquid Chromatography

PGC: Porous Graphitized Carbon

RI: Refractive Index (Detection)

RP: Reversed-Phase

RPLC: Reversed-Phase Liquid Chromatography

SCX: Strong Cation Exchange

SEC: Size Exclusion Chromatography

SFC: Supercritical Fluid Chromatography

SMB: Simulated Moving Bed

SMO: Statistical Model of Overlap

TCM: Traditional Chinese Medicine

TLC: Thin Layer Chromatography

UHPLC: Ultra High Performance Liquid Chromatography

UV: Ultra-Violet

V: Valve

Vis: Visible (reference to detection)

Zr: Zirconia

x

LIST OF SYMBOLS

General Symbols

Ai Amount of Sample Collected

Cf Concentration of Target Component

C1 Column 1

C18 Octadecylsilane (reference to stationary phase support modification)

C2 Column 2

C0i Injected Concentration of Sample Constituent

D1 Detector 1

D2 Detector 2

D3 Detector 3

dp Particle Diameter

ε Total Porosity

EffPri Effective Production Rate

k Retention Factor

i Component

Mi Mass per Injection

ni Amount of Sample Injected

N Number of Theoretical Plates

N Non-selective (displacement)

PracPri Practical Production rate

PPri Production rate

P1 Pump 1

xi

P2 Pump 2

pI Isoelectric point

s Sample dimensionality

Sa Cross Sectional Surface Area

S Selective (displacement)

S(1-8) Solvent reservoir 1 to solvent reservoir 8

t0 Retention time of void marker

ti Retention time of component of interest

tR Retention time

V(1-4) Valve 1 to Valve 4

V f Volume of Collected Fraction

α Selectivity or Separation Factor

Δt Entire retention time range

tc Cycle Time

Vs Sample volume

χa Normalised Retention Factor

Yi Recovery Yield

xii

LIST OF TABLES

Table 1.1: Types of discrete displacement combinations and their effect on two-dimensional

peak capacity

Table 1.2: Characteristics of multidimensional and gradient elution separations of component A

from Clerodendrum floribundum

Table 2.1: Chromatographic columns used in this study

Table 3.1: Number of diastereomers for oligostyrene (OLIGOSTYRENE) oligomer with tert-

butyl end grouoligostyrene

Table 4.1: Characteristics of two-dimensional separation of target component n = 5 #2


Table 4.3: Collection volume and percentage of diastereomers #1-8 in 2nd dimension

Table 4.4: Characteristics of two-dimensional separation eight diastereomers for 500 µL

heart-cut



Table 5.1: Characteristics of two-dimensional separation of target component n=5 #2


Table 5.3: Characteristics of two-dimensional separation of target component n = 5 #1-8 500 µL

heart-cut

Table 5.4: Characteristics of two-dimensional separation of target component n = 5 #2 for 500

µL heart-cut

xiii

Table 5.5: Characteristics of two-dimensional separation of target component n = 5 #2 for 500

µL heart-cut

Table 6.1: Characteristics of two-dimensional separation of target component n =5 #2 for

different injection volumes

Table 6.2: Characteristics of two-dimensional separation of target component n=5 #2 for






xiv

LIST OF FIGURES

Figure 1.1: Illustration of sample dimensionality (a)PAHs (size and shape), and (b)

diastereoisomers of n-butyl oligostyrene with n = 2 to n = 5 styrene repeating units.

Figure 1.2: Illustration of the combinations of discrete selective (S) and non-selective (N)

displacements: (a) Two-dimensional S × SI displacement; (b) Two-dimensional S ×

Sc displacement; (c) Two-dimensional N × S displacement.

Figure 1.3: Normalised two-dimensional plot of the C18 (methanol)/CCZ (acetonitrile)

system in the separation of the 58 oligostyrene isomer mix. Each boxed section

represents isomeric components containing the same number of configurational

repeat units. The numbers adjoining the data points indicate the number of

components co-eluting.

Figure 1.4: 1D chromatogram of crude extract of C. floribundum. Column: CN (150 mm x

4.6 mm, 5 m). Mobile phase: gradient elution 95% water/5% MeOH - 100 %

MeOH over 18 minutes. F = 1.0 mL/min, injection volume 20 L. Detection UV 270

nm.

Figure 1.5: 1D chromatogram of crude extract of C. floribundum. Column: Luna CN (150

mm x 4.6 mm, 5 m). Mobile phase: isocratic elution of 45% water/60% ACN. F =

1.0 mL/min, injection volume 10 L. Detection UV 270 nm.

Figure 1.6: Chromatograms in the second dimension illustrate the relative change in the

concentrations of CF1 and CF2 following heart cutting from the first dimension.

Experimental conditions: 1D: Column: Luna CN (150 mm x 4.6 mm, 5 m). Mobile

phase, 40% water/60% ACN. 2D: Column: ValuePak C18 (250 mm x 4.6 mm, 5

xv

m). Mobile phase, 55% water/45% ACN. F = 1.0 mL/min, injection volume 10 L.

Heart cut volume 200 L.

Figure 1.7: Chromatograms in the second dimension illustrate the relative change in the

concentrations of CF1 and CF2 following heart cutting from the first dimension.

Experimental conditions: 1D: Column: Luna CN (150 mm x 4.6 mm, 5 m). Mobile

phase, 40% water/60% ACN. 2D: Column: ValuePak C18 (250 mm x 4.6 mm, 5

m). Mobile phase, 60% water/40% ACN. F = 1.0 mL/min, injection volume 10 L.

Heart cut volume 100 L.

Figure 1.8: 1D chromatogram of crude extract of C.floribundum. Column: Luna CN (150

mm x 4.6 mm, 5 m). Mobile phase: gradient elution of 95% MeOH/5% water- 100

% MeOH in 18 minutes. F = 1.0 mL/min, injection volume 20 L. Detection UV 270

nm.

Figure 1.9: Two dimensional chromatogram of target component of crude extract of

C.floribundum under overload conditions. Illustration of the cycle time and the

maximisation of separation space. (a) 1D separation: Column: CN (150 mm x 4.6

mm, 5 m). Mobile phase: water-ACN (30:70). F = 1.0 mL/min, injection volume

200 L. (b) 2D separation: Column: C18 (250 mm x 10.0 mm, 5 m). Mobile phase:

water-ACN (40:60). F = 1.0 mL/min, heart-cut volume 200 L. Detection UV 270

nm.

Figure 2.1: Diagram of the 2D HPLC system. P1-P2: Pumoligostyrene that deliver

solvents; V1-V4: six-port two-position switching valves; C1: 1D column; C2:2D

column. D1-D2 detectors. a) System configuration for elution on C1 and C2 and b)

system configuration for elution of a band from C1 onto sample loop.

xvi

Figure 3.1: Diagram of the configurational repeating unit of styrene.

Figure 3.2: n = 5 oligostyrene with tert-butyl end group, five configurational

repeat units and atactic stereochemistry.

Figure 3.3: Chromatogram of tert-butyl oligostyrene separation on C18 Varian Pursuit XRs

column (250 × 4.6 mm). Conditions: gradient elution 100:0 MeOH: DCM to 80:20

MeOH: DCM in 4 minutes at 1.0 mL/min at ambient temperature with 30 μL

injection volume. Detection UV 272 nm. Oligomers number 1-10 accordingly.

Figure 3.4: Chromatogram of tert-butyl oligostyrene separation on C18 Varian Pursuit XRs

column (250 × 4.6 mm). Conditions: 100% ACN at 1.0 mL/min at ambient

temperature with 10 μL injection volume. Detection UV 272 nm.

Figure 3.5: Chromatogram of tert-butyl oligostyrene separation on CCZ column (100 × 10.0

mm). Conditions: gradient elution 100:0 ACN: DCM to 0:100 ACN: DCM in 50

minutes at 3.0 mL/min at ambient temperature with 10 μL injection volume.

Detection UV 262 nm.

Figure 3.6: Chromatogram of gradient separation of oligomer n =6 on CCZ column (100 x

10 mm). Conditions: gradient elution 80-20 ACN: DCM to 0-100 ACN: DCM in 50

minutes at 3.0 mL/min with at ambient temperature 30 µL injection volume.







xvii

Figure 3.9: Chromatogram of gradient separation of oligomer n =8 on Hypercarb column

(100 x 4.6 mm). Conditions: gradient elution 100-0 ACN: DCM to 0-100 ACN:

DCM in 40 minutes at 1.0 mL/min with at ambient temperature 30 µL injection

volume.




Figure 4.1: Chromatogram of the separation of tert-butyl oligostyrene n = 5 separation on

CCZ column, target component peak #2. Conditions: CCZ column (50 mm × 4.6

mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 100 μL heart-cut volume.

Figure 4.2: Chromatogram of the separation of tert-butyl oligostyrene separation on C18

column displaying the overlap of n = 4, n = 5 and n = 6. Conditions: C18 column (50

mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection

volume.

Figure 4.3: Area that target component n = 5 #2 occupies in first dimension.

Figure 4.4: Overlap of n = 4 #3 and n = 5 #2 on C18 column for heart-cuts from the first

dimension. Conditions: C18 column (50 mm × 4.6 mm), 100 % ACN mobile phase

at flow rate 2.0 mL/min. 100 μL heart-cut volume.

Figure 4.5: Chromatogram of the separation of tert-butyl oligostyrene separation on CCZ

column showing recovery of target component. Conditions: CCZ column (50 mm ×

4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 100 μL heart-cut

volume.

xviii




volume.

Figure 4.7: Recovery plot of target component.

Figure 4.8: Chromatogram of purity for (a) 100 µL heart-cut and (b) 700 µL heart-cut.

Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate

2.0 mL/min. 10 μL injection volume.

Figure 4.9: Recovery plot of target component; (a) 2.05 minute centred heart-cuts, (b) 2.10

minute centred heart-cuts.

Figure 4.10: Purity vs total recovery; (a) 2.05 minute centred heart-cuts, (b) 2.10 minute

centred heart-cuts.

Figure 4.11: Close up of where diastereomers (a) #1-6 and (b) #7-8 were collected on CCZ

column. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at

flow rate 2.0 mL/min. 500 μL heart-cut volume

Figure 4.12: Chromatograms of purity for 500 µL heart-cut (a) #1, (b) #2, (c) #3 and 4, (d)

#5, (e) #6 and (f) #7. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN

mobile phase at flow rate 2.0 mL/min. 10 μL injection volume.




volume.

Figure 4.14: Recovery vs purity of target component.

xix

Figure 4.15: Chromatogram of purity for 500 µL heart-cut. Fraction collected and re-

injected onto CCZ.

Figure 4.16: Comparison of the strategies for the separation of target diastereomer

separation of the 300 µL heart-cuts

Figure 4.17: Comparison of the strategies for the separation of target diastereomer

separation of the 400 µL heart-cuts.

Figure 4.18. Comparison of the strategies for the separation of target diastereomer

separation of the 500 µL heart-cuts.


column. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at

flow rate 1.0 mL/min. 10 μL injection volume. Detection UV 272 nm. Oligomers

numbered 1-10 accordingly.

Figure 5.2: Time line of 1D and 2D separation.

Figure 5.3: Chromatogram of the five successive injections onto 1st dimension. Conditions:

C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0

mL/min. 10 μL injection volume.

Figure 5.4: Chromatogram of successive injections onto 1st dimension showing empty

separation space and heart-cut. Conditions: C18 column (50 mm × 4.6 mm), 100 %

MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection volume.

Figure 5.5: Chromatogram of the five successive injections onto 2nd dimension. Conditions:

CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 4.0 mL/min.

700 μL heart-cut volume.

xx

Figure 5.6: Overlap of n = 4#3 and n = 5#2 on C18 column for 300 µL heart-cuts from the

first dimension. (1) 1.95 minute centre band (2) 2.05 minute centre band (3) 2.10

minute centre band. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH

mobile phase at flow rate 1.0 mL/min. 10 μL injection volume.


first dimension. (1) 1.95 minute centre band (2) 2.05 minute centre band; (3) 2.10

minute centre band. Conditions as in Figure 5.6.


first dimension. (1) 1.95 minute centre band (2) 2.05 minute centre band (a); (3) 2.10

minute centre band (b). Conditions as in Figure 5.6.

Figure 5.9: Comparison of the practical production rate of the target diastereomer for 300,

400 and 500 µL heart-cuts.

Figure 6.1: Different injection volumes on C18 column. Mobile phase 100% MeOH.

Figure 6.2: Chromatogram of the separation of tert-butyl oligostyrene n = 5 separation on

CCZ column, target component peak #2. (a) 30 L injection volume to C18 (5cm),

(b) 50 L injection. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile

phase at flow rate 2.0 mL/min. 100 μL heart-cut volume.


column displaying area n = 5 occupies. Conditions: C18 column (50 mm × 4.6 mm),

100 % MeOH mobile phase at flow rate 1.0 mL/min. 30 μL injection volume.

xxi


column displaying area n = 5 occupies. Conditions: C18 column (50 mm × 4.6 mm),

100 % MeOH mobile phase at flow rate 1.0 mL/min. 50 μL injection volume.

Figure 6.5: Chromatogram of the separation of tert-butyl oligostyrene n =5 separation on

CCZ column, target component peak #2 for 30 μL injection volume. (a) 700 μL

heart-cut volume, (b) 1000 μL heart-cut volume. Conditions: CCZ column (100 mm

× 4.6 mm), 100 % ACN mobile phase at flow rate 1.0 mL/min.

Figure 6.6: Chromatogram of the separation of tert-butyl oligostyrene n =5 separation on

CCZ column, target component peak #2 for 50 μL injection volume. (i) 700 μL

heart-cut volume, heart-cut time 1.55-2.25 min. and (ii) 1000 μL heart-cut volume,

heart-cut time 1.30-2.30 min. Conditions: CCZ column (100 mm × 4.6 mm), 100 %

ACN mobile phase at flow rate 1.0 mL/min. 100 μL heart-cut volume.


column showing recovery of target component for 30 μL injection volume.


1.0 mL/min. 700 μL heart-cut volume.


column showing recovery of target component for 50 μL injection volume.


1.0 mL/min. 700 μL heart-cut volume (1.55-2.25 minutes).

Figure 6.9: Chromatogram of purity for 30 µL injection. 700 µL heart-cut. Fraction

collected and re-injected onto CCZ.

xxii

Figure 6.10: Chromatogram of purity for 50 µL injection. 700 µL heart-cut (1.55-2.25 min).

Fraction collected and re-injected onto CCZ.

Figure 6.11: Chromatogram of tert-butyl oligostyrene separation on C18 Sphereclone

column (150 × 4.6 mm). 100 % MeOH mobile phase at flow rate 0.83 mL/min, 50

μL injection volume. Oligomers number 1-10 accordingly.

Figure 6.12: Chromatogram of purity for 50 µL injection for 1328 µL heart-cut. Fraction


Figure 6.13: Chromatogram of tert-butyl oligostyrene separation on C18 Sphereclone

column (150 × 4.6 mm). 100 % MeOH mobile phase at flow rate 0.83 mL/min, 100

μL injection volume. Oligostyrene n = 5 numbered.





Figure 6.16: Comparison of the recovery and purity of the target diastereomer separation for

increased N in 2D from Section 6.3.2 for heart-cuts of 30 µL and 50 µL injections.

Figure 6.17: Comparison of the recovery and purity of the target diastereomer separation for

increased N in 1D and 2D from Section 6.3.3 for heart-cuts of 50 µL and 100 µL

injections.

Figure 6.18: Comparison of the practical production rate of the target diastereomer for

increased N in 2D; and increased N in 1D and 2D for heart-cuts of 30 µL, 50 µL and

100 µL injections.

xxiii

PREFACE Chromatography is a powerful separation technique that was initially developed for the isolation

of natural components in a highly purified form from complex mixtures. However, early

applications of chromatography that were preparative in nature were quickly surpassed by

analytical separations as the need for qualitative and quantitative information about components

present in simple and complex mixtures became the primary objective.

HPLC is the commonly used analytical separation technique for the determination of

components in complex mixtures as it offers high sensitivity and also high selectivity. In general,

HPLC is carried out in a single dimension using one primary retention mechanism dictated

largely by the stationary phase. Although modern HPLC column technology has made available

stationary phases with improved efficiency, particularly for the separations of complex mixtures,

there are limits to its use. In reality there is only so much space available for components in a

given sample to be separated into individual peaks, which is limited by the peak width and

determined by the efficiency of the column. This is highly dependent upon the number of

theoretical plates (N) available for the separation and therefore the peak capacity.

Multidimensional HPLC is a technique that is gaining appreciable support at the analytical level

due to the vastly expanded separation space that allows for increased resolution of components in

complex samples. The introduction of a second dimension, which offers a change in selectivity

xxiv

to that of the first dimension, is a means of increasing the total peak capacity of the separation

process and therefore expanding the separation space. Two-dimensional HPLC is an effective

separation technique for the analysis of complex mixtures if the sample‟s complexity can be

reduced as the separation mechanism of the first dimension may be tailored towards the sample‟s

multidimensionality and/or its physical characteristics such as size, polarity, charge and shape.

Reducing the complexity can be as simple as ensuring co-elution of the key components in the

first dimension, but then utilising the second dimension to resolve those components co-eluting

in the first dimension. Here the first dimension can be considered essentially as a „clean-up‟ step

in order to isolate the required targets in the second dimension. This essentially reduces the

required peak capacity of the first dimension and the separation space in the second dimension

more than compensates for this reduction as the second dimension is required to separate a lower

number of components.

Preparative HPLC has only seen resurgence in the last few decades with traditional methods such

as distillation, centrifugal extraction and crystallisation unsuitable for the problems encountered

by the various industries. The stringent regulations of governing bodies for the approval of

highly purified products to be released into a highly competitive market dictate largely the

advancement of chromatographic methods for preparative separations.

Chapter 1 is a general introduction with a brief history of chromatography presented. The theory

and practice of two-dimensional chromatographic separations and preparative chromatographic

separations are also discussed. Sample dimensionality and its importance to the determination of

orthogonality of separation steps are also considered.

xxv

Chapter 2 details the chemicals and equipment used to perform the experiments outlined

throughout the thesis. Specific general methods are also included here for reference.

Chapter 3 describes a model framework for the development of a two-dimensional liquid

chromatographic system and was reliant on low molecular weight oligostyrenes as they are

complex, are indefinitely stable and easily characterised. The ultra high resolution separation of

diastereomers of low molecular weight oligostyrenes on carbon adsorption stationary phases are

also discussed.

Chapter 4 examines the practical aspects in the optimisation of targeted isolations in two-

dimensional HPLC with emphasis on analytical scale analysis. The emphasis in this chapter was

on the isolation of „a‟ target analyte from „a‟ complex mixture, where effectively „a‟ represents a

generic sample, complex in nature. Chapter 4 also examines the practical aspects in the

optimisation of targeted isolations in two-dimensional HPLC with emphasis on analytical scale

analysis. However this chapter focuses on maximising the recovery of the target component at

analytical scale analysis.

Chapter 5 investigates the practical aspects in the optimisation of preparative scale two-

dimensional isolations by the establishment of a continuous batch-wise 2D purification process,

with the intent to preparative scale-up. In this chapter the experimental variables that effect the

production of a target component are introduced and the influences they have on the separation

are discussed. Chapter 5 also investigates the use of the system introduced at the preparative

level however high sample loads are now used to determine the effect of the recovery in both the

xxvi

first and second dimensions; the purity of the collected product; and finally the product recovery

yield, effective production rate and the practical production rate.

Chapter 6 focuses on improving the quality of the isolation process by increasing the peak

capacity of the second dimension firstly, followed by an increase in the peak capacity in both the

first and second dimensions. This improved the resolution and therefore reduced the number of

components transferred to the second dimension. This had the benefit of decreasing the

performance demand of the second dimension column.

Chapter 7 summarises all of the findings contained throughout this thesis.

1

Chapter 1

General Introduction

2

1.1 Introduction

In the century since the invention of chromatography there has been a tremendous technological

advancement that has sprung from this pioneering technique. Chromatography is the most

utilised separation tool that is available to scientists today in fields as diverse as the

petrochemical industry to the pharmaceutical industry.

1.1.1 History of Chromatography

Mikhail Twsett (1872-1919) the lauded inventor of chromatography was a botanist researching

the separation of plant pigments over a century ago. His Master‟s thesis was a comprehensive

study on the selective adsorption of chlorophyll leaf pigments by substrates using different

solvents (1902). He further investigated the adsorption characteristics of chlorophyll by more

than a hundred solid organic and inorganic substances packed into small tubes. By careful

selection of solvents he was able to successfully elute individual pigments that are in the

chlorophyll sample from the column. He first used the term chromatography in 1906 to describe

this method and even postulated the use of two-dimensional chromatography by developing

columns with another solvent after the first separation [1, 2].

Chromatography was given little attention in the following decades. In the early 1920s Leroy S

Palmer used Twsett‟s chromatographic technique on various natural products and published a

monograph in 1922 that detailed the use of chromatography [3]. However, it was not until the

early 1930s when Lederer examined the xanthophylls in egg yolks using chromatography [4, 5]

and the subsequent separation and isolation of α- and β-carotene [6] that interest from the wider

3

scientific community was generated. Essentially chromatography of this time was a preparative

technique to isolate components from complex mixtures for further investigation. The

separations were very slow as gravity or low applied pressure influenced the flow of mobile

phase through the column and columns were generally not reusable.

It was Martin and Synge in 1941 working for the Wool Industries Research Association in

England that modernised liquid chromatography with the development of partition

chromatography (liquid-liquid chromatography, LLC) for the separation of amino acids [7].

They also presented a mathematical treatment of the theory of chromatography, including the

theoretical plate concept. Due to the limitations of LLC, paper chromatography was developed

for the separation of chemical mixtures in the liquid phase. Silica gel had been used in LLC but

was replaced with paper in an effort to separate several dicarboxylic and basic amino acids [8].

Furthermore their study on paper chromatography demonstrated the feasibility of a two-

dimensional technique for improved separation power. Gas chromatography (GC) was invented

in 1941 [7] but not developed until 1952 [9]. The achievements of Martin and Synge transformed

chromatography and they were honoured by the Chemistry Nobel Prize in 1952.

Thin-layer chromatography (TLC) another planar chromatographic technique like paper

chromatography, was described in 1951 by Justus G Kirchner [10] and was further developed by

Stahl [11, 12]. Due to the variety of stationary phases available, its ease of use and speed, TLC is

still used as a standard analytical and preparative chromatographic method. Another liquid

chromatographic method that became an important analytical tool particularly in biochemistry

was size-exclusion chromatography (SEC). SEC was developed by Flodin and Porath in the late

1950s [13], the commercial gel Sephadex was introduced in 1959; and gel-permeation

4

chromatography (GPC) for the fractionation of synthetic polymers was reported by J.C Moore

[14]. Ion-exchange chromatography had been used during World War II in the large-scale

separation of cations of rare earth metals for nuclear purposes as part of the Manhattan Project

[15, 16]. Soon biological samples were separated by ion-exchange by Moore et al. [17, 18, 19]

and the first automated liquid instrumentation, the automatic amino acid analyser, was developed

by Spackman, Moore and Stein [20].

The theory of the chromatographic process was further described in 1956 by van Deemter et al.

in regards to the influence of diffusion and the resistance to mass transfer between the two

phases in GC [21]. GC had sophisticated instrumentation and reusable columns [9] and provided

a simple and sensitive method for the analysis of volatile compounds. GC was the most widely

used analytical technique for the separation of mixtures of organic compounds until the

modernisation of liquid chromatography (LC).

The development of modern LC had to overcome basic problems of slow diffusion in the liquid

mobile phase, which were resolved by the use of small and uniform particles that had short

diffusion paths, and high mobile phase velocities, with mobile phases delivered under high

pressure. Martin and Synge had already experimentally predicted this in 1941 [7] as well as

Giddings in 1963 [22]. However, it was not until 1967 that chromatographic experiments using

small particle sizes were reported by Horvath [23], Huber [24] and Scott [25]. In 1969 Jack

Kirkland working at DuPont developed pellicular materials that would soon become

commercially available [26]. Due to the high pressure generated by the use of smaller particles in

the stationary phase, specialist hardware was now required. Horvath had developed an integrated

5

instrument in the early 1960s akin to the gas chromatograph instrumentation and described an

LC system that would incorporate a dedicated solvent delivery system, an injector, an on-line

detector and recorder and a highly efficient column [23]. In 1968 Kirkland described how to

construct high performance liquid chromatography (HPLC) equipment by modifying existing

components [27] and by the early 1970s HPLC systems were commercially available that could

be operated under high pressure with continuous detection and high efficiency [28]. The rapid

progression of HPLC in the 1970s was attributed to the availability of small porous particles

(diameter of particles as small as ~ 3 µm as early as 1978) and the development of the “bonded”

reversed-phase (RP) stationary phases [29]. The reduction in diameter of the stationary phase

particles improved column efficiency and shorter columns could be used thus reducing the

analysis time and improving detection sensitivity. LC consequently evolved from a preparative

method into an analytical technique and due to its versatility and precision, became the method

of choice for analytical chemistry in a wide range of industries [30].

1.2 One-dimensional HPLC

The synthesis of novel products and the isolation and purification of natural substances provided

the impetus for the advancement of chromatography during the last century. This motivation also

drives the development of today‟s technologies in the chemical, pharmaceutical,

biotechnological and agrochemical industries among others, where the need for superior

separations of highly purified products, which are available faster and cheaper than ever before,

is paramount [31]. In the pharmaceutical industry for example it was estimated in 2001 that for

each new drug discovery, research and development costs were in excess of approximately

6

US$800 million [32, 33]. Part of that cost is often related to the identification and subsequent

extraction of key components in natural products. Therefore it is important to investigate more

efficient processes of identification and isolation. After screening for bioactivity, the active

components must be isolated and extracted. In plants, this typically results in a low yield product

of varying purity dependent upon the method used for separation. Once isolated, these

constituents can be analysed by further techniques for structure elucidation and for biological

mechanisms that may show promise as new leads for drug discovery. Typically the isolation of

bioactive components in natural products has been a long and arduous task that is made quite

difficult due to the complexity of the samples presented [31, 34].

Investigation into separation techniques that can improve the separation, isolation and structural

elucidation of natural products is therefore vital for the reduction the overall cost of new drug

development. Chromatographic separations are by far the most important; chromatographic

separation techniques available for the isolation of natural products are varied and include TLC,

GC, countercurrent chromatography (CCC) and HPLC [31]. Improvement in both HPLC

automation and column technology has seen the efficiency of the analysis and the isolation of

natural products improve significantly [31]. Therefore HPLC is a widely used technique in the

pharmaceutical and the natural products industries and is applied at all stages of drug discovery,

development and production and is a method of choice for the isolation of active components in

sample matrices [31]. However, as the complexity of the sample matrix increases, the

productivity of the isolation decreases as the useable peak capacity of the separation is exceeded,

limiting resolution and thus making it more difficult to bring about the collection of pure

7

quantities of the target analyte [31]. Pre-treatment processes are thus required to reduce the

complexity of the sample.

