Advances In Chromatography: Volume 44 (Advances in Chromatography)

Citation preview

VOLUME 44

Advances in CHROMATOGRAPHY

VOLUME 44

Advances in CHROMATOGRAPHY EDITORS:

ELI GRUSHK A Hebrew University of Jerusalem Jerusalem, Israel

NELU GRINBERG Boehringer-Ingelheim Pharmaceutical, Inc. Riverside, Connecticut, U.S.A.

Boca Raton London New York

A CRC title, part of the Taylor & Francis imprint, a member of the Taylor & Francis Group, the academic division of T&F Informa plc.

Published in 2005 by CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 © 2005 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International Standard Book Number-10: 0-0000-0000-0 (Hardcover) International Standard Book Number-13: 978-0-0000-0000-0 (Hardcover) Library of Congress Card Number XX-XXXXX This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data Authorxxxxxxxxxxxxxxxxxxxxxxxxx titlexxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx p. cm. xxxxxxxxxxxxx Includes bibliographical references and index. ISBN 0-0000-0000-0 xxxxxxxxxx 1xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxx xxxxxxxxxxx xxx xxx xxx xxxx xxxx xxxxxxxxxxxxx 2004 xxxxxx'xx--dc22

XX-XXXXX

Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com Taylor & Francis Group is the Academic Division of T&F Informa plc.

and the CRC Press Web site at http://www.crcpress.com

Foreword With Volume 44 of the Advances in Chromatography series I say goodbye, reluctantly, to Phyllis R. Brown, who served as one of the editors of the series. Advances in Chromatography was started by J. Calvin Giddings and Roy A. Keller in 1965. Very quickly, the series became an important tool for quick dissemination of new developments in the field. In Volume 12 of the series (1975) Jack Cazes and I joined Giddings and Keller as editors. In 1976 (Volume 14) Phyllis replaced Roy Keller. Jack Cazes left the series in the mid 1980s and Cal Giddings stepped down from his position as executive editor in 1993. Over the last 10 years or so Phyllis and I were the sole editors of the Advances in Chromatography series. Several years ago Phyllis retired from her position as a chemistry professor at the University of Rhode Island, and now she has decided that the time has come to hand over her editorial responsibilities to the younger generation. Nelu Grinberg was selected to replace Phyllis, and with this volume he makes his debut as an editor. Phyllis served as an editor for almost 30 years. In many respects she was the primary moving force of the series. The fact that Advances in Chromatography is still going strong after 40 years is due to her foresight, determination, diligence, and perseverance. Phyllis has a very good eye for recognizing the latest developments in chromatography and related techniques. Her purpose was always to “bring to our readers the latest developments and advances at the forefront of the field.” Using her many acquaintances and contacts, we succeeded in staying on top of the field and providing our readers with up-to-date reviews in all areas of chromatography and other elution techniques. Based on her own research, Phyllis recognized very early the usefulness of HPLC in life sciences. She saw to it that the Advances in Chromatography series educates its readers about the power of the technique as applied to the biological field. I want to thank Phyllis very much for her friendship as well as for her help, encouragement, and patience all through our years together as editors. I would like to welcome Nelu Grinberg as my new co-editor. I have known Nelu for many years and I am well familiar with the caliber of his work and with his commitment, as a scientist, to excellence. Nelu and I will strive to maintain the excellence of the series that was so important to Phyllis. Judging from the cooperation between us in producing the present volume of the Advances in Chromatography, the future of the series remains bright. Eli Grushka Jerusalem

Contributors Andrew J. Alpert PolyLC, Inc. Columbia, Maryland Gabriela Cimpan Sirius Analytical Instruments, Ltd. Forest Row, U.K. John Comer Sirius Analytical Instruments, Ltd. Forest Row, U.K. Mauricio Dantus Merck and Company Rahway, New Jersey Andrew G. Ewing Departments of Chemistry and Neural and Behavioral Sciences The Pennsylvania State University University Park, Pennsylvania Simion Gocan Analytical Chemistry Department Babes-Bolyai University Cluj-Napoca, Romania Michael J. Gray Nanoscale Organisation and Dynamics Group University of Western Sydney Sydney, Australia

Tyge Greibrokk Department of Chemistry University of Oslo Oslo, Norway Y.V. Kazakevich Seton Hall University South Orange, New Jersey R. LoBrutto Novartis Pharmaceutical Corporation East Hanover, New Jersey Elsa Lundanes Department of Chemistry University of Oslo Oslo, Norway Tracy L. Paxon Department of Chemistry The Pennsylvania State University University Park, Pennsylvania R. Andrew Shalliker Nanoscale Organisation and Dynamics Group University of Western Sydney Sydney, Australia Rodger W. Stringham Chiral Technologies, Inc. West Chester, Pennsylvania

Table of Contents Chapter 1 Separations in Multiple-Channel Microchips ...........................................................1 Tracy L. Paxon and Andrew G. Ewing Chapter 2 Temperature Effects in Liquid Chromatography ....................................................45 Elsa Lundanes and Tyge Greibrokk Chapter 3 Lipophilicity Measurements by Liquid Chromatography.......................................79 Simion Gocan, Gabriela Cimpan, and John Comer Chapter 4 Concepts and Practice of Multidimensional High-Performance Liquid Chromatography ....................................................................................................177 R. Andrew Shalliker and Michael J. Gray Chapter 5 High-Performance Liquid Chromatography in the Pharmaceutical Industry: Application, Validation, and Regulatory Issues Under the PAT Framework .......237 Mauricio Dantus Chapter 6 The Use of Polysaccharide Phases in the Separation of Enantiomers.................257 Rodger W. Stringham Chapter 7 Chaotropic Effects in RP-HPLC ...........................................................................291 R. LoBrutto and Y.V. Kazakevich Chapter 8 Chromatography of Difficult and Water-Insoluble Proteins with Organic Solvents....................................................................................................317 Andrew J. Alpert Index......................................................................................................................331

1

Separations in MultipleChannel Microchips Tracy L. Paxon The Pennsylvania State University, University Park

Andrew G. Ewing The Pennsylvania State University, University Park

CONTENTS 1.1 1.2

1.3

1.4 1.5

1.6

1.7

Introduction ......................................................................................................2 Microfabrication of Multiple-Channel Microchips .........................................3 1.2.1 Glass Substrates ...................................................................................3 1.2.2 Polymer Substrates...............................................................................3 Multiple-Channel Microchip Designs .............................................................4 1.3.1 Rectilinear Microchip Designs ............................................................4 1.3.2 Radial Microchip Designs ...................................................................6 1.3.3 Bioassay Microchip Designs ...............................................................8 1.3.4 Continuous Sample Introduction Microchip Designs .........................8 Channel Filling and Sample Loading..............................................................8 Injection Methods ............................................................................................9 1.5.1 T-Type Injections................................................................................10 1.5.2 Optically Gated Injection...................................................................13 1.5.3 Capillary Sample Introduction...........................................................15 Detection Methods .........................................................................................17 1.6.1 Single-Point Confocal Detection .......................................................18 1.6.2 Translation-Stage Confocal Detection...............................................18 1.6.3 Scanning Methods of Detection ........................................................19 1.6.3.1 Confocal Fluorescence Detection with a Galvanometric Scanner.......................................................19 1.6.3.2 Confocal Fluorescence Detection with a Rotary Scanner....................................................................19 1.6.3.3 Fluorescence Detection with an Acousto-Optical Deflection Scanner..............................................................22 1.6.4 Fluorescence Detection with a Charge-Coupled Device (CCD) ......22 Applications of Multiple-Channel Microchips..............................................23 1.7.1 DNA Separations ...............................................................................24

1

2

Advances in Chromatography

1.7.1.1

Separations Involving Restriction Enzyme Digests — DNA Fragment Analysis ....................................................24 1.7.1.2 Genotyping..........................................................................25 1.7.1.3 Sequencing in Multiple-Channel Microchips ....................27 1.7.2 Affinity Electrophoresis for Immunoassays in Microchips ..............29 1.7.3 Dynamic Analysis of Reactions with Multiple-Channel Electrophoresis ...................................................................................30 1.7.3.1 Application to DNA............................................................30 1.7.3.2 Application to RNA Reactions...........................................32 1.8 Conclusions and Future Perspectives ............................................................32 Acknowledgments....................................................................................................35 References................................................................................................................35

1.1 INTRODUCTION Capillary electrophoresis (CE) has been reviewed extensively for a variety of applications [1]. The technique provides fast separation times and small sample requirements; however, CE is a serial technique that significantly limits throughput. To address this, capillary array electrophoresis, in which separations are performed in an array of parallel silica capillaries, was introduced by Mathies [2] and subsequently has been used to perform high-speed, high-throughput deoxyribonucleic acid (DNA) sequencing [3, 4] and DNA fragment sizing [5]. This method has proven to be an invaluable technique for the completion of the human genome project. However, sample introduction into a large number of capillaries and the physical manipulation of these capillaries can be difficult and cumbersome. The miniaturization of capillary electrophoresis onto a microchip was thus realized in the early 1990s [6] and miniaturization has been used to increase performance by reducing analysis times and reagent volumes [6−9]. Since then, extensive advances have been made in the area of microchip technology, allowing for the analysis of a variety of substances including amino acids [10], small drug molecules [11], peptides [12], oligonucleotides [13], proteins [14, 15], and DNA fragments [16]. Initial microchip separation devices consisted of a single separation channel and a T-type injector. Developments have since been made to multiplex many separation channels on a single microchip, thereby providing a means to achieve high throughput separations [16−25]. However, multiple channels require multiple reservoirs, thus increasing the complexity of multi-channel microchip designs. Other factors that must be considered when dealing with multiplexed microchips include fabrication, bonding, filling, sample introduction, and complex electronics required to perform injections and separations, as well as simultaneous detection for multiple separations. The objective of this chapter is to present recent advances in multichannel microchip design, sample introduction, and detection of various analytes for high throughput analyses.

Separations in Multiple-Channel Microchips

3

1.2 MICROFABRICATION OF MULTIPLE-CHANNEL MICROCHIPS Microchip fabrication is based on techniques developed by the semiconductor industry and has traditionally been performed on glass substrates but polymeric microchips have also been produced. The multi-channel microchip fabrication process is similar to the process for single channel microchips; however, the masks become increasingly more complex with an increase in channel number. Additionally, fabrication of large numbers of channels on a single microchip requires the substrates to be essentially free of defects. Thus, the challenge is to fabricate defect-free multiplechannel microchips in a reproducible, cost-effective manner. Developments to this effect have been reviewed by Gao et al. [26], and therefore will be only briefly discussed here.

1.2.1 GLASS SUBSTRATES Channels and other microchip features are typically defined on glass plates using photolithography and wet chemical etching techniques [7, 9, 27−29]. Microchips have been fabricated on soda lime glass slides and, more commonly for multiplechannel microchips, on Borofloat glass wafers. The traditional photolithographic process normally begins with the deposition of various metals followed by layers of photoresist onto a clean glass substrate; however, amorphous silicon has recently been used as a sacrificial mask prior to photoresist deposition [16, 30]. In the modified protocol, glass substrates are pre-etched in hydrofluoric acid followed by deposition of a thin layer of amorphous silicon using a plasma-enhanced chemical vapor deposition system [30]. Spin-coating a layer of photoresist onto the plate and subsequent illumination of the substrate through a mask containing the multiplechannel pattern follows. The photoresist is developed and the channel pattern is transferred to the silicon film by reactive etching with SF6 plasma. The channels are then etched into the glass by a concentrated hydrofluoric acid solution. The photoresist layer is stripped off with high-pressure O2 plasma. Holes are drilled into either the etched plate or a second glass plate that has not been subjected to this procedure, to provide access holes to each channel. The plates are finally aligned and thermally bonded, thereby enclosing the channels [31].

1.2.2 POLYMER SUBSTRATES Polymeric materials have been employed as alternative substrates as they lend themselves to mass fabrication of affordable microfluidic devices [32−37]. Currently, the most widely used replication process to fabricate channel structures for separations is hot embossing [32, 33, 38, 39]. After fabrication of the master, the master and planar polymer substrate are heated, placed under vacuum and brought together with a controlled force. This force presses the channel structure into the polymer, creating the desired microstructures. Alternatively, injection molding is used in polymer microchip fabrication. In this process, the raw polymer material in granular form is melted and injected under high pressure into an evacuated cavity containing

4

Advances in Chromatography

the master [32, 40–42]. The molten polymer cools in a few seconds, allowing for rapid production of identical structures. Casting is another polymer microchip fabrication method in which a mixture of an elastomer precursor and its curing agent are poured over the master template [32, 34, 36, 37]. After a few hours, the solidified polymer is removed from the master, exposing the desired channel structure. Regardless of the fabrication technique used, the open channels must be bonded to a planar surface to form closed channels that can be used for separations. There are a variety of methods for bonding including lamination, gluing, heat, and pressure application, laser welding, and ultrasonic welding. For more information on these polymer methods, refer to Becker and Gartner [32].

1.3 MULTIPLE-CHANNEL MICROCHIP DESIGNS Significant advances have been made over the last 10 years in the area of multiplechannel microchips. Channel layouts have progressed from single channels, with as many as four reservoirs for sample introduction, to as many as 384 channels with 481 reservoirs for separation and sample introduction, respectively, as demonstrated by Mathies and coworkers [31]. Since the fabrication of multiple features on a glass microchip does not significantly increase the cost of production, as compared to fabrication of a single feature, parallelization is a cost effective way to increase throughput. All of the components of a microchip are fabricated on a single substrate. However, as the number of separation channels and reservoirs increases, so does the complexity of sample loading, electrical connections to apply injection and separation potentials, as well as parallel channel detection. Thus, microchip design is continually being improved to achieve higher throughput and address the challenges of chip complexity. Several chip designs have emerged from such development and include rectilinear, radial, and bioassay microchips.

1.3.1 RECTILINEAR MICROCHIP DESIGNS Parallel analysis on a microchip was demonstrated in 1994 by Wooley and Mathies when they performed ultra-high-speed DNA fragment separations on a parallelchannel microchip [43]. In this microchip design, 15 channels are etched into soda lime glass. The channel dimensions vary in size to include 3.5 cm long by 30, 50, and 70 μm wide channels, with 30, 70, and 120 μm cross-channels. Their work has characterized two different injection methods — stack injection and plug injection — and addresses both the effect of channel geometry and the effect of electric field on separation. Procedures to fill these channels with hydroxyethyl cellulose- (HEC) sieving matrices has been developed, and separations of FX174 Hae III fragments with excellent resolution achieved in less than 2 minutes. Additionally, ultra-highspeed DNA sequencing has been accomplished on similar microchips using fourcolor fluorescence detection [44]. In an attempt to simplify detection, Woolley and coworkers [24] have designed and fabricated a capillary array electrophoresis microchip (Figure 1.1a) that has the capacity to analyze 12 different samples in parallel in under 160 seconds. This microchip design consists of 12 independent pairs of injection (A, 8 mm long) and

Separations in Multiple-Channel Microchips

5

FIGURE 1.1 (a) Mask design for the 12-channel capillary array electrophoresis chip for highspeed DNA genotyping. (From Woolley, 1997.) (b) Mask design of the 16-channel capillary electrophoresis chip for parallel DNA sequencing. Sixteen identical 250 μm twin-T injectors are used for all channels. (Inset) Detail of the compensation of different channel lengths. (From Liu, 2000.) (c) Mask pattern for the 96-sample capillary array electrophoresis microplate for genetic analysis. (From Simpson, 1998.)

separation (B, 8 μm deep, 60 μm wide, and 60 mm long) channels, as well as two optical alignment channels (C), which are patterned and etched on a glass microscope slide. With just 12 channels, 41 access holes are required to provide sample, buffer, and waste reservoirs on this microchip. These studies provide a separation method in which high-speed, parallel separations of DNA restriction fragments and PCR products can be analyzed. Furthermore, this microchip design demonstrates that microfabricated multi-channel microchips provide a powerful method for rapid highthroughput DNA diagnostics in which ten base pair (bp) resolution has been achieved. Studies to perform DNA sequencing on a 16-channel microchip (Figure 1.1b) has been completed by Liu et al. [45] This microchip is fabricated on a 10-cm diameter wafer with two eight-channel groups, each with a common anode reservoir. Each channel has its own sample and waste reservoir in the twin-T injector configuration as well as its own cathode reservoir. The Mathies group has expanded upon previous work by creating a capillary array electrophoresis microchip in which two different samples are serially separated in 48 parallel channels for a total of 96 sample analyses [30]. The microchip design

6

Advances in Chromatography

employed in these experiments consists of 48 individual separation channels etched in a 150 μm periodic array in the detection region (Figure 1.1c) [30]. The channels branch out to form an 8 × 12 array of reservoirs. These arrays of reservoirs are spaced 9 mm apart such that samples and buffer can be loaded via an eight-channel pipettor. The separation channels are 10 cm in length, spanning the area between the injection region and the anode. Two injection reservoirs (B and C) are coupled to each separation channel, thus allowing for individual serial injections into each channel. In order to simplify the microchip design, the researchers group the channels such that only one waste reservoir (D) is required for every four channels. By the same rationale, the cathode reservoirs (E) are connected to groups of six or twelve channels. The anode (F) is placed off-center and is common to all 48 channels. Overall, 127 holes are drilled into the etched plate to provide access to the channels, followed by thermal bonding to a flat glass wafer. The investigators report high quality bonding over the entire surface of the microchip. Using this device, 96 samples have been separated in two runs of 48 samples (corresponding to the samples in reservoirs B and C) in less than 8 minutes. Nineteen different samples have been dispensed among the 96 sample reservoirs, thereby providing five-fold redundancy in sample analysis. This work further multiplexes separation channels on a microchip by combining reservoirs to reduce the number of access holes thereby creating a more simplified device; however, this design can result in sample cross-contamination. Serial injections are an effective method of increasing the sample throughput with a limited number of separation channels and will be discussed in detail later in this review.

