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 4.2: 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 4.5: Characteristics of two-dimensional separation of target component n = 5 #2
Table 4.6: Characteristics of two-dimensional separation of target component n = 5 #2
Table 5.1: Characteristics of two-dimensional separation of target component n=5 #2
Table 5.2: 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
different injection volumes
Table 6.3: Characteristics of two-dimensional separation of target component n =5 #2 for
different injection volumes
Table 6.4: Characteristics of two-dimensional separation of target component n =5 #2 for
different injection volumes
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.
Figure 3.7: Chromatogram of gradient separation of oligomer n =7 on CCZ column (100 x
10 mm). Conditions: gradient elution 80-20 ACN: DCM to 40-60 ACN: DCM in 80
minutes at 3.0 mL/min with at ambient temperature 30 µL injection volume.
Figure 3.8: Chromatogram of gradient separation of oligomer n =8 on CCZ column (100 x
10 mm). Conditions: gradient elution 80-20 ACN: DCM to 40-60 ACN: DCM in 40
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 3.10: Chromatogram of gradient separation of oligomer n =9 on CCZ column (100 x
10 mm). Conditions: gradient elution 100-0 ACN: DCM to 0-100 ACN: DCM in 20
minutes at 3.0 mL/min with at ambient temperature 200 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
Figure 4.6: 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. 700 μL heart-cut
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.
Figure 4.13: 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.
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.
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.
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.
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.
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.
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.
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.
xxi
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.
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.
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.
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).
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
collected and re-injected onto CCZ.
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 6.14: Chromatogram of purity for 100 µL injection for 700 µL heart-cut. Fraction
collected and re-injected onto CCZ.
Figure 6.15: Chromatogram of purity for 100 µL injection for 2075 µL heart-cut. Fraction
collected and re-injected onto CCZ.
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 Carbon clad zirconiaa
ZirChrom Separations
3 50 × 4.6 3, 4, 5, 6
4 Carbon clad zirconiaa
ZirChrom Separations
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)
Retention time (min)
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)
Retention time (min)
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
)
Retention time (min)
Figure 3.7: Chromatogram of gradient separation of oligomer n =7 on CCZ column (100 × 10 mm). Conditions: gradient elution 80:20 ACN: DCM to 40:60 ACN: DCM in 80 minutes at 3.0 mL/min with at ambient temperature 30 µL injection volume.
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)
Retention time (min)
Figure 3.8: Chromatogram of gradient separation of oligomer n =8 on CCZ column (100 × 10 mm). Conditions: gradient elution 80:20 ACN: DCM to 40:60 ACN: DCM in 40 minutes at 3.0 mL/min with at ambient temperature 30 µL injection volume.
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)
Retention time (min)
Figure 3.10: Chromatogram of gradient separation of oligomer n =9 on CCZ column (100 × 10 mm). Conditions: gradient elution 100:0 ACN: DCM to 0:100 ACN: DCM in 20 minutes at 3.0 mL/min with at ambient temperature 200 µL injection volume.
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.
4.2.2 Chromatographic separation
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 Results and Discussion
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
1
2
3
Figure 4.6: 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. 700 μL heart-cut volume.
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)
Retention time (min)
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)
Retention time (min)
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)
1st Dimension 2nd Dimension Total
Rec
over
y (%
)
Heart-cut volume (uL)
300 350 400 450 500
30
40
50
60
70
80
90
100(b)
1st Dimension 2nd Dimension Total
Rec
over
y (%
)
Heart-cut volume (uL)
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
(%)
Heart-cut volume (uL)
300 350 400 450 500
30
40
50
60
70
80
90
100(b)
Purity Total recovery
(%)
Heart-cut volume (uL)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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).
Table 4.6: 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 36 100 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)
Retention time (min)
Figure 4.13: 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.
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)
Retention time (min)
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
Practical Aspects in the Optimisation of
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.
5.3.2 Chromatographic separation
As described in Section 4.2.2.
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
5.4 Results and Discussion
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)
Retention time (min)
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
)
Retention time (min)
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.
Table 5.2: 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
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
)
Retention time (min)
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
)
Retention time (min)
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
)
Retention time (min)
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
Table 5.4: 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 0.0003 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
Practical Aspects in the Optimisation of
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
As described in Section 4.2.1.
6.2.2 Chromatographic separation
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
As described in Section 4.2.3.
6.2.4 Calibration
As described in Section 5.3.3
6.3 Results and Discussion
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)
Retention time (min)
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)
Retention time (min)
(a)
2 4 6 8 10 12 14 16 18 20
0.000
0.005
0.010
0.015
Inte
nsity
(m
V)
Retention time (min)
(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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (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.
(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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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
)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
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)
Retention time (min)
n =4#3n =5#1
n =5#2
Figure 6.14: Chromatogram of purity for 100 µL injection for 700 µL heart-cut. Fraction collected and re-injected onto CCZ.
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)
Retention time (min)
Figure 6.15: Chromatogram of purity for 100 µL injection for 2075 µL heart-cut. Fraction collected and re-injected onto CCZ.
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.