The example of natural product isolation is not unique to the pharmaceutical industry; the use of

HPLC is warranted for analytical and preparative separations across a diverse range of industries

where quantitative and qualitative information is to be ascertained about both natural and

synthetic samples [34]. Many methods exist in which a chromatographic approach is used for the

resolution of the sample matrix and detection methods such as mass spectrometry (MS), infrared

(IR) and nuclear magnetic resonance spectroscopy (NMR) are used to assist in the overall

analysis and identification. Although MS offers high selectivity, the overall sensitivity of MS in

the analysis of complex mixtures is restricted as the ionisation of trace compounds may be

suppressed by those of principal components. Diastereomers are also difficult to differentiate by

MS as the molecular ions are identical. Infrared (IR) detection, suffers from solvent interference

effects. NMR is expensive, relatively insensitive, but can be used to provide information that

relates directly to the chemical structure of a substance. In combination with LC, the three

resultant hyphenated methods of detection yield a combined process of analysis that is

unsurpassed in its ability to provide qualitative and quantitative sample information, yet as a

whole, only a few laboratories world-wide can afford to accommodate all three methods of

analysis.

HPLC is the commonly used analytical separation technique for the determination of

components in complex mixtures as it offers high sensitivity and also high selectivity. In general,

HPLC is carried out in a single dimension using one primary retention mechanism dictated

largely by the stationary phase. Although modern HPLC column technology has made available

8

stationary phases with improved efficiency, particularly for the separations of complex mixtures,

there are limits to their use [35]. In reality, the peak capacity of a column is limited by the peak

width and determined by the efficiency of the column. This is highly dependent upon the number

of theoretical plates (N) available for the separation [36]. The peak capacity is the measure of the

number of components that can be theoretically resolved side by side over the entire separation

space without peak overlap [37] and is generally proportional to the square root of the theoretical

plates available. However, one-dimensional HPLC is often limited by a maximum peak capacity

as predicted by Davis and Giddings [38], where the number of components that can be resolved

has shown to be notably less than the peak capacity of the column due to the statistical overlap of

the component peaks in complex samples [38]. This information becomes particularly important

when a complex mixture is presented for separation, as only a limited number of components

may be resolved dependent upon the peak capacity. If the selectivity of the column does not

allow for the separation of all components, some peaks may overlap and actually represent more

than one compound.

Traditionally, different selectivity steps are often employed to reduce the complexity of mixtures

and also improve the resolution of the separation [31]; these may include sample pre-treatments

and/or the use of more efficient columns, for instance columns with decreased stationary phase

particle size or increased column length [30, 31] (both of which increase the theoretical plate

count, N). Because N is directly proportional to the column length, L, the resolution and peak

capacity would increase in direct proportion to the square of the root of L [34]. However the

separation time may be increased appreciably and at some stage the upper limits of the column

pressure are reached [34]. As well these additional steps may add to labour and solvent

9

consumption costs [34]. When the sample complexity exceeds the point at which suitable

separation can be achieved, gradient elution is often employed as it can be an effective way to

increase the peak capacity and allow the elution of compounds with widely varying polarity

typical of a natural product [34]. Nevertheless, employing gradient elution results in an increase

in the separation down time as a consequence of re-equilibration between successive runs. If the

gain in peak capacity is still insufficient, additional separation steps may need to be employed.

These may include precipitation and centrifugation, or multiple selectivity separation steps [34].

Either of these approaches, however, results in an increase in the time and labour required to

bring about the isolation, unless the additional selectivity step are coupled on-line, which may

minimise the associated labour cost. Multidimensional chromatography is a way to increase peak

capacity, thus allowing for complete separation.

1.3 Multidimensional HPLC

Two or more different chromatographic steps are referred to as multidimensional HPLC.

Multidimensional HPLC is a technique that is gaining appreciable support at the analytical level

because it creates a vastly expanded separation space that allows for increased resolution of

components in complex samples [34]. The number of components that can be resolved in a one

dimensional (1D) system is substantially less than the maximum peak capacity because of the

randomness of the peak displacement [37]. As a result 1D HPLC is limited by the inadequate

space suitable for the resolution of all components in a complex sample as predicted by Davis

and Giddings [37, 38]. The introduction of a second dimension, which offers a change in

selectivity to that of the first dimension, is a means of increasing the total peak capacity of the

10

separation process and therefore expanding the separation space. Multidimensional HPLC

introduces a second dimension that ideally offers a retention mechanism that is very different to

that of the first dimension, where orthogonality is loosely used to describe these selectivity

differences between dimensions [39, 40]. Orthogonality is strictly a binary property, either it is

or is not orthogonal, but in the context of a descriptor of differences in retention behaviour

between systems, degrees of orthogonality had been implied as a measure of the separation

capability of a two-dimensional (2D) HPLC system and subsequently the higher the

orthogonality of a system the lower the correlation between separations on each dimension

becomes. When there is a low correlation between dimensions, the peak capacity of the system is

increased, and as a consequence, the probability of a component occupying a space that is unique

to that component is improved. In fact, the total peak capacity of an orthogonal two-dimensional

separation is equal to the product of the peak capacities in both dimensions.

In order to determine the orthogonality and maximise the separation space available for the

separation of complex samples, numerous approaches have been developed for two-dimensional

HPLC systems. Selectivity studies, where the comparison of different stationary phase and

mobile phase combinations are examined, are generally the most simplistic means of

determining the most orthogonal and least correlated systems and thus the most effective system

for separation. Selectivity can be controlled through changes in not only stationary phases but

also through the composition of mobile phases [41-46], mobile phase additives [45, 46], pH

modifications [45, 46] and through temperature adjustments [46, 47].

11

The coupling of reversed-phase (RP) dimensions can provide an orthogonal system as a result of

selectivity changes and has proved successful for the separation the complex mixture of

oligostyrene isomers that were previously unable to be resolved or were only partially resolved

in 1D HPLC systems [48-51]. Gray et al. [51] successfully separated between 46 and 49 of the

58 components in an oligostyrene sample using 2D RP HPLC system incorporating C18 and

carbon clad zirconia (CCZ) columns. This sample could not be resolved in a 1D HPLC system,

yet high resolution was achieved in the coupled 2D HPLC system. Ikegami et al. [52] developed

a 2D HPLC system using two RP C18 monolithic columns in both dimensions; the selectivity of

each dimension was achieved by varying the mobile phase compositions. Under these conditions

the theoretical peak capacity of 900 was reached (although true orthogonality did not exist due to

some correlation between dimensions) [52].

Even though only a fraction of the separation space of a highly correlated system may be

utilised, the separation may still be successful if the aim is to isolate and identify selected

components in a complex sample matrix since often a high degree of correlation between

dimensions results from similarity in the mobile phases. Most 2D HPLC systems do display

some degree of correlation, which greatly alleviates the difficulties associated with the

compatibility between dimensions. For example, it would be expected that 2D RP/RP

separations would be highly correlated but these typically use aqueous mobile phases in each

dimension that simplifies the interfacing between each of the RP dimensions and greatly

alleviates problems of immiscibility between dimensions.

12

2D HPLC generates two sets of retention data for a given sample, and the retention times in the

second dimension may be plotted against those in the first dimension and displayed as contour,

surface or scatter plots to provide distinctive patterns which may be useful for applications such

as chemical fingerprinting. Since a chemical signature of the sample is able to be produced [53,

54] the technique of 2D HPLC is useful for product profiling and evaluating whether changes in

the sample have taken place with time, due perhaps to storage or even to measure the effect of

climatic variables associated with the growing and harvest cycle.

2D RP-RP HPLC is a useful separation technique for the analysis of complex mixtures. The

sample‟s complexity can be reduced if the separation mechanism for the first dimension is

tailored towards the sample‟s multidimensionality and/or its physical characteristics such as size,

polarity, charge and shape. Reducing the complexity can be as simple as ensuring co-elution of

the key components in the first dimension, but then utilising the second dimension to resolve

those specific components co-eluting in the first dimension. The first dimension could be

considered as a „clean-up‟ step in order to isolate the required target compounds in the second

dimension. This essentially reduces the peak capacity required for the first dimension, but the

separation space in the second dimension more than compensates for this reduction because the

second dimension is then required to separate a lower number of components.

1.3.1 Sample Dimensionality

Sample dimensionality can be described as the number of unique features of the sample, and

these can be utilised for separation purposes. To separate an n- dimensional sample, an n-

13

dimensional chromatographic separation system should be employed [55], one dimension for

each sample attribute would ultimately yield the greatest degree of separation power and

maintain greatest control on separation order. Sample characteristics that could be considered as

„sample dimensionality‟ could be; molecular weight, pKa, and isomers (including structural,

diastereomers and enantiomers). For example, the length of an alkyl chain may be described by

the molecular weight or define the hydrophobicity of the molecule. If this chain is branched then

the degree and location of the branching may be a second and third dimension; double and triple

bonds, a fourth or fifth dimension. The position and/or number of certain functional groups,

perhaps on an aromatic ring or rings, even the bonding of these rings within the structure are

factors for describing the sample dimensionality and hence exploiting these sample attributes

from a separation sense.

Two simple examples of sample dimensionality are found for the samples of polycyclic aromatic

hydrocarbons (PAHs) and low molecular weight oligostyrene (Figure 1.1). In the case of the

PAHs, the molecular size of the homologous series: naphthalene, anthracene, 2,3-

benzanathracene and pentacene increases as the number of aromatic rings increase– the first

sample dimension. The second sample dimension could be represented by the structural isomers

of these PAHs, for example the four ring homologues: chrysene, pyrene, 2,3-benzanthracene and

benz[a]anthracene.

In the same way the sample characteristics of low molecular weight oligostyrenes can be

described [51]. Firstly, the number of repeat units that make up the chain determines the

polymer‟s molecular weight and hence the first sample dimension. The tacticity of the polymer

14

can be used to describe the second dimension. The third dimension could be described by the

enantiomers of each of the diastereomers.

Figure 1.1: Illustration of sample dimensionality (a) PAHs (size and shape), and (b) diastereoisomers of n-butyl oligostyrene oligomers with n = 2 to n = 5 styrene repeating units.

C

H

H

CH2

C H

CH2

C H

CH2

C H

CH2

C H

CH2

CH2CH2CH2CH3

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

C H

CH2

CH

CH2

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

C H

CH2

CH

CH2

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

CH

CH2

CH

CH2

C H

H

C

CH2

H

CH2

CH

CH2

CH2CH2CH2CH3

CH

CH2

CH

CH2

C H

H

C

CH2

H

CH2

CH

CH2

CH2CH2CH2CH3

C H

CH2

CH

CH2

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

CH

CH2

C H

CH2

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

C H

CH2

C H

CH2

C H

H

C H

H

CH2

C H

CH2

CH

CH2

CH

CH2

CH2CH2CH2CH3

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

CH

CH2

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

C H

CH2

C H

H

C

CH2

H

CH2

CH

CH2

CH2CH2CH2CH3

CH

CH2

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

C H

H

C

CH2

H

CH2

C H

CH2

CH2CH2CH2CH3

C H

H

C

H

H

CH2

C H

CH2

CH2CH2CH2CH3

n = 2 n = 3 n = 4 n = 5

(a)

(b)

15

Irrespective of how the sample is described, the key to separation is then choosing a system so

that each sample attribute chromatographically. If the sample attribute cannot be

chromatographically expressed, then essentially that sample attribute does not exist. For

example, enantiomers co-elute in most chromatographic environments, except those that are

chiral, and even then, the chiral environment must be sensitive to the sample. Therefore, in an

achiral environment, this sample dimension does not exist, and will not exist unless a chiral

environment is included in the separation process.

1.3.2 Selective and non-selective displacements

Displacements in two-dimensional column HPLC are sequential; that is the first dimensional

separation must be followed by the analysis of discrete sections of the first dimension in the

second dimension. An important example of sequential displacements are discrete displacements

in which a small discrete sample is applied to a corner of the two-dimensional separation plane,

akin to an injection onto the first dimension. Consequently separation occurs along each resulting

axis, creating discrete elliptical zones [56].

There are two types of one-dimensional displacements fundamental to two-dimensional

displacements [56]: Selective (S) and Non-selective (N). Selective displacements occur when

separation of the sample components occur in each subsequent phase of the multidimensional

system, that is the selectivity factors (α) observed for the sample components are greater than 1

(Figure 1.2 a-b). Selective displacements are separative displacements. Non-selective

displacements result in no separation where α = 1. These S and N displacements can be

16

combined in a number of ways (Table 1.1) corresponding to the displacement along each axis of

the two-dimensional plane [56] .

Table 1.1: Types of discrete displacement combinations and their effect on two-dimensional peak capacity.

From reference [57]

Maximum separation in the two-dimensional plane occurs when each dimension offers selective

displacement, particularly when the separation mechanisms are totally independent. Figure 1.2a

illustrates an S SI (I = independent) separation where all components have successfully been

separated [56]. So although components may co-elute in a one-dimensional system the

components can be well resolved in two-dimensions. With retention mechanisms that are

correlated (S Sc) the availability of separation space is limited and the separation will be

equivalent to a one-dimensional separation. This is evident by the alignment of data along the

main diagonal as shown in Figure 1.2b [56]. With partial correlation between the dimensions the

available space in the second dimension is reduced and the separated components also align

together although less evident than in Figure 1.2b. As a result the resolution and peak capacity

between sample components would therefore decline. Discrete N S displacements offer no gain

in the selectivity factor in the first dimension as no components are separated in the first

dimension. The column of separated components in the second dimension appear at an identical

17

separation time in the first dimension for discrete N S displacements (Figure 1.1c). The

effectiveness of these processes is shown in Table 1.1.

Figure 1.2: Illustration of the combinations of discrete selective (S) and non-selective (N) displacements: (a) Two-dimensional S × SI displacement; (b) Two-dimensional S × Sc displacement; (c) Two-dimensional N × S displacement [56, 57].

In many instances, a multidimensional sample may experience both S and N displacement within

a single chromatographic environment: the retention behaviour being dominated by a single

sample attribute. Thus the sample essentially migrates as if it contained only a single sample

attribute, or the sample dimensionality equals 1. While Table 1.1 predicts that an N S

18

displacement will not lead to an increase in separation power, it does not predict that a non-

selective displacement with respect to say one of three sample attributes can in fact lead to an

improved separation process because the order of component elution can be more predictable,

and this is very important in the separation of complex samples.

For example consider the two-dimensional separation of a mixture of 58 low molecular weight

oligostyrenes [51]. S displacement was observed for the diastereomers belonging to the

oligostyrenes of varying molecular weights, and also within these groups oligostyrene selectivity

differences were observed between oligostyrenes with different end groups, either tert-, sec-, or

n-butyl. From plotting the normalised retention data for both dimensions (Figure 1.3), distinct

columns were evident for the diastereomers indicating a N displacement in the first dimension

for the diastereomers resolved in the second dimension. Essentially the first dimension was

capable of separating according to two sample attributes: that of molecular weight and that of

end group and the second dimension able to resolve according to stereoselectivity, with minor

end group selectivity [51]. Thus the molecular weight dimensionality of the sample was

expressed only in one dimension. Hence the concept of the two-dimensional system operating as

HPLC×HPLC, whereby the second dimension was the selective diastereomer analyser [51].

Alcohol ethoxylates provide another good example of retention related to sample dimensionality.

These polymer compounds have both distributions in ethylene oxide units and also in the length

of hydrophobic (alkyl) end group [58]. In the separation of Neodol 25-12, Murphy et al. [58]

demonstrated that S displacement on a normal phase first dimension occurred based upon the

distribution of ethylene oxide while a N displacement occurred in the RP second dimension

19

based upon the length of hydrophobic alkyl chains within the alcohol ethoxylates, which are

resolved in the RP second dimension. It was therefore necessary in both instances to combine

selective and non-selective displacements so as to deter chaotic two-dimensional component

separation.

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

n = 5

n = 4n = 3

n = 22

]2]2

24

CC

Z M

eOH

C18 MeOH

Figure 1.3: Normalised 2D plot of the C18(MeOH)/CCZ (ACN) system in the separation of the 58 oligostyrene isomer mix. Each boxed section represents isomeric components containing the same number of configurational repeat units. The numbers adjoining the data points indicate the number of components co-eluting [51].

20

1.4 Two-Dimensional Chromatographic Systems

1.4.1 Comprehensive and heart-cutting separations

Depending upon the goal of the analysis, two-dimensional separations can be carried out using

either a heart-cutting process or comprehensively [34]. Either of these techniques could feasibly

be employed in a screening process, where the goal of the separation is perhaps to search for the

appearance of certain chemicals and less interest is paid to compounds of no relevance to the

analyst. In such a situation the 2D separation could be fine tuned to target those particular

compounds, sacrificing total peak capacity, but maximising resolution in the region that is most

important.

The process of heart cutting involves the transport of a discrete area of interest from the first

dimension to the second dimension for further separation. This may even involve several heart-

cut fractions from the first dimension being transported to the second dimension [34]. The

advantage of a targeted heart cutting approach is that only the components of interest are

specifically analysed, be they predetermined contaminants whose analysis is dictated by

regulatory authorities, or components within the sample whose presence largely describes the

quality of the sample. Applied in a targeted approach, heart cutting is useful for improving the

resolution of components by simplifying the matrix, as only the bands of interest are cut from the

first dimension and transported to the second dimension; but since not all the sample is analysed

the speed in analysis is somewhat faster. A limitation of this technique is, however, that some

previous knowledge of the components of interest may be required to ascertain the area(s) to be

heart-cut and may require some additional pre-work. Hence this form of two-dimensional HPLC

21

is very useful for continuous screening of samples, or for the targeted isolation of specific

components.

Comprehensive chromatography involves the transfer of the entire first dimension to the second

dimension for further separation [34]. However, the transportation of the entire sample from the

first dimension to the second has many disadvantages, the most significant of which is the

physical limitations associated with undertaking the second dimension separation within a time

frame appropriate for the first dimension. The second dimension must therefore be fast in order

to avoid wrap around effects where solutes from the later cuts from the first dimension becomes

mixed with the analytes from previous cuts in the second dimension. This results in chaotic band

displacement, and potential co-elution of compounds that were previously separated in the first

dimension – negating the power of the 2D system. In order to gain speed in the second

dimension, peak capacity is often sacrificed, with the affect of these being less discrimination in

the fingerprinting result. As such there is a delicate balance that may be played between how

many peaks can be separated and how many peaks need to be separated in order to show a

chemical signature match. Is a higher resolution 2D separation of fewer components better than a

lower resolution 2D separation containing more components, or visa versa? In the case of

fingerprinting, however, comprehensive separations are considered more suitable as the entire

sample is generally subjected to two-dimensional analysis, unless the analytical result can be

substantiated with less information, in which case a heart cutting approach could be feasible. A

way of compromise is to employ a modified version of the comprehensive approach – an off-line

comprehensive 2DHPLC. In this technique sample fractions from the first dimension are

collected, stored and then when convenient, run in the second dimension. In this mode of

22

operation, the second dimension can have a high peak capacity as there is no time limitation

associated with the analysis. However, the drawback is that sample fractions collected from the

first dimension may need to be re-concentrated prior to injection into the second dimension.

There is also the risk of labile compounds degrading while waiting for analysis in the second

dimension, a problem commonly encountered in natural product research. Furthermore, the off-

line approach is slow, but here speed is sacrificed for peak capacity.

A variation in the off-line approach mentioned above, is the comprehensive or incremental heart

cutting approach. Whelan et al. [59] successfully separated carboxylic acids from humic

substances in Bayer liquor. In this technique, the sample was injected in the first dimension; a

small aliquot from the first dimension was transported to the second dimension via appropriate

switching valves, and then analysis takes place in the second dimension. Once separation in both

dimensions was complete, another aliquot of sample was loaded into the first dimension, and this

time a different fraction was heart-cut to the second dimension, and the process repeated.

Depending on how many samples are injected into the first dimension and how small the aliquot

sampled to the second dimension is; and how much of the first dimension is actually sampled –

will determine the quality of the chemical signature. An advantage of this process, however, is

that very high theoretical peak capacities can be obtained since there is no speed limitation in the

second dimension, other than that dictated by the patience of the operator.

A disadvantage of the comprehensive approach, either on-line or off-line is the creation of the

enormous amount of information that is collected, and hence must be analysed because many

components, not just those of interest in the sample, would be resolved. This creates more

23

complicated chromatograms as large amounts of data need to be converted from the recorded

data acquisitions, which may prove time consuming and problematic and in our modern world of

information collection, it is sometimes the analysis of the data that proves to be the limiting

factor.

1.4.2 Two-dimensional System Designs

The use of an automated on-line two-dimensional system eliminates the need for manual sample

handling such as fraction collection or re-injection of sample. However, some consideration to

the system design and experimental objectives are required. The most common way of

interfacing columns for two-dimensional HPLC systems is using 4-, 6-, 8- or 10- port, two

position automated switching valves [57, 60-65]. The switching valves essentially allow the

dimensions to operate independently from one another without loss of the resolution achieved in

the first dimension. Two-dimensional configurations generally incorporate either two sample

loops and switching valves; two sample traps and switching valves; or switching valves with a

dual or quad column configuration in the second dimension. The use of sample loops allows

eluent to be collected from the first dimension while eluent held on an additional loop is loaded

on the second dimensional column. This process is controlled by the precise timing of the

switching valves and is generally computer controlled on-line. Almost any HPLC system can be

converted to a two-dimensional system through the addition of switching valves.

One of the earliest 2D HPLC systems was developed by Erni and Frei [60] where two loops were

connected to an eight-port switching valve. While one loop was being filled from the first

24

dimension, the second loop was being loaded onto the second dimensional column, alternating

with each valve switch. Many systems have been developed based on this pioneering concept,

with adaptations of this design by Bushey and Jorgenson [61] being used for comprehensive 2D

separations [62]. This design may be problematic, because of differences in the retention times of

certain components, particularly when loop sizes become large as the sample is forward flushed

on one loop and reversed flushed onto the other loop [63]. Ten-port switching valves with two

loops [63, 64] and six-port switching valves are also used for comprehensive 2D systems [65].

This allows for continuous collection and re-injection of the first dimensional eluent. A 2D

system comprised of four six-port switching valves that connected the two dimensions was

developed for the separation of oligostyrenes [49]. By reducing the six-port switching valves to

two this system is capable of operating under heart cutting conditions.

1.4.3 Data Collection

Another important factor relates to how data is collected and then subsequently represented as

2D chromatographic information. If a heart cutting or „semi-comprehensive‟ approach is

employed then this is achieved through detection of sample in each dimension and the data is

collected and either represented as conventional uni-dimensional chromatograms or transferred

to a spreadsheet and converted to a contour plot or a 2D plot for visual interpretation, depending

on how many „cuts‟ are analysed in the second dimension. When a comprehensive approach is

employed it is necessary only to detect the output from the second dimension. A typical output is

intensity as a function of frequency data set. This information is collected in a continuous output

over the entire duration of the 2D separation. At the end of the separation the unidimensional

data stream is converted to a matrix according to the frequency of sample modulating from the

first to the second dimension. This information is then presented as a contour or surface plot.

25

Some programs not commercially available are also able to measure the degree of orthogonality

between dimensions and also offer enhanced baseline discrimination of low level components

[66].

1.5 Applications of 2D HPLC

1.5.1 Pharmaceutical

The identification and detection of impurities in pharmaceutical ingredients and their chemical

processes requires the use of stringent analytical techniques that comply with regulatory

requirements. As pharmaceutical ingredients may contain related structural impurities, increasing

the need for highly selective methods of analysis are actively sought. Sheldon et al. [67]

incorporated the use of 2D HPLC coupled to MS for further detection and method development

of the chromatographic system for the analysis of pharmaceutical compounds. Xue et al. [68]

developed a fully automated comprehensive orthogonal method evaluation technology

(COMET) system employing orthogonal HPLC separations and hyphenated UV-MS detection

for impurities in pharmaceutical drugs. The system was capable of tracking each impurity over

all chromatograms of a drug sample recorded under different chromatographic conditions by

automatically assigned molecular weights. The retention data was normalised and a radar plot

constructed where each axis represented each COMET method. Crossover of the data

represented changes in elution order between neighbouring methods and is indicative of the

systems orthogonality. Initial testing of the automated peak tracking yielded 80% success rate for

26

over 500 drug impurities, although some impurities were not detected due to the low ionization

efficiencies.

1.5.2 Natural Products

Proteomic analysis is quite challenging and recently 2D HPLC methods have proved useful.

Proteins and peptides may be separated according to their physical properties such as isoelectric

point (pI) and comparative molecular mass, size, charge and hydrophobicity. 2D HPLC for

proteomic analysis usually couple strong cation exchange (SCX) with RP HPLC, size-exclusion

(SEC) and RP [69] and more recently RP-RP [45, 70]. Dobrev and co-workers [70] developed a

2D HPLC system for the determination of phytohormones, in this instance auxin (IAA) and

abscisic acid (ABA). The 2D system was developed to provide another option for the pre-

existing purification steps to eliminate the co-elution of IAA and ABA and also to reduce the

high amounts of UV-absorbing and fluorescing contaminants. There was also no hormone loss

due to irreversible binding and recovery yields were 95%. The purification potential of this

system was then tested on plant extracts of developing wheat grains at different days after

anthesis. The 2D system provided a means of purification and a reliable quantitative method

comparative to that of GC-MS, although having a longer analysis time, due to the gradient

conditions.

Blahova and co-workers [71] employed a comprehensive 2D HPLC system to for the separation

of phenolic antioxidants. The total analysis time was 80 minutes and eleven compounds were

successfully resolved of the seventeen compounds known to exist in the phenolic antioxidant

standard. Using a stop-flow method the first dimension, while the second dimension continues

27

with its separation, is referred to as „peak parking' as demonstrated by Kohne and co-workers

[72, 73] and in this instance there was no evidence of adverse band broadening effects. They then

applied this serially coupled system to beer samples and hop extracts for the analysis of phenolic

antioxidants. The low concentration of the antioxidants, however, precluded the use of the UV

detector, so coulometric detection was employed. The downside of this, however, was that a

gradient elution approach could not be taken as the dual-cell detector that was employed did not

permit gradient elution and therefore only an isocratic approach using the serially coupled

columns only could be employed. In general the serially connected columns offered an

improvement in the overall separation as each column had an independent separation

mechanism. Nevertheless serially coupling may generate co-elution of peaks previously resolved

on the first column after transport to the second column [74]; however this did not appear to be

an issue with this separation.

Cacciola and co-workers [75] investigated the use of four different comprehensive 2D HPLC

systems for the separation of phenolic antioxidants. The first dimensional column was a PEG-

silica column and in the second dimension three different columns were used, a Chromolith

SpeedRod, a Discovery Zr-CARBON column and an ACE 3 C18 column. A ten-port switching

valve coupled the two dimensions and transport of fractions was facilitated either by sample

loooligostyrene or trapping columns dependent upon the system utilised. Eighteen of twenty

compounds could be separated using this system, only the peaks of ferulic acid and vanillic acids

overlapped in the second dimension. For the separation of the pilot beer sample, six compounds

were successfully identified based on their retention times and UV spectra. This system was also

successfully applied for the analysis of hop, beer and tea samples.

28

1.5.3 Traditional Chinese Medicines

Herbal medicines and traditional Chinese medicines (TCM) are complex samples comprising of

many components, and in mixtures that contain multiple herbs considerable complexity is evident.

Chemical fingerprinting is customarily used for the analysis and identification of herbal medicines

and is a useful method for the identification and validation of components in these mixtures and as

such is also an effective method for quality control. Chromatographic methods such as HPLC are

often used with highly efficient detection techniques such as DAD [76] and MS. Currently there is

no universal methodology for the analysis of the different herbal and traditional medicines,

however, 2D HPLC has shown promise as an emerging technology with many examples [77, 78]

of high resolution separations of components that were once unable to be resolved from these

complex samples.