1.3.2 RADIAL MICROCHIP DESIGNS The design, fabrication, and operation of a radial capillary array electrophoresis microchip and scanner for high-throughput DNA analysis was presented in 1999 and has been applied in various DNA applications including genotyping, sequencing and single nucleotide polymorphism typing [16, 30, 46–49]. A 96-channel radial microplate (Figure 1.2a) has more recently been used for the detection of 2,4,6trinitrotoluene (TNT) [50]. These applications are discussed later in this review. This microchip consists of 96 individual channels in a radial design with a common anode reservoir in the center of a circular 10 cm-diameter wafer and an array of channels extending toward injector units at the perimeter of the wafer [16]. The distance from the injector to the detection point is 33 mm. Channels are grouped into four 24channel quadrants, each terminating at the center of the microchip. Access holes for each of the reservoirs have been drilled prior to thermal bonding. This radial design is preferred over former rectilinear designs [24, 30] because it is easier to fabricate and all of the channels are straight, thereby eliminating resolution problems associated with curved channels [51]. Additionally, the radial design allows for the use of a radial scanner for detection purposes. This will be discussed further in the section on detectors [16]. Additional studies by the Mathies group have led to the development of a 96folded-channel radial microchip (Figure 1.2b) in order to improve separation efficiency [49]. Initial studies considered the parameters that cause turn-induced band

Separations in Multiple-Channel Microchips

7

FIGURE 1.2 (a) Mask pattern for the 96-channel radial capillary array electrophoresis microchip utilized for various DNA applications including genotyping and single nucleotide polymorphism typing, as well as the detection of 2,4,6-trinitrotoluene. (From Medintz, 2000.) (b) Mask pattern used to form a 96-channel radial capillary array electrophoresis microplate with folded channels for increased (16 cm) separation length. Enlarged view of the folded-channel layout and the novel hyper-turn geometry. (From Medintz, 2001.) (c) Layout of the 384-lane capillary array electrophoresis microchip on a 200 mm diameter wafer for ultra-high throughput genetic analysis. (From Emrich, 2002.)

broadening and investigated ways to modify channel geometry such that the broadening could be reduced [52]. These devices are a modified version of the aforementioned 96-channel microchips and have been fabricated using the same method. By using optimized folded channels, separation quality is maintained while the separation efficiency is significantly enhanced due to the increased channel length. The investigators suggest these devices will be useful in the development of optimized DNA-sequencing multi-channel microchips with enhanced performance, throughput, and device packing [47, 49, 52, 53]. In fact, the 96-folded-channel radial microchip has been used for DNA genotyping and sequencing [47, 49]. The pioneering work with 96-channel radial microchips has enabled the design, fabrication, and implementation of a 384-channel radial microchip (Figure 1.2c)

8

Advances in Chromatography

[31] for massively parallel genetic analysis, providing a new tool for accelerating gene mapping, pharmacogenomic screening, forensics, and proteomics. This microchip design consists of 4 quadrants of 24 quartets distributed radially around a common anode reservoir. Each channel has its own unique sample reservoir and each quartet shares a common cathode. This microchip contains no waste reservoirs; therefore, injections are made directly onto the separation lanes. The elimination of waste reservoirs allows for more tightly packed channels but also results in less well-defined sample plugs, as described below. The researchers report that using a multiple-step electrokinetic injection method along with short injection times helps to minimize the size-biased injection effect of the less well-defined sample plug [31].

1.3.3 BIOASSAY MICROCHIP DESIGNS To date, DNA analysis has been the major focus of the multiplexed microchip designs presented but work to integrate immunoassays onto microchips is underway in several laboratories. Initial studies have focused on single channel devices [54–61], although there has been a recent desire to incorporate multiple bioassay channels onto a single device [50, 62–66]. To this end, Cheng et al. [62] have designed a sixchannel microfluidic immunoassay device. This device consists of six independent mixing, reaction, and separation channels as well as two channels for alignment purposes (Figure 1.3) [62]. The separation channels are 9.3 cm in length and bent around two or three corners in order to fit on the substrate. The corners are rounded to a curvature of 300 μm such that sample plug distortion is minimized [51]. Using this device, simultaneous direct immunoassays for ovalbumin and for anti-estradiol have been performed. Mixing, reaction, and separation have been performed within 60 seconds in all cases and in less than 30 seconds when optimized conditions are employed. Another high-throughput homogeneous immunoassay for the sensitive detection of TNT has been developed by Bromberg and Mathies using the 96-channel radial capillary array electrophoresis microdevices previously described [50].

1.3.4 CONTINUOUS SAMPLE INTRODUCTION MICROCHIP DESIGNS Several microchip designs have been developed to facilitate sample introduction and will be discussed in detail in the injection methods section. Briefly, the Ewing group has utilized straight channels for multiplexing separations using optically gated sample introduction as well as straight, open channels with a specialized capillary sample introduction scheme [66–69]. These unique methods of sample introduction have allowed dynamic reactions to be studied with minimal user intervention.

1.4 CHANNEL FILLING AND SAMPLE LOADING In single channel microchip electrophoresis, channel filling is straightforward. The separation media is placed in one reservoir and is either vacuum- or pressure-pumped through the lane. Multiplexing this process is not a trivial task but the concept remains the same as for single channel devices. Different methods have been developed to facilitate channel filling and sample loading. A single pipette is often used

Separations in Multiple-Channel Microchips

9

FIGURE 1.3 Illustration of the six-channel layout of the immunoassay device referred to as SPIDY. Each reaction cell has reservoirs for sample (S1), antibody (S2), sample injection waste (SW), and running buffer (B). Expansion on the left illustrates the sample loading double-T injector design. The bottom right expansion illustrates the antibody (Ab) and antigen (Ag) mixer in the reaction cell. The lower left expansion illustrates the pattern of channels at the buffer waste (BW) reservoirs used during separations. Expansion in the upper right of the detection zone, across which a laser beam is swept, illustrates the separation of two components in the size separation channels. (From Cheng, 2001.)

to fill a single channel or multiple channels on a simple microchip; however, as the number of channels on the microchip increases, this process becomes increasingly difficult. In one approach, an 8-channel pipette has been used to load sample onto a 16-channel microchip from a microtiter plate (a second transfer is required to load all 16 samples) [45]. In another approach, a capillary array for filling 96 reservoirs (Figure 1.4) has been developed by the Mathies group [16], providing a simple method for simultaneously transferring separation medium into a radial 96-channel microchip. In this method, pressure is applied and all 96 channels are filled simultaneously. Samples are loaded onto this microchip in a similar manner. The advantage of capillary sample introduction in this method is that the rectilinear format of the microtiter plate can be coupled to the radial design of the previously described highdensity microchips.

1.5 INJECTION METHODS Sample injection is a straightforward process on a microchip but as the number of channels increases, so does the complexity of the electronics required to perform the injections. For a thorough review of sample introduction techniques for

10

Advances in Chromatography

FIGURE 1.4 Schematic of the 96-sample capillary loader. Pressurization of the rectilinear microtiter dish chamber (left) is used to transfer 96 samples to the sample reservoirs of the radial microplate (right). (From Shi, 1999.)

microfabricated separations devices, readers are directed to Roddy et al. [67] Three common methods of sample injection — T-type, optical gating, and capillary sample introduction — will be discussed briefly below.

1.5.1 T-TYPE INJECTIONS T-type injection was originally developed for single lane microchips [6, 7, 70]; however, it has also been extensively applied in the area of multiplexed microchips [31, 71]. Typically, a single channel with a side arm (sample channel) is used to perform these injections. A potential is first applied across the sample channel to load the sample, and the potential is switched and then applied across the separation channel to accomplish separation (Figure 1.5a). Various other T-type injectors have been developed to improve the control of fluid flow within the microchip channels and to increase the reproducibility of sample plug volumes, including the cross-T [24, 25, 43, 72, 73], twin-T [16, 46, 47, 49, 74], and double cross-T [30, 62]. In the cross-T injection design, as developed by Harrison and coworkers, potential is applied across the sample channel, filling the region where the channels intersect. The potential is then switched and applied across the separation channel to accomplish separation (Figure 1.5b) [27]. As shown in Figure 1.1a and Figure 1.1b, the cross-T has been applied to multiplexed microchips; however, the large number of reservoirs required for this approach quickly becomes too cumbersome to individually address. Effenhauser and coworkers designed a double-T injector that has been more recently applied to multi-channel microchips [9, 13, 30, 62, 75] and requires fewer reservoirs. The Mathies group has utilized a double-T injection scheme on their rectilinear 48-channel microchip (Figure 1.6) [30] in an attempt to minimize the number of separation channels. As a result, twice as many samples can be analyzed with half the number of separation channels. The development of radial microchips has allowed for a significant reduction of reservoirs by incorporating a single common anode at the center of the microchip [16, 31, 46, 48–50, 53, 74]. Each of the radial

Separations in Multiple-Channel Microchips

Buffer reservoir

11

Buffer reservoir +

+ Sample reservoir +

Sample reservoir Sample waste +

Waste reservoir

Buffer waste

Injection

Separation (a)

Injection

Separation (b)

FIGURE 1.5 (a) Typical microchip schematic including sample injection (left) and sample separation (right) using T-type injection. Potential is applied between positive and ground. The arrows indicate the direction of flow in the channel. (Adapted from Manz, 1992.) (b) Typical microchip schematic including sample injection (left) and separation (right) using a cross T-injector. Potential is applied between positive and ground. The arrows indicate the direction of flow in the channel. (From Harrison, 1993.)

microchips discussed in this chapter utilize various T-type injection schemes. Both the 96-straight-channel microchip and the 96-folded-channel microchip function with twin-T injectors [16, 46–49, 53, 74] (Figure 1.7a) [74]. However, the 384channel microchip utilizes the simpler, more compact T-type injector (Figure 1.7b) [31]. Additional studies have focused on investigating both the T-type and cross-T injection configurations in narrow sample channels [71, 72] to eliminate the need for leakage control during the injection process (Figure 1.8) [72]. In this design, eight parallel separation channels with different injection designs are employed with a common anode and a common cathode for the application of separation potential. The sample channels are of smaller dimension than traditional sizes used and have been studied to optimize injection on microfabricated devices. The fabrication of these microchips involves production of a glass master and subsequent pouring of poly(dimethylsiloxane) (PDMS) onto the master to form the channel structures. Each of the eight channels is individually equipped with a set of sample injection and waste reservoirs by punching holes in the PDMS layer at both ends of the sample channel. In this design, the sample channel widths are one fifth that of the separation channel. Using these narrow sample channels, a 10-fold increase in sensitivity and a 2-fold increase in column efficiency over conventional injection methods have been reported. Additionally, the narrow sample channels decrease injection volumes; thus, no leakage control is required and this narrow channel injection method provides a simplified version of traditional cross and T-type injectors [62]. Additional studies by Beard and coworkers have investigated field-amplified sample stacking of biogenic amines on microfabricated electrophoresis devices [71]. The stacking method utilizes T-type injection through narrow sample channels on a six-channel microchip where all channels share both the anodic and cathodic

12

Advances in Chromatography Sample 1

Sample 2

B

C

Cathode

Anode D

Waste

Cathode

Anode Sample 1

Sample 2

B

C (a)

Waste

Waste

Cathode

Anode

Cathode

Anode

Sample 1

Sample 2

Sample 1

Sample 2

(b)

(c)

Waste

Waste

Cathode

Anode

Cathode

Anode

Sample 1

Sample 2

Sample 1

Sample 2

(d)

(e)

FIGURE 1.6 (a) Layout of the double-T sample injector. The sample injector includes four sample reservoirs, two separation columns, and one common waste reservoir, labeled as in Figure 1.1c. (b–e) Fluorescence images illustrating the operation of the injector with fluorescein. (From Simpson, 1998).

reservoirs (Figure 1.9a) [71]. The distances between the injection intersection and the buffer inlet determine the length of the sample plug, and voltage switching on a microchip with a common anode and cathode provides a simple method of simultaneously stacking samples in multiple channels for subsequent separation (Figure 1.9b) [71].

Separations in Multiple-Channel Microchips

13

33 mm

Sample

Waste Cathode

Detection Common area anode

Twin-T injector (a)

8.0 cm

Detection point (b)

Cathode 0

S3

S2

5 S1

S4 10 (mm) (c)

FIGURE 1.7 (a) Twin-T injector design for 96-channel radial capillary array electrophoresis microplate. Two individual channels share common cathode, waste, and anode reservoirs. (From Mednitz, 2000.) (b) Single quartet of channels with their T-type injectors for 384channel radial capillary array electrophoresis microplate. All channels in a quartet share a common cathode reservoir located closest to the edge of the wafer. The T-type injector design is shown in detail below. (From Emrich, 2002.)

1.5.2 OPTICALLY GATED INJECTION Optical gating, first developed for liquid chromatography in small diameter capillaries [76, 77] and for capillary electrophoresis [78], provides an alternative to traditional T-type injection on a microchip. This injection method requires no potential switching and requires only two reservoirs per channel, lending itself nicely to multiplexed microchip electrophoresis. This injection technique, introduced on a single channel microchip by Lapos and Ewing [79], has been used to demonstrate fast, serial, and reproducible injections on a single-channel microchip. In order to

14

Advances in Chromatography

Power supplies

Platinum wire

Buffer inlet

Injectors

Detection window

Conventional cross injector

NSC cross injector

NSC tee injector

Buffer outlet

Power supplies

FIGURE 1.8 Schematic layout of a chip composed of eight channels with different injection designs. On the right are the expanded views of three typical injectors: conventional cross, narrow sample channel cross, and narrow sample channel T-injectors. (From Zhang, 2001.)

achieve optically gated sample introduction, the laser is split into two beams. The more intense gating beam is used to perform sample introduction by time-discriminated photobleaching of the sample, and the less intense beam is used for detection (Figure 1.10a). To perform an injection, potential is applied and fluorescently-labeled sample continuously flows from the sample reservoir to the waste reservoir. Initially, the gating laser beam continuously photobleaches fluorescently labeled analytes as they pass through the channel. The gating beam is then briefly blocked, thereby allowing a small, unphotobleached sample plug to move past the injection window. This plug is separated and finally detected by laser-induced fluorescence detection as it passes through the probe beam. Through the entire injection, separation, and detection process, the potential never needs to be switched or toggled; therefore, the electronics required for optical gating remain extremely simple compared to traditional injection methods. As previously mentioned, optically gated sample introduction requires fewer reservoirs than other injection methods and does not require side arms for injection. Therefore, optically gated injection should allow easier integration of a large number of separation channels onto a single microchip than traditional injection methods. In a demonstration of this principle, a schematic of a multiple-channel optical gating microchip (Figure 1.10b) shows five 8-cm channels that converge in the center of the microchip to facilitate detection [22]. Since the advent of this microchip injection method, optical gating has been applied to the analysis of amino acids [22, 79], DNA fragments [23, 80], and enzyme reactions [61].

Separations in Multiple-Channel Microchips

15

Buffer outlet (floating)

Power supply *Diagram not to scale

1

100 um

Buffer

Buffer outlet Sample reservoir

Detection window

Sample

Sample loading

Platinum wires 1

2

3

4

5

Buffer inlet (ground) Buffer outlet (ground) Buffer

6 2

100 um

Sample

Buffer inlet

Sample reservoir Stack & separate (floating)

Buffer

Buffer inlet

(a)

(b)

FIGURE 1.9 (a) Schematic layout of the six-channel capillary electrophoresis array microdevice used to investigate sample stacking on microfabricated electrophoresis devices. (b) Illustration of the stacking process within the channel network with supporting CCD footage: (1) Sample loading. (2) Sample stacking and separation. (From Beard, 2003.)