Ma et al. [79] used the combination of a size exclusion column (SEC) and a RP C18 column in a

comprehensive 2D HPLC mode to separate Qingkailing, a treatment derived from TCM

composed of eight materials or their extracts. Qingkailing is a very complex sample that may

contain many constituents. The 2D HPLC design incorporated the use of a second autosampler

that injected the collected samples transported from the first dimension via a switching valve, to

the second dimensional column. Hence, strictly speaking this was not an example of a coupled

column system, but rather represents a modern approach of the old school technique involving

multiple phase separations followed by fraction collection then re-injection. Following the

application of the 2D separation, the expanded separation space yielded a separation with much

greater resolution. Further detection for peak identification was performed using ion trap MS

analysis. This 2D system enabled the researchers to successfully separate 54 components of

29

interest that was not possible in a 1D RP system where only 30 peaks were evident. Although the

analysis time was quite long at 18.5 hours, the peak capacity in this 2D system was increased

allowing for greater qualitative analysis of complex TCM chemical constituents. Chen et al. [77]

utilised a comprehensive 2D HPLC system coupling a cyano (CN) and C18 column for the

separation of components in a commonly used the TCM Rhizoma chuanxiong. The second

dimension column was connected to a DAD, which was connected directly to APCI-MS.

Following the application of the 2D separation further detection for peak identification was

performed using ion trap MS analysis. The 2D system used enabled the researchers to

successfully separate 11 components of interest in less than 215 minutes. Continuing on from the

previous study, in an effort to improve the conditions for the separation of components in TCM,

Hu et al. [78] utilised a 2D HPLC system for the separation of the components in two TCMs, R.

chuanxiong and Angelica sinensis. The effectiveness of the monolithic column as a second

dimensional column, able to be operated at high flow rates, is represented by the shorter analysis

time of 5 minutes compared to 20 minutes for the CN column used in the first dimension. The

sample from the first dimension was transported to the second dimension and then transported

for detection by diode array and MS. Normalised peaks heights were used which was achieved

by reducing the highest peak by one-sixth to allow for the detection of peaks. This method of

normalisation was developed so that low-abundant components can be determined more easily.

Approximately 120 components were separated in R. chuanxiong using this system. This 2D

system was then applied for the separation of Angelica sinensis where approximately 100

components were successfully separated. Also the number of components detected in R.

chuanxiong almost doubled presenting a system that can provide a fast and powerful separation

system for these complex mixtures.

30

More recently Click dipeptide/C18 RP/RP 2D HPLC systems have shown promise for the

separation of TCM‟s with orthogonality studies carried out for aromatic compounds [80] and for

R. Palmatum L., a complex TCM sample [81]. Recently a NP/RP 2D system detected 876 peaks

for a complex TCM sample, Zhenghan pill, reaching a peak capacity of 1740 [82].

1.5.4 Forensic applications

The identification and accurate association of oil spills and petroleum products to their source

has become increasingly important in recent years. This is not only significant in the assessment

of environmental damage but for settling questions of liability and for the application to the field

of forensic science as circumstantial evidence. The identification and source of a sample is made

difficult due to the complex nature of crude oil and its refined petroleum products.

Many methods have been developed for the identification of oil spill sources including chemical

fingerprinting, analysis of source-specific marker compounds and determination and comparison

of diagnostic ratios [83, 84]. Because PAHs are found in environmental and forensic samples as

extremely complex mixtures they may require analysis by the use of selective detection and/or

use of multidimensional HPLC techniques to accurately quantify individual PAHs [85].

Murahashi et al. [86] used comprehensive 2D HPLC to separate PAHs in gasoline and gasoline

exhaust. This technique could achieve both separation and identification and was proposed as a

technique for the separation of PAHs in environmental samples [86]. It was shown that the

coupling of two dimensions provided a substantial amount of information when compared to

single dimensional HPLC [86]. Goodpaster et al. [87] concluded that the quantitative and

31

qualitative determination of PAHs allow for motor oil to be profiled allowing for the unique

identification of motor oil from a particular vehicle or engine [87].

2D HPLC may also have applications for the analysis of gasoline in criminal investigation of

arson and petroleum products found at crime scenes or in oil spills. The analysis of petroleum

products can also provide circumstantial links in criminal investigations where chemical

fingerprints may prove useful in providing a positive association between motor oil and a

suspect‟s vehicle, petroleum products and a suspected arson scene, and petroleum based lubricants

and sexual assault [87]. Human fingerprints can be used for identification purposes and to place a

person at the scene of a crime [88]. Similarly, chemical fingerprints can link petroleum products

found at the scene of a crime to a particular suspect. This is useful in criminal investigations of

arson where petroleum products are used as accelerants and in motor vehicle accidents where

motor oil may come into contact with a victim.

1.6 Preparative Chromatography

Chromatography is a powerful separation technique that was initially developed for the isolation

of natural components in a highly purified form from complex mixtures. However, early

preparative applications of chromatography were quickly surpassed by analytical separations, as

the need for qualitative and quantitative information about components present in simple and

complex mixtures became the primary objective. Preparative HPLC has only seen resurgence in

the last few decades with traditional methods such as distillation, centrifugal extraction and

32

crystallisation unsuitable for the problems encountered by the various industries. The stringent

regulations of governing bodies for the approval of highly purified products to be released into a

highly competitive market dictate largely the advancement of chromatographic methods for

preparative separations.

The distinction between preparative and analytical chromatography lies in the purpose of the

separation; analytical chromatography is generally aimed at identification and quantitative

information whereas preparative chromatography is essentially aimed at isolating an amount of

purified material for other purposes. These purified materials may be needed as standards,

synthesis intermediates and for screening purposes to name a few. Preparative chromatography is

generally carried out by two processes; either as a continuous sample feed or by batch process.

Simulated moving bed (SMB) is an example of an effective large-scale continuous

chromatographic process particularly suited to the separation of binary mixtures where the

chromatogram is effectively split into two halves. It has been utilised successfully in the

petrochemical [89] and sugar industries [90] and more recently for the separation and

purification of chiral molecules [91], antibodies [92] and nucleosides [93]. Annular

chromatography is also another continuous chromatographic process [94, 95, 96], however, both

SMB and annular techniques are complex processes limited in their use to simple sample

matrices. For more complex samples batch type processes are more frequently utilised. Batch

chromatography essentially can be operated in two modes; displacement and elution (either

isocratic or gradient). The simplest mode is that of isocratic elution where the mobile phase has

constant composition and is suited for a simple sample matrix. Gradient elution is more suitable

for complex mixtures as the incremental increase in solvent strength improves the peak capacity

33

and therefore allows for the separation of a greater number of components. In displacement

chromatography a strong adsorbing component is applied to the column after the sample has

been injected thus displacing the feed component in an isotachic train. Both gradient elution and

displacement mode require regeneration of the columns between runs therefore increasing the

time of the separation, this can negatively influence the production and in general isocratic

elution is the preferred mode.

Preparative isolations can be performed in a non-overloaded manner, but preference is given to

overloaded or non-linear chromatography since sample yields are higher. For preparative

chromatography rarely is it necessary to isolate all the components in the sample instead the aim

is usually focused on the maximising recovery of targeted compounds. Therefore multistage or

multidimensional chromatographic separation steps are employed in a manner whereby the first

dimension effectively reduces the complexity of the sample matrix, while the next and

subsequent dimensions aims to resolve the target compound from the less complicated sample

matrix that was transferred to the second or subsequent dimensions. In on-line 2D preparative

HPLC (PHPLC) there is often a high degree of correlation between each dimension, at least with

respect to the solvent flow stream, as this facilitates speed in separation and simplicity in

operation, avoiding effects such as viscous fingering and other solvent mismatch phenomena.

The 2D PHPLC system may even be reduced to a column switching process.

Wong et al. [97] employed a 2D RP HPLC system, operated in a heart-cutting mode for the

isolation of two bioactive components of interest from an Australian native plant, Clerodendrum

floribundum (CF). The complexity of the crude extract of CF is illustrated by the 1D gradient

34

elution on a C18 column in Figure 1.4, where the active components are labelled CF1 and CF2.

The long analysis time required bringing about the elution of strongly retained components, plus

the additional re-equilibration essentially meant that using this type of 1D separation approach

would limit the effectiveness of the separation for the purpose of sample purification. The

separation of the crude extract of CF on a CN column yielded poor separation, using either

isocratic (Figure 1.5) or gradient elution.

Figure 1.4: 1D chromatogram of crude extract of C. floribundum. Column: CN (150 mm x 4.6 mm, 5 m). Mobile phase: gradient elution 95% water/5% MeOH - 100 % MeOH over 18 minutes. F = 1.0 mL/min, injection volume 20 L. Detection UV 270 nm. [ref 97]

Due to the 1D limitations the authors employed 2D heart-cutting HPLC for the isolation of the

bioactives in a crude extract of CF. The column in the first dimension was a CN column and the

column in the second dimension was a C18 column. Both separation dimensions were operated

35

isocratically, which greatly reduced total separation time. Heart-cut fractions of 200 µL were

transported from the first dimension to the second dimension through the use of a two six-port

valves and a sample loop. The second dimensional chromatograms are shown in Figure 1.6 and

Figure 1.7, with the peaks labelled a-e illustrating the corresponding change in sample recovery

associated with the location of the cut fraction from the first dimension.

Figure 1.5: 1D chromatogram of crude extract of C. floribundum. Column: Luna CN (150 mm x 4.6 mm, 5 m). Mobile phase: isocratic elution of 45% water/60% ACN. F = 1.0 mL/min, injection volume 10 L. Detection UV 270 nm. [ref. 97]

Figure 1.6 is a larger heart-cut fraction from the first dimension to the second dimension at 200

L and Figure 1.7 is a 100 L heart-cut. At the higher heart-cut volume of 200 L the peak

shapes were somewhat broadened so a smaller analytical scale injection volume of 100 L was

investigated which provided improved peak shape. However as the aim was to eventually scale-

up the isolation of the bioactive components in CF using an overloaded method the broadening

of peaks was deemed acceptable; particularly since the recoveries decreased in the analytical

mode to 38% and 46% for CF1 and CF2 respectively.

36

Figure 1.6: Chromatograms in the second dimension illustrate the relative change in the concentrations of CF1 and CF2 following heart cutting from the first dimension. Experimental conditions: 1D: Column: Luna CN (150 mm x 4.6 mm, 5 m). Mobile phase, 40% water/60% ACN. 2D: Column: ValuePak C18 (250 mm x 4.6 mm, 5 m). Mobile phase, 55% water/45% ACN. F = 1.0 mL/min, injection volume 10 L. Heart-cut volume 200 L. [ref. 97]

Figure 1.7: Chromatograms in the second dimension illustrate the relative change in the concentrations of CF1 and CF2 following heart cutting from the first dimension. Experimental conditions: 1D: Column: Luna CN (150 mm x 4.6 mm, 5 m). Mobile phase, 40% water/60% ACN. 2D: Column: ValuePak C18 (250 mm x 4.6 mm, 5 m). Mobile phase, 60% water/40% ACN. F = 1.0 mL/min, injection volume 10 L. Heart cut-volume 100 L. [ref. 97]

37

The analytical version of the 2D heart cutting system developed by Wong et al. [53] was then

scaled to overload conditions for the isolation of the bioactive component (Component A) in the

crude extract of CF [45]. The 2D system comprised a CN column in the first dimension and a

C18 column in the second dimension. The separation of the crude extract of CF on the CN

column in the first dimension is illustrated in Figure 1.8, with the target component highlighted

as component A. This column was coupled via two 2-position, 6-port switching valves and

sample isolation loop to a semi-preparative C18 column. Injection volumes were 200 L and the

heart-cut volumes were varied between 1.2 mL and 2.0 mL. This resulted in recoveries between

75% and 100 % respectively depending on the desired degree of purity. For example, the product

purity for the 2.0 ml band cut resulted in an unacceptably low purity whereas for the 1.2 mL

band cut the purity was in excess of 99%.

Figure 1.8: 1D chromatogram of crude extract of C.floribundum. Column: Luna CN (150 mm x 4.6 mm, 5 m). Mobile phase: gradient elution of 95% MeOH/5% water- 100 % MeOH in 18 minutes. F = 1.0 mL/min, injection volume 20 L. Detection UV 270 nm. [ref. 53]

38

The aim of their study was to illustrate how the production rate could be enhanced using the 2D

system in comparison to a 1D system for these types of complex samples. The production rate

(Pri) of component i, describes the amount of a purified component that is turned into product,

per unit column cross–section area, per unit time, in mg cm-2 s-1. The production rate for a

preparative scale separation can be manipulated according to the variables described by equation

1.1 [98].

ca

iis

r tSYCV

Pi

0

(1.1)

Where Vs is the sample volume, C0i is the injected concentration of the sample constituent, ε is

the total porosity ,Sa is the cross sectional surface area ,Yi is the recovery yield and tc is the

cycle time.

The production rate as applied to a 2D HPLC system is partly limited by the overload conditions

on the column in the first dimension, and in order to maximise recovery the 2D column should

have a larger internal diameter than the 1D column [53]. However, gains in production rate can

be made by full utilisation of the dead time within the system. That is because isocratic mobile

phases can be employed in both dimensions for even very complex samples as there is no re-

equilibration time. Hence injections can be made more frequently in the first dimension to

correspond with the period of time the second dimension is waiting for sample to be cut from the

first dimension. An example is shown in Figure 1.9, which illustrates how the second dimension

39

is fully utilised with little wasted separation time. Hence the cycle time was drastically reduced

to 7.3 minutes compared to 22 minutes in a 1D gradient system [53].

Figure 1.9: 2D chromatogram of target component of crude extract of C.floribundum under overload conditions. Illustration of the cycle time and the maximisation of separation space. (a) 1D separation: Column: CN (150 mm x 4.6 mm, 5 m). Mobile phase: water-ACN (30:70). F = 1.0 mL/min, injection volume 200 L. (b) 2D separation: Column: C18 (250 mm x 10.0 mm, 5 m). Mobile phase: water-ACN (40:60). F = 1.0 mL/min, heart-cut volume 200 L. Detection UV 270 nm. [ref. 53]

Furthermore, the data in Table 1.2 details how more efficiently the 2D system was able to isolate

the desired product at higher purity than the 1D gradient elution RPLC systems. The overloaded

1D separation had a recovery yield of 95% for a purity of 98% and the final production rate was

40

0.085 g min -1 cm -2, which was 6.8 times less than the comparative 2D system for the same level

of purity. When the desired purity was increased to 99% the recovery yield of the 1D system was

substantially reduced, which further decreased the production rate of the 1D system in

comparison to that of the 2D system.

Table 1.2 Characteristics of multidimensional and gradient elution separations of component A from Clerodendrum floribundum

From reference [53].

1.7 Objectives

The objectives of this current work were to develop and evaluate a 2D HPLC system for the

separation and consequent isolation of a targeted species from a complex sample. An

41

oligostyrene sample was chosen for this study as it is extremely complex having many hundreds

of components and also because its sample dimensionality is well characterised. The first part of

this study investigated the capabilities of a 2D HPLC system for high resolution separations of

this complex sample. A heart cutting approach was used to reduce the overall wrap-around effect

and also because the focus was on separating targeted components only. The second part of this

study again involved targeted separation; however, the focus was on investigating the

experimental parameters that affect the purity and recovery of an isolated component from a

complex mixture. Finally the third part looked at the scale-up of the sample at a preparative

scale.

42

Chapter 2

General Methods

43

2.1 Chemicals

HPLC grade MeOH, ACN, chloroform and DCM were obtained from Lomb Scientific,

Australia. Oligostyrene standard (molecular mass = 114,200 Da) was purchased from Polymer

Laboratories. tert-Butyl oligostyrene (molecular mass ~580 Da) was synthesised using anionic

polymerisation of styrene initiated with tert-butyl lithium.

Stationary phase used for the preparation of chromatography columns were (1) Carbon clad

zirconia (Zirchrom-CARB- 3 m dp) purchased from ZirChrom Separations, Inc., Anoka, MN,

USA and (2) Nucleosil C18 (10 m dp) was obtained from Alltech Associates Pty. Ltd.,

Baulkham Hills, NSW, Australia).

2.2 Chromatography columns

The details of the chromatography columns, utilised throughout this study, which includes

column format, particle diameter and supplier, and the relevant Chapters in which they were

employed are listed in Table 2.1.

44

Table 2.1: Chromatographic columns used in this study I.D Column type Supplier Particle

diameter (m) Dimensions

(mm) Chapter location

1 C18 Pursuit XRs

Varian 10 250 × 4.6 3

2 Carbon clad zirconiaa

ZirChrom Separations

3 30 × 4.6 3



3 50 × 4.6 3, 4, 5, 6



3 100 × 10.0 3

5 Hypercarb ZirChrom 5 100 × 4.6 3 6 Nucleosil C18a Phenomenex 10 50 × 4.6 4, 5, 6, 7 C18

Sphereclone Phenomenex 5 150 × 4.6 6

a Packed in-house with stationary phase supplied by manufacturer as detailed in section 2.2.2

2.2.1 Chromatography column packing conditions

Carbon clad zirconia (CCZ) columns were packed in-house using a downward slurry packing

technique in which 7.5 g of stationary phase was slurried in 70 % isopropanol and 30 % hexane

(35 mL). The slurry was stirred for 30 minutes followed by 20 minutes of ultrasonification then a

further 10 minutes of stirring. A DCM displacement solvent was employed in column assembly

(amount, length and diameter of sections dependent upon column requirements) and the column

was packed at 7500 p.s.i using isopropanol packing solvent. Packing continued until 45 mL of

isopropanol passed through the bed. The column assembly was then end-capped and after one

hour the middle sections were removed and the columns were constructed.

Nucleosil C18 columns were prepared in-house using a similar slurry packing technique as

mentioned above, but employing different solvents. For the Nucleosil columns 4.0 g of stationary

phase was slurried in 30 mL of acetone. The slurry was stirred for 30 minutes followed by 20

45

minutes of ultrasonification then a further 10 minutes of stirring. A DCM displacement solvent

was employed in column assembly (5 × 50 mm sections) and the column was packed at 7000

p.s.i using MeOH packing solvent. Packing continued until 200 mL of MeOH passed through the

bed. The column assembly was then end-capped and after one hour the three middle sections

were removed and the columns were constructed.

2.3 Equipment

The chromatographic separations were performed on a Waters LC system that incorporated two

600 controllers, 717plus autosampler, two 2487 dual wavelength UV detectors and Millenium32

Version 4.00 software running on a Compaq EVO D500 Pentium 4 1.6 GHz personal computer

with 256 Mb RAM (Waters Associates). Column switching was achieved using two 6-port, 2-

position switching valves fitted with micro-electric two position valve actuators (Valco

Instruments, Houston, TX, USA) controlled via the onboard Millenium32 software. The two-

dimensional HPLC system utilised in this study is illustrated in Figure 2.1.

46

Figure 2.1: Diagram of the 2D HPLC system. P1-P2: Pumoligostyrene that deliver solvents; V1-V4: six-port two-position switching valves; C1: 1D column; C2:2D column. D1-D2 detectors. a) Figure 2.1: Diagram of the 2D HPLC system. P1-P2: Pumps that deliver solvents; V1-V4: six-port two-position switching valves; C1: 1D column; C2:2D column. D1-D2 detectors. a) System configuration for elution on C1 and C2 and b) system configuration for elution of a band from C1 onto sample loop.

2.4 Chromatographic separations

tert-Butyl oligostyrene was dissolved in MeOH:chloroform (80:20) at a concentration indicated

in text. Mobile phases were sparged continually with helium. The column in the second

dimension was thermostatted as indicated in text using Braun Thermomix® M circulation

thermostat (B. Braun, Melsungen, Germany). UV detection was 272 nm for oligostyrene analysis

and 262 nm for isomer analysis. Specific separation details are listed in the relevant chapters.

47

Chapter 3

Ultra High Resolution Separations of

Diastereomers on Carbon Adsorption

Stationary Phases

48

C C

H H

H

n

3.1 Introduction

The need for increased resolving power, driven by the demands of industry has been in part the

driving force behind the development of multidimensional HPLC. By judicious selection of the

various separation steps with consideration to the nature of the sample, the separation can be

tuned to the various sample attributes accordingly. Ultimately this type of separation process can

lead to very high levels of selectivity and hence the probability of component overlap in the two-

dimensional domain decreases.

Of interest in this study is the separation of diastereomers of low molecular weight oligostyrenes

with the tert-butyl end-group. The styrene configurational repeating unit is shown in Figure 3.1.

The sample dimensionality was previously described in section 1.31, and consists of variations in

molecular weight, with a corresponding increase in the number of diastereomers, at a rate of 2(n-2)

where n = the number of repeat units.

Figure 3.1: Diagram of the configurational repeating unit of styrene.

49

Figure 3.2 illustrates the n =5 oligostyrene of five configurational repeat units and a tert-butyl

end group; eight diastereomers are possible, since there are four stereochemical carbon atoms on

the oligostyrene backbone. These diastereomers have the orientation; one isotactic, one

syndiotactic and six atactic. The spatial orientation of the molecule thus allows separation to

occur when the chromatographic environment is such that it offers shape sensitivity. Table 3.1

lists the number of diastereomers for each of the oligostyrenes.

Figure 3.2: n = 5 oligostyrene with tert-butyl end group, five configurational repeat units and atactic stereochemistry.

C

H

H

CH2

C

CH2

H

C

CH2

H

C

CH2

H

C

C

H

CH3H3C

CH3 End-group

Number of configurational repeat units

Stereochemistry- Spatial orientation

50

Table 3.1: Number of diastereomers for oligostyrene with tert-butyl end group Number of configurational repeat units

(n)

Sites of Stereochemistry Number of Diastereomers

2 1 1

3 2 2

4 3 4

5 4 8

6 5 16

7 6 32

8 7 64

9 8 128

10 9 256

3.1.1 Selectivity of C18 and carbon clad zirconia phases

Generally, silica based RP stationary phases, such as the C18, gives retention that, at least for

homologues, is dependent upon the molecular weight, according to Martin‟s Rule (Equation 3.1)

[36]:

In k = Bn + In A (3.1)

In which A and B are empirical coefficients and k increases exponentially with the number of

repeat units, n, within a homologous series. This dependence upon molecular weight often

inhibits high resolution separations of diastereomers since mixtures of diastereomers have

identical molecular weights. To some degree, the stereochemistry of isomers may be controlled

or expressed by the use of different mobile phases. The C18 phase gives good separation of low

molecular weight oligostyrenes using a MeOH mobile phase [99, 100] and by changing the

51

mobile phase to ACN the oligomeric separation also exhibits some diastereomer resolution [99,

100]. However, the extent of the selectivity for diastereomers is often limited due to the retention

being dominated by the molecular weight. Hence the underlying resolution of the isomeric

components must occur within a distinct region defined by their molecular weight, and as a

consequence the separation capacity is limited.

A more useful chromatographic method for the separation of diastereomers is adsorption

chromatography, referred to as Liquid Solid Chromatography (LSC). This mode of separation is

usually very sensitive to molecular shape since solutes interact with adsorption sites, and their

interactions are largely shape dependent. Examples of these phases, which can operate in RP

mode, include the porous graphitized carbon (PGC) and CCZ columns.

Carbon stationary phases offer the potential for unique retention and present an alternative

mechanism of separation to conventional bonded phase RP stationary phases. The extensive

delocalised network on these carbon adsorption surfaces allows for the establishment of

electronic (-) bonding [101] and offers stereoselectivity of diastereomers that are capable of

undergoing these - type interactions.

These types of carbon adsorption phases, which includes not only the CCZ but also PGC

stationary phases were developed as alternatives to the bonded phase supports [102, 103]. These

phases offer greater shape selectivity, particularly for stereoisomers [104]. The CCZ is a carbon

coated zirconia particle prepared by chemical vapour deposition (CVD) of hydrocarbons over

porous zirconia microspheres at elevated temperatures [105-107]. This method offers a

52

mechanically and chemically stable support [108]. CCZ has been used for the separation of

derivatised enantiomers (diastereomers) and was shown to offer superior resolution when

compared to conventional bonded phase RP surfaces [109]. CCZ has also demonstrated high

resolution of diastereomers of low molecular weight oligostyrene [50, 110-111]. However, a

limitation of the surface is poor reproducibility in the manufacturing process [112], and this in

part could explain the limited applications that have been reported in the literature.

Another carbon phase is that of the PGC, which is thought to consist of porous carbon particles

comprised in flat sheets of hexagonally arranged carbon atoms. The surface is stable across a

large pH and temperature range [103]. The preparation of this surface is achieved by multiple

steps that involve the chemical and thermal treatment of a polymerised resin impregnated within

the pores of a silica gel template that is then subjected to carbonisation, the dissolution of the

silica template, and finally graphitisation. The spherical shape and particle size of the original

silica template are thus retained by the carbon [112]. PGC has proven useful for the separation of

isomers, where the spatial arrangement of the diastereomers rather than molecular weight of the

isomer can determine selectivity. Examples of diastereomer separation on PGC include the

quantitation of diastereomers in plasma [113], the separation of cis- and trans-stilbenes [114],

cis-trans isomers of potential anti-asthma agents [115], cis-trans isomers of proline-containing

dipeptides [116], and diastereomeric gluronides of almakalant [117]. However, poor mechanical

stability, low surface area, the heterogeneous nature of the surface and the non-uniform pore

structure (which have effects on the loading capacity) have been listed as possible shortcomings

of PGC [105, 106].

53

CCZ has also demonstrated high resolution of diastereomers of low molecular weight

oligostyrene using the CCZ column [44, 49, 50, 110, 111, 118-121]. This chapter focuses on the

high resolving power of the carbon adsorption phase. The CCZ stationary phase and the PGC

stationary phase (Hypercarb) are utilised to demonstrate the separation of oligostyrene

diastereomers.

3.2 Experimental

3.2.1 Chemicals

tert-Butyl oligostyrene (molecular mass ~580 Da) was dissolved in 80:20 MeOH:chloroform.

3.2.2 Chromatographic separation

The diastereomer separations were undertaken using a Waters 2D LC system. Columns were: (1)

C18 Varian Pursuit XRs column (250 × 4.6 mm), which was used with various MeOH: DCM

mobile phase gradient conditions and also 100% ACN (ACN), (2) several CCZ columns (30 ×

4.6 mm, 50 × 4.6 mm and 100 mm × 10.0 mm) that were used to separate the diastereomers of

oligomer fractions using various ACN: DCM mobile phase gradient conditions. (3) Hypercarb

(100 mm × 4.6 mm) using ACN: DCM mobile phase gradient conditions. All experiments were

conducted under ambient temperature unless otherwise stated and injection volumes onto the

C18 column were 30 L unless otherwise stated. UV detection was 272 nm for oligostyrene

analysis and 262 nm for isomer analysis.

54

3.3 Results and Discussion

3.3.1 Peak profiles for separation of oligostyrenes

When dealing with complex samples, a reduction in the complexity of the sample is usually

warranted in order to overcome the limitation in peak capacity. 1D gradient elution is often used

in chromatography when complex sample mixtures cannot be separated in an isocratic manner or

where resolution achieved is not adequate. In gradient elution the mobile phase composition is

changed in a pre-determined and continuous manner. The advantage of this method for complex

sample mixtures is that the resolution between components is greatly improved as components

that are usually weakly retained (k < 3) or strongly retained (k > 3) are able to be separated in

one run. Gradient elution of complex mixtures can be completed in a reasonable amount of time,

however, regeneration time between runs can greatly increase the overall analysis time.

Gray et al. [44] employed the C18 column with MeOH mobile phase and a CCZ column with

ACN as the mobile phase both under isocratic conditions; resulting in an acceptable 2D

separation of oligostyrene with different end groups including tert-butyl. To engender a higher

resolution separation of the tert-butyl oligomers and its isomers for this study, gradient

conditions were employed using DCM with MeOH and ACN. DCM alone is not suitable as a

mobile phase for the separation of oligomers on the C18 and CCZ phases due to the oligomers

being unresolved and unretained [44]; however, it has successfully been used in conjunction with

other mobile phases in isocratic elution [121] and gradient elution of oligostyrenes [122-124].

55

The separation of the tert-butyl oligostyrene using a C18 column under linear gradient elution of

100:0% MeOH: DCM to 80:20 MeOH: DCM 4 minutes at 1.0 mL/min is shown in Figure 3.3.

The resolution was predominated by molecular weight with retention increasing incrementally

with the addition of each configurational repeat unit; ten oligomers were observed to elute. No

diastereomer resolution was evident. The separation of oligostyrenes according to molecular

weight has previously been described using both normal-phase (NP) [121, 122, 125] and RP

[126-130] HPLC. Changing the mobile phase to 100 % ACN (isocratic elution), distortion of the

oligomeric peaks was apparent due to the expression of the isomeric sample attribute under these

elution conditions, as shown in Figure 3.4. However, the separation was primarily governed by

molecular weight. The expression of isomeric resolution has also been previously observed in

both NP [105, 110, 117] and RP separations [36, 106, 110] and such separations, at least to the

degree of resolution afforded here are trivial.

Changing the column to CCZ and operating under linear gradient conditions of 100:0 ACN:

DCM to 0:100 ACN: DCM in 50 minutes at 3.0 mL/min, diastereomer selectivity was observed

to dominate over that of molecular weight. The apparent chaotic band displacement shown in

Figure 3.5 shows no systematic displacement pattern associated with the oligomeric fractions

seen in the separations on C18. The CCZ phase was instead stereoselective more so than size

selective. In spite of the differing selectivity of the CCZ phase complete resolution of the

isomeric content was not feasible, even under gradient conditions because the sample contained

too many components, saturating the column peak capacity. This was the case for both the CCZ

and C18 phases. As such a 2D approach to this separation must be undertaken. That is,

56

separation in one dimension according to molecular weight and in the other dimension,

according to stereochemistry.

0 5 10 15 20 25 30

0.00

0.05

0.10

0.15

0.20

0.25

0.30

1

1098

7

6

5

43

2

Inte

nsity

(mV)

Retention Time (min)

Figure 3.3: Chromatogram of tert-butyl oligostyrene separation on C18 Varian Pursuit XRs column (250 × 4.6 mm). Conditions: gradient elution 100:0 MeOH: DCM to 80:20 MeOH: DCM in 4 minutes at 1.0 mL/min at ambient temperature with 30 μL injection volume. Detection UV 272 nm. Oligomers number 1-10 accordingly.

57

0 5 10 15 20 25 30 35 40 45 50 55 60

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

Inte

nsity

(mV)

Retention time (min)

Figure 3.4: Chromatogram of tert-butyl oligostyrene separation on C18 Varian Pursuit XRs column (250 × 4.6 mm). Conditions: 100% ACN at 1.0 mL/min at ambient temperature with 10 μL injection volume. Detection UV 272 nm.

0 5 10 15 20 25

0.00

0.01

0.02

0.03

0.04

0.05

Inte

nsity

(mV)


Figure 3.5: Chromatogram of tert-butyl oligostyrene separation on CCZ column (100 × 10.0 mm). Conditions: gradient elution 100:0 ACN:DCM to 0:100 ACN:DCM at 3.0 mL/min at ambient temperature with 10 μL injection volume. Detection UV 262 nm.

58

3.3.2 Ultra high resolution separations of oligostyrene isomers on CCZ

Under the chromatographic conditions described for the first dimension (gradient elution 100:0

MeOH: DCM to 80:20 MeOH: DCM 4 minutes at 1.0 mL/min) it was possible to separate the

tert-butyl oligomers (n = 2 to n = 10) with baseline resolution according to molecular weight

selectivity. Such separations have been extensively reported in the literature and require no

further discussion [29, 99, 100, 121-123]. In a 2D system diastereomer separations have been

limited to oligomer n = 5 [44, 48-50, 110, 111, 120, 127]. Of interest to this work was the ultra

high resolution of the diastereomers of oligomers n = 6, n = 7, n = 8. These oligomers were

heart-cut from the first dimension to the second dimension where their diastereomers were

separated on the CCZ column. The number of diastereomers possible is given in Table 3.1. An

oligomer containing six configurational repeating units, for example, has 16 diastereomers;

seven configurational repeating units, 32 diastereomers; eight configurational repeating units, 64

diastereomers; nine configurational repeating units, - 128 diastereomers.

Many different gradients were explored; such as linear gradient 80:20 to 40:60 ACN:DCM in 40

and 80 minutes; 100:0 to 0:100 ACN:DCM over 40 and 50 minutes; linear gradient 100:0 to

0:100 ACN:DCM over 60 minutes; linear gradient 0:100 to 100-0 ACN:DCM over 120 minutes.

The individual elution conditions given in the figure captions. The chromatogram illustrated in

Figure 3.6 shows the separation of all 16 diastereomers of the tert-butyl n = 6 oligostyrene on the

CCZ column. Figure 3.7 shows the separation of 28 of the diastereomers of the tert-butyl n = 7

oligostyrene. The insert of Figure 3.7 shows an expanded region between ~10 to 32 minutes,

more clearly defining the resolution of the isomers. In previous studies by Mourey et al. [121],

NP LSC was employed for the separation of diastereomers of low molecular weight

59

oligostyrenes. They used two 250 mm × 4.6 mm serially coupled silica columns running a linear

gradient increasing at 0.1 %/mL from 89% to 11% n-hexane: DCM at a flow rate of 1.0 mL/min.

They resolved a total of 11 diastereomers for the n =7 oligomer. The efficiency of the RP LSC

separation on CCZ relative to that of the NP LSC separation on silica is quite remarkable given

the bed length in the RP CCZ mode was 10 cm, compared to 50 cm in the NP mode and better

separation was observed on the CCZ phase.

0 5 10 15 20 25 30 35

0.000

0.005

0.010

0.015

0.020

Inte

nsity

(mV)


Figure 3.6: Chromatogram of gradient separation of oligomer n =6 on CCZ column (100 × 10 mm). Conditions: gradient elution 80:20 ACN: DCM to 0:100 ACN: DCM in 50 minutes at 3.0 mL/min with at ambient temperature 30 µL injection volume.

60

0 10 20 30 40 50 60 70 80 90

0.000

0.001

0.002

0.003

0.004

Inte

nsi

ty (

mV

)



Figure 3.8 shows the separation of the diastereomers of the eighth oligomer on the CCZ column,

where 42 of the diastereomers were resolved. An insert expanding the 10 minute time period

between 30 and 40 minutes illustrates the high resolution of the diastereomers. This is a

remarkable separation given the column length was 10 cm. Comparable separations are not

possible using conventional C18 surfaces. In Figure 3.9 the separation was performed on a

Hypercarb column, also renowned for its shape selectivity, however, at least for the oligostyrene

diastereomers the separation was not as powerful as for the CCZ stationary phase. Development

of ultra performance liquid chromatography (UPLC) may give the higher resolution required to

give similar high resolution diastereomer separations, but even then it is doubtful.

61

0 10 20 30 40 50 60 70 80 90 100

-0.001

0.000

0.001

0.002

0.003

Inte

nsity

(mV)



Even with the excellent resolving power that was evident in the separations shown in Figures

3.6, 3.7 and 3.8, the peak capacity of this column was exceeded for the separation of the

diastereomers of the ninth oligomer fraction, as shown by the separation in Figure 3.10. The 128

diastereomer mixture eluted essentially as a continuum and hence a larger capacity column

would be required for resolution of the components.

62

0 10 20 30 40 50 60 70 80 90

0.000

0.005

0.010

0.015

0.020

Inte

nsity

(mV)

Retention (min)

Figure 3.9: Chromatogram of gradient separation of oligomer n =8 on Hypercarb column (100 × 4.6 mm). Conditions: gradient elution 100:0 ACN: DCM to 0:100 ACN: DCM in 40 minutes at 1.0 mL/min with at ambient temperature 30 µL injection volume.

0 5 10 15 20 25 30

0.000

0.002

0.004

0.006

0.008

0.010

0.012

Inte

nsity

(mV)



63

3.4 Conclusion

In order to demonstrate the superb diastereomer selectivity of the carbon based CCZ stationary

phase, the complex oligostyrene sample matrix required simplification. This was readily

achieved using a 2D HPLC system whereby discrete oligomer fractions could be isolated on the

C18 dimension and then analysed on the CCZ phase. The separations presented here

demonstrated the superb diastereomer selectivity that was gained using a carbon adsorption

stationary phase. Analysts who seek to resolve similar complex mixtures should consider re-

investigating this now often overlooked separation mode by employing carbon supports that are

designed for the rigours of HPLC.

64

Chapter 4

Practical Aspects in the Optimisation of

Targeted Isolations in Two-dimensional

HPLC:

Analytical Scale Analysis

65

4.1 Introduction

The studies in Chapter 3 showed that 2D HPLC serves to reduce the complexity of extremely

complex mixtures, and in the specific example of the oligostyrenes, expression of the

diastereomer sample attribute was achieved by incorporating into the system a shape selective

separation dimension. While for the most part 2D HPLC has been employed for high resolution

separations and sample profiling, the focus of this and following chapters is on targeting the

isolation and purification of specific compounds within the complex mixture. In this chapter the

isolation of targeted components at analytical scale analysis was investigated. The principle

reason behind such a separation protocol may be for purposes of identification, more so than

collection and subsequent re-employment of the target analyte. In this chapter, the effect of the

recovery in both the first and second dimensions and the purity of the collected product were

investigated. The emphasis in this chapter was on the isolation of „a‟ target analyte from „a‟

complex mixture, where effectively „a‟ represents a generic sample, complex in nature. Low

molecular weight oligostyrenes have been used here, because they are complex, are indefinitely

stable and easily characterised. In effect, the separation performance was „de-tuned‟ from that

shown in Chapter 3 so as to mimic a more complex and crowded separation space that would be

apparent in real natural product type samples, but here with the advantage of absolute stability in

the recovered analyte. Furthermore, this separation problem represents a scenario of extracting a

minor constituent from the complex multicomponent bulk sample.

66

4.2 Experimental

4.2.1 Chemicals

tert-Butyl oligostyrene (molecular mass ~580 Da) was dissolved in 80:20 MeOH: chloroform.

For calibration purposes oligostyrene standard of molecular mass ~114,200 Da was dissolved in

100% chloroform.


The diastereomer separations were undertaken using the Waters system detailed in Chapter 2.

The columns utilised in this work were (1) a C18 column (50 mm × 4.6 mm), which was used

with 100% MeOH for the separation of oligomers, and (2) a CCZ column (50 mm × 4.6 mm),

which was used for the separation of the diastereomers within the oligomer fractions that were

transported to the second (CCZ) dimension. The CCZ column was operated using 100 % ACN

mobile phase. The C18 column was operated at ambient temperature, while the CCZ column was

thermostatted to 45°C. Injection volumes onto the C18 column (1st dimension) were 10 L

unless otherwise stated.

4.2.3 Determination of product purity and recovery

The target component purity was determined by the duplicate re-injection of the final product

both onto the C18 column and the CCZ column and measured using the Waters UV-Vis detector

at 262 nm and 272 nm respectively. Product recovery yield was determined through the

integrated peak area using the onboard LC software as detailed in Section 2.3.

67


4.3.1 Determination of target component

In 1D HPLC, the target analyte is recovered in a single step separation process. However, in 2D

HPLC the target analyte is transported from the first separation dimension to a second separation

dimension, which affects a new selectivity towards the sample, allowing isolation from the

underlying impurities. The heart cutting 2D HPLC system that was employed in this study is

shown in Chapter 2, Figure 2.1. The use of a C18 column with MeOH as the mobile phase in the

first dimension allowed for separation of the oligomers, which was dependent almost exclusively

on molecular weight selectivity. In the second dimension selectivity was dependent on solute

shape, or the spatial orientation of the diastereomers of each oligomer fraction, in a system

comprising a carbon adsorption phase and an ACN mobile phase. The chromatographic

separations were performed under isocratic conditions because the separation targeted a specific

solute, and resolution beyond the target analyte was not required. Hence a high peak capacity, as

would be obtained in gradient applications was not warranted. As a result, resolution of the

sample in its entirety was sacrificed for speed, with only the isolation of the target analyte being

required, and since isocratic elution was employed, no column re-equilibration between runs was

necessary.

Sample loads were restricted to 10 μL injections, and the tert-butyl oligostyrene n = 5 in which

the target diastereomer #2 occupies, were heart-cut to a sample loop where the target

diastereomer was isolated from the bulk before loading onto the CCZ column. Both dimensions

had autonomous operation and separations in both dimensions could occur simultaneously and

68

were operated on-line. The separation obtained in the second dimension, illustrated in Figure 4.1,

shows the resolution of the target analyte from the neighbouring diastereomers, all from the n = 5

oligostyrene. The n = 5 oligostyrene was chosen as the target oligomer as it has eight

diastereomers that elute in the second dimension in a reasonable time of eighteen minutes under

the conditions described in Figure 4.1. Clearly, diastereomer number 2 (#2) is poorly resolved,

and this presents as a serious challenge in the isolation of this targeted component from this

complex mixture. While baseline resolution of this target analyte could have been obtained using

a column with greater N, the point of this exercise, was rather, to illustrate the process of

recovery for solutes with such limited separation, and to do so as quickly as possible, with the

greatest yield and in the purest form. The retention range for the target diastereomer (#2) was

determined to be between 3.36 minutes and 3.52 minutes in a total volume of 340 μL.

2 4 6 8 10 12 14 16 18

0.000

0.005

0.010

Target component

1

Inte

nsity

(m

V)


8

7

6

5

43

2

Figure 4.1: Chromatogram of the separation of tert-butyl oligostyrene n = 5 separation on CCZ column, target component peak #2. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 100 μL heart-cut volume.

69

Maximum recovery of any target component is achieved by maximising the recovery from both

dimensions. This requires that the elution region of the target analyte in the first dimension is

firstly determined. The elution region was identified by systematically heart cutting small

fractions from the first dimension across the region of expected elution into the second

dimension, and then measuring the quantity of target analyte that was transported to the second

dimension. The region that was heart-cut to the second dimension was between 1.40 and 2.50

minutes in the first dimension. In total 11 fractions (100 L each) were sequentially transported

to the second dimension. The region of separation space that the target component occupied in

the first dimension is shown in Figure 4.2 and was between 1.60 minutes to 2.30 minutes with a

maximum and central retention time of 1.95 minutes. Other components also eluted within this

region of separation space in the first dimension. These included other diastereomers of the n = 5

oligomer, and overlap from the n = 4 and n = 6 oligomers. The n = 4 oligomer occupied the

region between 1.40 to 2.20 minutes and the n = 6 oligomer occupied the region between 2.17

and 2.63 minutes. This co-elution further complicated the process of isolation.

The total elution volume of the target component from the first dimension was 700 µL. The area

of the target component, as a function of the heart-cut time (volume) was used to ascertain the

recovery from the first dimension (Figure 4.3). From this data seven heart-cut fractions that had a

central retention time of 1.95 minutes were selected. The volume of these heart cut fractions

increased from 100 µL to 700 µL (Figure 4.3). The 700 µL heart-cut fraction had a recovery

greater than 99 % of the target component; subsequently as the heart-cut fractions decreased in

volume, so too did the recovery of the target component, reaching 36 % for the smallest heart-cut

of 100 µL. The second dimension recoveries were calculated from the area pertaining to the

70

target band collected in the second dimension, the target fraction being approximately 340 µL in

volume (in the second dimension) for all heart-cuts.

1.0 1.5 2.0 2.5 3.0

0.0

0.1

0.2

0.3

0.4

0.5

6

43

2

5

n=6n=4

n=5

Inte

nsity

(m

V)


Figure 4.2: Chromatogram of the separation of tert-butyl oligostyrene separation on C18 column displaying the overlap of n = 4, n = 5 and n = 6. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection volume.

Maximising analyte purity is important, and notionally maintaining purity greater than 99 % is

desirable [98]. However, this is often not possible at the required target production rate as it

places unfavourable constraints on the separation (resolution, time etc) of complex mixtures.

Because of the complexity of the sample employed in this work, it was difficult to maintain a

high level of purity and at the same time achieve a high productivity. The overlap of the main

impurity (n = 4#3 oligostyrene (3rd diastereomer of the fourth oligomer fraction)) in the first

71

dimension with the target component is shown in Figure 4.4. This demonstrates the difficulty

that arises in trying to obtain a highly purified product, because co-elution in the first dimension

exists, not only for the eight diastereomers of the n = 5 oligostyrene, but also, the n = 4

oligostyrenes, of which the #3 (n = 4) diastereomer is the most important since it also co-elutes

with the #1 (n = 5) diastereomer in the second dimension (both have retention times in second

dimension of ~ 1.47 minutes) and they tail into the n = 5 #2 diastereomer.

1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4

-20000

0

20000

40000

60000

80000

100000

700 µL600 µL

500 µL

400 µL300 µL

200 µL

100 µL

Pea

k ar

ea (

mV

*sec

)

Heart-cut time (min)

Figure 4.3: Area that target component n = 5 #2 occupies in first dimension.

72

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

0.0

0.2

0.4

0.6

0.8

1.0

n =5#2n =4#3

Impurity Target

Nor

mal

ised

pea

k ar

ea (

mV

*sec

)

1st Dimension retention time (min)

Figure 4.4: Overlap of n = 4 #3 and n = 5 #2 on C18 column for heart-cuts from the first dimension. Conditions: C18 column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 100 μL heart-cut volume. 4.3.2 Isolation of target component: Maximising purity

The objective of this section is to isolate the target analyte, maximising the purity. To bring

about this aim consideration had to be given to two aspects:

The region of recovery in the first dimension,

The region of recovery in the second dimension.

73

The following sections examine „optimising the purity‟ of the isolation employing two different

strategies:

(1) By recovering the target component from the central area of the band in both

dimensions,

(2) By recovering the target component from off-centre of the band in the first dimension

and from the central area of the band in the second dimension.

(1) Recovering the target component from central area of band in 1D and 2D

As the focus of this work was on collecting a pure component, the central area of the target band

in both dimensions was recovered to minimise impurities arising from adjacent diastereomers

from not only the oligostyrene of interest but also from the early eluting n = 4 oligostyrene.

Figure 4.5 shows the area that was recovered on the CCZ column of the target component for the

100 μL heart-cut volume.

In the second dimension the target component was collected from peak valley to valley, the exact

volume depending on the volume transported from the first dimension. Recovery of the target

component in the second dimension followed the opposite trend to that of the first dimension

with the smaller heart-cuts having higher recovery. The recovery loss was greatest for the 400

µL heart-cut fraction with a recovery of 76 % of the target analyte (Table 4.1). In comparison the

recovery from the 100 µL heart-cut fraction was 96 %. This was because recovery of the target

component in the second dimension (at high purity) was dependent on the number of closely

74

eluting components that were subsequently transferred to the second dimension. Purity thus

decreased as the heart-cut volume increased because of the overlap of adjacent bands transferred

to the second dimension. This is illustrated by comparing the second dimension separations

resulting from a 700 µL heart-cut fraction in Figure 4.6 to that of the 100 µL heart-cut fraction in

Figure 4.5. The smaller heart-cut volume fraction resulted in better resolution of the target

analyte in the second dimension since less contaminating species were transferred to the second

dimension. Consequently, the smaller heart cut fractions resulted in higher purity of the target

analyte, but at the cost of reduced recovery.

Table 4.1: Characteristics of two-dimensional separation of target component n = 5 #2 Injection

volume

Cut time

Cut

volume

Recovery

1D

Recovery

2D

Total Recovery Purity

(µL) (min) (µL) (%) (%) (%) (%)

10 1.90-2.00 100 36 96 34 90

1.85-2.05 200 66 84 55 83

1.80-2.10 300 83 78 65 84

1.75-2.15 400 91 76 69 83

1.70-2.20 500 96 77 74 82

1.65-2.25 600 98 79 77 65

1.60-2.30 700 99 80 79 59

75

2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4-0.002

0.000

0.002

0.004

0.006

0.008

0.010

0.012

43

21

Inte

nsity

(m

V)


Figure 4.5: Chromatogram of the separation of tert-butyl oligostyrene separation on CCZ column showing recovery of target component. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 100 μL heart-cut volume.

3.6 3.8 4.0 4.2 4.4 4.6

0.00

0.01

0.02

0.03

0.04

0.05

0.06

4

Inte

nsity

(m

V)


1

2

3


76

This purification strategy resulted in total recoveries of the target component ranging from 34 to

79 % (Table 4.1) with the recovery being dependent on the loss of sample from the first

dimension separation. As larger heart-cuts were transferred from the first dimension, the area

that the target component occupied encroached on that of the earlier oligostyrenes, as some

overlap was essential to ensure close to 100 % recovery of the target, although it was detrimental

to the recovery of the target component from the first dimension. Total recovery increased for all

heart-cuts as a function of the cut volumes as demonstrated in Figure 4.7.

100 200 300 400 500 600 70030

40

50

60

70

80

90

100

1st Dimension 2nd Dimension Total

Rec

over

y (%

)

Heart-cut volume (uL)

Figure 4.7: Recovery plot of target component.

77

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.0000

0.0001

0.0002

0.0003

0.0004

n=5 #3

n=5 #2

n=5 #1n=4 #2

Inte

nsity

(mV)


1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.0000

0.0002

0.0004

0.0006

0.0008

0.0010

n=5 #2

n=5 #2

n=5 #1n=4 #2

Inte

nsity

(mV)


Figure 4.8: Chromatogram of purity for (a) 100 µL heart-cut and (b) 700 µL heart-cut. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 10 μL injection volume. The purity displayed an opposite trend to the total recovery with purity of the target component

ranging from 90 % to 59 %. The 700 µL heart-cut yielded a purity of 59 % much lower than

that of the 100 µL band cut at 90 % purity. The chromatograms in Figure 4.8 highlight the

impurities that arise from the 100 µL heart-cut (a) and 700 µL heart-cut (b). When the heart-cut

(b)

(a)

78

volume increased so too did the concentration of impurities, particularly that of the impurity n

=4#3. This was not surprising as previously mentioned the n =4#3 overlap with the target

diastereomer was considerable (Figure 4.4). Impurities arising from the n = 5

oligostyrene/diastereomer (n = 5#1 and 5#3) were also present for the larger heart-cut volumes,

although in much smaller concentrations than the n = 4#3 impurity. The most significant

decrease in purity occurred for the 600 µL and 700 µL heart-cut volume fractions. However,

purity greater than 80 % for the heart-cuts between 100 µL and 500 µL, although not perfect,

was sufficiently high enough to satisfy the aim of this strategy.

(2) Recovering the target component off-centre in 1D

The aim of this strategy was to optimise the purity of the target analyte by focusing on the

recovery of the target component off-centre in the first dimension and recovery of the central

area of the band in the second dimension. In the first approach above the first dimension was

largely the limiting factor in determining the purity of the final product because of the transport

to the second dimension of numerous contaminating species, which became more prevalent as

the heart-cut volume increased. One of the key contaminates was the n = 4#3 diastereomer,

which eluted from the first dimension between 1.40 and 2.20 minutes, overlapping the target

analyte at 1.60 to 2.20 minutes. To minimise contamination from this impurity the region heart-

cut from the first dimension was shifted from the location of the peak maxima of the n = 5

oligomer to offset this contaminating species in the second dimension.

Heart cutting was subsequently undertaken with centralised retention times of 2.05 minutes and

2.10 minutes for 300, 400 and 500 µL heart-cut volumes (Table 4.2). These heart-cuts were

79

between 1.80 minutes and 2.40 minutes in total. Two separate central retention times were used

to investigate if moving the heart-cut further away from the impurities could make further gains

in the purity of the target component. Shifting the heart-cut section as such decreased the overlap

between the two oligostyrenes. The overlap between n = 5 and n = 6 occurred at 2.17–2.63

minutes, but no diastereomer from the n = 6 oligostyrene co-eluted with the n = 5#2 target

analyte in the second dimension.

Table 4.2: Characteristics of two-dimensional separation of target component n=5 #2 Injection

volume

Cut time Cut volume Recovery 1D

Recovery

2D

Total

Recovery Purity

(µL)

(min) (µL) (%) (%) (%) (%)

(a) Centre band 2.05 min

10 1.90-2.20 300 69 78 53 91

1.85-2.25 400 84 82 69 89

1.80-2.30 500 93 78 72 93

(b) Centre Band 2.10 min

2.00-2.30 300 35 85 29 93

1.95-2.35 400 53 87 46 90

1.90-2.40 500 71 80 57 93

The results in Table 4.2 detail the purity of the targeted component, with respect to the recovery

from the first and second dimensions. While overlap with the limiting impurity was still evident

the extent of contamination was greatly reduced. For the off centred heart cut at 2.05 minutes,

the purity remained generally constant at ~ 92 %, irrespective of the volume transported to the

second dimension, i.e. between 300 and 500 L, Furthermore, the total recovery was as high as

80

72 %. This compared to a 34% recovery at 90 % purity for the first employing the centralised

heart cutting method corresponding to the peak maxima of the target analyte in the first

dimension. For the off-centred heart cut at 2.10 minutes, the purity was again ~ 92 % irrespective

of the heart cut volume, however, at the central location of 2.10 minutes, the recovery of the

target was greatly reduced (57 % compared to 72 % at the peak maxima 2.05 minutes).

Clearly off-centering the location of the heart-cut section, away from the peak maxima and the

limiting impurity positively impacted on the purity, and at the same time yielded improved

resolution with the limiting impurity (compared to similar impurity) concentrations. A slight

reduction in the first dimension recovery for all heart-cuts was observed in the off-centred heart

cutting, when compared to the heart cutting at 1.95 minute in the centre of the band. In the

second dimension recovery was again set between valley to valley. The recoveries from the

second dimension improved as the heart cutting was off centred, largely as a result of there being

less contamination from the limiting impurity.

Total recovery increased monotonically as a function of the heart-cut volume, irrespective of the

location of the central cutting region, i.e. 2.05 or 2.10 minutes (Figure 4.9). This corresponded

to a monotonic increase in the recovery from the first dimension as a function of heart-cut

volume, even though the recover from the second dimension was not monotonic. The non-

monotonic relationship between heart-cut volume and recovery in the second dimension was

affected by the increase in quantity of sample contaminants transported to the second dimension

as the heart-cut volume increased, which consequently increased the valley height between the

target and contaminants. Nevertheless there was an increase in the total recovery of the target

81

component as the heart-cuts increased and the purity stayed consistent for both the 2.05 and 2.10

minute centred heart-cuts, Figure 4.10 illustrates this trend.