1.5.3 CAPILLARY SAMPLE INTRODUCTION In contrast to most of the injection methods reviewed here, the capillary sample introduction technique is another scheme that does not require the toggling of voltages to perform injection. The use of a capillary for sample introduction eliminates the need for individual sample and waste reservoirs, making this microchip design quite simple (Figure 1.11a). Additionally, capillary sample introduction provides an effective means to interface the microchip with the macroscopic world. In this approach, the lanes of the microchip are filled with linear polyarylamide and plugs of sample are injected into each of the lanes (Figure 1.11b) either electrokinetically or by pressure [66, 68, 69]. Without interrupting the separation potential, injections are performed via a computer-controlled micromanipulator. In one example using capillary introduction, a five-channel microchip specifically fabricated to accommodate automated capillary sample introduction in a continuous manner has been used. The microchip design involves five open-ended channels such that one common anode and one common cathode can be employed [68]. Traditional photolithographic techniques have been used to create a microchip with parallel channels 12.7 cm in length, 500 to 700 μm wide, and 250 to 350 μm deep. The etched bottom plate is thermally bonded to a flat top plate, creating five enclosed straight channels for the analysis of various biological reactions. Given reasonable parameters, it is

16

Advances in Chromatography

Gating beam

Probe beam

Sample reservoir

Waste reservoir

1

2

3

4

e

5 (a)

2 mm

3 cm 8 cm (b)

Separations in Multiple-Channel Microchips

17

FIGURE 1.10 (a) Schematic illustrating optically gated injection. Fluorescently labeled sample is continuously electrophoresed through the channel. The gating beam is used to perform sample introduction by time-discriminated photobleaching of the sample. The gating beam continuously photobleaches fluorescently labeled analytes as they pass through the channel (1). An injection is accomplished by blocking the gating for a specific time (2); thus, a plug of unphotobleached samples moves past the injection window (3), is separated (4) and detected (5) by laser-induced fluorescence detection as it passes through the probe beam. (b) Schematic of five-channel optical gating microchip. (From Xu, 2002.)

theoretically possible to carry out 550 separations in 1 hour by increasing the number of lanes on the microchip to 37 [68]. The increased throughput and the simplicity of the chip design and electronics make capillary sample introduction an attractive alternative to the traditional methods of sample introduction.

1.6 DETECTION METHODS Electrochemical detection, mass spectrometry, chemical luminescence, and laserinduced fluorescence have been the most popular methods for detecting separation analytes on single-channel microchips. Readers are directed to a review by Uchiyama et al. [81] for more information on these techniques. While these detection methods

FIGURE 1.11 (a) Chip design for capillary sample introduction with five open separation channels. (A) Microfabricated separation channel. (B) Glass partition. (b) Digital photos in chronological order (A–H) of manipulator-controlled movement of a capillary in and out of two channels of a chip. Injection of sample occurs at (C) and (H). (From Smith, 2001.)

18

Advances in Chromatography

are commonly applied to single channel microchips, detection on multi-channel microchips requires more complicated instrumentation. As a result, laser-induced fluorescence detection is by far the most common detection method for multi-channel microchips due to its inherent sensitivity and adaptability [2, 6]. Many variations of confocal laser-induced fluorescence detection, as developed by the Mathies group in the early 1990s [2, 3, 82], have been adapted for different microchip designs and have been reviewed extensively [3, 4, 26, 83, 84]. Confocal methods, scanning methods, and charge-coupled device (CCD) detection methods will be reviewed in the sections below.

1.6.1 SINGLE-POINT CONFOCAL DETECTION Single-point confocal detection has been previously described [3–5, 82, 85] and provides the basis for all other types of confocal detection. Briefly, a laser beam is collimated and subsequently reflected by a dichroic beam splitter through an objective lens. The beam is focused on the center of the microchip channel at the point of desired detection. Emitted fluoresence and scattered excitation laser light from the channel are collected and collimated by the objective lens. The light is then passed through a dichroic beam splitter and the emission wavelength is focused through an achromatic lens and pinhole that is confocal with the achromatic lens. Only fluorescence from the separation channel in the microchip is passed through the pinhole while scattered light is blocked. This increases the signal-to-noise ratio and allows sensitive detection of a variety of fluorescent molecules. Typically, photomultiplier tubes (PMT) or CCDs are used to detect the emitted fluorescence passing through the pinhole. Confocal detection has been applied to multi-channel microchips for detection in individual channels where multiple confocal detection units have been set up for each channel [45]. The adaptations for multi-channel microchip detection utilizing only a single detector will be discussed below [16, 24, 30, 31, 43, 46, 47, 49, 53, 74, 86].

1.6.2 TRANSLATION-STAGE CONFOCAL DETECTION The development of confocal detection with a translation stage was originally applied to capillary array electrophoresis and called a confocal-fluorescence capillary array scanner [2]. In this method, a computer-controlled stage is used to translate the capillary array past the optical system such that data can be collected from all capillaries in the array. Similarly, the translation stage has been adapted for use with multi-channel microchips where the microchip is moved back and forth through the excitation laser beam [22–24, 30, 61]. In one particular application, the 12-channel rectilinear microchip shown in Figure 1.1a is placed on a computer-controlled translation stage [24]. The detection laser beam is focused on the detection region of the microchip, and the microchip is translated through the beam. Fluorescence is collected by the objective, passed through a dichroic beamsplitter, and divided into two fluorescence channels with a second dichroic beamsplitter. The signal from each channel is then focused with a lens onto a confocal spatial filter prior to detection with a PMT (Figure 1.12a) [24]. This detector has been further improved to include

Separations in Multiple-Channel Microchips

19

the detection of four independent fluorophores; however, the four-color work has been completed on a capillary array [87]. In another application, the five-channel optical gating microchip shown in Figure 1.7b has been traversed through both the gating and detection (probe) beams, and fluorescence has been collected in the traditional confocal manner (Figure 1.12b) [23]. In this arrangement, the microchip is fixed to a translation stage that is attached to a voice coil actuator. The actuator is used to move the microchip back and forth at a repetition rate of 10 Hz, causing the two beams to scan repeatedly across all five channels on the microchip. This technique has been used to investigate amino acid [22] as well as oligonucleotide [23] separations. In both of these applications, the microchip is attached to a translation stage and moves through the stationary detection beam.

1.6.3 SCANNING METHODS

OF

DETECTION

A variety of detection methods exist in which the microchip remains stationary while the laser beam is manipulated such that all channels on a multi-channel device can be illuminated and thus detection of multiple channels can be achieved. Galvanometric scanner confocal, rotary scanner confocal, acousto-optical deflection (AOD) scanner, and CCD detection will be reviewed in the following sections. 1.6.3.1 Confocal Fluorescence Detection with a Galvanometric Scanner High-speed, high-throughput separations are characteristic of microchip separations. Detection methods are therefore continuously being improved to allow for the detection of entire peak profiles as a fluorescent analyte is detected. The galvanometric scanner (Figure 1.13a) [30] has been developed and used for high throughput genetic analysis on the microfabricated 96-sample capillary array electrophoresis microchip of the design shown in Figure 1.1c [30]. The laser beam is focused through a 0.33 numerical aperture scan lens system designed for quantitative field analysis to a 5-μm region [88]. The beam is scanned across 48 stationary channels at 40 Hz and data are collected through the detection optics including a dichroic beamsplitter, emission filter, lens, pinhole, and PMT in the same manner as traditional confocal detection. DNA sequencing using four-color detection on a similar instrument has also been carried out with this system [87]. 1.6.3.2 Confocal Fluorescence Detection with a Rotary Scanner The scan speed for detection with a translation stage is limited; therefore, a new confocal detection system with a rotary scanner has been developed (Figure 1.13b) [16]. This detection system consists of a rotating objective head coupled to a fourcolor confocal detection unit. A laser is directed through a series of mirrors and dichroic beamsplitters through the rotating central hollow shaft of a stepper motor. The beam is then displaced by a rhomb prism and focused into the microchip by a

20

Advances in Chromatography

Computer Amplifier

Mirror Spatial filter Lens 645 mm LP filter

PMT

Dichroic beam splitier 488 mm laser

Spectral filter

Mirror Dichrole beam splitier Objective

CAE chip

Translation stage Detection zone

Buffer reservoirs (a)

CPU

Mirrors

Lens Shutter

Argon ion laser

Motor controller

IIV

Beam splitter Gating beam

Chip fixed on a translation stage Objective

Probe beam

Collimator/expander Dichroic mirror

Keithley Voice coil actuator

Filter Photomultiplier tube

(b)

FIGURE 1.12 (a) Schematic of the laser-excited confocal-fluorescence scanner and a capillary array electrophoresis chip. (From Woolley, 1997.) (b) Multichannel system setup with optically gated sample introduction and laser-induced confocal-fluorescence detection. (From Xu, 2004.) The laser beam is stationary in both of these detection schemes while the chip is attached to a stage that translates each channel through the beam.

Separations in Multiple-Channel Microchips

21

FIGURE 1.13 (a) Schematic of the laser-excited galvanoscanner and the capillary array electrophoresis microplate assembly. (From Simpson, 1998.) (b) Schematic of the rotary confocal-fluorescence scanner used to detect separations on radial capillary array electrophoresis microplates. (From Shi, 1999.)

22

Advances in Chromatography

rotating microscope objective. This objective also serves to collect the emission wavelength derived from the sample. The fluorescence is passed along the reverse optical path through the dichroic beamsplitter to the four-color confocal fluorescence detector [16, 87]. This detection system has proven capable of scanning and collecting information at rates up to 20 revolutions per second with 5000 points collected per revolution per channel. In this detection scheme, the microchip is stationary while the microscope objective rotates and has been employed for various DNA applications as well as for an immunoassay. 1.6.3.3 Fluorescence Detection with an Acousto-Optical Deflection Scanner Thus far, all of the fluorescence detection methods described require the physical movement of either the microchip or the laser beam. Conversely, an AOD scanner does not require physical movement of either the microchip or a laser (Figure 1.14), resulting in scanning rates as high as 30 Hz [73]. This method of detection, as developed by Landers and coworkers [25, 73], allows flexible, high-speed, selfaligning scanning for fluorescence detection in multi-channel microchip electrophoresis. The scanning unit of the AOD scanner is dictated by a transparent crystal that changes the direction of laser light based on the frequency of the acoustic wave propagating through it [21, 25, 73]. In brief, the laser light is refracted as it travels through the crystal and the degree of refraction is adjusted by changing the frequency of the acoustic wave generated within the crystal using a piezoelectric transducer. A computer-controlled voltage-to-frequency converter controls this transducer, moving the laser from channel to channel. The AOD scanner is unique in that it is possible to address individual, nonadjacent channels and thus data may be collected from any number of channels in any order. Using this method, a dichroic beamsplitter, an objective, and a bandpass filter have been used to collect emission wavelength light, while a CCD camera and a video capture device have been used in place of a PMT for detection purposes for the investigation of amino acids as well as DNA digest samples.

1.6.4 FLUORESCENCE DETECTION DEVICE (CCD)

WITH A

CHARGE-COUPLED

Clearly, addressing each channel on a microchip can be challenging. Most detection methods reviewed thus far have utilized some variation of confocal detection. Confocal methods, however, are not the only way to monitor fluorescence detection on a multiple-channel microchip. In the past, CCDs have been thought to be incompatible with multi-channel microchip separations [26]; however, CCDs have been used successfully on capillary array electrophoresis systems [89–91] and are now being applied to microchip separations. Ewing and coworkers have used a detection scheme in which an argon ion laser beam is expanded and focused into a line across all channels on a microfabricated microchip (Figure 1.15) [68]. This is similar to detection schemes developed earlier for ultrathin slab gel electrophoresis [92–97].

Separations in Multiple-Channel Microchips

23

FIGURE 1.14 Acousto-optic detection system and multichannel chip. (a) Deflection of the incident laser beam by the acousto-optic crystal. Deflection angle is determined by the frequency of the acoustic wave. (b) Optical setup. (c) Eight-channel microchip configuration. (d) Detection region of the microchip for laser beam scanning. (From Ferrance, 2001.)

As sample passes through the laser line, a CCD camera perpendicular to the microchip detects the emission fluorescence [66, 68, 69]. Pictures are taken at predetermined intervals and compiled such that a complete image of the separations in all lanes is obtained. This detection method allows simultaneous detection of analytes from separations occurring in each channel, thereby providing another methodology that physically eliminates the need for translating the microchip or scanning the laser beam across all channels.

1.7 APPLICATIONS OF MULTIPLE-CHANNEL MICROCHIPS Microfluidic devices continue to be utilized for an increasing number of applications. For instance, single-channel devices have been applied to the analysis of DNA for sizing, genotyping, and sequencing as well as for the integration of PCR on a microchip. Many single-channel devices have also been used for the analysis of proteins, peptides, tryptic digests, enzyme assays, immunoassays, catecholamines, and many other biological analyses that have been reviewed by Bruin [98]. The current chapter has presented multiple-channel microchip designs, injection and detection methods, and will now discuss several bioanalytical applications of these multiplexed devices.

24

Advances in Chromatography

FIGURE 1.15 Charge-coupled device detection. Schematic of the chip setup with capillary sample introduction. Laser-induced fluorescence detection is achieved by focusing an argon ion laser into a line across the chip. A liquid nitrogen-cooled CCD is used to collect the fluoresence. (Adapted from Smith, 2001.)

1.7.1 DNA SEPARATIONS 1.7.1.1 Separations Involving Restriction Enzyme Digests — DNA Fragment Analysis Restriction enzymes are DNA-cutting enzymes found in bacteria. Because they cut within the molecule, they are often called restriction endonucleases. A restriction enzyme recognizes and cuts DNA only at a particular sequence of nucleotides, typically four to eight base pairs in length [99]. Restriction enzymes are essential for analyzing chromosome structures, mapping and sequencing very long DNA molecules, isolating genes, and creating new DNA molecules for cloning purposes. DNA fragments produced by the action of a restriction enzyme can serve as a fingerprint of an overall DNA molecule. Multiple-channel microchips provide a method for the rapid analysis of these DNA fragments. Woolley and Mathies carried out a separation of FX174 Hae III DNA restriction fragments on a 15-channel microchip, proving that microchips are capable of highresolution electrophoretic separations [43]. These separations have been completed in only 120 seconds [43]. Additionally, this process demonstrate the high-speed sizing of PCR-amplified HLA-DQa alleles [43]. This experiment laid the groundwork for high-speed, high-throughput DNA separations on multiple-channel microchips. Simultaneous separation and detection of the double-stranded DNA (dsDNA) fragments of pBR322 has been accomplished by Huang et al. on an eight-channel microchip using AOD [25]. Further studies conducted by the Mathies group employ

Separations in Multiple-Channel Microchips

25

a radial 96-channel microchip — initially using one-color detection — for the separation of 96 pBR322 MspI samples labeled with TOTO [16]. In addition to separations of short DNA fragments, there are examples in the literature of microchip separations of long DNA molecules. A nanofluidic channel device, consisting of many entropic traps, has been designed and fabricated for the separation of long DNA molecules [100]. This nongel-based method for the separation of DNA has been used to separate a T2–T7 DNA mixture [100]. The device used is composed of multiple channels with narrow constrictions and wider regions that cause size-dependent trapping of DNA at the onset of a constriction. Under these conditions, electrophoretic mobility differences are created and separation without the use of a gel matrix or pulsed electric fields could be accomplished. With this technology, samples of long DNA molecules have been efficiently separated, and multiple-channel devices operating in parallel have been demonstrated [100]. Further development of this technique has demonstrated that a microfabricated entropic trap array can be useful in separating large (5 to 200 kbp) DNA molecules efficiently (within 30 minutes) by DC electrophoresis on a microchip platform without a sieving matrix [101]. The investigators report further development of the technique, with emphasis on the optimization of separation selectivity and resolution for these DNA, and comment on the requirement and feasibility of separating megabase-pair DNA with the entropic trap array device [101]. Membrane-mediated restriction digestion and rapid analysis of cleaved DNA fragments has been accomplished using a 96-channel gel microchip electrophoresis separation system [58]. Complete digestion of an FX174 plasmid DNA by Hae III (a 101 base pair PCR product of the pUC19 cloning vector by Sma I) as well as the GFP/pABarMCS construct by Bst EII, Hind III, Bam HI and by the combination of Sac I and Sma I enzymes, have been accomplished in a few minutes on a membrane. Membrane-mediated sample loading [102] has been used to inject these samples into the separation device. The researchers report the possibility of the analysis of up to 6400 samples per day, particularly when robotic liquid handling is employed [58]. Continuous monitoring of a restriction enzyme digest of DNA on a five-channel microchip has also been demonstrated; however, this digestion was monitored dynamically and will be presented in a different application later in this review [66]. A variety of multiple channel electrophoresis devices have been utilized for the analysis of DNA fragments and as technology progresses these multiplexed microchips will be employed for various other DNA applications. 1.7.1.2 Genotyping The development of safer and more effective individualized medicine is possible through the increased understanding of how a particular genotype interacts with a potential drug. Traditional methods of genotyping, typically slab-gel electrophoresis and more recently capillary electrophoresis, are serial techniques that make them both cumbersome and labor intensive. Thus, these techniques are not well suited for large-scale genetic testing. Multi-channel microchip electrophoresis has proven to be a high-throughput, cost-effective alternative to traditional methods.