300 350 400 450 50030

40

50

60

70

80

90

100

(a)


Rec

over

y (%

)


300 350 400 450 500

30

40

50

60

70

80

90

100(b)


Rec

over

y (%

)


82

Figure 4.9: Recovery plot of target component; (a) 2.05 minute centred heart-cuts, (b) 2.10

minute centred heart-cuts.

300 350 400 450 50030

40

50

60

70

80

90

100

(a)

Purity Total recovery

(%)


300 350 400 450 500

30

40

50

60

70

80

90

100(b)

Purity Total recovery

(%)


Figure 4.10: Purity versus total recovery; (a) 2.05 minute centred heart-cuts, (b) 2.10 minute centred heart-cuts.

83

4.3.3 Multicomponent Isolations

The studies in Section 4.3.2 examined the isolation of a single target component only from

within the complex mixture. In this section the focus is on isolating all eight diastereomers of the

n = 5 oligostyrene (5#1-8). This is of importance because often more than one component may

be of interest, perhaps because they have similar chemical attributes, or they work symbiotically.

This may be particularly true for pharmaceutical and natural product separations where samples

are screened for viable products and components have known or unknown symbiotic

relationship. Polymer separations also may benefit from separation of all components in a

mixture as their properties may show similar behaviour. Of interest also in this section of study is

the effect that the collection of all eight diastereomers (5#1-8) has on the isolation of the target

component as this still remains as the primary targeted species with the other seven

diastereomers being secondary targets.

To achieve maximum recovery of the target component from the first dimension and transport it

to the second dimension it is imperative to collect as much of the target as possible. In order to

achieve this goal it was necessary to determine the target regions of all diastereomers in the first

dimension as detailed in Section 4.3.1. Therefore the area that all eight diastereomers occupied

was determined. The area that the target components were found to occupy in the first dimension

were between 1.60 minutes and 2.30 minutes, however, the maximum retention time for each

target within this region differed. Again the target components displayed overlap with adjacent

oligomers; the n = 4 whose range was 1.40-2.20 minutes and the n = 6 oligomer whose range

was 2.17-2.63 minutes.

84

Essentially the separation was identical to that of Section 4.3.2 where a series of tests were

undertaken whereby the target band n = 5 (Figure 4.2), was cut into sections of 100, 200, 300,

400, 500, 600 and 700 μL and the resulting section transferred to the second dimension. The

target components were then fractionated sequentially in the second dimension and their

presence verified by the corresponding retention times from re-injection on both the C18 column

and the CCZ columns. The purity of the target components were determined by the method

detailed in section 4.2.3. The impurities occurring from other diastereomers of both oligostyrene

n = 4, n = 5 and n = 6 could be monitored and determination of final purities were ascertained.

The chromatogram in Figure 4.11 illustrates the resolution of the eight diastereomers and their

collection starting and finishing points on the second dimension CCZ column. Note that for the

collection of the diastereomers 1 to 6, the collection period was between the minima of each

respective component, whereas, diastereomers 7 and 8 were always baseline resolved. Table 4.3

lists the individual collection volumes for each diastereomer and their percentage of the total

diastereomers collected for the n = 5 oligostyrene. The retention range for the eight targeted

diastereomers occurred between 3.55 minutes and 18.69 minutes with a total collection volume

for all diastereomers of 30.02 mL. The diastereomer #2 had the lowest percentage of overall

volume composition at 1 % and diastereomer #8 comprised the largest percentage at 47 % (by

volume) of the total diastereomers for the n = 5 collected.

85

3.5 4.0 4.5 5.0 5.5 6.0 6.5

0.00

0.01

0.02

0.03

5

2

Inte

nsity

(m

V)


43

6

1

(a)

6 8 10 12 14 16 18

0.000

0.002

0.004

0.006

0.008

Inte

nsity

(m

V)


7

8

(b)

Figure 4.11: Close up showing where diastereomers (a) #1-6 and (b) #7-8 were collected from CCZ column. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 500 μL heart-cut volume.

86

Table 4.3: Collection volume and percentage of diastereomers #1-8 in 2nd dimension Diastereomer # Collection volume (mL) Volume %

1 0.40 1.33

2 0.32 1.07

3 0.96 3.20

4 0.60 1.99

5 1.22 4.06

6 1.64 5.46

7 10.66 35.5

8 14.22 47.4

Table 4.4: Characteristics of two-dimensional separation eight diastereomers for 500 µL heart-cut

Injection

volume

Cut time

Diastereomer #

Recovery

1D

Recovery

2D

Total

Recovery Purity

(µL) (min) (%) (%) (%) (%)

10 1.85-2.05 1 97 71 69 90

2 98 63 62 75

3 91 85 78 32

4 91 85 78 60

5 96 93 93 65

6 98 70 69 79

7 97 100 92 100

8 96 100 96 100

87

In the first dimension the target components eluted in the region between approximately 1.60 and

2.30 minutes with a total volume of 700 µL. Heart cutting this volume to the second dimension

assured recovery greater than 90 % from the first dimension for all diastereomers and recoveries

in the second dimension ranging from 63 % to 100 % (Table 4.4).

Attempting to isolate all diastereomers of the n = 5 oligostyrene was not without difficulty as

impurities influenced the isolation of the product. Figure 4.12 illustrates the purities of

diastereomers #1-7 and impurities are evident for all diastereomers with the exception of

diastereomer #7 and #8. Impurities arise from the n = 4 oligostyrene/diastereomers as well as the

n = 6 oligostyrene/diastereomers. Figure 4.12 (d) is the purity of peak #3 and #4, these

diastereomers were collected together as they co-elute and were unable to be isolated separately.

As all diastereomers were collected sequentially there was the potential for overlap between the

bands impacting on the purity.

Regardless, purities were still quite fair particularly for #2 with a total recovery of 62 % and

purity at 75 %, considering that the focus was on eight components instead of one.

Diastereomers #7 and #8 were isolated with 100 % purity as they had entirely unique two-

dimensional retention times, independent of impurities from n = 4 and n = 6 with total recoveries

greater than 90 %.

88

1.0 1.2 1.4 1.6 1.8 2.0

-0.00005

0.00000

0.00005

0.00010

0.00015

0.00020

0.00025

n = 4#2

Inte

nsity

(m

V)


n = 5#1n = 4#3

(a)

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.00000

0.00005

0.00010

0.00015

0.00020

Inte

nsity

(m

V)


n = 5#1 n = 4#3

n = 5 #2(b)

1.2 1.4 1.6 1.8 2.0 2.2 2.4

-0.00005

0.00000

0.00005

0.00010

0.00015

0.00020

0.00025

0.00030 n = 5#4n = 6#1

Inte

nsity

(m

V)


n = 5#3

n = 5#1n = 4#3#2

(c)

n = 5#2

2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0

-0.00005

0.00000

0.00005

0.00010

0.00015

Inte

nsity

(m

V)


n = 6#2

n = 5#5 (d)

2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6

-0.00004

-0.00002

0.00000

0.00002

0.00004

0.00006

0.00008

0.00010

0.00012

Inte

nsity

(m

V)


n = 5#5

n = 5#6(e)

4.0 4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8 6.0

-0.00002

0.00000

0.00002

0.00004

0.00006

0.00008

0.00010

0.00012

0.00014

Inte

nsity

(m

V)


n = 5#7

(f)

Figure 4.12: Chromatograms of purity for 500 µL heart-cut (a) #1, (b) #2, (c) #3 and 4, (d) #5, (e) #6 and (f) #7. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 10 μL injection volume.

89

The recoveries and purities for all heart-cuts for the component diastereomer 5#2 are given in

Table 4.5. Recoveries from the first dimension were slightly higher than the recoveries recorded

for the experiment completed to optimise purity (see Section 4.3.2 and Table 4.1). The second

dimension recoveries, however, were lower for all heart-cuts; resulting in lower total recoveries.

The reduction in the second dimension recovery can be attributed to the collection volume being

slightly less than that of other the sections mentioned. Thus the purity was again affected by the

recurring impurities from the n = 4 and n = 6 oligostyrenes. Unique to these results is the affect

the sequential collection of the target diastereomers had on the overall purities as a result of some

cross-over between bands due to some contamination occurring between collections of all

fractions.

Table 4.5: Characteristics of two-dimensional separation of target component n = 5 #2 Injection

volume

Cut time

Cut

volume

Recovery

1D

Recovery

2D

Total Recovery Purity

(µL) (min) (µL) (%) (%) (%) (%)

10 1.90-2.00 100 38 72 28 80

1.85-2.05 200 70 62 43 80

1.80-2.10 300 87 62 54 87

1.75-2.15 400 95 64 61 81

1.70-2.20 500 99 63 63 75

1.65-2.25 600 100 54 53 66

1.60-2.30 700 100 47 47 77

90

4.3.4 Isolation of Target Component: Maximising Recovery

The studies in the first part of this chapter (Section 4.3.2) focussed on the optimisation of product

purity by employing two different strategies; (1) by recovering the target component from the

central area of the band in both dimensions; and (2) recovering the target component from off-

centre of the band in the first dimension and from the central area of the band in the second

dimension. The smaller heart-cut fractions resulted in higher purity of the target analyte, but at

the cost of reduced recovery. However, purity greater than 80 % for the heart-cuts between 100

µL and 500 µL was sufficiently high enough to satisfy the aim of maximising the purity.

This section therefore addresses the need to maximise recovery of the target component at an

analytical scale, in preparation for scale up to preparative isolations. In doing so, the relationship

between recovery and purity, with respect to both the level of contamination and the number of

components contaminating the target analyte was examined. The number of contaminating

species is an important consideration because, a target analyte that is 95% pure or 25% pure

following isolation from the complex mixture present essentially the same challenge to polishing

the product if the number of contaminating species remains constant. However, if the analyte that

is 25% pure contains substantially more contaminants then the polishing step for the target may

be somewhat more complex than for the 95% pure analyte. Hence, recovery and purity would

then need to be balanced according to the economics of the isolation process. In this section the

recovery from the first dimension was in accord with the previous studies undertaken as the

target component occupied the same area in the first dimension separation. Effectively, the target

component eluted from the first dimension in a period between 1.60 and 2.30 minutes, with a

total volume of 700 µL. Heart cutting this volume to the second dimension assured 99 %

91

recovery from the first dimension. Smaller heart-cut volumes from the first dimension had much

lower recoveries from the first dimension (down to 36 % for the 100 µL heart-cut) (Table 4.6).


Injection

volume

Cut time Cut

volume Recovery 1D Recovery 2D

Total

Recovery

Purity

(µL)

(min) (µL) (%) (%) (%) (%)

10 1.90-2.00 100 36 100 36 26

1.85-2.05 200 65 100 65 27

1.80-2.10 300 83 100 83 28

1.75-2.15 400 92 100 92 28

1.70-2.20 500 96 100 96 26

1.65-2.25 600 97 100 97 22

1.60-2.30 700 99 100 99 14

The recovery of the analyte in the second dimension was varied to assess the relationship

between purity and the number of components contaminating the sample. The target analyte was

fractionated from the second dimension in an area that expanded uniformly from the centre of

the band (Figure 4.13). The identity of the target analyte was confirmed by re-injection of the

collected fraction into both the C18 column and the CCZ columns. The purity of the target

component was determined by the method detailed in section 4.2.3. The level and identity of the

92

impurities in the collected sample were determined following the re-injection of the sample into

the C18 and CCZ 1D systems.

2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4

0.000

0.005

0.010

0.015

0.020

5

432

1

Inte

nsity

(m

V)



In the second dimension the target component eluted with a total volume of approximately 1200

µL for all heart-cuts except the 600 µL and 700 µL, which were 1400 µL. This volume was

nearly four times that of the heart-cut section performed in Section 4.3.2, since the sample in the

93

second dimension was then collected only between the valleys of neighbouring bands. In this

study recovery of the target component from the second dimension was 100 % for all heart-cut

sections, irrespective of the degree of overlap presented by the contaminating species that co-

eluted in the second dimension. As the heart-cut volume from the first dimension increased more

contaminating species were also transferred to the second dimension, with the result being

greater overlap with the target analyte in the second dimension. Figure 4.14 illustrates the

relationship between heart-cut volume and recovery and the purity of the final product. Up to a

heart-cut volume as high as 500 µL (recovery ~ 95%), product purity remained essentially

constant: The contaminating species arising from the overlap in the second dimension, which

became more substantial after heart cutting volumes greater than 500 µL. At 700 µL, for

example the purity was as low as 14 % at 99 % recovery. As a consequence of collecting 100 %

of the target component in the second dimension the purity is therefore low.

94

100 200 300 400 500 600 700

20

40

60

80

100

Purity Recovery

(%)

Heart-cut (uL)

Figure 4.14: Recovery versus purity of target component.

Purity was a limiting factor in the isolation of the target component. This does suggest that

although there is a gain in the recovery of the target component particularly from the second

dimension, the loss in purity will affect the overall process of recovering the final product as the

level of impurity is such that there were additional components present in the sample. This can

be seen clearly in Figure 4.15 where the purity of the target component was severely

compromised not only by the recurring n = 4 #3 but also n = 5 #1, n = 5 #3 and n = 5 #4; and n =

6 #1 are also present. Due to the presence of five contaminating species the polishing steps

required for these samples would be much more complex and detrimental to the overall process.

This may involve the re-injection of the recovered product onto the CCZ column either in a

recycling mode where the sample is continuously recycled and the product recovered close to

100 % pure by shaving the component peak of impurities at each cycle; or by the manual

95

collection of the target and consequent re-injection and collection of the pure product [98]. These

additional steps to re-purify the product may impact on the production [98], however, the target

component is still in a much more simplified form than the original separation with only six

components evident in comparison to 2054 components. This may not be a difficult task as

essentially it requires minimal solvent and time and the current system could easily be modified

to perform these tasks at the end of production resulting in a highly purified product.

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.0000

0.0001

0.0002

0.0003

n = 6#1

n = 5#4

n = 5#3n = 5#2

n = 5#1n = 4#3

Inte

nsity

(m

V)


Figure 4.15: Chromatogram of purity for 500 µL heart-cut. Fraction collected and re-injected onto CCZ.

96

4.3.5 Summary

Two strategies were employed to optimise the purity in sections 4.3.2. The most predominant

impurity was the n = 4#3 diastereomer, which was however easily minimised by applying off-

centred heart cutting from the first dimension (2.10 min. centre) and high purities were then

obtained. Although the target analyte was frequently contaminated with the n = 4#3

diastereomer, it was easily removed in a polishing step using a C18 column, and thus in the

overall final result posed little difficulty in obtaining a high final purity. Non off-centred heart

cutting from the first dimension resulted in lower final yields and purity. Multicomponent

isolations were also undertaken, with the result being lower overall recovery and purity of each

of the diastereomers, except those that were baseline resolved in the second dimension.

Figure 4.16 details the recovery both in the first and second dimension, the total recovery and

the purity of the final product for all sections for the 300 µL heart-cut. What can be observed is

that the purity of the final product is greatly influenced by the recovery of the target component.

Only the 1.95 min centred heart-cuts (Section 4.3.2) and the multicomponent heart-cuts (Section

4.3.3) had higher recoveries in the first dimension compared to the second dimension; all other

heart-cuts experienced the opposite trend with lower recoveries in the first dimension compared

to the second dimension. The total recoveries were less than 65 % for all sections with the

exception of the 1.95 min heart-cuts (Section 4.3.2), where the aim was to achieve high purity

and only the centre of the band was collected. Total recovery appeared to be more greatly

influenced by the first dimension recovery than that of the second dimension as all recoveries in

the second dimension were relatively high for all experiments. The shift in centre saw a

reduction in the total recovery for both 2.05 and 2.10 minute centred bands of (Section 4.3.2)

97

between 53 % and 73 % and 29 % and 57 % respectively, in comparison with the 1.95 minute

centred band with total recoveries ranging between 65% and 74 %. Shifting the heart-cuts did not

see an overall improvement in the recovery of the target component for this set of heart-cuts;

however it did result in a much higher purity for both centred bands than for the corresponding

heart-cuts with the 1.95 minute centred heart-cuts. This was because the off-setting of the heart-

cuts minimised the overlap of the n = 4 oligostyrene impurity that was a recurrent problem in

these isolations. However the loss of sample, at best 43 % and worst at 71 %, may not be

justified if the starting material is expensive and when the amount of injections may be

excessively high to compensate for the loss of sample, even at the higher purity. This pattern was

observed for all heart-cuts and Figure 4.17 and 4.18 illustrate the 400 µL and 500 µL heart-cuts.

Figure 4.16: Comparison of the experiments for the separation of target diastereomer separation of the 300 µL heart-cuts.

0

200

400

600

80

100

Purity Total Recovery

2DDRecovery

1D Recovery

(%)

Maximised purity (1.95 min) Maximised recovery ((4.3.4) Section 5.3 Maximised purity (2.05 min) Maximised purity (2.10 min) Multicomponent

98

Figure 4.17. Comparison of the strategies for the separation of target diastereomer separation of the 400 µL heart-cuts.

Figure 4.18: Comparison of the strategies for the separation of target diastereomer separation of the 500 µL heart-cuts.

0

20

40

60

80

100

Purity Total Recovery

2D Recovery

1D

Recovery

(%)

Maximised purity (1.95 min) Maximised recovery Maximised purity (2.05 min) Maximised purity (2.10 min Multicomponent

0

20

40

60

800

100

Purity

Total Recovery

2D Recovery

1D Recovery

(%)

Maximised purity (1.95 min) Maximised recovery Maximised purity (2.05 min) Maximised purity (2.10 min) Multicomponent

99

The focus of the third section of this chapter was to maximise recovery (Section 4.3.4), this

resulted in the highest recoveries for both dimensions, however, this also resulted in very low

purity (<30 %). The number of contaminating species increased as a consequence of larger

fractions collected in the second dimension and posed the most serious problem in obtaining a

highly purified final product as the polishing step was subsequently more difficult. The recovery

in the first dimension for all heart-cut fractions was higher than in the second dimension. Total

recovery was between 83 % and 96 %. The highest total recovery, however, also resulted in very

low purities as a result of many more contaminating species being present. This was a result of

the fraction collected in the second dimension encompassing a much larger volume and therefore

more impurities overlapping the target analyte. The final product, nevertheless, was a much less

complex mixture than the original sample, which contained over 2000 components.

4.4 Conclusion

The isolation of targeted components at an analytical scale was investigated in this chapter. The

effect of the recovery from the first and second dimension and the purity of the collected product

were investigated. The results have revealed that the highest purity product was a result of off-

centering the heart-cut at a central retention time of 2.05 minutes to be transported to the second

dimension the farthest away from the recurring impurity in Section 4.3.2. The final product

however had the lowest recovery of all strategies. Intermediate results could be achieved with

total recovery and purity for both the 1.95 minute heart-cut (Section 4.3.2) and the 2.10 minute

centred heart-cut (Section 4.3.2) which saw the heart-cut being off-centred to reduce overlap of

100

impurity. The highest total recovery however was the heart-cut from Section 4.3.4; however this

also resulted in the lowest purity of all strategies.

101

Chapter 5


Preparative Scale Two-dimensional

Isolations:

Low Sample Loads

102

5.1 Targeting the isolation and purification of specific compounds within the complex mixture at the preparative level 5.2 Introduction The studies in Chapter 3 showed that 2D HPLC serves to reduce the complexity of extremely

complex mixtures, and in the specific example of the oligostyrenes, expression of the

diastereomer sample attribute was achieved by incorporating into the system a shape selective

separation dimension. While for the most part 2D HPLC has been employed for high resolution

separations and sample profiling, the focus of the following chapter is on targeting the isolation

and purification of specific compounds within the complex mixture at the preparative level. The

aim at the preparative level is usually very different to that at the analytical level. Rarely is it

necessary to isolate all the components in the sample in the manner that would yield a chemical

signature, for example. Instead the aim is usually focused on maximising the recovery of targeted

compound/s. Therefore 2D preparative HPLC (2D PHPLC) systems may be employed in a

manner whereby the first dimension effectively reduces the complexity of the sample matrix,

while the second dimension then resolves the target compound from the less complicated sample

matrix that was transferred to the second dimension. In this instance a higher degree of

correlation between each dimension can be tolerated, sometimes deliberately introduced in order

to facilitate speed in separation; the system may in fact even be reduced to the level of a simple

column switching process in which there is very little difference in selectivity between each

103

dimension [127]. This, however, was not the case in the present study, as both dimensions were

almost orthogonal for this sample matrix.

The production rate (See equation 1.1) is an important measure of the performance in PHPLC,

albeit a measure that is largely academic as it does not take into account true working

environments. As applied to 2D PHPLC the production rate is partly limited by the overload

conditions on the column in the first dimension and also by the fact that two columns, instead of

one, are required for operation – hence increasing solvent consumption, and if not correctly

operated could lead also to an increase in the cycle time [53]. However, a key aspect of

optimisation is to make full use of the total system dead time in order to increase the production

rate [53]. For this to occur, isocratic mobile phases should ideally be employed in both

dimensions even for very complex samples since in isocratic separations there is no re-

equilibration time and hence the dead time can be minimized. When the aim of the purification

process is orientated specifically towards a key target component from within the complex

mixture, isocratic elution is possible for even very complex samples. The aim of the first

dimension is to bring about elution of all components quickly, with minimal retention. The target

analyte is then transferred to the second dimension, but in a far less complex sample matrix.

Since the sample that is separated in the second dimension is far less complex, there is a

decreased demand on the available peak capacity in the second dimension. Hence higher

resolution, in an isocratic system that is tuned for optimal retention of the target analyte can be

obtained. This decreased sample complexity in the second dimension results in faster separation

times, hence reducing dead time as gradients are not required. If gradient elution where to be

used in the second dimension, this would necessitate the use of two columns in the second

104

dimension so that while one column is regenerating the other is used for separation. The

disadvantage of this is increased solvent consumption and operational complexity.

5.2.1 Production rate variables

Although the production rate was designed for operation in a 1D system its experimental

variables are still valid for 2D PHPLC and thus provides a starting point in which to model

purification. As the focus is now the isolation of a targeted component/s at a preparative scale,

the experimental variables that can maximise productivity and yield and at the same time

minimise mobile phase consumption and total operational costs need to be examined. The

choices of experimental variables to be used for optimisation strategies depends on the

requirement of the preparative separation and are usually associated with the production costs of

the starting material, the solvent costs, the cost of the final product and the purity criteria. Since

the experimental variables are interconnected, when one variable is optimised it can have a

detrimental effect on another variable, for instance when the product purity is maximised the

final productivity is generally lowered. However, reducing the cycle time can have a positive

effect by increasing the productivity and also decreasing the solvent consumption. Impurities that

elute close to the target component also place constraints on optimisation parameters such as

production rate, product recovery yield and purity.

5.2.1.1 Sample volume and sample concentration

For a non-overloaded column the volume of sample to be injected generally should not exceed

one-third the volume of the earliest peak of interest to generate maximum resolution [36].

105

However in preparative chromatography where large amounts of purified product are required it

is usual for the column to be operated under overloaded conditions and this results in distortion

of the band and a reduction in the resolution [36]. The recovery of the target component in a 2D

PHPLC system is therefore in effect determined by the band broadening that occurs due to the

overloading in the first dimension.

The volume of the sample that can be introduced by injection onto a column is dependent upon

the column internal diameter and length, the solubility of the sample and the resolution criteria

[36]. The internal column diameter determines the amount of sample that can be loaded onto a

column, so as diameter increases the load also increases without a reduction in the

chromatographic efficiency [36]. The column length determines the number of theoretical plates

of the chromatographic separation and thus resolution [36]. The solubility of the sample is also

important as precipitation of concentrated material on the column inlet frit, detectors and

chromatographic plumbing will be detrimental to chromatographic separations [36].

A unique consideration with respect to productivity in 2D HPLC isolations is the effect that the

volume of the heart-cut plug has on the second dimensional separation as a result of the increase.

The larger the area that the target component occupies in the first dimension the larger the heart-

cut required to be transported to the second dimension. Issues of solvent compatibility between

the mobile phases of each dimension are important, more so as the transfer volume increases.

106

5.2.1.2 Product recovery yield Yi

The product recovery yield Yi, is defined as the ratio between the amount of the desired

component i recovered in the collected fraction at the required purity to the amount of the

component in the injected sample [98]. The product recovery yield for component, i, is given by

equation 5.1:

iniAin

iY (5.1)

Where ni is the amount of sample injected and Ai is the amount of component collected in the

purified fraction. The product recovery yield is the fraction of the component recovered in the

purified fraction as the final product. In a 2D PHPLC system the product recovery yield becomes

a function of the recovery of the heart-cut from the first dimension and the target component

collected from the second dimension [53]. The recovery of the target component in 2D PHPLC

systems therefore is in effect determined by the band broadening in the first dimension and the

subsequent volume of heart-cuts that can be transported to the second dimension, which comes

back to how much sample can be injected onto the first dimension column as previously

discussed.

A more simplified product recovery yield is the mass of purified product collected per injection

as ascertained from a calibration curve of the sample. This is more suited to 2D PHPLC systems

as here very complex mixtures are employed where the target is only one component of many

thousands. The mass per injection Mi can be given by equation 5.2:

fV

fCiM (5.2)

107

where Cf is the concentration of the target component (units of g/L) of the final product and V f

is the volume of the collected fraction.

5.2.1.3 Cross-sectional surface area and total porosity

The production rate is assumed to be proportional to the cross-sectional surface area of a column

[98], so it follows that if there were an increase in the surface area there would be an increase in

the production rate since the amount of sample that can be loaded onto the column increases. In a

2D system it is the column in the first dimension that is overloaded and limits the sample load,

because the column in the second dimension only receives a fraction of the sample in the form of

a heart-cut from the first dimension; it therefore does not exhibit overloaded tendencies and

behaves in a linear manner akin to analytical scale.

5.2.1.4 Cycle time

Another important variable is the cycle time. Reducing the cycle time can result in a great

improvement in the final production rate. The cycle time is the time that separates two successive

injections. An important consideration in minimizing the cycle time is to ensure that all of the

separation space is utilized all of the time. As a rule of thumb, the second dimension separation

should be completed by the time the next heart-cut fraction from the next injection onto the first

dimension is ready to be transported to the second dimension. As each dimension has

independent flow the second dimension effectively does not become operational until the target

analyte is heart-cut to the second dimension. Ideally the empty separation space that results from

this period of waiting should be utilized and hence the injection of the sample into the first

108

dimension should be timed to take advantage of this empty separation space in the second

dimension, thus reducing the cycle time.

5.2.1.5 Purity

Although not a variable of the production rate, the purity is a very important experimental

parameter that is usually the main limitation of the preparative separation. The purity

requirements determine how much of the fraction is collected and therefore impact on the yield,

the production rate and the economics of the process. Although purity is generally desired at

greater than 99 % sometimes this is not achievable and is a limiting factor when faced with

complex mixtures particularly. Additional polishing steps may need to be employed to bring

about further purification; such steps may involve the re-injection of the target component for the

removal of impurities or in cases where purity is severely compromised techniques such as

recycling may be beneficial.

5.2.1.6 Effective and practical production rate

Each experimental parameter can impact on the other variables and thus have serious

implications for the production rate. As this study has a fixed column format (volume of sample,

concentration of sample, porosity and cross-sectional surface area) with the only experimental

variables that are to be optimized are the product recovery yield and cycle time. The production

rate can be further simplified to the effective production rate [129](equation 5.3):

109

tcm

irEffP (5.3)

where m is the mass of the product recovered and tc is the cycle time. This allows for the

examination of these variables and their implications not only to the production rate but also the

other experimental variables.