26

Advances in Chromatography

Hereditary hemochromatosis (HHC) is associated with an increased risk of cirrhosis, cardiomyopathy, and hepatic cancer; thus, early detection through rapid and low-cost screenings can allow early diagnosis and treatment, and a normal life span for patients. Towards this goal, genotyping of HLA-H, a marker gene for the diagnosis of HHC, has been performed on a 12-channel microchip to demonstrate the rapid analysis of biologically relevant samples [24]. Further studies have been conducted to demonstrate the functionality of a microfabricated 96-sample capillary array electrophoresis microchip for the separation and fluorescence detection of a restriction fragment marker for the diagnosis of HHC [30]. This restriction fragment marker for HHC is derived from the HFE gene and separations of 96 HFE amplicons have been performed on a 48-channel microchip [30]. A total of 19 different samples have been used for a 5-fold redundancy in sample analysis. Thus, serial injections proved to be an effective method of increasing the sample throughput with a limited number of separation channels [30]. Furthermore, this work illustrates the utility of multi-channel microchips for genetic disease screening. In an attempt to reduce the cost of high-throughput applications such as population screening for genetic diseases, Sassi and coworkers have demonstrated rapid, parallel separations of D1S80 alleles in a plastic 16-channel microchip [86]. D1S80 alleles are amplified from genomic DNA extracted from whole blood and separated. One of every four channels contains the reference alleles. Ultimately, the authors have demonstrated the feasibility to perform high-throughput genotyping in injection-molded, plastic multi-channel microchips. They report that with appropriate internal standards and optimization of microchip design and separation conditions, the accuracy of typing on these polymeric microchips could be improved to match the accuracy of conventional slab-gel electrophoresis [86]. In another genotyping experiment, the 96-channel radial microchip described earlier in this review has been used to perform an analysis of methylenetetrahydrofolate reductase (MTHFR) allelic variation. To accomplish this, PCR amplicons of the MTHFR gene are digested with Hinf I and detected in two-color mode [16]. The same 96-channel device is used to perform high-speed single nucleotide polymorphism (SNP) typing of an HHC mutation [74]. In these studies, fluorescently labeled sequence-specific primers are used to generate allele-specific products for the S65C polymorphism. The PCR product is subsequently separated and typed. Thus, the authors have demonstrated the potential of using covalently labeled sequence specific PCR amplicons with multi-channel microdevices for high-throughput SNP genotyping analyses [74]. Another study has focused on genotyping energy-transfer-cassette-labeled, short-tandem-repeat amplicons with four-color detection [46]. Covalent labeling with energy-transfer (ET) tags can be used to increase the throughput of genotyping analyses using multi-channel microchip electrophoresis. In this study, 122 samples are separated and genotyped in 96 channels in less than 8 minutes [46]. Thus, the work demonstrates the powerful combination of energy-transfer-cassette labeling and multi-channel electrophoresis for high-performance, short-tandem-repeat analysis. Similarly, demonstration of the combined use of an allele-specific polymerase chain reaction (ASPCR) with energy transfer primers and multiple-channel microchip electrophoresis has also been conducted on the 96-channel radial microchip

Separations in Multiple-Channel Microchips

27

[48]. Previous analyses have employed one- or two-color detection; however, this study demonstrates the use of four-color DNA analysis for genotyping using ET labels [48]. A different type of 96-channel gel microchip has been used for high-throughput genotyping of repeat polymorphisms in the regulatory region of the serotonin transporter gene (5-HTTLPR) [103]. To accomplish this, the injection end of the gel features a straight edge with no individual wells. Samples are injected onto the gel microchip by membrane-mediated loading technology using 96-tab membranes. Automated spotting of the loading membrane is accomplished with a robot directly from a 96-well PCR plate [103]. The researchers report separation and visualization of the PCR amplicons from either the 5-HTTLPR short allele or the 5-HTTLPR long form in a few minutes [103]. Massively parallel genetic analysis is attractive for high-throughput evaluation of genetic variation and multi-channel microchips are a valuable tool in this endeavor. A microfabricated 384-channel microchip has been developed and utilized for massively parallel genetic analysis. For instance, the simultaneous genotyping of 384 individuals for the common hemochromatosis-linked H63D mutation in the human HFE gene has been achieved in only 325 seconds (Figure 1.16) [31]. The H63D mutation is detected by digestion of a PCR product with Mbo I. Mutant amplicons remain undigested while wild-type amplicons are cleaved into two fragments [31]. This study introduces a powerful new tool for accelerating gene mapping, pharmacogenomic screening, forensics and proteomics as well as providing a massively high-throughput bioanalyzer [31]. Significant advances have been made in the area of genotyping on multi-channel microchips in the recent past. The ability to perform rapid high-throughput population screening on integrated multiplexed devices in the clinical setting can be envisioned in the near future. 1.7.1.3 Sequencing in Multiple-Channel Microchips DNA sequencing involves the determination of the precise sequence of nucleotides in a sample of DNA and thus requires excellent resolution of separated peaks such that strands that differ by only a single nucleotide can be differentiated. Single capillaries, capillary arrays, and single channel microchips have been employed for DNA sequencing and have been reviewed elsewhere [104]. Multiple-channel microchips have been applied to DNA sequencing and provide higher throughput than traditional methods [26]. Optimization of high-speed DNA sequencing has been performed in a fourchannel microchip [105]. The separation matrix, separation temperature, channel length and depth, injector dimensions, and injection parameters have been optimized such that one-color sequencing separations of single-stranded M13mp18 DNA could be accomplished [105]. Additionally, four-color sequencing separations have been performed in 7-cm long channels, and experiments completed in under 20 minutes with 99.4% accuracy to DNA sizes up to 500 bp [105]. These optimization studies have significantly advanced sequencing on microfabricated devices. A 48-channel microchip has been utilized to separate single-stranded DNA fragments up to 500 bases in length [106]. Using this separation device, a 400-base

28

Advances in Chromatography

FIGURE 1.16 DNA microchip application. Electropherograms collected for massively parallel genetic analysis on the 384-lane microcapillary array electrophoresis chip. RFLP fragments are highlighted in lanes 1 (wild type) and 193 (heterozygote). Genotyping of the samples is possible in 379 lanes (98.7% success rate). Electrophoresis is completed within 325 seconds. (From Emrich, 2002.)

read has been obtained with 97% accuracy and 440 bases are sequenced with 93% accuracy on a 10 cm channel. These experiments utilize a novel transmission imaging spectrograph for detection and provides an increase in sample throughput by color multiplexing [106]. Throughput has been significantly increased but accuracy suffers in these experiments. Additional sequencing data has been collected using a 16-channel microchip as shown in Figure 1.1b. After optimization of injection time, temperature, and template

Separations in Multiple-Channel Microchips

29

concentration, four-color DNA sequencing has been performed in 16 microfabricated channels [45]. Greater than 99% accuracy is observed for the base-called sequence of 543 bases. Separations carried out in less than 15 minutes typically resolve 457 ± 35 bases with 99% accuracy and 512 ± 44 bases with 98% accuracy [45]. DNA sequencing is highly dependent on the separation resolution; hence, several groups have employed longer channels to increase resolution. Devices containing 48 independent straight microchannels, 50 cm in length, have been fabricated [107]. Single-base resolution for read lengths greater than 600 bases has been achieved with this device and the sequence has been called to 640 bases with 98% accuracy [107]. Additionally, a 96-channel hyper-turn microchip has been developed [47, 49]. This microchip is suited for DNA sequencing where long separation lengths are required. The 16-cm-long hyper-turn channel device initially provides 96 parallel, high-quality DNA sequencing separations to approximately 500 bases in less than 30 minutes [47]. More recently, high-throughput DNA sequencing has been performed and coupled to four-color fluorescence detection [49]. Data have been successfully collected from 95 of the 96 channels where an average read length of 430 bp has been obtained. The probability of incorrectly calling or identifying a base for over 80% of the read length is 0.01% [49]. This technology is capable of producing sequencing data at a rate of 1.7 kbp/min, which the authors report is a 5-fold increase over current commercial capillary array electrophoresis instruments [49]. An interesting hybrid device consisting of a microchip containing multiple twinT injectors attached to an array of capillaries that serve as separation channels has been used to sequence over 1000 real-world samples [108]. The investigators report an average Phred20 read of 675 bases in about 70 minutes with a success rate of 91% as compared to similar samples on MegaBACE 1000 where the average Phred20 read is about 550 to 600 bases in 120 minutes with a success rate of about 80 to 90% [108]. Sequencing with a high degree of accuracy is highly desirable when searching for minute differences in DNA sequences. Thus far, multiplexed microchip electrophoresis has proven to be an efficient platform for increasing throughput for sequencing studies.

1.7.2 AFFINITY ELECTROPHORESIS MICROCHIPS

FOR IMMUNOASSAYS IN

The immunoassay is one of the most important analysis methods for biological molecules. Thus, it is clear why so many groups are interested in integrating immunoassay on a microchip, particularly for its application in point-of-care analysis. There are several methods for conducting an immunoassay on a microchip; however, most have been accomplished on single channel devices. A small-volume heterogeneous immunoassay system has been demonstrated by Hayes et al. in microchannels exploiting magnetic manipulations of small paramagnetic particles [56]. Kitamori’s group has integrated heterogeneous immunoassay systems with microfabricated separation devices as well [64, 109, 110]. Most recently, a bead-bed immunoassay system suitable for simultaneous assay of multiple samples has been constructed on a microchip [64]. The microchip used in these experiments utilized branching

30

Advances in Chromatography

channels and four reaction and detection regions. This device has been used for the simultaneous analysis of four interferon γ samples by a three-step sandwich immunoassay [64]. The assay time for the four samples is 50 minutes, a significant improvement over the 35 minutes it takes for one sample to be analyzed in the single channel assay [64]. The Harrison group has been at the forefront of immunoassay on microfabricated devices [111–116] and has recently developed a multi-channel microfluidic analysis system employing affinity capillary electrophoresis for immunoassay [62]. In this application, a six-channel immunoassay device (depicted in Figure 1.3) has been used to complete simultaneous direct immunoassays for ovalbumin and for antiestradiol. Mixing, reaction, and separation are performed in less than 60 seconds, and in 30 seconds under optimized conditions. This work demonstrates the multiplexing of mixing and reaction steps as well as injection and separation, thus providing increased throughput for complex sample processing [62]. Additional information on enzyme assays and immunoassays on microfabricated devices can be found in a review by Guijt and coworkers [57]. Recently, Bromberg and Mathies developed a homogeneous immunoassay for the detection of TNT and some of its analogues utilizing a single channel microfabricated device [117]. Following this, a high-throughput homogeneous immunoassay for the detection of TNT has been developed using 46 channels of the already described 96-channel radial microchip (Figure 1.2a) [50]. This assay is based on competition between the fluorescein-labeled trinitrobenzene derivative (TNB-Fl) and TNT for antibody binding sites. Sample preparation, reagent mixing, and incubation have been performed off the microchip, whereas injection and separation of the reactants and detection of the resulting complex have been performed on the microchip. Data collected from 46 channels showed nine different TNT sample concentrations, and each trace represented three serial injections (Figure 1.17a) [50]. A calibration curve has been constructed by plotting the ratio of signal intensity for free TNB-Fl to the signal intensity for the antibody-TNB-Fl complex as a function of the TNT concentration (Figure 1.17b) [50]. The equilibrium binding constant, binding rate constant, and dissociation rate constant of the labeled TNT have been determined. Consequently, this study demonstrates the feasibility of using multichannel microfabricated devices to simultaneously analyze multiple samples with homogeneous immunoassay methods. In the future, this technology could be used to probe for the presence of explosives in environmental samples or to perform highthroughput clinical assays [50].

1.7.3 DYNAMIC ANALYSIS OF REACTIONS WITH MULTIPLE-CHANNEL ELECTROPHORESIS 1.7.3.1 Application to DNA While the function of restriction enzymes has been explored in the laboratory, much remains to be learned about their structures, their binding characteristics, and their mechanisms of cleavage. A separations-based method of continuously sampling from a DNA restriction enzyme digest has been developed by Roddy et al., providing a

Separations in Multiple-Channel Microchips

1

5

2

10

8

15

Relative fluorescence

31

3 6 4

25

Ratio

20

4

0.2

5 0.1

2

30

6

0 0

35

7

0

100

0

1 200

2

3 300

TNT (ng/ml) (b)

40

8 45

0

9 100 Time (sec)

200

(a)

FIGURE 1.17 TNT microchip bioassay. (a) Multichannel homogeneous TNT immunoassay performed on a microfabricated capillary array electrophoresis device. Nine samples are dispensed to 46 wells (numbered to the left side of the figure). TNT sample concentrations: 0, 0.3, 1, 3, 10, 20, 50, 100, and 300 ng/mL for samples 1–9, respectively. Each trace presents three consecutive injections. The first peak marks the beginning of the 30-second injection, the second peak is the Ab-TNB-FL complex and the third peak is the free TNB-Fl. (b) Calibration curve for TNT assay obtained by plotting the ratio of signal intensity of free TNBFl to the signal intensity of the Ab-TNB-Fl complex as a function of the TNT concentration. Each point is an average of signals obtained from wells of the same TNT concentration. (From Bromberg, 2004).

means to monitor changes in substrate and product concentrations over the course of the reaction [66]. This method utilizes a capillary as a sampling device to transfer sample from a single reaction mixture to a microfabricated device with an array of separation lanes and has been initially optimized by separating Hae III-digested pBR322 [68]. More recently, this methodology has been applied to the analysis of a restriction enzyme digest during the course of digestion. Specifically, a fluorescently-labeled 62 base pair double-stranded DNA fragment containing a single Kpn

32

Advances in Chromatography

I site has been chosen to demonstrate the ability of this multi-channel separation device to monitor the progress of a dynamic reaction [66]. In brief, the reaction is initiated offline and as the sample enteres the microchip, the reaction is quenched on-line. Therefore, each injection into the microchip captures a time-point during the reaction such that all of the fragments are observed [66]. The results of a digest carried out at 45˚C using one unit of Kpn I per 13 ng of DNA have been reconstructed by plotting the fluorescence signal at the laser line collected by the CCD camera over time (Figure 1.18a) [66]. A cross-section of intensity vs. time for the separations shown in lane five yields a series of electropherograms (Figure 1.18b) [66]. A distinct change in sample composition has been observed as time progresses. This methodology has provided continuous analysis of the products of a digest reaction, thereby contributing to the fundamental understanding of restriction enzyme binding and activity as well as enhancing practical knowledge for the use of restriction enzymes as essential tools in the laboratory [66]. 1.7.3.2 Application to RNA Reactions RNA enzymes, or ribozymes, behave similarly to protein-based enzymes; thus, there is heightened interest in understanding the mechanism of cleavage of these molecules. The applications of RNA catalysts can be expanded as kinetic information is ascertained. Kinetic analysis of ribozymes typically involves the tedious process of collecting and quenching reaction time points and then fractionating by polyacrylamide gel electrophoresis (PAGE). As a way to automate and simplify this process, continuous analysis of ribozyme reactions have been demonstrated by Paxon et al. [69] Completely automated capillary sample introduction on a five-channel microchip with laser-induced fluorescence detection is employed to examine a 30-nucleotide ribozyme cleavage reaction that is sensitive to the lead ion. The ribozyme cleavage is initiated by 10 μM Pb2+ and the progress of this reaction is monitored for 60 minutes. A cross-section of a single channel is shown as a relative intensity vs. separation time (min) electropherogram in Figure 1.19a [69]. The first peak in each doublet set of peaks represents the fluorescently labeled 6-mer elution while the second peak is indicative of the 30-mer elution. Raw data from Figure 1.19a have been plotted as peak height vs. injection time and are shown in Figure 1.19b. Product formation and starting material cleavage showed mirror image behavior. The presence of only two peaks has shown to be consistent with a simple cleavage reaction as well as the absence of side reactions, RNA degradation, and alternative conformations. Fraction cleaved vs. injection time has been plotted in Figure 1.19c and kinetic information about this ribozyme has been obtained. This work demonstrated, through the use of a simple ribozyme model, the potential of multi-channel microchips with capillary sample introduction to provide valuable kinetic information for biologically relevant RNA and protein enzymes.

1.8 CONCLUSIONS AND FUTURE PERSPECTIVES Microchip separations have evolved considerably since their inception in the early 1990s, particularly in the area of multiplexing. There are numerous areas where

Separations in Multiple-Channel Microchips

33

0 500

Time (s)

1000 1500 2000 2500 3000 3500 0

200

0

400 600 Pixel number

800

1000

500 1000 1500 2000 2500 3000 3500

Intensity (a)

2500

Intensity

2000 1500 1000 500 0 0

10

20

30

40

50

60

Time (min) (b)

FIGURE 1.18 DNA digestion assay. (a) Reconstructed plot of time vs. pixel number across the width of the chip. The shaded bar represents fluorescence intensity collected by the CCD. The capillary moved right to left depositing samples into the chip, so the lane farthest to the right is lane 1. (b) Cross section through lane 5 of the plot in (a) showing intensity vs. time. (From Roddy, 2003.)

34

Advances in Chromatography

Relative intensity

7000 5250 3500 1750 0 0

10

20

40 30 Separation time (min)

50

60

70

(a) 1

7000

0.8

5000 4000

Feleaved

Peak height

6000

3000

0.6 0.4

2000 6-mer 30-mer

1000 0 0

20

40

Injection time (min) (b)

60

0.2 0

0

10 20 30 40 50 60 70 Injection time (min) (c)

FIGURE 1.19 RNA ribozyme cleavage assay. (a) Cross-section through lane two of a leadzyme reaction in the presence of 10 μM lead acetate illustrating fluorescence intensity vs. separation time. Asterisk highlights 5′ product (6-mer). (b) Plot of peak height for the 6mer and 30-mer strands vs. injection time for the experiment shown in (a). (c) Plot of fraction cleaved vs. time of injection. (From Paxon, 2004.)

multi-channel microchip separations require new advances to be made including detection methods and application to real-world samples. Indeed, a number of applications from DNA to proteins to immunoassays have been developed around this technology to date. Detection methods for multiple-channel microchips have been limited to laser induced fluorescence detection. However, work has been published on detection methods on single channel devices including sheath flow detection [118], electrochemical detection [119–125], and mass spectrometry [18, 126–134]. One can envision any number of detection methods applied to multiplexed microchips for increased sensitivity when analyzing biological samples present at very low concentrations or in limited amounts. Multiple-channel microchips hold the promise of higher throughput separations and, as capillary array electrophoresis replaced single capillary and slab-gel electrophoresis in the high throughput separations that were needed to sequence the

Separations in Multiple-Channel Microchips

35

human genome, multi-channel microchips might play an equally important role in the future of separations on chips. In addition, these microchip designs are a probable precursor to integrated systems requiring multiple channels of any kind, and the fundamental science and engineering developed in this area should be invaluable. An important part of the future of multi-channel microchips appears to be the use of these systems to follow reaction dynamics, and this will lead to more detailed and cellular applications in biology. In another context, multi-channel microchips could be used to mimic the concentration profiles in biological systems with different amounts of solutes in different channels representing spatially differing concentrations in cells, organs, or circulatory systems. In still another direction, the Orwar group has developed a multi-channel microchip for screening drugs with a cell sensor that is scanned at the outlet of multiple channels [135–138]. Although this has not yet been used for separations in the traditional sense, the use of multiple channels allows different reagents or drugs to be transported to where a cell acts as a biological detector. This is yet another example of the versatility of the multi-channel microchip concept. This chapter has presented recent advances in multi-channel microchip designs, sample introduction, and detection. Furthermore, high-throughput analyses of complex analytes have been demonstrated utilizing these multiplexed microchips. This chapter has outlined the enormous potential and current challenges for the multichannel microchip community and discussed a variety of applications that illustrate the immense growth in the field of multiplexed separations.

ACKNOWLDGMENTS The authors would like to acknowledge the NSF and NIH for funding, in part, the research presented here. We thank our co-workers past and present for work that is referenced herein. Additionally, we thank current members of the Ewing research group for helpful comments about this manuscript.

REFERENCES 1. Issaq, H.J. A decade of capillary electrophoresis. Electrophoresis, 2000, 21(10), 1921–1939. 2. Mathies, R.A. and Huang, X.C. Capillary array electrophoresis — An approach to high-speed, high-throughput DNA sequencing. Nature, 1992, 359(6391), 167–169. 3. Huang, X.H.C., Quesada, M.A., and Mathies, R.A. DNA sequencing using capillary array electrophoresis. Analytical Chemistry, 1992, 64(18), 2149–2154. 4. Huang, X.H.C., Quesada, M.A., and Mathies, R.A. Capillary array electrophoresis using laser-excited confocal fluorescence detection. Analytical Chemistry, 1992, 64(8), 967–972. 5. Clark, S.M. and Mathies, R.A. High-speed parallel separation of DNA restriction fragments using capillary array electrophoresis. Analytical Biochemistry, 1993, 215(2), 163–170.

36

Advances in Chromatography 6. Harrison, D.J., Manz, A., Fan, Z.H., Ludi, H., and Widmer, H.M. Capillary electrophoresis and sample injection systems integrated on a planar glass chip. Analytical Chemistry, 1992, 64(17), 1926–1932. 7. Manz, A., Harrison, D.J., Verpoorte, E.M.J., Fettinger, J.C., Paulus, A., Ludi, H., and Widmer, H.M. Planar chips technology for miniaturization and integration of separation techniques into monitoring systems — Capillary electrophoresis on a chip. Journal of Chromatography, 1992, 593(1-2), 253–258. 8. Harrison, D.J., Glavina, P.G., and Manz, A. Towards miniaturized electrophoresis and chemical-analysis systems on silicon — An alternative to chemical sensors. Sensors and Actuators B-Chemical, 1993, 10(2), 107–116. 9. Effenhauser, C.S., Manz, A., and Widmer, H.M. Glass chips for high-speed capillary electrophoresis separations with submicrometer plate heights. Analytical Chemistry, 1993, 65(19), 2637–2642. 10. von Heeren, F., Verpoorte, E., Manz, A., and Thormann, W. Micellar electrokinetic chromatography separations and analyses of biological samples on a cyclic planar microstructure. Analytical Chemistry, 1996, 68(13), 2044–2053. 11. Ramseier, A., von Heeren, F., and Thormann, W. Analysis of fluorescein isothiocyanate derivatized amphetamine and analogs in human urine by capillary electrophoresis in chip-based and fused-silica capillary instrumentation. Electrophoresis, 1998, 19(16-17), 2967–2975. 12. Zhang, B.L., Foret, F., and Karger, B.L. A microdevice with integrated liquid junction for facile peptide and protein analysis by capillary electrophoresis/electrospray mass spectrometry. Analytical Chemistry, 2000, 72(5), 1015–1022. 13. Effenhauser, C.S., Manz, A., and Widmer, H.M. Manipulation of sample fractions on a capillary electrophoresis chip. Analytical Chemistry, 1995, 67(13), 2284–2287. 14. Liu, Y.J., Foote, R.S., Jacobson, S.C., Ramsey, R.S., and Ramsey, J.M. Electrophoretic separation of proteins on a microchip with noncovalent, postcolumn labeling. Analytical Chemistry, 2000, 72(19), 4608–4613. 15. Liu, Y.J., Foote, R.S., Culbertson, C.T., Jacobson, S.C., Ramsey, R.S., and Ramsey, J.M. Electrophoretic separation of proteins on microchips. Journal of Microcolumn Separations, 2000, 12(7), 407–411. 16. Shi, Y.N., Simpson, P.C., Scherer, J.R., Wexler, D., Skibola, C., Smith, M.T., and Mathies, R.A. Radial capillary array electrophoresis microplate and scanner for highperformance nucleic acid analysis. Analytical Chemistry, 1999, 71(23), 5354–5361. 17. Manz, A., Graber, N., and Widmer, H.M. Miniaturized total chemical-analysis systems — A novel concept for chemical sensing. Sensors and Actuators B-Chemical, 1990, 1(1-6), 244–248. 18. Xue, Q.F., Foret, F., Dunayevskiy, Y.M., Zavracky, P.M., McGruer, N.E., and Karger, B.L. Multichannel microchip electrospray mass spectrometry. Analytical Chemistry, 1997, 69(3), 426–430. 19. Figeys, D., Ning, Y. B., and Aebersold, R. A microfabricated device for rapid protein identification by microelectrospray ion trap mass spectrometry. Analytical Chemistry, 1997, 69(16), 3153–3160. 20. Figeys, D. and Aebersold, R. High sensitivity analysis of proteins and peptides by capillary electrophoresis tandem mass spectrometry: Recent developments in technology and applications. Electrophoresis, 1998, 19(6), 885–892. 21. Huang, Z.L., Jin, L.J., Sanders, J.C., Zheng, Y.B., Dunsmoor, C., Tian, H.J., and Landers, J.P. Laser-induced fluorescence detection on multichannel electrophoretic microchips using microprocessor-embedded acousto-optic laser beam scanning. IEEE Transactions on Biomedical Engineering, 2002, 49(8), 859–866.

Separations in Multiple-Channel Microchips

37

22. Xu, X.M., Roddy, T.P., Lapos, J.A., and Ewing, A.G. Parallel analysis with optically gated sample introduction on a multichannel microchip. Analytical Chemistry, 2002, 74(21), 5517–5522. 23. Xu, H.W., Roddy, E.S., Roddy, T.P., Lapos, J.A., and Ewing, A.G. Parallel separations of oligonucleotides with optically gated sample introduction on multi-channel microchips. Journal of Separation Science, 2004, 27(1-2), 7–12. 24. Woolley, A.T., Sensabaugh, G.F., and Mathies, R.A. High-speed DNA genotyping using microfabricated capillary array electrophoresis chips. Analytical Chemistry, 1997, 69(11), 2181–2186. 25. Huang, Z., Munro, N., Huhmer, A.F.R., and Landers, J.P. Acousto-optical deflectionbased laser beam scanning for fluorescence detection on multichannel electrophoretic microchips. Analytical Chemistry, 1999, 71(23), 5309–5314. 26. Gao, Q.F., Shi, Y.N., and Liu, S.R. Multiple-channel microchips for high-throughput DNA analysis by capillary electrophoresis. Fresenius Journal of Analytical Chemistry, 2001, 371(2), 137–145. 27. Harrison, D.J., Fluri, K., Seiler, K., Fan, Z.H., Effenhauser, C.S., and Manz, A. Micromachining a miniaturized capillary electrophoresis-based chemical-analysis system on a chip. Science, 1993, 261(5123), 895–897. 28. Jacobson, S.C., Hergenroder, R., Koutny, L.B., and Ramsey, J.M. High-speed separations on a microchip. Analytical Chemistry, 1994, 66(7), 1114–1118. 29. Manz, A., Verpoorte, E., Effenhauser, C.S., Burggraf, N., Raymond, D.E., Harrison, D.J., and Widmer, H.M. Miniaturization of separation techniques using planar chip technology. HRC: Journal of High Resolution Chromatography, 1993, 16(7), 433–436. 30. Simpson, P.C., Roach, D., Woolley, A.T., Thorsen, T., Johnston, R., Sensabaugh, G.F., and Mathies, R.A. High-throughput genetic analysis using microfabricated 96-sample capillary array electrophoresis microplates. Proceedings of the National Academy of Sciences of the United States of America, 1998, 95(5), 2256–2261. 31. Emrich, C.A., Tian, H.J., Medintz, I.L., and Mathies, R.A. Microfabricated 384-lane capillary array electrophoresis bioanalyzer for ultrahigh-throughput genetic analysis. Analytical Chemistry, 2002, 74(19), 5076–5083. 32. Becker, H. and Gartner, C. Polymer microfabrication methods for microfluidic analytical applications. Electrophoresis, 2000, 21(1), 12–26. 33. Rotting, O., Ropke, W., Becker, H., and Gartner, C. Polymer microfabrication technologies. Microsystem Technologies, 2002, 8(1), 32–36. 34. Duffy, D.C., McDonald, J.C., Schueller, O.J.A., and Whitesides, G.M. Rapid prototyping of microfluidic systems in poly(dimethylsiloxane). Analytical Chemistry, 1998, 70(23), 4974–4984. 35. Ford, S.M., Kar, B., McWhorter, S., Davies, J., Soper, S.A., Klopf, M., Calderon, G., and Saile, V. Microcapillary electrophoresis devices fabricated using polymeric substrates and x-ray lithography. Journal of Microcolumn Separations, 1998, 10(5), 413–422. 36. Effenhauser, C.S., Bruin, G.J.M., Paulus, A., and Ehrat, M. Integrated capillary electrophoresis on flexible silicone microdevices: Analysis of DNA restriction fragments and detection of single DNA molecules on microchips. Analytical Chemistry, 1997, 69(17), 3451–3457. 37. McDonald, J.C., Duffy, D.C., Anderson, J.R., Chiu, D.T., Wu, H.K., Schueller, O.J.A., and Whitesides, G.M. Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis, 2000, 21(1), 27–40.

38

Advances in Chromatography 38. Heckele, M., Bacher, W., and Muller, K.D. Hot embossing — The molding technique for plastic microstructures. Microsystem Technologies, 1998, 4(3), 122–124. 39. Roetting, O., Heckele, M., and Bacher, W. Reactive ion etching for the production of metal microstructures by hot embossing. Microsystem Technologies, 1999, 6(1), 11–14. 40. McCormick, R.M., Nelson, R.J., AlonsoAmigo, M.G., Benvegnu, J., and Hooper, H.H. Microchannel electrophoretic separations of DNA in injection-molded plastic substrates. Analytical Chemistry, 1997, 69(14), 2626–2630. 41. Hooper, H., McCormick, R., Alonso, G., Wang, H., and Benvegnu, D. Microchannel electrophoresis in injection molded substrates. Abstracts of Papers of the American Chemical Society, 1997, 213, PMSE-333. 42. Spesny, M. and Foret, F. Microdevice for separation and quantitative fraction collection. Electrophoresis, 2003, 24(21), 3745–3747. 43. Woolley, A.T. and Mathies, R.A. Ultra-high-speed DNA fragment separations using microfabricated capillary array electrophoresis chips. Proceedings of the National Academy of Sciences of the United States of America, 1994, 91(24), 11348–11352. 44. Woolley, A.T. and Mathies, R.A. Ultra-high-speed DNA-sequencing using capillary electrophoresis chips. Analytical Chemistry, 1995, 67(20), 3676–3680. 45. Liu, S.R., Ren, H.J., Gao, Q.F., Roach, D.J., Loder, R.T., Armstrong, T. M., Mao, Q.L., Blaga, I., Barker, D.L., and Jovanovich, S.B. Automated parallel DNA sequencing on multiple channel microchips. Proceedings of the National Academy of Sciences of the United States of America, 2000, 97(10), 5369–5374. 46. Medintz, I.L., Berti, L., Emrich, C.A., Tom, J., Scherer, J.R., and Mathies, R.A. Genotyping energy-transfer-cassette-labeled short-tandem-repeat amplicons with capillary array electrophoresis microchannel plates. Clinical Chemistry, 2001, 47(9), 1614–1621. 47. Medintz, I.L., Paegel, B.M., and Mathies, R.A. Microfabricated capillary array electrophoresis DNA analysis systems. Journal of Chromatography A, 2001, 924(1-2), 265–270. 48. Medintz, I., Wong, W.W., Berti, L., Shiow, L., Tom, J., Scherer, J., Sensabaugh, G., and Mathies, R.A. High-performance multiplex SNP analysis of three hemochromatosis-related mutations with capillary array electrophoresis microplates. Genome Research, 2001, 11(3), 413–421. 49. Paegel, B.M., Emrich, C.A., Weyemayer, G.J., Scherer, J.R. ,and Mathies, R.A. High throughput DNA sequencing with a microfabricated 96-lane capillary array electrophoresis bioprocessor. Proceedings of the National Academy of Sciences of the United States of America, 2002, 99(2), 574–579. 50. Bromberg, A. and Mathies, R.A. Multichannel homogeneous immunoassay for detection of 2,4,6-trinitrotoluene (TNT) using a microfabricated capillary array electrophoresis chip. Electrophoresis, 2004, 25(12), 1895–1900. 51. Culbertson, C.T., Jacobson, S.C., and Ramsey, J.M. Dispersion sources for compact geometries on microchips. Analytical Chemistry, 1998, 70(18), 3781–3789. 52. Paegel, B.M., Hutt, L.D., Simpson, P.C., and Mathies, R.A. Turn geometry for minimizing band broadening in microfabricated capillary electrophoresis channels. Analytical Chemistry, 2000, 72(14), 3030–3037. 53. Medintz, I.L., Paegel, B.M., Blazej, R.G., Emrich, C.A., Berti, L., Scherer, J.R., and Mathies, R.A. High-performance genetic analysis using microfabricated capillary array electrophoresis microplates. Electrophoresis, 2001, 22(18), 3845–3856.