If the solvent consumption where to be factored in as a requirement of the economic costs of the

preparative separation, the practical production rate would then be the amount of solvent

consumed per unit amount of purified product prepared [91]. It is an important contributor to the

entire cost of the production and should be given consideration to the overall cost of a

preparative separation. The amount of solvent used during a cycle is the product of the cycle

time and the flow rate. The practical production rate [98] is given by equation 5.4:

urEffP

iracP Pr (5.4)

Along with the effective production rate and the product recovery yield, the practical production

rate can be used to optimise the experimental parameters.

The purpose of the present study is to illustrate the application of 2D PHPLC for the isolation of

one or several target diastereomers that belongs to a family of 2054 compounds that only differ

in their molecular weight and spatial orientation. Several of the experimental parameters that are

related to the production rate will be investigated in this chapter and the following chapter to

define a multidimensional preparative system model. In Chapter 4 purifications undertaken at

110

low sample loads, using small injection volumes were examined. Recovery was assessed at

varying levels of targeted purity. Low mass recovery purifications of this nature are useful for

targeted isolations from complex species, where perhaps the aim is to undertake further

examination of the substance, possibly by LC-NMR and consequently large „in-hand‟ portions

are not required, rather a quick and efficient isolation is preferred. In this chapter, the primary

interest is in the minimisation of the cycle time leading to maximising the recovery yield. The

sample load remained constant, and the total porosity and the cross sectional surface areas of the

columns were fixed parameters of the separation; accordingly only the cycle time and product

recovery yield were considered.

5.3 Experimental

5.3.1 Chemicals

As described in Section 4.2.1.



5.3.3 Calibration

A calibration curve was constructed by injecting an oligostyrene standard at several known

concentrations and computing response factors based upon the linear regression of a plot of peak

area versus concentration.

111


The target diastereomer n = 5#2 eluted from the first dimension between 1.60 and 2.30 minutes.

The last oligostyrene (n = 10) eluted from the first dimension after 6.3 minutes (Figure 5.1).

Following a 100 μL heart-cut fraction of the target analyte (n = 5#2) into the second dimension,

the last eluting diastereomer was n = 5#8 with an elution volume to the peak tail of 14.0 minutes.

Consequently there was 7.7 minutes difference in the time period between the completion of the

separations in the first and second dimensions. This dead time in the first dimension is thus

wasted time and productivity in the isolation process of the target analyte.

0 1 2 3 4 5 6 7 8

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

109

8

7

6

5

43

2

1Inte

nity

(m

V)


Figure 5.1: Chromatogram of the separation of tert-butyl oligostyrene separation on C18 column. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection volume. Detection UV 272 nm.Oligomers numbered 1-10 accordingly.

112

As cycle time has an enormous impact on the effective and practical production rate it is

essential to maximise any vacant separation space that can reduce the cycle time, therefore

increasing production. Consideration into reducing the cycle time was dependent upon

limitations of the individual columns and operating in the confines of the 2D PHPLC system. As

the CCZ column in the second dimension was operating at the lower end of its pressure limits it

was therefore feasible to increase the flow rate without difficulty ensuring that the second

dimension was sufficiently fast so that the dead time was eliminated from the isolation process.

The flow rate in the second dimension was subsequently increased to 4.0 mL/min. with the first

dimension flow rate set at 1.0 mL/min. The last diastereomer 5#8 in the second dimension

subsequently eluted within 7.0 minutes.

The void times of both dimensions; less than one minute in the first dimension and 2.3 minutes

in the second dimension, the injection delay for a 10 µL injection was one minute (the required

time for the autosampler to go through the process of injection), and the elution in the second

dimension of all components in 7.0 minutes; combined reduced the cycle time to 7.3 minutes.

This ensured that the separation space was fully utilised and that a continuous operation would

permit the injection of the sample every seven minutes and that the target component could be

collected 8.5 times per hour. The time line in Figure 5.2 demonstrates the void times in both

dimensions and the last eluting peaks from each dimension; it also illustrates the points of

injections and where the sample is heart-cut.

113

Time (min) Figure 5.2: Time line of 1D and 2D separation.

To demonstrate that the isolation of the target component could be achieved in a preparative

mode, the target was isolated and collected in a continuous manner consisting of five successive

injections. Figure 5.3 is the separation in the first dimension for five successive injections. The

separation space in the first dimension visually appears to still have scope for improvement as

there was some apparent empty separation space. In Figure 5.4 the chromatogram has been

expanded in the first dimension to illustrate that in fact this empty space contained the elution of

higher order oligomer fractions, present at low concentrations. Hence, in order to avoid

contamination following scale-up to higher sample loads, this separation spaced remained

unusable.

Void 1D

0 1.0 2.3 6.3 7.0 7.3

Void 2D

Last eluting peak 2D

Inject

Heart-cut

Last eluting peak 1D Inject

Void 1D

114

0 5 10 15 20 25 30 35-500

0

500

1000

1500

2000

2500

3000

3500

Inte

nsity

(m

V)

Retention time (minutes)

Figure 5.3: Chromatogram of the five successive injections onto 1st dimension. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection volume.

0 1 2 3 4 5 6 7

-100

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

1300

1 .0 1 .5 2 .0 2 .5 3 .0 3 .5 4 .0 4 .5 5 .0 5 .5 6 .0 6 .5 7 .0

-3 5

-3 0

-2 5

-2 0

-1 5

-1 0

-5

0

5

1 0

Inte

ns

ity

(m

V)

R e te n tio n t im e (m in )

Inte

nsi

ty (

mV

)


5

6

7

8

910

1112

13

Figure 5.4: Chromatogram of successive injections onto 1st dimension showing empty separation space and heart-cut. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection volume.

Heart-cut

115

Figure 5.5 shows the 2D separation of the five successive injections. The doubling of the flow

rate in the second dimension allowed for a fast separation and all the separation space to be

utilised. In this preparative mode the target component eluted every 7.3 minutes. This repetitive

injection process, with maximum utilisation of the separation space in both dimensions allowed

for the automation of the 2D preparative system.

0 5 10 15 20 25 30 35 40

-50

0

50

100

150

200

Inte

nsity

(m

V)

Retention time (minutes)

Figure 5.5: Chromatogram of the five successive injections onto 2nd dimension. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 4.0 mL/min. 700 μL heart-cut volume.

116

5.4.1 Recovery yield of the target component

The strategies previously examined for purity and recovery of the target component in Chapter 4

were investigated in this chapter for recovery yield (at low sample injection volume) of the target

component in the final product.

5.4.1.1 Product recovery of the target component from central area of band in 1D and 2D

The product recovery yield is the only experimental parameter that needs to be considered as all

other parameters were fixed, including purity. The product recovery yield as defined for this

complex separation is the mass of the target component that is recovered in the purified fraction

as the final product (equation 5.1). From calibration data determined in Section 5.3.3 the

concentration of the target component for each heart-cut was determined. This was multiplied by

the volume of the fraction collected to give the mass per injection of the targeted component.

Results in Table 5.1 detail the mass of target per injection, the effective production rate, the

practical production rate and the purity. The mass for the 100 µL heart-cut fraction transferred to

the second dimension and then subsequently collected from the second dimension was the lowest

at 6.0 10-4 mg (per injection) and the highest mass collected was for the 500 µL and 600 µL

heart-cut fractions (each yielding 1.6 10-3 mg per injection). Subsequently, the effective

production rate (equation 5.3) for the 500 and 600 µL heart-cut fractions was 1.3 10-2 mg/h.

The practical production rate (equation 5.4) for the 600 µL heart cut fraction, which takes into

consideration the solvent consumption was 4.4 10-2 mg/h/L. For the 600 µL heart-cut fraction,

which had a total recovery of 77 % this would result in 1.058 mg of sample being produced per

day (based on the column format used for these trials) if continuous operation were feasible. The

117

purity, however, for the 600 µL heart-cut fraction was only 65 %. A more attractive option

would be the 500 µL heart-cut fraction, which had a total recovery of the target component at 79

% and a practical production rate of 4.4 10-2 mg/h/L with purity at 82%. The 100 µL heart-cut

fraction yielded the highest purity of collected product (90%), but the loss of target analyte was

76% (i.e. 24% recovery), and this reduced the practical production rate 1.7 10-2 mg/h/L.

Subsequently, the trade off in purity resulted in twice the productivity for a reduction in purity by

only 14 (and importantly, a constant number of impurities). Considering each sample contained

the same impurities, the more cost effective option was to maximize the production rate,

irrespective of the purity because the same final polishing step would be applied to both

products.

Although >99 % purity is the ideal aim, this cannot be achieved for this type of separation

problem, unless a substantial sacrifice is made with respect to yield (i.e. 24 % recovery resulted

in 90 % purity). For the larger heart-cut volumes the total recovery increased, but at the cost of

purity. This begs the question, is it worth the loss of sample for a gain in purity, not to mention

the extra time it would take to achieve the same amount of sample and the additional solvent

required. When the original sample is quite expensive the loss of sample particularly at such

high levels would not be financially viable and would notably impact on the cost of production.

At this point, polishing steps were not undertaken to improve the product purity. Nonetheless a

highly purified product with high recovery may be obtained through other means, such as re-

injection into both the C18 and CCZ systems, whereby the C18 removes the contamination by

other oligomers (most notably n = 4) and the CCZ removes the contamination of the

118

diastereomers (5#1 and 5#3). Such a process could feasibly be implemented in a recycling

system, but this was not tested here.

Table 5.1: Characteristics of two-dimensional separation of target component n =5 #2

Injection

volume

Cut time

Cut

volume

Mass of target

per injection

Effective

production

rate

Practical

production

rate

Purity

(µL)

(min)

(µL)

(mg) (mg/h)

(mg/h/L)

(%)

10 1.90-2.00

100 0.0006 0.0052 0.017 90

1.85-2.05

200 0.0012 0.010 0.034 83

1.80-2.10

300 0.0014 0.011 0.037 84

1.75-2.15

400 0.0014 0.011 0.038 83

1.70-2.20

500 0.0016 0.013 0.044 82

1.65-2.25

600 0.0016 0.013 0.044 65

1.60-2.30

700 0.0015 0.013 0.042 59

119

5.4.1.2 Off-centre in 1D

In an attempt to improve product purity and practical production rate, offset heart cutting was

implemented. Results in Table 5.2 detail the mass of target per injection, the effective production

rate, the practical production rate and the purity. The mass of target analyte per injection was

highest for the 400 µL heart-cut for the 2.05 minute centred band at 1.6 10-3 mg and the 300

µL heart-cut for the 2.10 minute centred band had the lowest mass per injection at 5.0 10-4 mg.

Note larger „off-centred‟ heart cut sections did not yield higher recovery since volumes beyond

500 µL exceeded the elution region of the target analyte in the first dimension. The effective

production rate was also highest for the 400 µL heart-cut (2.05 min) at 1.3 10-2 mg/h which

also had the highest practical production rate at 4.4 10-2 mg/h/L. The lowest practical

production rate was for the 300 µL heart-cut with a centre band of 2.10 minute centred band at

1.4 10-2 mg/h/L. The practical production rate, which takes into account the overall solvent

consumption was highest for the heart-cuts of 2.05 min with heart-cuts of 2.10 minutes being

measurably lower.

Purities were consistently higher and greater than 89 % for all heart-cuts. Off-centering the heart-

cut fraction of the target analyte away from the limiting impurity (n = 4 #3) in the first dimension

resulted in a highly purified product with high practical production rates. This is in spite of the

fact that a large percentage of the target component was lost to waste (between 28 % to 71 %

depending on the cutting volume and location of the offset heart-cut time period – see Table 4.2).

Also just by shifting the heart-cuts another 0.10 minutes away from the n = 4#3 not only

dramatically reduced the overlap but resulted in very pure and high practical production of the

target component. Therefore, the first dimension can be optimised to bring about higher purity

120

and yield of the target component but once again this is dependent upon the aim of the

preparative isolation.


Injection

volume

Cut time

Cut volume

Mass of

target per

injection

Effective

production

rate

Practical

production

rate

Purity

(µL)

(min)

(µL)

(mg) (mg/h)

(mg/h/L)

(%)

10

Centre Band 2.05 min

1.90-2.20

300 0.0015 0.012 0.042 91

1.85-2.25

400 0.0016 0.013 0.044 89

1.80-2.30

500 0.0014 0.012 0.039 93

Centre Band 2.10 min

2.00-2.30

300 0.0005 0.0042 0.014 93

1.95-2.35

400 0.0010 0.0080 0.027 90

1.90-2.40

500 0.0009 0.0076 0.025 93

121

To demonstrate the overlap of n = 4#3 and n = 5#2 in the first dimension, Figures 5.6, 5.7 and

5.8 highlight the region that the target n = 5#2 occupies (1.60-2.30 minutes). The overlap

between n = 4#3 and n = 5#2 occurs between 1.60-2.20 minutes. Also illustrated are the heart-

cuts with the different central retention times (1.95, 2.05 and 2.10 min) numbered 1,2 and 3

respectively. These figures demonstrate that the overlap of the impurity with the target

component is a significant factor in the recovery, purity and therefore the practical production

rate of the target. The 300 µL heart-cut (2.05 min) had overlap of 50 % (by time) with the n =

4#3 impurity (Figure 5.6) and encompassed 69 % of the n = 5#2 target component; total recovery

was 53 % and purity 91 % (Table 4.2) with practical production at 4.2 10-2 mg/h/L (Table 5.2).

The 300 µL heart-cut (1.95 min) (Section 5.4.1.1) had overlap of 50 % (Figure 5.6) and

encompassed 83 % of the target component; total recovery was 65 % and purity 84 % (Table

4.1), with practical production at 3.7 10-2 mg/h/L (Table 5.1). So even though they both had 50

% overlap with n =4#3 impurity, by decreasing the area that the target encompassed by shifting

the heart-cut (2.05 min) away from the impurity saw an increase in purity compared to the same

heart-cut fraction (1.95 min). The 300 µL heart-cut (2.10 min) had an overlap of 33 % (Figure

5.7) and encompassed 35 % of the target component with total recovery of 29 % and purity of 93

% (Table 4.2), with practical production of 1.4 10-2 mg/h/L (Table 5.2).

The 400 µL heart-cut (2.05 min) had the greatest practical production rate at 4.4 10-2 mg/h/L

(Table 5.2). Production for the 400 µL heart-cut (2.05 min) would result in 1.06 mg of sample

being produced per day if continuous operation were feasible. The 400 µL heart-cut (2.05 min)

had 58 % overlap with the n = 4#3 impurity (Figure 5.7) encompassing 84 % of the target

component, total recovery was 69 % at 89 % purity (Table 4.2). The 400 µL heart-cut (1.95 min)

122

had 67 % overlap (Figure 5.7) encompassing 91 % of the target component with 69 % total

recovery and 83 % purity (Table 4.1) and a production rate of 3.7 10-2 mg/h/L (Table 5.2).

Clearly shifting the heart-cuts away from the n = 4#3 impurity had benefits for improving the

purity as the overlap was reduced and the purity increased. However, the 400 µL heart-cut (2.10

min) had 42 % overlap (Figure 5.7) encompassing 53 % of the target component with 46 % total

recovery at 90 % purity (Table 4.2), but the practical production rate was reduced to 2.7 10-2

mg/h/L (Table 5.2). So even though the overlap was reduced to 42 % the area of the target

component for this heart-cut occupied was reduced to 53 %; recovering less than half of the

target.

1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

0.0

0.2

0.4

0.6

0.8

1.0

32

1

n =4#3 n =5#2

ImpurityTarget

Nor

mal

ised

pea

k ar

ea (

mV

*sec

)


Figure 5.6: Overlap of n = 4#3 and n = 5#2 on C18 column for 300 µL heart-cuts from the first dimension. (1) 1.95 minute centre band (2) 2.05 minute centre band (3) 2.10 minute centre band. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 10 μL injection volume.

123

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

0.0

0.2

0.4

0.6

0.8

1.0

32

1

n =4#3 n =5#2

ImpurityTarget

Nor

mal

ised

pea

k ar

ea (

mV

*sec

)


Figure 5.7: Overlap of n = 4#3 and n = 5#2 on C18 column for 400 µL heart-cuts from the first dimension. (1) 1.95 minute centre band (2) 2.05 minute centre band; (3) 2.10 minute centre band. Conditions as in Figure 5.6.

The 500 µL heart-cut (2.05 min) had the next best practical production rate at 3.9 10-2 mg/h/L

with an overlap of 67 % (Figure 5.8) and encompassing 93 % of the target component, total

recovery was 72 % and purity 93 % (Table 4.2). In comparison to the same heart-cut (1.95 min)

that had overlap of 83 % (Figure 5.8) and encompassed 96 % of the target component, total

recovery was 74 % and purity 82 % (Table 4.1), production rate at 4.4 10-2 mg/h/L (Table 5.1).

The 500 µL heart-cut (2.10 min) had 50 % overlap (Figure 5.8), encompassing 71 % of the target

component with 57 % recovery at 93 % purity (Table 4.2) with the practical production rate at

2.5 10-2 mg/h/L (Table 5.2). Consequently for the 500 µL heart-cuts (2.05 and 2.05 min) the

124

reduction in the practical production is due to the heart-cut encompassing slightly less of the

target component, therefore reducing the total recovery.

1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

0.0

0.2

0.4

0.6

0.8

1.0

32

1

n =4#3 n =5#2

ImpurityTarget

Nor

mal

ised

pea

k ar

ea (

mV

*sec

)


Figure 5.8: Overlap of n = 4#3 and n = 5#2 on C18 column for 500 µL heart-cuts from the first dimension. (1) 1.95 minute centre band (2) 2.05 minute centre band (a); (3) 2.10 minute centre band (b). Conditions as in Figure 5.6.

125

5.4.1.3 Multicomponent

Production for the eight diastereomers for the 500 µL heart-cuts is given in Table 5.3.

Diastereomer 5#8 has the highest effective production at 4.8 10-2 mg/h and practical

production of 1.60 10-1 mg/h/L and diastereomer 5#2 had the lowest practical production at 3.9

10-2 mg/h/L. This was understandable since 5#8 could be isolated in 96 % recovery at 100 %

purity (Table 4.4), while 5#2 had total recovery of 62 % at 75 % purity (Table 4.4). Even though

analyte 5#3 was less pure than 5#2 (32 % compared to 75 %) (Table 4.4) the practical production

of 5#3 at 5.9 10-2 mg/h/L was greater than 5#2 at 3.9 10-2 mg/h/L, as it was present within

the sample at higher concentrations. This in itself is an interesting side story, as the

diastereomers of the oligostyrene were not equally concentrated. Correcting for the variation in

diastereomer distribution resulted in almost constant production rates for all diastereomers (

0.8% R.S.D).

For the primary target diastereomer, 5#2, the mass of target per injection was highest for the 400

µL and 500 µL heart-cut at 1.4 10-3 mg as well as the effective production rate at 1.2 10-2

mg/h (Table 5.4). The 400 µL heart-cut also had the highest practical production rate at 3.9 10-

2 mg/h/L. The lowest practical production rate was for the 100 µL heart-cut at 9.4 10-3 mg/h/L.

Production for the 400 µL heart-cut would result in 0.9456 mg of sample being produced per day

if continuous operation were feasible. The 400 µL heart-cut volume had total recovery of 61 % at

81 % purity (Table 4.4).

126

Table 5.3: Characteristics of two-dimensional separation of target component n = 5 #1-8 for 500 µL heart-cut

Injection

volume

Cut time

Diastereomer

Mass of

target per

injection

Effective

production

rate

Practical

production

rate

Purity

(µL)

(min)

(#)

(mg) (mg/h)

(mg/h/L)

(%)

10 1.85-

2.05

1 0.0020 0.018 0.058 90

2 0.0014 0.012 0.039 75

3 0.0020 0.018 0.059 32

4 0.0039 0.033 0.11 60

5 0.0043 0.037 0.12 65

6 0.0037 0.032 0.11 79

7 0.0049 0.042 0.14 100

8 0.0056 0.048 0.16 100

127


Injection

volume

Cut

time

Cut volume

Mass of target

per injection

Effective

production

rate

Practical

production

rate

Purity

(µL)

(min)

(µL)

(mg) (mg/h)

(mg/h/L)

(%)

10 1.90-

2.00

100 0.0003 0.0028 0.0094

80

1.85-

2.05

200 0.0010 0.0083 0.028

80

1.80-

2.10

300 0.0012 0.010 0.035

87

1.75-

2.15

400 0.0014 0.012 0.039

81

1.70-

2.20

500 0.0014 0.012 0.039

75

1.65-

2.25

600 0.0011 0.0096 0.032

66

1.60-

2.30

700 0.0013 0.011 0.036

77

128

5.4.1.4 Maximising recovery

In an attempt to maximise the recovery of the target analyte, a larger area was collected from the

second dimension with the same centralized heart-cut from the first dimension as in Sections

5.4.1.1 and 5.4.1.2. From Table 5.5 the mass per injection of the target component for the 100 µL

heart-cut was the lowest at 6.0 10-4 mg and highest was for the 700 µL heart-cut at 2.1 10-3

mg per injection. The effective production rate was highest for the 700 µL heart-cut at 1.78 10-

2 mg/h and when the solvent consumption was taken into account 5.9 10-2 mg/h/L of the target

component could be produced. For the 700 µL heart-cut, which had a total recovery of 99 % with

a final purity of 14 % (Table 4.6) production would result in 1.41 mg of sample being produced

per day if continuous operation were feasible. However, under these conditions, contamination

was due to not only larger concentrations of the existing contaminants, but also due the presence

of more contaminating species. Hence, this would further complicate the necessary polishing

steps. For the larger heart-cut volumes, although the total recoveries were much higher, the

purity (Table 4.6), was severely compromised largely due to the unavoidable impurities that

were a result of the chemical nature of the oligostyrene sample.

129

Table 5.5: Characteristics of two-dimensional separation of target component n = 5 #2 for 500 µL heart-cut

5.4.2 Summary

The comparison of the practical production rate for each of the strategies for the 300, 400 and

500 µL heart-cut used from all sections in this chapter is shown in Figure 5.9. The practical

Injection

volume

Cut

time

Cut

volume

Mass of target

per injection

Effective

production

rate

Practical

production

rate

Purity

(µL)

(min)

(µL)

(mg)

(mg/h)

(mg/h/L)

(%)

10 1.90-

2.00

100

0.0006 0.0053 0.018 26

1.85-

2.05

200

0.0015 0.012 0.040 27

1.80-

2.10

300

0.0018 0.015 0.048 28

1.75-

2.15

400

0.0018 0.015 0.050 28

1.70-

2.20

500

0.0019 0.016 0.052 26

1.65-

2.25

600

0.0020 0.016 0.054 22

1.60-

2.30

700

0.0021 0.018 0.059 14

130

production increased as the heart-cuts increased with the exception of the 2.05 and 2.10 minute

heart-cuts (Section 5.4.1.2); again this was due to the heart-cuts extending away from the target

component as to lessen the impurity. The highest practical production for all heart-cuts were

found when the recovery was maximised (Section 5.4.1.4) with production rates ranging from

4.8 10-2 mg/h/L to 5.2 10-2 mg/h/L (Table 5.5). However, these heart-cuts also had extremely

low purities ranging from 26% to 28 % (Table 5.5).

The highest purity product (> 90 %) was a result of off-centering the heart-cut to be transported

to the second dimension the farthest away from the recurring impurity for the 2.10 min centred

bands (Table 4.2). The final product, however, had the lowest recovery (Table 4.2) and lowest

practical production rate of all strategies ranging from 1.41 10-2 mg/h/L to 3.67 10-2 mg/h/L

as demonstrated in Figure 5.9.

The 500 µL heart-cut with central retention times of 1.95 minute (Section 5.4.1.1) and the 400

µL heart-cut with central retention time of 2.05 minute (Section 5.4.1.2) had equivalent practical

production rates at 4.4 10-2 mg/h/L. The recoveries were 74 % and 69 % respectively; the

purities were similar 82 % and 89 % respectively (Tables 4.1 and 4.2). Clearly off-setting the

heart-cut bands even for a smaller heart-cut resulted in a slightly more purified fraction and

therefore a more highly concentrated target component within the final product. The practical

production for the 2.10 minute centered bands (Section 5.4.1.2) however had the lowest practical

production rates, as their recoveries were much lower (Table 4.2) although the purities were

greater than 90 %.

131

Figure 5.9: Comparison of the practical production rate of the target diastereomer for 300, 400 and 500 µL heart-cuts. 5.5 Conclusion

Low sample loads were examined in this chapter investigating the experimental parameters that

affected the preparative isolation of the target component. The highest practical production rate

of the final product was a result of collecting the highest total recovery as in Section 5.4.1.4;

however, this also resulted in the lowest purity of all strategies. The highest purity product was a

0.00

0.01

0.02

0.03

0.04

0.05

Multi- Component 5.4.1.3

Off-centre 2.10 min 5.4.1.2

Off-centre 2.05 min 5.4.1.2

High Recovery 5.4.1.4

High

Purity 5.4.1.1

(mg/h/L)

300 u L heart-cut 400 u L heart-cut 500 u L heart-cut

132

result of off-centering the heart-cut to be transported to the second dimension the farthest away

from the recurring impurity in Section 5.4.1.2 (2.10 min). The final product, however, had the

lowest recovery and lowest practical production rate of all strategies. Intermediate results were

achieved with total recovery, purity, and production rate for both Sections 5.4.1.1 and 5.4.1.2

(2.05 min) that saw the heart-cut being off-centred to reduce overlap of impurity. A consequence

of the low sample loads was that the first dimension was operated essentially under analytical

conditions and the potential to load the first dimension column with sufficient sample was not

reached.

133

Chapter 6


Preparative Scale Two-dimensional

Isolations: High Sample Loads

134

6.1 Introduction

The focus of study in Chapter 5 was to implement continuous operation of the batch-wise

purification process, minimising the cycle time at low sample loads. The focus of the study in

this chapter is to employ the continuous batch-wise 2DHPLC system for preparative scale

isolations, using high sample loads. Specifically, the recovery in both the first and second

dimensions; the purity of the collected product, the product recovery yield, the effective

production rate and the practical production rate will be investigated.

6.2 Experimental

6.2.1 Chemicals



The diastereomer separations were undertaken using a Waters 2D LC system. Columns were: (1)

A C18 SphereClone (50 × 4.6 mm) column, which was used with a 100% MeOH mobile phase

in the first dimension, and (2) a CCZ column (either 50 x 4.6 mm or 100 × 4.6 mm as stated in

the text when appropriate), which was used with a 100% ACN mobile phase in the second

dimension. All experiments on the C18 column were conducted under ambient temperature,

while the CCZ column was thermostated to 45°C. Injection volumes into the C18 column are as

stated in the text.