Separations in Multiple-Channel Microchips

39

54. Martynova, L., Locascio, L.E., Gaitan, M., Kramer, G.W., Christensen, R.G., and MacCrehan, W.A. Fabrication of plastic microfluid channels by imprinting methods. Analytical Chemistry, 1997, 69(23), 4783–4789. 55. Wang, J., Chatrathi, M.P., Tian, B.M., and Polsky, R. Microfabricated electrophoresis chips for simultaneous bioassays of glucose, uric acid, ascorbic acid, and acetaminophen. Analytical Chemistry, 2000, 72(11), 2514–2518. 56. Hayes, M.A., Polson, N.A., Phayre, A.N., and Garcia, A.A. Flow-based microimmunoassay. Analytical Chemistry, 2001, 73(24), 5896–5902. 57. Guijt, R.M., Baltussen, E., and van Dedem, G.W.K. Use of bioaffinity interactions in electrokinetically controlled assays on microfabricated devices. Electrophoresis, 2002, 23(6), 823–835. 58. Guttman, A., Ronai, Z., Barta, C., Hou, Y.M., Sasvari-Szekely, M., Wang, X., and Briggs, S.P. Membrane-mediated ultrafast restriction digestion and subsequent rapid gel microchip electrophoresis of DNA. Electrophoresis, 2002, 23(10), 1524–1530. 59. Kricka, L.J. Prospects for microchips in immunoassays and immunoassay tests at the point-of-care. Journal of Clinical Ligand Assay, 2002, 25(4), 317–324. 60. Lim, T.K., Ohta, H., and Matsunaga, T. Microfabricated on-chip-type electrochemical flow immunoassay system for the detection of histamine released in whole blood samples. Analytical Chemistry, 2003, 75(14), 3316–3321. 61. Xu, H.W. and Ewing, A.G. A rapid enzyme assay for beta-galactosidase using optically gated sample introduction on a microfabricated chip. Analytical and Bioanalytical Chemistry, 2004, 378(7), 1710–1715. 62. Cheng, S.B., Skinner, C.D., Taylor, J., Attiya, S., Lee, W.E., Picelli, G., and Harrison, D.J. Development of a multichannel microfluidic analysis system employing affinity capillary electrophoresis for immunoassay. Analytical Chemistry, 2001, 73(7), 1472–1479. 63. Xue, Q.F., Wainright, A., Gangakhedkar, S., and Gibbons, I. Multiplexed enzyme assays in capillary electrophoretic single-use microfluidic devices. Electrophoresis, 2001, 22(18), 4000–4007. 64. Sato, K., Yamanaka, M., Takahashi, H., Tokeshi, M., Kimura, H., and Kitamori, T. Microchip-based immunoassay system with branching multichannels for simultaneous determination of interferon-gamma. Electrophoresis, 2002, 23(5), 734–739. 65. Jiang, X.Y., Ng, J.M.K., Stroock, A.D., Dertinger, S.K.W., and Whitesides, G.M. A miniaturized, parallel, serially diluted immunoassay for analyzing multiple antigens. Journal of the American Chemical Society, 2003, 125(18), 5294–5295. 66. Roddy, E.S., Price, M., and Ewing, A.G. Continuous monitoring of a restriction enzyme digest of DNA on a microchip with automated capillary sample introduction. Analytical Chemistry, 2003, 75(15), 3704–3711. 67. Roddy, E.S., Xu, H.W., and Ewing, A.G. Sample introduction techniques for microfabricated separation devices. Electrophoresis, 2004, 25(2), 229–242. 68. Smith, E.M., Xu, H.W., and Ewing, A.G. DNA separations in microfabricated devices with automated capillary sample introduction. Electrophoresis, 2001, 22(2), 363–370. 69. Paxon, T.L., Brown, T.S., Lin, H.N., Brancato, S.J., Roddy, E.S., Philip C. Bevilacquaa, P.C., and Ewing, A.G. Continuous monitoring of enzyme reactions on a microchip: Application to catalytic RNA self-cleavage. Analytical Chemistry, 2004, 76(23), 6921–6927.

40

Advances in Chromatography 70. Seiler, K., Harrison, D.J., and Manz, A. Planar glass chips for capillary electrophoresis — Repetitive sample injection, quantitation, and separation efficiency. Analytical Chemistry, 1993, 65(10), 1481–1488. 71. Beard, N.R., Zhang, C.X., and deMello, A.J. In-column field-amplified sample stacking of biogenic amines on microfabricated electrophoresis devices. Electrophoresis, 2003, 24(4), 732–739. 72. Zhang, C.X. and Manz, A. Narrow sample channel injectors for capillary electrophoresis on microchips. Analytical Chemistry, 2001, 73(11), 2656–2662. 73. Ferrance, J. and Landers, J.P. Exploiting sensitive laser-induced fluorescence detection on electrophoretic microchips for executing rapid clinical diagnostics. Luminescence, 2001, 16(2), 79–88. 74. Medintz, I., Wong, W.W., Sensabaugh, G., and Mathies, R.A. High speed single nucleotide polymorphism typing of a hereditary haemochromatosis mutation with capillary array electrophoresis microplates. Electrophoresis, 2000, 21(12), 2352–2358. 75. Effenhauser, C.S., Paulus, A., Manz, A., and Widmer, H.M. High-speed separation of antisense oligonucleotides on a micromachined capillary electrophoresis device. Analytical Chemistry, 1994, 66(18), 2949–2953. 76. Monnig, C.A., Dohmeier, D.M., and Jorgenson, J.W. Sample gating in open tubular and packed capillaries for high-speed liquid chromatography. Analytical Chemistry, 1991, 63(8), 807–810. 77. Monnig, C.A. and Jorgenson, J.W. On-column sample gating for high-speed capillary zone electrophoresis. Analytical Chemistry, 1991, 63(8), 802–807. 78. Moore, A.W. and Jorgenson, J.W. Study of zone broadening in optically gated highspeed capillary electrophoresis. Analytical Chemistry, 1993, 65(24), 3550–3560. 79. Lapos, J.A. and Ewing, A.G. Injection of fluorescently labeled analytes into microfabricated chips using optically gated electrophoresis. Analytical Chemistry, 2000, 72(19), 4598–4602. 80. Roddy, E.S., Lapos, J.A., and Ewing, A.G. Rapid serial analysis of multiple oligonucleotide samples on a microchip using optically-gated injection. Journal of Chromatography A, 2003, 1004(1-2), 217–224. 81. Uchiyama, K., Nakajima, H., and Hobo, T. Detection method for microchip separations. Analytical and Bioanalytical Chemistry, 2004, 379(3), 375–382. 82. Quesada, M.A., Rye, H.S., Gingrich, J.C., Glazer, A.N., and Mathies, R.A. Highsensitivity DNA detection with a laser-excited confocal fluorescence gel scanner. Biotechniques, 1991, 10(5), 616–622, 624–625. 83. Dolnik, V., Liu, S., and Jovanovich, S. Capillary electrophoresis on microchip. Electrophoresis, 2000, 21(1), 41–54. 84. Lee, T.T. and Yeung, E.S. Capillary electrophoresis detectors: Lasers. Methods in Enzymology, 1996, 270, 419–449. 85. Zhu, H.P., Clark, S.M., Benson, S.C., Rye, H.S., Glazer, A.N., and Mathies, R.A. High-sensitivity capillary electrophoresis of double-stranded DNA fragments using monomeric and dimeric fluorescent intercalating dyes. Analytical Chemistry, 1994, 66(13), 1941–1948. 86. Sassi, A.P., Paulus, A., Cruzado, I.D., Bjornson, T., and Hooper, H.H. Rapid, parallel separations of D1S80 alleles in a plastic microchannel chip. Journal of Chromatography A, 2000, 894(1-2), 203–217. 87. Kheterpal, I., Scherer, J.R., Clark, S.M., Radhakrishnan, A., Ju, J.Y., Ginther, C.L., Sensabaugh, G.F., and Mathies, R.A. DNA sequencing using a four-color confocal fluorescence capillary array scanner. Electrophoresis, 1996, 17(12), 1852–1859.

Separations in Multiple-Channel Microchips

41

88. Alexay, C., Kain, R., Hanzel, D., and Johnston, R. Fluorescence scanner employing a macro scanning objective. Proceedings of SPIE: The International Society for Optical Engineering, 1996, 2705 (Fluorescence Detection IV), 63–72. 89. Kambara, H. and Takahashi, S. Multiple-sheathflow capillary array DNA analyzer. Nature, 1993, 361(6412), 565–566. 90. Quesada, M.A. and Zhang, S.P. Multiple capillary DNA sequencer that uses fiberoptic illumination and detection. Electrophoresis, 1996, 17(12), 1841–1851. 91. Pang, H.M., Pavski, V., and Yeung, E.S. DNA sequencing using 96-capillary array electrophoresis. Journal of Biochemical and Biophysical Methods, 1999, 41(2-3), 121–132. 92. Hietpas, P.B., Bullard, K.M., Gutman, D.A., and Ewing, A.G. Ultrathin slab gel separations of DNA using a single capillary sample introduction system. Analytical Chemistry, 1997, 69(13), 2292–2298. 93. Ewing, A.G., Gavin, P.F., Hietpas, P.B., and Bullard, K.M. Continuous separations in microfabricated channels for monitoring ultrasmall biological environments. Nature Medicine, 1997, 3(1), 97–99. 94. Bullard, K.M., Hietpas, P.B., and Ewing, A.G. Capillary sample introduction of polymerase chain reaction (PCR) products separated in ultrathin slab gels. Electrophoresis, 1998, 19(1), 71–75. 95. Bullard, K.M., Gavin, P.F., and Ewing, A.G. Electrophoretic separation schemes in ultrathin channels with capillary sample introduction. Trac-Trends in Analytical Chemistry, 1998, 17(7), 401–410. 96. Hietpas, P.B., Bullard, K.M., and Ewing, A.G. Characterization of electrophoretic sample transfer from a capillary to an ultrathin slab gel. Journal of Microcolumn Separations, 1998, 10(6), 519–527. 97. Bibeau, D.R., Smith, K.B., Smith, E.M., and Ewing, A.G. Separations of DNA fragments with a coated 25-μm capillary coupled to a 25-μm high open channel. Journal of Microcolumn Separations, 1999, 11(8), 567–575. 98. Bruin, G.J.M. Recent developments in electrokinetically driven analysis on microfabricated devices. Electrophoresis, 2000, 21(18), 3931–3951. 99. Pingoud, A. and Jeltsch, A. Structure and function of type II restriction endonucleases. Nucleic Acids Research, 2001, 29(18), 3705–3727. 100. Han, J. and Craighead, H.G. Separation of long DNA molecules in a microfabricated entropic trap array. Science, 2000, 288(5468), 1026–1029. 101. Han, J.Y. and Craighead, H.G. Characterization and optimization of an entropic trap for DNA separation. Analytical Chemistry, 2002, 74(2), 394–401. 102. Guttman, A. Sample stacking during membrane-mediated loading in automated DNA sequencing. Analytical Chemistry, 1999, 71(16), 3598–3602. 103. Nemoda, Z., Ronai, Z., Szekely, A., Kovacs, E., Shandrick, S., Guttman, A., and Sasvari-Szekely, M. High-throughput genotyping of repeat polymorphism in the regulatory region of serotonin transporter gene by gel microchip electrophoresis. Electrophoresis, 2001, 22(18), 4008–4011. 104. Carrihlo, E. DNA sequencing by capillary array electrophoresis and microfabricated array systems. Electrophoresis, 2000, 21, 55–65. 105. Liu, S.R., Shi, Y.N., Ja, W.W., and Mathies, R.A. Optimization of high-speed DNA sequencing on microfabricated capillary electrophoresis channels. Analytical Chemistry, 1999, 71(3), 566–573.

42

Advances in Chromatography

106. Simpson, J.W., Ruiz-Martinez, M.C., Mulhern, G.T., Berka, J., Latimer, D.R., Ball, J.A., Rothberg, J.M., and Went, G.T. A transmission imaging spectrograph and microfabricated channel system for DNA analysis. Electrophoresis, 2000, 21(1), 135–149. 107. Backhouse, C., Caamano, M., Oaks, F., Nordman, E., Carrillo, A., Johnson, B., and Bay, S. DNA sequencing in a monolithic microchannel device. Electrophoresis, 2000, 21(1), 150–156. 108. Liu, S.R., Elkin, C., and Kapur, H. Sequencing of real-world samples using a microfabricated hybrid device having unconstrained straight separation channels. Electrophoresis, 2003, 24(21), 3762–3768. 109. Sato, K., Tokeshi, M., Odake, T., Kimura, H., Ooi, T., Nakao, M., and Kitamori, T. Integration of an immunosorbent assay system: Analysis of secretory human immunoglobulin A on polystyrene beads in a microchip. Analytical Chemistry, 2000, 72(6), 1144–1147. 110. Sato, K., Tokeshi, M., Kimura, H., and Kitamori, T. Determination of carcinoembryonic antigen in human sera by integrated bead bed immunoasay in a microchip for cancer diagnosis. Analytical Chemistry, 2001, 73(6), 1213–1218. 111. Colyer, C.L., Tang, T., Chiem, N., and Harrison, D.J. Clinical potential of microchip capillary electrophoresis systems. Electrophoresis, 1997, 18(10), 1733–1741. 112. Chiem, N.H. and Harrison, D.J. Monoclonal antibody binding affinity determined by microchip-based capillary electrophoresis. Electrophoresis, 1998, 19(16-17), 3040–3044. 113. Chiem, N.H. and Harrison, D.J. Microchip systems for immunoassay: An integrated immunoreactor with electrophoretic separation for serum theophylline determination. Clinical Chemistry, 1998, 44(3), 591–598. 114. Jiang, G.F., Attiya, S., Ocvirk, G., Lee, W.E., and Harrison, D.J. Red diode laser induced fluorescence detection with a confocal microscope on a microchip for capillary electrophoresis. Biosensors & Bioelectronics, 2000, 14(10-11), 861–869. 115. Qiu, C.X. and Harrison, D.J. Integrated self-calibration via electrokinetic solvent proportioning for microfluidic immunoassays. Electrophoresis, 2001, 22(18), 3949–3958. 116. Taylor, J., Picelli, G., and Harrison, D.J. An evaluation of the detection limits possible for competitive capillary electrophoretic immunoassays. Electrophoresis, 2001, 22(17), 3699–3708. 117. Bromberg, A. and Mathies, R.A. Homogeneous immunoassay for detection of TNT and its analogues on a microfabricated capillary electrophoresis chip. Analytical Chemistry, 2003, 75(5), 1188–1195. 118. Ertl, P., Emrich, C.A., Singhal, P., and Mathies, R.A. Capillary electrophoresis chips with a sheath-flow supported electrochemical detection system. Analytical Chemistry, 2004, 76(13), 3749–3755. 119. Woolley, A.T., Lao, K.Q., Glazer, A.N., and Mathies, R.A. Capillary electrophoresis chips with integrated electrochemical detection. Analytical Chemistry, 1998, 70(4), 684–688. 120. Martin, R.S., Gawron, A.J., Lunte, S.M., and Henry, C.S. Dual-electrode electrochemical detection for poly(dimethylsiloxane)-fabricated capillary electrophoresis microchips. Analytical Chemistry, 2000, 72(14), 3196–3202. 121. Lunte, S.M., Martin, R.S., Gawron, A., Lacher, N., Fogarty, B., and Reagan, F. Bioanalytical applications of microchip capillary electrophoresis with electrochemical detection. Abstracts of Papers of the American Chemical Society, 2001, 221, U102–U102.

Separations in Multiple-Channel Microchips

43

122. Lapos, J.A., Manica, D.P., and Ewing, A.G. Dual fluorescence and electrochemical detection on an electrophoresis microchip. Analytical Chemistry, 2002, 74(14), 3348–3353. 123. Fogarty, B.A., Lacher, N.A., Pasas, S., Vanadeever, W.R., Martin, R.S., Davies, M.I., Heppert, K,. and Lunte, S. Bioanalytical applications of microchip electrophoresis with electrochemical detection. Abstracts of Papers of the American Chemical Society, 2003, 225, U151–U151. 124. Osbourn, D.M. and Lunte, C.E. On-column electrochemical detection for microchip capillary electrophoresis. Analytical Chemistry, 2003, 75(11), 2710–2714. 125. Lacher, N.A., Lunte, S.M., and Martin, R.S. Development of a microfabricated palladium decoupler/electrochemical detector for microchip capillary electrophoresis using a hybrid glass/poly(dimethylsiloxane) device. Analytical Chemistry, 2004, 76(9), 2482–2491. 126. Ramsey, R.S. and Ramsey, J.M. Microchip devices for fluidic manipulation and esi mass spectrometry. Abstracts of Papers of the American Chemical Society, 1997, 214, ANYL-150. 127. Xue, Q.F., Dunayevskiy, Y.M., Foret, F., and Karger, B.L. Integrated multichannel microchip electrospray ionization mass spectrometry: Analysis of peptides from onchip tryptic digestion of melittin. Rapid Communications in Mass Spectrometry, 1997, 11(12), 1253–1256. 128. Meng, Z.J., Qi, S.Z., Soper, S.A., and Limbach, P.A. Interfacing a polymer-based micromachined device to a nanoelectrospray ionization Fourier transform ion cyclotron resonance mass spectrometer. Analytical Chemistry, 2001, 73(6), 1286–1291. 129. Lazar, I.M., Ramsey, R.S., Sundberg, S., and Ramsey, J.M. Subattomole-sensitivity microchip nanoelectrospray source with time-of-flight mass spectrometry detection. Analytical Chemistry, 1999, 71(17), 3627–3631. 130. Lazar, I.M., Ramsey, R.S., Jacobson, S.C., Foote, R.S., and Ramsey, J.M. Novel microfabricated device for electrokinetically induced pressure flow and electrospray ionization mass spectrometry. Journal of Chromatography A, 2000, 892(1-2), 195–201. 131. Lazar, I.M., Ramsey, R.S., and Ramsey, J.M. On-chip proteolytic digestion and analysis using “wrong-way-round” electrospray time-of-flight mass spectrometry. Analytical Chemistry, 2001, 73(8), 1733–1739. 132. Svedberg, M., Pettersson, A., Nilsson, S., Bergquist, J., Nyholm, L., Nikolajeff, F., and Markides, K. Sheathless electrospray from polymer microchips. Analytical Chemistry, 2003, 75(15), 3934–3940. 133. Ssenyange, S., Taylor, J., Harrison, D.J., and McDermott, M.T. A glassy carbon microfluidic device for electrospray mass spectrometry. Analytical Chemistry, 2004, 76(8), 2393–2397. 134. Yang, Y.N., Kameoka, J., Wachs, T., Henion, J.D., and Craighead, H.G. Quantitative mass spectrometric determination of methylphenidate concentration in urine using an electrospray ionization source integrated with a polymer microchip. Analytical Chemistry, 2004, 76(9), 2568–2574. 135. Sinclair, J., Pihl, J., Olofsson, J., Karlsson, M., Jardemark, K., Chiu, D.T., and Orwar, O. A cell-based bar code reader for high-throughput screening of ion channel-ligand interactions. Analytical Chemistry, 2002, 74(24), 6133–6138. 136. Olofsson, J., Karlsson, M., Pihl, J., Sinclair, J., and Orwar, O. Microfluidic chips for time-resolved high-throughput patch-clamp studies. Biophysical Journal, 2003, 84(2), 488a–488a.