135

6.2.3 Determination of product purity and recovery


6.2.4 Calibration

As described in Section 5.3.3


6.3.1 Sample load limitations on 1D and 2D columns

Low sample loads were examined in Chapter 5 investigating the experimental parameters that

affected the preparative isolation of the target component. A consequence of the low sample

loads was that the first dimension was operated essentially under analytical conditions and the

potential to load the first dimension column with sufficient sample was not reached. Therefore

the ability of the 2D isolation process to function at a preparative level was tested at higher

sample loadings. Figure 6.1 illustrates the impact that sample load volume had on the resolution

of the oligostyrenes on the first dimension C18 column. The target analyte is indicated by the

arrows. The chromatographic elution profiles following a 10 L injection volume were almost

normally distributed with good oligomer to oligomer resolution. As the injection volume

increased resolution decreased substantially. At 30 L injection volume overloading was

evident; oligostyrene bands were less distinctive and band broadening increased. For the 50 L

injection volume the oligostyrenes eluted as a continuum, with no peak to peak resolution of the

oligomer fractions visible.

136

As sample size increased there was a decrease in the retention time and a decrease in the sample

resolution due to overloading. In addition to the overload on the first dimension, the effect the

heart cutting process had on the separation in the second dimension was also considered.

Increasing the sample load in the first dimension increased the elution volume of the target

analyte in the first dimension. Hence, transport to the second dimension demanded that the

second dimension column handle; (a) higher mass loadings and (b) higher volume heart-cuts. If

the second dimension cannot operate efficiently under these conditions then the recovery of the

target analyte from the first dimension could be severely diminished. The effects of overloading

from the first dimension therefore may have repercussions for the second dimension separation,

particularly since increasing heart-cut volumes into the second dimension would be necessary to

capitalise on the larger sample loads applied to the first dimension.

The elution region of the target diastereomer in each of the three sample loadings was

determined using incremental 100 L heart-cutting sections from the first dimension, and

subsequently analysing the response in the second dimension. Figure 6.2 illustrates the influence

that the larger sample loadings in the first dimension had on the resolution of the diastereomers

in the second dimension. As a consequence of the higher sample loading, the target elution

region in the first dimension was contaminated by many more components than were present at

the lower sample loadings depicted in Chapter 5. Even when the heart-cut volume from the first

dimension was limited to 100 μL the level of contamination and co-elution made it practically

impossible to obtain any degree of target component isolation at a level of purity to make the

task worthwhile, as seen in Figure 6.2. The impact on recovery of the target component and the

137

negative influence on purity were evident. As a result larger heart-cut volumes were not further

investigated.

0 1 2 3 4 5 6

0.0

0.5

1.0

1.5

2.0

2.5

50 uL

Inte

nsity

(m

V)


Target Fraction

10 uL

30 uL

Figure 6.1: Different injection volumes on C18 column. Mobile phase 100% MeOH.

138

2 4 6 8 10 12 14 16 18

0.00

0.01

0.02

0.03

0.04

0.05

0.06

Inte

nsity

(m

V)


(a)

2 4 6 8 10 12 14 16 18 20

0.000

0.005

0.010

0.015

Inte

nsity

(m

V)


(b)

Figure 6.2: Chromatogram of the separation of tert-butyl oligostyrene n = 5 separation on CCZ column, target component peak #2. (a) 30 L injection volume to C18 (5cm), (b) 50 L injection. Conditions: CCZ column (50 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 2.0 mL/min. 100 μL heart-cut volume.

Consequently, further investigations focused on how to improve the separation in the second

dimension, such that it could handle the higher component loading obtained by the overload in

the first dimension. As such, firstly the number of theoretical plates N, in the second dimension

2

2

139

was increased, by way of increasing the column length; and secondly N, in the first and second

dimension was increased, also by increasing the column length.

6.3.2 Increase in N in the second dimension.

The efficiency or rather N in the second dimension was increased by increasing the column

length from 50 mm (as employed for the separation shown in Figure 6.2) to 100 mm. This

increased the number of theoretical plates in the second dimension to 2000 plates (compared to

1000 plates). The effect of this increase in N for the targeted isolation of the diastereomers 5#2

was tested for sample loadings in the first dimension of 30 and 50 μL.

For an injection volume of 30 μL in the first dimension, the target diastereomer 5#2 occupied an

elution region between 1.50 to 2.50 minutes, with a maximum and central retention time of 2.10

minutes. The area of the target oligostyrene n = 5 was then plotted as a function of the cut time

so that maximum recovery from the first dimension could be ascertained. The area that n = 5

oligostyrene occupied in the first dimension for a 30 μL injection volume is shown by the

highlighted section in Figure 6.3. The same process to determine the elution region of the target

diastereomer in the first dimension was performed for the 50 μL injection volume; the elution

region of the target (5#2) was between 1.30 to 2.30 minutes with a maximum and central

retention time of 1.90 minutes in the first dimension as illustrated in Figure 6.4.

140

0 1 2 3 4 5 6 7 8

0.0

0.2

0.4

0.6

0.8

1.0

1.2

n=5

INte

nsity

(mV)


Figure 6.3: Chromatogram of the separation of tert-butyl oligostyrene separation on C18 column displaying area n = 5 occupies. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 30 μL injection volume.

141

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

0.0

0.5

1.0

1.5

2.0

n=5

Inte

nsity

(mV)


Figure 6.4: Chromatogram of the separation of tert-butyl oligostyrene separation on C18 column displaying area n = 5 occupies. Conditions: C18 column (50 mm × 4.6 mm), 100 % MeOH mobile phase at flow rate 1.0 mL/min. 50 μL injection volume.

As a result of increasing N in the second dimension larger heart-cut volumes were able to be

transported to the second dimension and at the same time maintain the appropriate resolution of

the target analyte suitable for recovery.

The separation on the CCZ column with the target analyte highlighted (5#2) is illustrated in the

chromatogram shown in Figure 6.5 following an injection volume in the first dimension of 30

142

μL. Figure 6.5a is the separation that is achieved using a 700 μL heart cut volume (87 %

recovery of target analyte), while Figure 6.5b, illustrates the separation obtained in the second

dimension following a 1000 μL heart cut volume (100 % recovery of target analyte). Likewise,

the separation of the target analyte in the second dimension following a 50 μL injection load in

the first dimension is illustrated in Figure 6.6a (for a 700 μL heart cut volume – 93 % recovery)

and Figure 6.6b (for a 1000 μL heart-cut volume – 100 % recovery). It is worth comparing the

second dimension separations obtained with 2000 plates, at these higher sample loadings and

large heart-cut volumes, to the separations obtained on 1000 plates as demonstrated in Figure

6.2, in which the heart-cut volume was limited to 100 μL. For the separations obtained on the

1000 plate 50 mm CCZ column, resolution of the target analyte was apparent, however as a

consequence of the higher sample loading, the target elution region in the first dimension was

contaminated by many more components than were present at the lower sample loadings

separations. There were however limi tations of this application (increase in N in the second

dimension ) in that the retention was increased, by virtue of the increased column length, and the

pressure effectively doubled, and that ultimately resulted in the flow rate being halved to 1

mL/min. The result was a four-fold increase in the time required to complete the second

dimension separation.

143

0 5 10 15 20 25 30 35 40 45 50

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

87

6

5

43

2

1

Inte

nsity

(mV)


0 5 10 15 20 25 30 35 40 45 50

0.00

0.05

0.10

0.15

0.20

1

2

34

5

67

8

Inte

nsity

(mV)


Figure 6.5: Chromatogram of the separation of tert-butyl oligostyrene n =5 separation on CCZ column, target component peak #2 for 30 μL injection volume. (a) 700 μL heart-cut volume, (b) 1000 μL heart-cut volume. Conditions: CCZ column (100 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 1.0 mL/min.

(a)

(b)

144

0 5 10 15 20 25 30 35 40 45 50

0.00

0.05

0.10

0.15

0.20

0.25

0.30

1

2

34

5

6

87

Inte

nsity

(mV)


0 5 10 15 20 25 30 35 40 45

0.0

0.1

0.2

0.3

0.4

0.5

0.6 1

2

3

4

5

6

78

Inte

nsity

(mV)


Figure 6.6: Chromatogram of the separation of tert-butyl oligostyrene n =5 separation on CCZ column, target component peak #2 for 50 μL injection volume. (i) 700 μL heart-cut volume, heart-cut time 1.55-2.25 min. and (ii) 1000 μL heart-cut volume, heart-cut time 1.30-2.30 min. Conditions: CCZ column (100 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 1.0 mL/min. 100 μL heart-cut volume.

(a)

(b)

145

The results presented in Table 6.1 detail the efficiency of the isolation process. Firstly, for the 30

μL injection load, with a 700 μL heart-cut volume to the second dimension, recovery from the

first dimension was 86 %. Subsequent collection of the target analyte in the second dimension

was 72 % (collection region shown in Figure 6.7). This resulted in a total recovery of 62 %. The

purity of the final product was 70 %. By increasing the heart-cut volume to 1000 μL, the

recovery in the first dimension increased to 100 %, but the recovery in the second dimension was

reduced to 66 % (collection region illustrated in Figure 6.8), with a subsequent total recovery of

66 % for product with 60 % purity. For a 50 μL injection load in the first dimension, and a heart-

cut volume of 700 μL, the first dimension recovery was 93 %, with a second dimension recovery

of 68 %, for a total recovery of 63 % at a purity of 66 %. Increasing the heart-cut volume to 1000

μL ensured 100 % recovery from the first dimension, but the recovery in the second dimension

decreased to 59%, for a total recovery of 59 % at a purity of 51%.

Table 6.1: Characteristics of two-dimensional separation of target component n =5 #2 for different injection volumes

Injection

volume

Centre of

band

(min)

Cut time Cut

volume

Recovery

1D

Recovery

2D

Total

Recovery

Purity

(µL) (min) (µL) (%) (%) (%) (%)

30 2.10 1.55-2.25 700 86 72 62 70

2.10 1.50-2.50 1000 100 66 66 60

50 1.90 1.55-2.25 700 93 68 63 66

1.90 1.30-2.30 1000 100 59 59 51

146

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

Inte

nsity

(m

V)


Figure 6.7: Chromatogram of the separation of tert-butyl oligostyrene separation on CCZ column showing recovery of target component for 30 μL injection volume. Conditions: CCZ column (100 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 1.0 mL/min. 700 μL heart-cut volume.

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0

0.00

0.05

0.10

0.15

0.20

Inte

nsity

(m

V)


Figure 6.8: Chromatogram of the separation of tert-butyl oligostyrene separation on CCZ column showing recovery of target component for 50 μL injection volume. Conditions: CCZ column (100 mm × 4.6 mm), 100 % ACN mobile phase at flow rate 1.0 mL/min. 700 μL heart-cut volume (1.55-2.25 minutes).

147

The greatest overall recovery (66 %), expressed as a percentage of target sample loaded, was

obtained following a 1000 µL heart-cut volume from a 30 µL injection volume in the first

dimension. The lowest recovery (59 %) was obtained for the 1000 µL heart-cut volume from a

50 μL injection on the first dimension.

Not surprisingly, the purities were higher for the 30 µL injections than the 50 µL injections with

the 700 µL heart-cut having a purity of 70 % and the 1000 µL heart-cut band cut at 60 % purity.

The chromatograms in Figure 6.9 illustrate the level of contamination within the sample

following the 700 µL heart-cut volume for the 30 µL injection in the first dimension. Likewise,

the chromatogram shown in Figure 6.10 illustrates the level of contamination for the 700 µL

heart-cut volume following a 50 μL injection in the first dimension. In both cases, the

contamination was limited to two impurities, both of which would be easily removed in a single

step polishing process.

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4-0.0002

0.0000

0.0002

0.0004

0.0006

0.0008

0.0010

0.0012

0.0014 n=5 #2

n=4 #3n=5 #1

Inte

nsity

(mV)


Figure 6.9: Chromatogram of purity for 30 µL injection. 700 µL heart-cut. Fraction collected and re-injected onto CCZ.

148

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.0000

0.0005

0.0010

0.0015

0.0020

n=5 #2

n=4 #3n=5 #1

Inte

nsity

(mV)


Figure 6.10: Chromatogram of purity for 50 µL injection. 700 µL heart-cut (1.55-2.25 min). Fraction collected and re-injected onto CCZ.

6.3.2.1 Production

The recovery and level of purity are only two factors important in the isolation process,

particularly since, irrespective of the degree of recovery or the sample loading, the purity was

only marginally affected, and then, only by the concentration of the contaminants, rather than the

number of contaminants. Hence, if a polishing step is required to yield the desired final product,

the same procedure is required if the target is 51% pure, as it is if it is 70% pure, given the same

contaminating species are present. Hence, the mass of target per injection or rather the

production rate is an important indicator of the purification process. The mass recovery was

highest for the 1000 µL heart-cut volume for the 50 µL injection (9.5 × 10-3 mg per injection),

which yielded an effective production rate at 3.0 × 10-2 mg/h. Taking into consideration the

solvent consumption, the practical production rate was 1.6 × 10-1 mg/h/L. Production of the

149

target diastereomer on the column formats employed here following an injection load of 50 µL

and a heart-cut volume of 1000 µL would thus result in 3.9 mg of sample being produced per day

if continuous operation was applied. The lowest practical production rate was, however,

observed for the 1000 µL heart-cut volume following an injection volume in the first dimension

of 30 µL injection. In this case the practical production rate was 8.4 × 10-2 mg/h/L. While it is not

surprising that the lower injection volume in the first dimension resulted in a decrease in

practical production rate, it was somewhat surprising that the 1000 µL heart-cut volume was less

efficient than the 700 µL heart-cut volume. This was largely due to the fact that the 700 µL

heart-cut volume resulted in a more efficient recovery yield in the second dimension and a

substantial improvement in the purity of the final product, compared to the 1000 µL heart-cut

scenario at the same first dimension sample load. Overall the larger sample loadings into the first

dimension resulted in considerable gains for the practical production rate of the target component

with moderate purities.

150

Table 6.2: Characteristics of two-dimensional separation of target component n=5 #2 for different injection volumes.

6.3.3 Increasing the peak capacity in the first dimension

A limiting factor in the separations obtained for an increase in N in the second dimension

(Section 6.3.2) was the quality of the first dimension separation. Essentially the polymer eluted

as a continuum, with the result being that heart cutting even small volumes to the second

dimension resulted in a large number of components/contaminants being transferred to the

second dimension. Subsequently this placed high demands on the separation performance of the

second dimension column – hence the need to increase the number of theoretical plates. The

Injection

volume

Cut time

Heart cut

volume

Mass of target

per injection

Effective

production rate

Practical

production

rate

(µL)

(min)

(µL) (mg)

(mg/h)

(mg/h/L)

30 1.55-2.25 700 0.0052 0.015 0.085

1.50-2.50 1000 0.0051 0.015 0.0834

50 1.55-2.25 700 0.0079 0.025 0.14

1.50-2.20 700 0.0077 0.024 0.13

1.30-2.30 1000 0.0095 0.030 0.16

151

increase in column length in the second dimension improved the quality of the purification

process. The cost of such an improvement was an increase in pressure, thus a decrease in flow

rate was required, which subsequently increased the cycle time. As the cycle time was increased,

caused only by the new operating conditions in the second dimension there was subsequently

down time in the first dimension because sample could not be reloaded onto the first dimension

until the second dimension was able to accommodate a new sample loading. Hence, in order to

alleviate the high peak capacity demands placed on the second dimension column as a result of

the higher component loading to the second dimension at these higher sample loadings, the first

dimension peak capacity was increased. This had a minimal affect on the cycle time since the

longer column format could largely be accommodated within the existing dead time associated

with the longer runtime in the second dimension. The advantage of increasing the peak capacity

in the first dimension was, however, greater resolution in the first dimension, and hence fewer

components transferred to the second dimension. In order to test this, a 150 mm C18 column was

used instead of the 50 mm column with the same internal diameter. The injection volumes

investigated were 50 μL and 100 μL.

6.3.3.1 Recovery, Purity and Production Rate as a Function of Sample Injection Volume on

the 150 mm C18 column.

6.3.3.1.1 Injection volume: 50 L

The oligomeric separation obtained following a 50 μL injection onto the 150 mm C18

SphereClone column is shown in Figure 6.11. In contrast to the separation obtained on the 50

152

mm C18 column, resolution to ~0.9 was obtained between the n = 4 and n = 5 diastereomers,

with essentially baseline resolution being achieved between n = 5 and n = 6. On the 150 mm C18

column the target diastereomer (5#2) eluted between 6.40 and 8.00 minutes, with a maximum

and central retention time of 7.20 minutes. The volumetric flow rate was 0.83 mL/min, hence

this equated to a peak volume of the oligomer fraction being 1328 μL, which was also the same

as the peak elution volume of the target diastereomer (as in previous Chapters, an incremental

heart-cutting process was used to determine the recovery of the target analyte (plot of heart-cut

region versus area of peak heart-cut to the second dimension)). To ensure 100 % recovery of the

target analyte from the first dimension the heart-cut volume was 1328 μL. As a result the total

recovery of the analyte was 92%, with a subsequent purity of the final product being 91%. The

purity of the product was improved to 94% by decreasing the volume of the heart cut from the

first dimension to 750 µL, however, the total recovery decreased to 62% (68% in the first

dimension). As a result a small gain in purity was observed, but with a substantial loss of sample

(see Table 6.3).

153

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28

0.0

0.2

0.4

0.6

0.8

1.0

6

Inte

nsi

ty (

mV

)


1098

7

5

4

3

2

Figure 6.11: Chromatogram of tert-butyl oligostyrene separation on C18 Sphereclone column (150 × 4.6 mm). 100 % MeOH mobile phase at flow rate 0.83 mL/min, 50 μL injection volume. Oligomers number 1-10 accordingly.

Increasing the length of the first dimension column had a substantial effect on the practical

production rate for the sample load of 50 µL. For a similar mass of product collected per

injection (for the same sample load) on a 50 mm column in the first dimension (Section 6.3.2)

there was a 25 % decrease in the practical production rate (1.6 × 10-1 mg/h/L (Table 6.2)) when

the 150 mm column was employed (1.2 × 10-1 mg/h/L Table 6.3)). This was largely as a result of

the increased solvent consumption associated with the increased elution volume required to elute

from the column, and a slight increase in the cycle time of the system given increased elution

time. The advantage, however, was that the product purity when the 150 mm column was

employed was much higher (91% versus 59%). The purity of the collected product following a

50 µL injection in the first dimension and a 1328 µL heart-cut of the target to the second

154

dimension is illustrated in Figure 6.12. The same two species contaminated the sample, as was

the case for injections onto a 50 mm column (Section 6.3.2), which as a result presents the same

challenges for final polishing of the product.

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.0000

0.0005

0.0010

0.0015

0.0020n=5 #2

n=4 #3n=5 #1

Inte

nsity

(mV)


Figure 6.12: Chromatogram of purity for 50 µL injection for 1328 µL heart-cut. Fraction collected and re-injected onto CCZ.

6.3.3.1.2 Injection volume: 100 µL

The same procedure as described above for the 50 µL injection volume was repeated for a 100

µL sample injection onto the 150 mm C18 column. The oligomeric separation that resulted is

illustrated in Figure 6.13. A substantial loss in resolution was observed in comparison to the 50

µL injection. The resolution of the n = 5 oligomer was reduced to approximately ½ peak height.

The target diastereomer eluted in the region between 5.62 and 8.12 minutes with a maximum and

central retention time of 5.77 minutes. Subsequently the peak elution volume of the n = 5

155

oligomer fraction was 3000 µL and that of the target diastereomer was 2075 µL (as determined

using the incremental heart-cutting process previously discussed).

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Inte

nsity

(m

V)


5

Figure 6.13: Chromatogram of tert-butyl oligostyrene separation on C18 Sphereclone column (150 × 4.6 mm). 100 % MeOH mobile phase at flow rate 0.83 mL/min, 100 μL injection volume. Oligostyrene n = 5 numbered.

The effect of heart-cutting volume on recovery, purity and practical production rate was tested

using heart cutting volumes of 700 µL, 1000 µL and 2075 µL. The resulting recoveries from the

first dimension were 33%, 48% and 100% respectively. Subsequently the total recoveries were

24%, 31% and 68% respectively, with the purity being 63%, 58% and 56% respectively (see

Table 6.4). Largely, the purity remained almost constant, but with major differences in the loss

of sample analyte. That is, an increased improvement in purity was observed for the 700 µL

heart-cut section, compared to the 2075 µL heart-cut, but with a loss of 44% of sample. As a

result, the practical production rate was almost three times higher for the 2075 µL heart-cut

section (see Table 6.4). Such a loss is likely to be unacceptable, given that the same number of

156

impurities was present in both samples, only their concentrations differed (compare Figure 6.14

and 6.15). Thus the same polishing step would apply to both products; hence the recovery is a

more important operational consideration than purity.

Table 6.3: Characteristics of two-dimensional separation of target component n =5#2 for different injection volumes

Injection

volume

Centre of

band

(min)

Cut time Cut

volume

Recovery

1D

Recovery

2D

Total

Recovery

Purity

(µL) (min) (µL) (%) (%) (%) (%)

50 6.80 6.75-7.65 750 68 92 62 94

6.80 6.40-8.00 1328 100 92 92 91

100 5.70-6.54 700 33 71 24 63

5.59-6.77 1000 48 64 31 58

5.62-8.12 2075 100 68 68 56

Table 6.4: Characteristics of two-dimensional separation of target component n =5 #2 for different injection volumes

Injection volume

Cut time

Heart cut volume

Mass of target per injection

Effective

production rate

Practical

production rate

(µL)

(min)

(µL)

(mg)

(mg/h) (mg/h/L)

50 6.75-7.65 750 0.0043 0.0101 0.0594

6.40-8.00 1328 0.0087 0.0203 0.1193

100 5.70-6.54 700 0.0069 0.0194 0.1144

5.59-6.77 1000 0.0088 0.0248 0.1460

5.62-8.12 2075 0.0186 0.0524 0.3034

157

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

-0.0002

0.0000

0.0002

0.0004

0.0006

0.0008

0.0010

0.0012

0.0014

0.0016

Inte

nsity

(m

V)


n =4#3n =5#1

n =5#2


Overall, the practical production rate of the target analyte following an injection volume of 100

µL and a heart-cut volume of 2075 µL from the first dimension was 7.28 mg of sample being

produced per day if continuous operation were undertaken.

158

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

0.000

0.001

0.002

0.003

0.004

0.005

n=5 #2

n=4 #3n=5 #1

Inte

nsity

(mV)



Since the capacity of the first dimension was increased, the total recovery of the target analyte

was more sensitive to loss of sample from the first dimension: In contrast to the findings where N

only was increased in the second dimension (Section 6.3.2), whereby the sample recovery was

dependent upon the separation in the second dimension. This difference is largely attributed to

the fact that fewer contaminating species were transferred to the second dimension when the

column length was increased to 150 mm, and this reduced the level of performance required

from the second dimension.

It was somewhat surprising, and perhaps a testament to the retention mechanism of the CCZ

column, that such a small internal diameter column was able to effectively handle heart-cutting

volumes up to 2075 µL.

159

6.3.4 Summary

In this chapter strategies were employed to optimise the production of the target component.

Effectively this entailed increasing the peak capacity of the second dimension (Section 6.3.2) and

both the first and second (Section 6.3.3) dimensions. The increase in peak capacities in both

these dimensions, systematically allowed for an increase in sample load. The cost of increasing

the peak capacity was at the expense of increased cycle time and increased solvent consumption,

but improvements were gained in the recovery of the analyte and the purity of the final product.

The effect of increasing sample loading differed depending upon whether the peak capacity of

the first or second dimension was increased. Maintaining a constant peak capacity in the first

dimension, but increasing the sample load, subsequently increased the number of components

that were transferred to the second dimension, that is, more contaminating species. Hence a

greater operational requirement to isolate the target analyte from a more complex mixture was

required in the second dimension column. Therefore increasing the length of the column in the

second dimension improved the quality of the isolation of the target analyte.

Increasing the peak capacity of the first dimension column, resulted in the transfer of a less

complex sample to the second dimension. Hence less demand was placed on the separation

performance of the second dimension, since there were fewer components to separate. Rather

than test the effect of increased peak capacity in the first dimension at constant second dimension

peak capacity, the sample load was further increased in the first dimension. This resulted in more

cross contamination in the heart-cut fractions transported to the second dimension, but the peak

160

capacity of the second dimension column was able to perform better at these increased levels of

sample load (increase in N 1D), in comparison to the lower sample loads where there was an

increase in N 1D and 2D (Section 6.3.3). Hence, in short, increasing the peak capacity in the first

dimension decreased the separation demand in the second dimension, resulting in increased load

capacity. Details of the separation performance are summarised in Figure 6.16 to 6.17.

Figure 6.16 illustrates the separation performance for both injection volumes where there was an

increase in N in the second dimension (Section 6.3.2). The recovery in the first dimension is

substantially greater than that of the second dimension however recovery is comparable for all

heart-cuts for both injection volumes. The total recoveries ranged between 59 % and 66 %.

Purity ranged between 51 % and 70 %.

Figure 6.17 illustrates the separation performance for both injection volumes where there was an

increase in N in the both the first and second dimensions (Section 6.3.3). The recovery in the first

dimension increases with increasing heart-cuts for both volumes and in the second dimension

recovery was constant for the 50 µL injection volume; the 100 µL injection volume second

dimension recovery decreased with increasing heart-cuts. Therefore the total recovery was

influenced by the first dimension recoveries. The purity also decreased with increasing heart-cuts

for both injection volumes, the purity for the 50 µL injection was substantially higher at greater

than 90 %. And the 100 µL injection volume had purities ranging from 24 % to 68 %.

161

0

20

40

60

80

100

PurityTotal Recovery

2DRecovery

1DRecovery

(%)

700 ul heart-cut 30 uL inj. 1000 ul heart-cut 30 uL inj. 700 ul heart-cut (a) 50 uL inj. 700 ul heart-cut (b) 50 uL inj. 1000 ul heart-cut 50 uL inj.

Figure 6.16: Comparison of the recovery and purity of the target diastereomer separation for increased N in 2D from Section 6.3.2 for heart-cuts of 30 µL and 50 µL injections. The comparison of the practical production rate for each of the strategies used from this chapter

is shown in Figure 6.18. The practical production was constant for the 30 µL injection where

there was an increase in N in the second dimension (Section 6.3.2). Thereafter the practical

production increased as the heart-cuts increased, the 2075 µL heart-cut from where there was an

increase in N in the both the first and second dimensions (Section 6.3.3) had the highest practical

production at 0.30 mg/h/L. The practical production for where there was an increase in N in the

second dimension (Section 6.3.2) ranged from 8.4 × 10-2 mg/h/L to 0.16 mg/h/L, where there was

162

an increase in N in the both the first and second dimensions (Section 6.3.4) the range was 5.9 ×

10-2 mg/h/L to 0.15 mg/h/L (2075 µL heart-cut excluded).

0

20

40

60

80

100

PurityTotalRecovery

2DRecovery

1DRecovery

(%)

750 uL heart-cut 50 uL inj. 1328 uL heart-cut 50 uL inj. 700 uL heart-cut 100 uL inj. 1000 uL heart-cut 100 uL inj. 2075 uL heart-cut 100 uL inj.

Figure 6.17: Comparison of the recovery and purity of the target diastereomer separation for increased N in 1D and 2D from Section 6.3.3 for heart-cuts of 50 µL and 100 µL injections.

163

0.00

0.05

0.10

0.15

0.20

0.25

0.30

2075

uL

h

eart

-cu

t

1000

uL

hea

rt-c

ut

700

uL h

eart

-cu

t

1328

uL

hea

rt-c

ut

750

uL h

eart

-cu

t

700

uL h

eart

-cu

t (b

)

1000

uL

hea

rt-c

ut

700

uL h

eart

-cu

t (a

)

1000

uL

hea

rt-c

ut

700

uL h

eart

-cu

t

Increase N1D and 2D100 uL inj.