44

Advances in Chromatography

137. Pihl, J., Sinclair, J., Olofsson, J., Sahlin, E., Karlsson, M,. and Orwar, O. A microfluidic tool for studies of multi-component concentration dependent ion channel effecters. Biophysical Journal, 2004, 86(1), 482a–482a. 138. Farre, C., Sinclair, J., Pihl, J., Olofsson, J., and Orwar, O. Microfluidic chip solutions for patch-clamp-based high-content-screening. Biophysical Journal, 2004, 86(1), 482a–482a. 139. Sinclair, J., Pihl, J., Olofsson, J., Farre, C., and Orwar, O. Microfluidic devices for high-content ion channel screening. Biophysical Journal, 2004, 86(1), 355a–355a.

2

Temperature Effects in Liquid Chromatography Elsa Lundanes University of Oslo, Norway

Tyge Greibrokk University of Oslo, Norway

CONTENTS Abstract ....................................................................................................................45 2.1 Introduction ....................................................................................................46 2.2 Temperature Effects on Retention and Selectivity ........................................46 2.3 Conventional Analytical LC...........................................................................49 2.3.1 Columns and Stationary Phases.........................................................49 2.3.2 Temperature Programming and Instrumentation ...............................51 2.3.3 Applications .......................................................................................52 2.4 Capillary LC...................................................................................................56 2.4.1 Columns and Stationary Phases.........................................................56 2.4.2 Temperature Programming and Instrumentation ...............................63 2.4.3 Applications .......................................................................................65 References................................................................................................................66

ABSTRACT Temperature has a large effect on retention, selectivity, and column efficiency and has long been accepted as an important parameter in liquid chromatography (LC). Despite this fact, temperature has not been very actively utilized in the past, mainly because of reported stability problems of the most commonly used stationary phases. However, more interest in the application of temperature for retention control has come of late because of the trend of miniaturization in chromatography and the availability of temperature-stable stationary phases. This work gives an overview of temperature effects on retention and selectivity in chromatography, especially on reversed-phase columns. Instrumental requirements, especially with respect to performing temperature gradient elution, are discussed and applications on both conventional-sized analytical columns and capillary columns are included.

45

46

Advances in Chromatography

2.1 INTRODUCTION Even though temperature influences parameters like solubility, diffusivity, mobile phase polarity, and viscosity, temperature has not been very actively used in liquid chromatography and most analyses are carried out at ambient temperature. However, in recent years there has been an increasing awareness of the potential of temperature for achieving better chromatographic resolution by either improvement in selectivity or chromatographic efficiency. At high temperatures, the viscosity is reduced and the diffusion rate is increased; thus, the mass transfer rate between the stationary and mobile phase is increased as well as the efficiency [1, 2] and it is possible to decrease the analysis time substantially [3, 4]. The temperature effect upon retention was extensively studied in reversed-phase systems as can be seen in review articles [2, 5–7] and some recent publications [4, 8–15]. However, temperature has also been actively used for retention control in normal-phase [16, 17], ion exchange [18, 19], and hydrophobic interaction chromatography [20], as well as on temperature-responsive stationary phases [21–24]. Despite the advantages of using high temperatures, the stability of the stationary phases commonly used in liquid chromatography (LC) today has hampered the development of temperature-optimized analytical methods. This review will elaborate upon the influence of temperature upon retention and selectivity as well as methods using temperature actively to optimize the separations both in conventional sized columns and smaller-diameter columns. This review will not include temperature effects in electrochromatography but information on this topic can be found elsewhere [6, 25]. Also, note that temperature regulation of retention may have been used in papers not included in this review, since the focus is on studies where temperature has been actively and explicitly used.

2.2 TEMPERATURE EFFECTS ON RETENTION AND SELECTIVITY Because reversed-phase chromatography is the separation mode used in about 80 to 90% of LC methods, the majority of studies involving temperature effects have been on reversed-phase stationary phases. The relationship between the retention factor, k, and the column temperature is well known [26, 27] and can be described by the equation: ln k = – ΔH/RT + ΔS/R + ln Φ

(2.1)

where R is the universal gas constant, T is the temperature, ΔH and ΔS are the enthalpy and entropy of a solute transfer from the mobile phase to the stationary phase, respectively, and Φ is the phase ratio of the column (the stationary to mobile phase volume ratio). The linear relationship between ln k and 1/T found in most studies with reversed-phase chromatography is also found in normal-phase as well as in cation-exchange chromatography [28]. The retention behavior in reversed-phase LC has been discussed on the basis of enthalpy−entropy compensation (EEC) between ΔH and ΔS (see [29] and references

Temperature Effects in Liquid Chromatography

47

3.5 3

In k or α

2.5 2

Benzene Toluene Selectivity

1.5 1 R2 = 0.989

0.5 0 0.0028

0.0030

84

60

0.0032 0.0034 1/T (1/K) 40

21

0.0036

0.0038

5

−10

T(ºC)

FIGURE 2.1 Van’t Hoff plots of benzene, toluene, and their difference, which is the methylene selectivity. The phase−ratio-dependent plots of benzene and toluene are non-linear while the phase-ratio-independent methylene selectivity plot is linear. Again, this suggests a phaseratio dependence in the van’t Hoff plots of benzene and possibly toluene. Experimental conditions: water-methanol (60:40) mobile phase, ODS stationary phase, at a flow-rate of 1.00 mL/min. From [41].

therein). ΔH and ΔS associated with the retention mechanism can be estimated by using the van’t Hoff equation (Equation 2.1). In reversed-phase chromatography, the value of ΔH is typically 7 to 15 kJ/mol for small solutes [2, 29–38], while larger molecules like polycyclic aromatic hydrocarbons (PAH) [28] usually have larger (negative) enthalpies. A common intersection point has been found for the ln k vs. 1/T lines of homologues in reversed-phase systems [39]. However, because of changes in the molecular conformation, plots of ln k vs. 1/T (van’t Hoff plot) often appear nonlinear for biopolymers like proteins [40]. However, Chester and Coym presented mathematical evidence and showed that phase ratio changes can also cause deviation from linearity for smaller molecules in van’t Hoff plots in the temperature range –5 to 80˚C [41] (see Figure 2.1). A wider temperature range was explored by Guillarme et al. [4], who studied the effect of temperature on the solute behavior on reversed-phase stationary phases that are thermally stable in the temperature range from 30 to 200˚C. All van’t Hoff plots were linear in the narrow temperature range 25 to 80˚C on all the types of stationary phases; however, within a larger range of temperature, a quadratic model was required. Nonlinear van’t Hoff plots were observed in the case of acetonitrile−water mixtures or water as mobile phase, while the plots of log k vs. 1/T for the Hypercarb column (20 to 180˚C) was linear whatever the organic solvent (methanol or acetonitrile) used. Jinno et al. [42] studied the effect of temperature in reversed-phase chromatography down to 50˚C and found that the retention mechanism in the lower temperature range is similar to that of the

48

Advances in Chromatography

normal temperature range when the water content of the mobile phase exceeds 10%. Mobile phases of pure water at superheated temperatures, between 100 and 200˚C, have also been explored and significant differences in the enthalpy of the retention process at high temperatures (–30 to –40 kJ/mol) were found, as compared to that at low temperatures of 15 to 55˚C (–6 to –13 kJ/mol) [43]. In a recent paper, the effect of temperature on the retention of ionizable compounds was addressed [44] and the authors pointed out that differences in ionization enthalpies of the buffers resulted in different eluent pH when the temperature was raised. Li [45] also found that buffers can play a very important role in affecting retention and selectivity with changes in temperature. The retention times of some bases increased with increasing temperature in the investigated range up to 70˚C using a mobile phase pH of about 7.8 [46]. The peak shapes also significantly improved, especially for weak bases (pKa ≈ 8). The increase in retention and improvement in efficiency were explained by the reduction in the pKa of bases with temperature, leading to reduction in protonation. Quaternary ammonium compounds showed normal behavior, i.e., reduced retention with increasing column temperature. The relationship in Equation 2.1 shows that for the same change in temperature, the retention factor of solutes with large ΔH is more strongly affected than that of solutes with small ΔH, providing different selectivity at different temperatures. Thus, a change in temperature can give a separation of solutes not obtained at the initial temperature explored. The effect of temperature upon retention is, however, rather small compared to the effect provided by an increase in mobile phase strength. A 4˚C increase in temperature was found to correspond to a 1% methanol increase [47], while another study showed that a 5˚C change in column temperature had the same effect on retention as a 1% change in acetonitrile concentration [48]. Similar values for neutral solutes were later found in another study using reversed-phase columns [49]. This result means that the elution strength window of temperature is much less than that of an organic modifier in a reversed-phase system and temperature programs cannot generally replace solvent gradients. In a series of papers, Snyder et al. [50–53] optimized reversed-phase separations of different solutes by changing the temperature or the solvent gradient to improve the selectivity, while Dolan et al. varied the temperature and solvent gradient steepness for maximizing the resolution using computer simulations [54]. These researchers showed that simultaneous changes in temperature and either gradient time (tG) or solvent strength (% B) is effective for control of selectivity in reversed-phase LC. When a solvent gradient is undesirable, isocratic separation can be optimized by varying the temperature and the mobile phase strength [55]. Neue and Mazzeo performed a theoretical study on optimization of solvent gradients at elevated temperatures and showed that the reduction in viscosity and improvement in diffusivity — and hence the mass transfer — made it possible to increase the speed of the separation [56]. The temperature dependence on retention using different mobile phases to assess the validity of hydrophobic theory as the driving force for retention has been addressed [57], as well as the influence of the temperature (23 to 73˚C) on the equilibrium isotherm [58].

Temperature Effects in Liquid Chromatography

49

2.3 CONVENTIONAL ANALYTICAL LC 2.3.1 COLUMNS

AND

STATIONARY PHASES

In almost all the papers mentioned above, silica-based reversed-phase columns have been used, mostly C18 columns. The mechanism of separation using C1, C4, C8, and C18 has been shown to be the same in the temperature range 0 to 77˚C [59] and changes in relative retention with temperature were approximately constant for different values of the gradient time tG for different C18 and C18 reversed-phase columns [60]. Raman spectroscopy has been used to study the temperature-induced surface changes in high-density C18 stationary phases [61]. Conformations and interchain coupling of the alkyl components of these phases were found to be sensitive to temperature and surface coverage, and each of the high-density C18 stationary phases exhibited subtle temperature-dependent phase changes. Linear van’t Hoff plots were produced for various monomeric and polymeric C18 phases over temperature ranges of about 30 K using water-containing mobile phases [27, 35, 62–64]. ΔS˚ values were provided only in a few studies [63, 64]; however, in the temperature range –5 to 80˚C, the entropic contribution to retention becomes more significant with respect to the enthalpic contribution when the stationary phase bonding density is increased for monomeric C18 phases [65] and the stationary phase structure gives an important contribution to selectivity, particularly at subambient temperatures [66]. In an early study, Chmielowiec and Sawatzky also showed that entropy effects can be large for molecules like polycyclic aromatic hydrocarbons on these phases [67]. The influence of temperature upon separation on C30 phases has been addressed in several studies [68, 69] and Albert et al. [68, 70] found that the strongly temperature-dependent separation characteristics of polymeric C30 phases correlated with the alkyl chain arrangement at the silica surface. By FTIR studies, titania-based self-assembled monolayer C30 phases were found to possess a higher degree of conformational order than zirconia- and silica-based C30 [71]. The stability of the commonly used silica-based reversed-phase materials has been a limiting factor for exploiting temperature for retention and selectivity control and the benefits of using higher temperatures for separation have not been achieved. Silica-based materials have been used up to 100˚C [72, 73] as can be seen in Figure 2.2, even though temperatures above 60˚C are usually not recommended [4, 15]. In addition to the temperature, the mobile phase composition, including the type of buffer used and the pH, influences the column lifetime of silica-based phases. In a recent paper, Claessens and van Straten [74] reviewed the chemical and thermal stability of reversed-phase stationary phases. Different silica-based, zirconia-based, and polymer columns are compared and they concluded that for many separation needs, specific silica reversed phases offered sufficient thermal stability from subambient and up to 90˚C. However, they also pointed out that reversed-phase stationary phases based on other inorganic substances and polymers may be needed for methods requiring higher thermal stability. In the recent study by Guillarme et al. [4], silica-based columns were used only in the temperature range 20 to 100˚C due to thermal instability, while the porous carbon, PS-DVB, and zirconia-based columns

50

Advances in Chromatography

mAU 80 70

(a)

60 50 40 30 20 10 0

0

5

10

15

20

25

30

min

10

15

20

25

30

min

10

15

20

25

30

min

mAU 80 70

(b)

60 50 40 30 20 10 0 0

5

mAU 80 70

(c)

60 50 40 30 20 10 0 0

5

FIGURE 2.2 Separation of polymers of average mass (A) 30,000, (B) 15,000, and (C) 8,000 on a non-porous Hytach C18 column operated at 100˚C with a gradient of 0 to 100% B in 60 min, where A is 400 mM NaClO4 and solvent B is 400 mM NaClO4 in acetonitrile-water (20:80). From [73].

could be used up to 180˚C. That the carbon columns and polymeric columns display better thermal stability has also been reported by others [4, 15, 75] and the influence of temperature on retention and selectivity on the polymeric columns has also been studied using the solvation parameter model [76]. An alumina-based C18 material was used up to 150˚C [77] but these materials have not been widely used. As mentioned above, zirconia-based stationary phases have been reported to be stable at high temperatures (see also [3, 78, 79]). The effect of temperature on the thermodynamic and dynamic properties of the zirconia-based stationary phases were investigated by Carr et al. [80] who found that, depending on the solute, the temperature can be a very important variable to also improve the selectivity on these

Temperature Effects in Liquid Chromatography

51

3

400

(b) 4 300

2 1

200

5

Absorbance (mAU)

100

0 500

0.0

0.2

3

0.4

0.6

0.8

1.0 (a)

400

4

2 1

300 200

5 100 0 0

3

6

9

12

15

Time (min)

FIGURE 2.3 Chromatograms of a reversed-phase mixture on PBD-coated zirconia. Plot A is the chromatogram at 30˚C and 1 mL/min and plot B is the chromatogram at 100˚C and 5 mL/min. Solutes: 1, uracil; 2, p-nitroaniline; 3, methyl benzoate; 4, phenetole; and 5, toluene. From [3].

phases. By using high temperatures, very fast separations using high flow rates are possible [3]. An 18-fold improvement relative to that at lower temperatures and normal flow rates was reported on these phases (see Figure 2.3).

2.3.2 TEMPERATURE PROGRAMMING

AND INSTRUMENTATION

A majority of the studies on the influence of temperature in analytical-sized columns has been performed using conventional equipment and isothermal conditions. The column thermostatting devices used vary widely in their design, and the actual temperature in the column may be quite different when two different LC systems are used [81, 82]. In most cases, mobile phase preheating is performed by either placing a preheating tube into a column liquid bath or an air circulation oven. A water or oil bath is more efficient than the air bath [83]; however, a recent study

52

Advances in Chromatography

concludes that it is quite possible to use an air system with analytical sized columns up to 150˚C and with columns with I.D. of 1 mm or less up to 200˚C [4]. Spearman et al. [82] monitored the effective column temperature by exploiting the shape selectivity and hydrophobicity of selected solutes, and suggested a method to distinguish between temperature and mobile phase composition differences that cause nonreproducibility in transferring methods between instruments and laboratories. They also found a marked increase in efficiency as the mobile phase temperature was reduced below the column temperature, in contrast to claims that thermal equilibration is essential [81, 83, 84]. The use of temperature gradients for retention control and selectivity optimization has not been widely used in analytical LC even though a temperature-programmed separation on alumina was published in 1966 [85], as well as a temperature-programmed (15 to 50˚C) separation of PAHs on a 4.6 mm I.D. C18 column in 1979 [67]. One of the reasons for the near-absence of temperature programming in analytical sized systems is that this methodology has been associated with radial temperature gradients and subsequent band broadening [81]. Poppe and Kraak [83] showed that the difference in temperature between the incoming mobile phase and the column was detrimental to the efficiency, a problem confirmed by others [81, 84]. Wolcott et al. suggested that the temperature difference between the entering mobile phase and the column oven should be within ±6˚C [81], and Djordjevic et al. [86] showed that analytical sized columns can be used with temperature programming if a mobile phase preheater is installed. In a recent study, Molander et al. [87] demonstrated that the column I.D. is not a critical limiting factor when performing temperature programming in LC, at least for columns narrower than 4.6 mm I.D. in the temperature range 30 to 90˚C. However, the relative apparent efficiency was inversely proportional to the column I.D., making the capillary columns generally more attractive for use with temperature gradients with respect to efficiency, as will be elaborated upon later in this review. Others have also found that with narrow-bore columns, the thermal mismatch effects are reduced significantly and can be neglected due to the fast heat transfer [47, 84]. However, despite these reports, temperature gradient elution was used in a few studies using microbore [88, 89] and conventional sized columns [11, 15, 90]. A thermal gradient technique that combined flow and temperature programming was suggested by Moore and Synovec [91] for microbore LC.