Increase N1D and 2D50 uL inj.

Increase N2D50 uL inj.

Increase N2D30 uL inj.

Pra

ctic

al p

rodc

utio

n (m

g/h/

L)

Figure 6.18: Comparison of the practical production rate of the target diastereomer for increased N in 2D; and increased N in 1D and 2D for heart-cuts of 30 µL, 50 µL and 100 µL injections.

From this data we can conclude that:

The highest purity product (94 %) was obtained for a 750 µL heart-cut fraction following

a 50 µL injection onto the 150 mm C18 column. The final product, however, had the

lowest practical production rate.

The highest practical production rate of the final product was a result of the 2075 µL

heart-cut fraction following a 100 µL injection volume onto the 150 mm C18 column.

The final product, however, had second lowest purity, but the second highest total

recovery.

164

The highest overall recovery was obtained when a 50 µL injection volume as injected

into the 150 mm C18 column and the heart-cut volume to the second dimension was

100%. This also yielded the second highest level of purity.

If the purity is not taken into account and only the total recovery and practical production

are considered then the highest practical production rate was achieved when the sample

injection volume was 100 µL onto the 150 mm C18 column, and a 2075 µL heart-cut

volume was transferred to the second dimension. The practical production rate was

0.3034 mg/h/L, with a loss of 32 % of the target component. In contrast, the highest total

recovery was obtained when an injection volume of 50 µL was applied to the 150 mm

C18 column, and the heart-cutting volume to the second dimension was 1328 µL. This

resulted in a loss of sample of only 8 % of target component for half the sample load. The

practical production rate was 0.1193 mg/h/L. The purity of 2075 µL heart-cut fraction

was 56 % compared to 92% for the 1328 µL heart-cut fraction.

An increased peak capacity in the second dimension saw the smaller injection volumes

have a slightly higher recovery in the second dimension with higher purity for smaller

heart-cuts, decreasing as the heart-cuts increased.

Increased peak capacity in the second dimension with the higher injection volumes in the

first dimension saw an increase in the practical production but with a decrease in total

recovery and product purity.

Increasing the peak capacity in the first dimension resulted in improvement in sample

recovery and purity (at constant sample load). However, increasing the sample load lead

to an increase in practical production rate, but at the expense of purity and recovery,

165

6.4 Conclusion

This would lead to the conclusion that injecting high sample loads in the first dimension greatly

improves the product recovery yield of the final product, increasing as the length of the first

dimension column increases. As the aim was to recover as much sample as possible and because

there are only ever two contaminants to remove in a polishing step, it would appear it is better to

inject higher sample loads in the first dimension as to maximise practical production whatever

the purity may be. Therefore the practical production is proportional to the injection volume and

as the injection volume increases so too does the practical production for increasing heart-cuts

regardless of purity. Another benefit of the higher sample loads is that they also have the highest

total recoveries for the largest heart-cuts an important financial outcome for any preparative

separation.

166

Chapter 7

General Conclusion

167

7.1 General Conclusion

Chromatography is a powerful separation technique that was initially developed for the isolation

of natural components in a highly purified form from complex mixtures. The intention of this

thesis was to develop a 2D HPLC system in which the optimised separation of a complex sample

could be achieved. A 2D approach was investigated due to the limitations that exist for a 1D

system particularly for complex samples. The limitations exist because of deficient separation

space or the peak capacity available for the separation of all components in complex mixtures.

Multidimensional HPLC introduces a second dimension that ideally offers a retention

mechanism that is very different to that of the first dimension and is a means of increasing the

total peak capacity of the separation process and therefore expanding the separation space.

The need for increased resolving power, driven by the demands of industry has been in part the

driving force behind the development of multidimensional HPLC. By judicious selection of the

various separation steps with consideration to the nature of the sample, the separation can be

tuned to the various sample attributes accordingly. Ultimately this type of separation process can

lead to very high levels of selectivity and hence the probability of component overlap in the two-

dimensional domain decreases. 2D HPLC is an effective separation technique for the analysis of

complex mixtures where the sample‟s complexity can be reduced as the separation mechanism of

the first dimension may be tailored towards the sample‟s multidimensionality and/or its physical

characteristics such as size, polarity, charge and shape.

The low molecular weight polymers used in this study are multidimensional complex mixtures

described according to two dominant sample attributes: Molecular weight and tacticity. Of

168

interest here was the separation and isolation of diastereomers of low molecular weight

oligostyrenes with the tert-butyl end-grouoligostyrene. The spatial orientation of the molecule

allowed separation to occur when the chromatographic environment is such that it offers shape

selectivity.

The coupling of RP dimensions can provide an orthogonal 2D system as a result of selectivity

changes and has proved successful for complex mixtures that were previously unable to be

resolved or were only partially resolved in 1D HPLC systems. The first part of this study

investigated the capabilities of a 2D HPLC system for high resolution separations of this

complex sample. Chapter 3 focussed on such a system for the separation of OLIGOSTYRENE

diastereomers with up to nine configurational repeating units where the high resolving power of

the carbon clad zirconia stationary phase was utilised in the second dimension. Carbon stationary

phases offer the potential for unique retention and present an alternative mechanism of

separation to conventional bonded phase RP stationary phases. The extensive delocalised

network on these carbon adsorption surfaces allowed for the establishment of electronic (-)

bonding and offers stereoselectivity of diastereomers that are capable of undergoing these -

type interactions.

The second part of this study involved a targeted separation however the focus was on

investigating the experimental parameters that affect the purity of an isolated component from a

complex mixture. Chapter 4 investigated the isolation of targeted components at analytical scale

optimising the purity of the isolation by employing two different strategies; (1) recovering the

target component from the central area of the band in both dimensions; and (2) recovering the

169

target component from off-centre of the band in the first dimension and from the central area of

the band in the second dimension. Low molecular weight oligostyrenes have been used here,

because they are complex, are indefinitely stable and easily characterised. The separation

performance mimicked a more complex and crowded separation space than in Chapter 3 that

would be apparent in real natural product type samples, but with the advantage of absolute

stability in the recovered analyte. This separation problem represented a scenario of extracting a

minor constituent from the complex multicomponent bulk sample. The smaller heart-cut

fractions resulted in higher purity of the target analyte, but at the cost of reduced recovery.

However, purity greater than 80 % for the heart-cuts between 100 µL and 500 µL was

sufficiently high enough to satisfy the aim of the first strategy. Off-centering the location of the

heart-cut section, away from the peak maxima and the limiting impurity positively impacted on

the purity (>89 %), and at the same time yielded improved resolution with the limiting impurity

(compared to similar impurity) concentrations.

The second part of Chapter 4 investigated the experimental parameters that effect high recovery

of the targeted components. Because of the complexity of the sample employed in this work, it

was difficult to maintain a high level of purity and at the same time achieve a high productivity.

Chapter 4 therefore focussed on maximising the recovery of the target component at analytical

scale analysis and consequently evaluated the effect this had on product purity, with respect to

both the level of contamination and the number of components contaminating the target analyte.

The number of contaminating species was an important consideration; because, a target analyte

that is 95% pure or 25% pure following isolation from the complex mixture present exactly the

same challenge to polishing the product if the number of contaminating species remains constant.

170

However, if the analyte that is 25% pure contains substantially more contaminants then the

polishing step for the target may be somewhat more complex than for the 95% pure analyte.

Hence, recovery and purity would then need to be balanced according to the economics of the

isolation process. The aim of high recovery of the target component had been achieved although

at a loss in purity, as low as 14% for the 700 µL heart-cut. Purity yet again was a limiting factor

in the isolation of the target component. This does suggest that although there is a gain in the

recovery of the target component particularly from the second dimension, the loss in purity will

affect the overall process of recovering the final product.

Chapter 5 investigated the establishment of a continuous batch-wise 2D purification process,

with the intent to preparative scale-up. Chapter 5 introduced the production rate and the

experimental variables that influence production. Of particular interest were the cycle time and

the product recovery yield. The experimental variables were interconnected; when one variable

was optimised it would have a detrimental effect on another variable, for instance when the

product purity was maximised the final productivity was generally lowered. However, reducing

the cycle time can have a positive effect by increasing the productivity and also decreasing the

solvent consumption. Impurities that elute close to the target component also placed constraints

on optimisation parameters such as production rate, product recovery yield and purity. Low

sample loads were examined in investigating the experimental parameters that affected the

preparative isolation of the target component. A consequence of the low sample loads was that

the first dimension was operated essentially under analytical conditions and the potential to load

the first dimension column with sufficient sample was not reached.

171

Chapter 6 investigated the scale-up of the 2D system used in Chapter 5 at the preparative level.

The ability of the 2D isolation process to function at a preparative level was tested at higher

sample loadings. High sample loads were utilised to determine the effect of the recovery in both

the first and second dimensions; the purity of the collected product; and finally the product

recovery yield, effective production rate and the practical production rate. The production rate

was studied as a function of the sample load, with a subsequent requirement to increase the

number of theoretical plates in the second dimension in order to handle the increased number of

contaminating species transferred to the second dimension. As such, the number of theoretical

plates N, in the second dimension was increased, by way of increasing the column length. An

improvement in the preparative separation of the target component was observed as a

consequence of increasing the number of theoretical plates, N, in second dimension. A limiting

factor in the separations obtained however were the quality of the first dimension separation.

Essentially the polymer eluted as a continuum, with the result being that heart-cutting even small

volumes to the second dimension resulted in a large number of components/contaminants being

transferred to the second dimension. Subsequently this placed high demands on the separation

performance of the second dimension column. The focus of Chapter 6 was also to improve the

quality of the isolation process by increasing the peak capacity of the first dimension, thus

improving resolution and therefore reducing the number of components transferred to the second

dimension. Consequently decreasing the performance demand of the second dimension column.

In order to test this, a 150 mm C18 column was used instead of the 50 mm column used in

Chapter 5 (with the same internal diameter). The injection volumes investigated were 50 μL and

100 μL. Since the capacity of the first dimension was increased, the total recovery of the target

analyte was more sensitive to loss of sample from the first dimension this was in contrast to the

172

findings in Chapter 5, whereby the sample recovery was dependent upon the separation in the

second dimension. This difference was largely attributed to the fact that fewer contaminating

species were transferred to the second dimension when the column length was increased to 150

mm, and this reduced the level of performance required from the second dimension.

In Chapter 6 strategies were also employed to optimise the production of the target component.

Effectively this entailed increasing the peak capacity of the second and the first dimensions. The

increase in peak capacities in both these dimensions, systematically allowed for an increase in

sample load. The cost of increasing the peak capacity was at the expense of increased cycle time

and increased solvent consumption, but improvements were gained in the recovery of the analyte

and the purity of the final product.

The effect of increasing the sample load differed depending upon whether the peak capacity of

the first or second dimension was increased. Maintaining a constant peak capacity in the first

dimension, but increasing the sample load, subsequently increased the number of components

that were transferred to the second dimension, that is, more contaminating species. Hence a

greater operational requirement to isolate the target analyte from a more complex mixture was

required in the second dimension column. Therefore increasing the length of the column in the

second dimension improved the quality of the isolation of the target analyte.

Increasing the peak capacity of the first dimension column resulted in the transfer of a less

complex sample to the second dimension. Hence less demand was placed on the separation

performance of the second dimension, since there were fewer components to separate. Rather

173

than test the effect of increased peak capacity in the first dimension at constant second dimension

peak capacity, the sample load was further increased in the first dimension. This resulted in more

cross contamination in the heart-cut fractions transported to the second dimension, but the peak

capacity of the second dimension column was able to perform better at these increased levels of

sample load, (Chapter 6) in comparison to the lower sample loads in Chapter 5. Hence, in short,

increasing the peak capacity in the first dimension decreases the separation demand in the second

dimension, resulting in increased load capacity.

174

References

175

[1] M Tswett, Ber. Dtsch. Botan. Ges., 24 (1906) 316-326.

[2] M Tswett, Ber. Dtsch. Botan. Ges., 24 (1906) 384-393.

[3] L.S Palmer, Carotinoids and related pigments:The Chromolipids (Chemical Catalog Co.,

New York, USA, 1922).

[4] R Kuhn and E Lederer, Naturwissenschaften., 19 (1931) 306.

[5] R Kuhn and E Lederer, Ber. Deut. Chem. Ges., 64 (1931) 1349.

[6] R Kuhn, A. Winterstein, E Lederer, Z.Phys.Chem., 197 (1931)141.

[7] A.J.P Martin and R.L.M Synge, Biochem. J., 35 (1941) 1358-1368.

[8] R. Consden, A.H. Gordon, A.J.P Martin, Biochem. J., 38 (1944) 224-232.

[9] A.T James and A.J.P Martin, Biochem, J., 50 (1952) 679-690.

[10] J.G Kirchner, J.M Miller, G.Keller, Anal. Chem., 2 (1951) 420-425.

[11] E.Stahl, Pharmazie., 11, (1956) 633-637.

[12] E.Stahl, Chem.Ztg., 82 (1958) 323-328.

[13] J.Porath, P.Flodin, Nature (London)., 183 (1959) 1657-1659.

[14] J.C Moore, J.Polymer. Sci., A2, (1964) 835-843.

[15] F.H Spedding, Disc. Faraday Soc., 7 (1949) 214-231.

[16] E.R. Tompkins, Disc. Faraday Soc., 7 (1949) 232-237.

[17] S.Moore, W.H Stein, J.Biol. Chem., 192 (1951) 663-681.

[18] S.Moore, W.H Stein, J.Biol. Chem.,211 (1954) 893-906.

[19]S.Moore, W.H Stein, J.Biol. Chem.,211 (1954) 907-913.

[20] D.H Spackman, W.H Stein, S.Moore, Anal. Chem., 30 (1958)1190.

[21] J.J van Deemter, F.J Zuiderweg, A. Klinkenberg, Chem. Eng. Sci., 5 (1956) 271.

[22] J.C Giddings, Anal. Chem., 35 (1963) 2215-2216.

176

[23] Cs. Horvath, B.A Preiss, S.R Lioligostyreneky, Anal. Chem., 39 (1967) 1422.

[24] J.F.K Huber, J.A.R.J Hulsman Anal. Chim. Acta., 38 (1967) 305.

[25] C.D Scott, J.E Attrill, N.G Anderson, Proc.Soc.Exp.Biol.Med., 125 (1967)181.

[26] J.J.Kirkland, Anal. Chem., 41 (1969) 218-220.

[27] J.J Kirkland, Anal.Chem., 40 (1968) 391.

[28]L.S Ettre, Chapters in the Evolution of Chromatography. Imperial College Press, London,

2008.

[29] J.J Kirkland J.J De Stefano, J.Chromatogr.Sci., 8 (1970) 309.

[30] U.D Neue, HPLC Columns. Theory, Technology and Practice. Wiley-VCH, New York,

1997.

[31] K Hostettmann, A. Marston and M. Hostettmann, Preparative Chromatography

Techniques-Applications in Natural Product Isolation. Springer, Berlin, 1997.

[32] D.J Newman, G.M Cragg and K.M Snader, J. Natural Products., 66 (2003) 1022.

[33] M.D Rawlins, Nature Reviews, 3 (2004).

[34] L. Mondello, A.C Lewis and K.D Bartle, Eds. Multidimenisonal Chromatography. John

Wiley and Sons, Ltd, West Sussex, 2002.

[35] G. Guiochon, J. of Chromatogr. A., 1126 (2006) 6.

[36] L.R. Snyder and J.J. Kirkland, Introduction to Modern Liquid Chromatography. 1976,

New York: John Wiley and Sons Inc.

[37] J.C Giddings, Anal Chem., 39 (1967) 1027.

[38] J.M Davis and J.C Giddings, Anal. Chem., 55 (1983) 418.

[39] E Van Gyseghem, S Van Hemelryck, M Daszykowski, F Questier, D.L Massart,

Y Vander Heyden, J. of Chromatogr. A., 988 (2002) 77.

177

[40] A Guttman, M Varoglu, J Khandurina, Drug Discovery Today., 9 (2004) 136.

[41] C.J Venkatramani, Y Zelechonok, Anal. Chem., 75 (2003) 3484.

[42] T Ikegami, T Hara, H Kimura, H Kobayashi, K Hosoya, K Cabrera, N Tanaka, J.

Chromatogr. A., 1106 (2006) 112.

[43] N Tanaka, H Kimura, D Tokuda, K Hosoya, T Ikegami, N Ishizuka, H Minakuchi, K

Nakanishi, Y Shintani, M Furuno, K Cabrera, Anal.Chem., 76 (2004) 1273.

[44] M.J Gray, A.P Sweeney, G.R Dennis, P Wormell, R. A Shalliker, J. Liq. Chromatogr. &

Rel. Tech., 27 (2004) 2905.

[45] M Gilar, P Olivova, A.E Daly, J.C Gebler, J. Sep. Sci., 28 (2005) 1694.

[46] W Bashir, E Tyrrell, O Feeney, B Paull, J Chromatogr. A., 964 (2002)113.

[47] D Bolliet, C.F Poole, Analyst., 123 (1998) 295.

[48] M.J Gray, G.R Dennis, P.J Slonecker, R.A Shalliker, J. Chromatogr. A., 1028 (2004) 247

257.

[49] M.J Gray, G.R Dennis, P.J Slonecker, R.A Shalliker, J. Chromatogr. A., 1041 (2004) 101.

[50] A.P Sweeney, V Wong, R.A Shalliker, Chromatographia., 54 (2001) 24- 30.

[51] M. J Gray, G.R Dennis, P.J Slonecker, R.A Shalliker, J. Chromatogr. A., 1015 (2003) 89.

[52] Ikegami, J. Chromatogr. A., 1106 (2006) 112.

[53] V Wong and R.A Shalliker, J. Chromatogr. A., 1036 (2004) 15-24.

[54] A Felinger and G Guiochon, J. Chromatogr. A., 796 (1998) 59.

[55] Gidding, J.C; J.Chrom A., 703 (1995) 3.

[56] J.C Giddings, Anal. Chem., 56 (1984) 1258A.

[57] R.A Shalliker, M.J Gray, Adv. in Chromatogr., 44 (2006) 177.

178

[58] M.R Murphy, M Schure, J.P Foley, Anal. Chem., 70 (1998) 4353.

[59] T-J Whelan, R. A Shalliker, C McIntyre and M.A. Wilson. Ind. Eng. Chem. Res., 44 (8),

2005.

[60] F Erni, R.W Frei, J. Chromatogr., 149 (1978) 561.

[61] M.M Bushey, J.W Jorgeneson, Anal. Chem., 62 (1990) 161.

[62] L Chen, X Kong, H Su, J Fu, R Ni, Zhao, H Zou, J. Chromatogr. A, 1040 (2004) 169.

[63] A van der Horst, P.J Schoenmakers, J. Chromatogr. A., 1000 (2003) 693.

[64] P Dugo, O Favoino, R Luppino, G Dugo, L Mondello, Anal. Chem., 76 (2004) 2525

[65]Murahashi, T.; Kawabata, M.; Sugiyama, H.; Hasei, T.; Watanabe, T.; Hirayama, T.; J.

Health Sci., 50 (2004) 635.

[66] P.G Stevenson, A Soliven, G.R Dennis, R.A Shalliker, J. Sep. Sci., 33(22) 2009 3880-3889.

[67] E.M Sheldon, J.Pharm. & Biomed. Anal., 31 (2003) 1153-1166.

[68] G Xue, A.D Bendick, R Chen, S.S Sekulic, J. Chromatogr. A., 1050 (2004) 159.

[69] G.J Opiteck, S.M Ramirez, J.W Jorgenson, M.A Moseley, Anal. Biochem., 258(2) (1998)

349.

[70] P.I Dobrev, L Havlicek, M Vagner, J Malbeck and M Kaminek, J.Chromatogr. A., 1075

(2005) 159.

[71] E Blahova, P Jandera, F Cacciola and L Mondello, J. Sep. Sci., 29 (2006) 555.

[72] A P Kohne and T Welsch. J. Chromatogr. A., 845 (1999) 463.

[73] A P Kohne, U Dornberger and T Welsch. Chromatographia. 48 (1998) 9.

[74] H.-J. de Geus, J. de Boer and U.A.Th. Brinkman, Multidimensionality in Gas

Chromatography. Trends Anal. Chem., 15(5) (1996) 398-408.

[75] F Cacciola, P Jandera, E Blahova and L Mondello, J.Sep. Sci., 29 (2006) 2500.

179

[76] J Yan, J. Chromatogr. A., 1090 (2005) 90.

[77] X Chen, L Kong, X Su, H Fu, J Ni, R Zhao and H Zou, J. Chromatogr. A., 1040 (2004) 169.

[78] L Hu, X Chen, L Kong, X Su, M Ye and H Zou, J. Chromatogr. A., 1092 (2005) 191.

[79] S Ma, L Chen, G Luo, K Ren, J Wu and Y Wang, J. Chromatogr. A; 1127 (2006) 207.

[80]Y.M Lui, M Guo, Y Jin, C Chu, Yu jin, X Liang, J. Chromatogr. A., 1216 (2009) 8623-

8629.

[81] Y.M Lui, M Guo, Y Jin, X Xue, Q Xu, F.F Zhang, M Liang, J. Chromatogr. A., 1206

(2008) 153.

[82] Y.Wei, T Lan, T Tang, L Zhang, F Wang, T Li, Y Du,W Zhang, J. Chromatogr. A., 1216

(2009) 8623-8629.

[83] Z. Wang and M. Fingas, Envir. Sci. & Tech., 29 (1995) 2842.

[84] Z. Wang and M. Fingas and D.S Page, J. Chromatogr. A., 843 (1999) 369.

[85] S.A Wise, L.C Sander and W.E May, J. Chromatogr. A., 642 (1993) 329

[86] T Murahashi, The Analyst; 128 (2003) 611.

[87] J.V Goodpaster, S.B Howerton and V.L McGuffin, J. Forensic Sci., 46 (6) (2001) 1358.

[88] B.L Murphy and R.D Morrison, Introduction to Environmental Forensics; Academic Press,

New York, 2002.

[89] D.B Broughton, G.G Gerholg, US Patent 2 (1961) 985.

[90] P.E Barker, Chem. Eng. Sci. 13 (1960) 82.

[91] R.M Nicoud, Sepn. Sci and Tech., 2 (2000) 476-501.

[92] N. Gottsclich, V. Kasche, J. Chromatogr. A. (1997) 201-206.

180

[93] S. Abel, M. Babler, C. Arpagaus, M. Mazzotti, J. Stadler, J. Chromatogr. A., (2004) 201-

210.

[94] F. Hilbrig, R. Freitag, J. Chromatogr. B., 790 (1-2) (2003) 1.

[95] J. Brozio, H. Bart, J. Chem. Eng. Tech., 27(9) (2004) 962.

[96] R. Schlegl, R. Necina, A. Jungbauer, J. Chem. Eng. Tech., 28(11) (2005) 45.

[97] V Wong, A P Sweeney and R.A Shalliker, J. Sep. Sci., 27 (2004) 47.

[98] G Guiochon, S.G Shirazi and A.M Katti, Fundamentals of Preparative and Nonlinear

Chromatography. Academic Press, Boston, 1994.

[99] L.-C Men, H Na, L.B Rogers, J. Chromatogr. 314 (1984) 55-63.

[100] L.-C Men, L.B Rogers, J. Chromatogr. 347 (1985) 39-50.

[101] B. J Bassler, R Kaliszan, R.A Hartwick. J. Chromatogr. 461 (1989) 139-147

[102] M.T Gilbert, J.H Knox, B Kaur, Chromatographia. 16 (1982)138-146.

[103] Knox, J.H.; Ross, P. Carbon-based packing materials for liquid chromatography.

Structure, performance and retention mechanisms. Advances in Chromatography; 37, Marcel

Dekker, Inc: New York, 1997; 73-119 pp.

[104] N Tanaka, T Tanigawa, K Kimata, K Hosoya, T Araki, J. Chromatogr. 549 (1991) 29-41.

[105] J Nawrocki, M.P Rigney, A McCormick, P.W Carr, J. Chromatogr. A., 657 (1993) 229-

282.

[106] T.P Weber, P.W Carr, Anal. Chem., 62 (1990) 2620-2625.

[107] T.P Weber, P.W Carr, E.F Funkenbusch, J. Chromatogr., 519 (1990) 31-52.

[108] T.P Weber, P.T Jackson, P.W Carr, Anal. Chem., 67 (1995) 3042-3050.

[109] P.T Jackson, T.- Y Kim, P.W Carr, Anal.Chem., 69 (1997) 5011-5017.

181

[110] A.P Sweeney, P Wormell, R.A Shalliker, Macromol. Chem. Phys., 203 (2002) 375-380.

[111] A.P Sweeney, S.G Wyllie, R.A Shalliker, J. Liq. Chromatogr. Relat. Technol., 24(17)

(2001) 2559-2581.

[112] J.H Knox, B Kaur, G.R Millward, J. Chromatogr. 352 (1986) 3-25.

[113] Y-Q Xia, M Jemal, N Zheng, X Shen, Rapid Commun. Mass Spectrom., 20 (2006) 1831-

1837.

[114] F Belliardo, O Chiantore, D Berek, I Novak, C Lucarelli, J. Chromatogr., 506 (1990) 371-

377.

[115] B.J Fish, J.Pharm. Biomed. Anal., 11(6) (1993) 517-521.

[116] A Wutte, G Gubitz, S Friebe, G.-J Krauss, J Chromatogr. A., 677 (1994) 186-191.

[117] M Stefansson, K-J Hoffman, Chirality., 4(8) (2004) 509-514.

[118] M. J Gray, G.R Dennis, P.J Slonecker, R.A Shalliker, J Chromatogr. A., 1073 (2005) 3-9.

[119]V Wong, A.P Sweeney, M Khurrum, R.A Shalliker, J. Liq. Chromatogr. Relat. Technol.,

25(3) (2002) 363-379.

[120] M.J Gray, A.P Sweeney, G.R Dennis, P.J Slonecker, R.A Shalliker, Analyst., 128 (2003)

598-604.

[121] T.H Mourey, Anal. Chem., 56 (1984) 1777-1781.

[122] T.H Mourey, G.A Smith, L.R Snyder, Anal. Chem., 56 (1984) 1773-1777.

[123] R.A Shalliker, P.E Kavanagh, and I.A Russell, J. Chromatogr., 558 (1991) 440-445.

[124] J.J Lewis; L.B Rogers: and R.E Pauls, J. Chromatogr., 264 (1983) 339-356.

[125] P. Jandera and J. Rozkošná, J. Chromatogr., 362 (1986) 325-343.

[126] D. Berek, J. Chromatogr. A., 1020 (2003) 219-228.

182

[127] R.A Shalliker, P.E Kavanagh, and I.A Russell, J. Chromatogr., 664 (1994) 221-228.

[128] J.P Larmann, J.J DeStefano, A.P Goldberg, R.W Stout, J. Chromatogr., 255 (1983) 163-

189.

[129] J.J Kirkland, Chromtographia, 8(12) (1975) 661-668.

[130] S Kayillo, G.R Dennis, P Wormell, R.A Shalliker, J. Chromatogr. A., 967(2) (2002) 173-

181.

Development of Two- Dimensional Liquid Chromatographic ...88… · Development of Two- Dimensional Liquid Chromatographic Systems For the Optimised Separation and Targeted Isolation

Documents