2.3.3 APPLICATIONS Even though most published studies utilizing temperature in LC have been of more or less fundamental character, temperature has also been actively used in conventional LC to solve separation needs, the goal of most analysts. In the majority of studies, C18 or C8 reversed-phase materials were used despite the fact that the largest influence of temperature has been found for long alkyl chain length stationary phases [7]. By optimizing the temperature, isothermal solvent gradient separation of 11 carotenoid isomers was obtained on a C30 column in the temperature range 15 to 25˚C [9]. The use of C30 stationary phases for determination of carotenoids, retinoids, and other related compounds in food has been reviewed [92].

Temperature Effects in Liquid Chromatography

53

Snyder et al. actively used temperature to optimize the separation of tryptic digests as well as other peptide and protein samples on C18 columns in the temperature range 20 to 60˚C [93, 94]. In a series of papers, Hodges et al. also studied the temperature selectivity effects of peptides in reversed-phase LC [13, 95–97] using both silica-based C8, CN [13, 95, 97] and PV-DVB [96] columns and a temperature range (isothermal) from 5 to 80˚C. They argued that variations in temperature can be used to affect significant changes in selectivity among peptide analogues and that a temperature-based approach to reversed-phase LC can be used to distinguish varying amino acid substituents at the same site of the peptide sequence. Lee et al. [12] studied the effect of temperature on the retention behavior of 14-residue cyclic and linear cationic peptides on a C18 column in the range 5 to 80˚C as a method to detect and measure self-association of small molecules. Temperature-controlled (35 to 60˚C) reversed-phase LC has also been used to obtain separation of phenylthiohydantoin-amino acids [98]. In this case, a temperature step program was used simultaneously with a solvent gradient. In a study on the effect of temperature (20 to 60˚C) and flow rate on the chromatography of basic compounds on a C18 column [8], the efficiency was improved at higher temperatures. However, the retention of bases often showed an anomalous increase with temperature, and the complex retention mechanisms were also dependent upon the buffer cation concentration. Guiochon et al. [10] studied the effect of temperature and pressure on protein adsorption equilibrium in reversed-phase C8 columns and found that the retention time increased substantially (up to 300%) when the pressure or the temperature increased in the pressure range 47 to 147 bar, and temperature range 25 to 50˚C. The relation between temperature and ln k is nonlinear with a parabolic shape. Dolfinger and Locke [99] studied the effect of temperature and eluent composition on the selectivity on different fluorinated and hydrocarbon reversed-phase columns for separation of 15 taxanes and found an increase of k for the xylosyl taxanes with increasing temperature in the range 25 to 55˚C on the C8 column, while the other taxanes behaved more normally. The same pattern was observed on one of the fluorinated phases. The separation of DNA fragments (21 to 587 bp) on nonporous C18 alkylated PS-DVB and alkylated wide-pore silica columns was studied in the temperature range 25 to 60˚C, and the optimum temperature was found between 40 and 50˚C using a solvent gradient [100]. Increasing the temperature changes the solvent polarity [101], and superheated water [102] and superheated deuterium oxide [11] have also been used as mobile phases in reversed-phase chromatography. Superheated water (170 to 200˚C) was used for the separation of steroids on a polymer-coated zirconia column and a complementary retention and retention mechanism compared to that on a traditional silica-based C18 and acetonitrile-water mobile phase at 40˚C was found; however, unique selectivity was obtained. Nuclear magnetic resonance (NMR) analysis of a ginger extract was obtained, on-line and off-line, by temperature programmed elution (50 to 130˚C) on a silica-based C18 column using superheated deuterium oxide (see Figure 2.4). The authors reported the efficiency was much better on the silica-based reversed-phase column than on a PBD-zirconia phase using superheated deuterium oxide as mobile phase. Dimethyl sulfoxide-modified subcritical water has been used for the separation of polar and nonpolar analytes such as polyhydroxybenzenes,

2.70:2

15.0

3.10:1

30.0

45.0

60.0

mU

Advances in Chromatography

72.5

54

min 47.0

40.0

35.0

30.0

25.0

20.0

15.0

10.0

5.0

−1.7 0.0

Zingerone

FIGURE 2.4 Separation of a 30-μL ginger extract on Xterra RP18 column using superheated deuterium oxide as mobile phase. Conditions: temperature 50 to 130˚C at 4˚C/min; eluent, deuterium oxide at 0.5 mL/min; detection, 280 nm. From [11].

phenol derivatives, alkylated benzenes, and PAHs using a silica-based C18 column, which was found to be stable in the temperature range 25 to 125˚C [103]. Methanol− water mixtures at elevated temperatures up to 140˚C have been used for the separation of nonpolar analytes like PAHs and polychlorinated biphenyls (PCB) on a silica-based C18 column [104]. By using ion-pair reversed-phase chromatography on a C8 column and optimizing the temperature in the range 35 to 55˚C, an almost complete separation of 23 trinitrotoluene metabolites and EPA Method 8338-compounds was achieved [14]. Increased temperature resulted in shorter retention time and better resolution, especially for the amino aromatics. The influence of temperature on the separation of steroids on a C18 column with a mobile phase modified by β-cyclodextrin was studied in the temperature range 5 to 80˚C [105] and the plot of ln k vs. 1/T was nonlinear in the investigated range. Ion-pair chromatography of aminoalcohol enantiomers on a carbon column using a mobile phase with N-derivatized peptides as the chiral selector was studied in the temperature range from –5 to 25˚C [106]. The

Temperature Effects in Liquid Chromatography

55

plots of ln k vs. 1/T were almost linear and the best separation was obtained at the lower temperatures. The effect upon retention in hydrophobic interaction chromatography (HIC) has also been studied, and nonlinear van’t Hoff plots are found for aromatic alcohol homologues on a silica-based polyethylene glycol column in the temperature range 0 to 80˚C [20]. The effect of temperature on the separation of polar compounds like urea, sucrose, and glycine on a diol column used in HIC has also been studied and ln k was found to decrease linearly with increasing temperature in the range 30 to 60˚C [107]. The influence of temperature in ion chromatography has not been widely examined. However, a study on the separation of aluminum and its fluoro, oxalate, and citrate species, where the temperature was varied in the range –10 to 55˚C on a cation exchanger, revealed that the distribution of the species varied with temperature and that temperatures below 10˚C could be used to suppress disintegration of the fluoride and oxalate ligands [18]. The retention and resolution of alkali metal on chelating stationary phases increased with decreasing temperature in the investigated range 1 to 75˚C and the dependence of ln K on 1/T could be approximated by second-order polynomials [108]. The temperature-responsive phases are made with the intention that the temperature should influence the retention. Hosoya et al. [109] developed a poly(N-isopropylacrylamide) polymer-based stationary phase that provided temperature-dependent separation selectivity in drug separation in the temperature range 30 to 50˚C. Separations of peptides and proteins were also explored on an N-isopropylacrylamide polymer-modified column in the temperature range 5 to 10˚C [21]. An N-isopropylacrylamide copolymer-modified silica column was developed and applied for the separation of five steroids having various hydrophobicities [110]. The retention increased with increasing temperature (5 to 50˚C); however, the van’t Hoff plots were not linear. The retention of some proteins was also studied at temperatures of 5 and 30˚C, the latter giving the highest retention. The poly(N-isopropylacrylamide)bonded stationary phase showed an increase in retention and column efficiency with an increase in temperature between 30 and 40˚C using benzene derivatives and alkyl phenyl ketones as model substances [22]. The temperature-responsive properties were reduced with increasing methanol content of the mobile phase and lost at 40% or higher. The temperature-responsive poly(N-isopropylacrylamide)-modified silica was also examined for its suitability to provide separation of steroids, alkaloids and substituted anilines [23]. In this case, the temperature-responsive phase showed changes in surface properties with an increase in temperature (5 to 55˚C) with a methanol content up to 20%. The temperature-responsive chiral polymer column composed of L-valine diamide derivatives provided enantiomeric separation of amino acid derivatives in the temperature range 0 to 70˚C in aqueous mobile phases (containing 1% methanol) [24]. The retention increased with increasing temperature and the enantioselectivity was also enhanced with temperature up to a critical temperature. The effect of temperature on the enantioseparation of 71 chiral compounds was studied on four macrocyclic glycopeptide chiral selectors — teicoplanin, its aglycone, ristocetin A and vancomycin — using both reversed-phase, normal-phase, and polar ionic mode

56

Advances in Chromatography

(PIM) mobile phases [111]. In the majority of cases, the enantioseparation decreased or even vanished when the temperature was increased. All van’t Hoff plots were linear, showing that the selector did not change in the investigated temperature range (5 to 45˚C). The retention of enantiomers on silica-based tris(1,3-dimethylphenylcarbamate)-derivatized cellulose was studied in the temperature range 12 to 40˚C in the reversed-phase mode using various mobile phase compositions [112]. The van’t Hoff plots were linear, the retention increased with decreasing temperature and the thermodynamic data revealed that the enantio separation was enthalpy-controlled. A human serum albumin (HSA) chiral stationary phase has also been explored for its potential to give separation of dansyl amino acids in the temperature range 3 to 27˚C [113]. Different van’t Hoff plot shapes were observed with different mobile phase pH, indicating a change in retention mechanism.

2.4 CAPILLARY LC The main advantage of narrow I.D. columns is the possibility to achieve lower detection limits when using concentration sensitive detectors, and this is especially advantageous when the sample amount is limited. The first attempts to study the effects of temperature with narrow-bore columns was in the 1980s [114–118] but it is the advent of packed capillary columns that has spurred the interest in using temperature as a parameter in liquid chromatography. In the following sections, the recent achievements in capillary LC utilizing the temperature for retention control and separation optimization are presented.

2.4.1 COLUMNS

AND

STATIONARY PHASES

The low thermal mass of the capillary columns results in a quick response to temperature changes, making rapid temperature programs possible. Thus, capillary columns with temperature gradients have recently been demonstrated in several studies (see Table 2.1). Most of the recent studies on temperature in capillary chromatography have been carried out using packed columns; however, both monolithic columns [119–121] and open capillary columns [122–126] have been utilized. So far, open tubular columns have not gained much interest because of either the requirement of a narrow I.D. or of very high temperatures. Monolithic polystyrene/divinylbenzene (PS-DVB) capillary columns have been used for separation of nucleic acids, proteins, and tryptic digests. Because of the polymerization-related variation in surface area of these monolithic columns, temperature regulation has been applied to optimize the resolution and obtain harmonized elution profiles for columns in parallel [119, 120] (see Figure 2.5). The inner diameter of these columns were 0.2 mm or less and they had a relatively short length of about 6 cm. Generally, temperatures above 40˚C are used isothermally [119, 120, 127–131] but temperaturegradient programming has also been used [121]. A methacrylate monolithic column intended for electrochromatography has also been used in the LC mode and showed a decrease in retention with increasing temperature (12 to 50˚C) [132]. The most common inner diameters of the packed capillary columns are 0.25 and 0.32 mm, but 0.50 mm I.D. has also been used (see Table 2.1). The column body

ACN/phosphate buffer (pH 7) 25–90 (I) MeOH/0.8% H3PO4 (56/44) 20–65 (G)

30–120 (G) 68–55 (inv. G) 30–80 (I)

0.05 × 150 mm nonporous PB-zirconia 0.30 × 150 mm Hypersil C18 BDS (3)

30–90 (G) 50–100 (G+I)

0.32+ × 150 mm Hypersil ODS (3) 0.50 × 100 mm PBD-ZrO2 (3)

0.25 × 200 mm Kromasil C18 (3.5) 0.50 × 100 mm Kromasil C18 (3.5) 0.20 × 100 mm Hypercarb (5)

15–70 (I) 30–150 (G+I) UV ELSD + ESITOFMS UV UV

ELSD

UV

UV and MS

Detector

Lab-made cLC-oven

UV UV

“Tailor-made” UV cLC-oven ELSD Col. oven UV

GC-oven GC-oven

cLC-oven GC-oven

30–120 (G)

ACN/water ACN/20 mM phosphate pH 7 (50/50) ACN ACN/THF/water (40/5/55) ACN/water; pH 2.2 and 12

GC-oven

50–120 (I+G)

22–70 (I)

Heating System

0.18 × 60 mm PLRP-S 300 PS-PVB (5) ACN/water/TFA 0.18 × 60 mm monolithic PS-PVB 0.32 × 350 mm Hypersil ODS (3) Ethyl acetate/ACN/TEA/AA (45.0/44.5/10.0/0.1) 0.32 × 100 mm PS-PVB SAX (10) pH-gradient of amine buffers 0.32 × 600 mm Kromasil C18 (3.5) Acetone/ACN

ACN/water/additive

Mobile Phase(s)

Temperature Range (°°C) (G = Gradient, I = Isothermal) Col. oven or circulating water bath GC-oven

0.20 × 60 mm monolithic PS-PVB

Dimension and Packing Particle Diameter (µm)

TABLE 2.1 Temperature Studies on Capillary Columns

Split 20

500a 100b 0.06

0.05 0.05

100 0.05–0.5

0.5

0.05

0.5

Inj. Vol. (µL)

Ref.

Polymer additives PEG 4000-3500 Sulfated glycosaminoglycan disaccharides Test compounds Chlorophenoxy acid herbicides

Hindered amine light stabilizers Proteins Polyglycerol fatty acid esters Alkylbenzenes Test compounds

Proteins

[140] [156]

[138] [155] [144]

[87] [139]

[19] [146]

[153]

[121]

Peptides, proteins [162]

Analyte(s)

Temperature Effects in Liquid Chromatography 57

0.32 0.32 0.32 0.32

× × × ×

400 400 300 250

mm mm mm mm

Kromasil C18 (3.5) Hypersil ODS (3) Hypersil ODS (3) Hypersil ODS (3)

0.30 × 150 mm Spherisorb ODS2 (3)

Temperature Range (°°C) (G = Gradient, I = Isothermal)

ACN/ammonium acetate pH 7 (69/31) ACN/DCM, THF/water, ACN/water ACN/water (30/70) MeOH/citric acid pH 3.1 (5/95)

Col. oven cLC-oven

51–57 (I) 10–90 (I+G)

80–30 (inv. G) 30–100 (G)

8–75 (G)

cLC-oven, GC-oven cLC-oven GC-oven

cLC-oven

GC-oven

80–95 (G)

15–70 (G)

GC-oven

cLC-oven GC-oven

Col. oven

Heating System

110–170 (G)

54–57 (I) ACN/(TEA+EDTA; pH 7) gradient 5–90 (G) ACN added TEA+FA ACN/ammonium formate pH 35–100 (G) 4 (25/75)

Mobile Phase(s)

0.32 × 1000 mm 4000Å Nucleosil silica TCB/isododecane/BuOH (5) 0.32 × 250 mm Hypersil ODS (3) ACN/ammonium formate pH 6 (40/60) 0.20 × 60 mm monolithic PS-DVB × 4 ACN/TEEA pH 7 0.32 × 200 mm C18 columnsc (3) MeOH and ACN/water pH 2–7

0.25 × 250 mm Kromasil C18 (3.5) 0.32 × 300 mm Kromasil C18 (3.5)

0.20 × 60 mm monolithic PS-DVB

Dimension and Packing Particle Diameter (µm)

TABLE 2.1 (CONTINUED) Temperature Studies on Capillary Columns

ELSD Electrochemical

UV

UV

LIF UV

UV

ELSD

UV UV

UV

Detector

0.55 0.06

0.2

5

1 0.05

1000b

0.06

500b at 5˚C 10

1

Inj. Vol. (µL)

PEG 1000 Tyrosines, DOPA

Diisocyanate derivatives DNA fragments Acidic, basic, neutral compounds Triazine herbicides Polystyrenes

Nucleic acids, proteins Irganox 1076 Citalopran, fluoxetine, paroxetine + metabolites Polymer blend

Analyte(s)

[163] [151]

[149]

[158]

[119] [49]

[135]

[17]

[136] [157]

[120]

Ref.

58 Advances in Chromatography

cLC-oven GC-oven GC-oven GC-oven

GC-oven

10–70 (G) 30–150 (G) 25–70 (I) 40–150 (I+G)

50–150 (G)

ACN/dichloromethane (70/30) 0.32 × 400–700 mm Hypersil ODS (3/5) ACN/DMF and other nonaqueous solvents 0.32 × 250 mm Suplex pKb-100 (5) ACN/5% ammonium acetate/AA (94.9/5/0.075) 0.32 × 230–435 mm Hypersil ODS (5) ACN or 0.32 × 230 mm Kromasil C18 (5) MeOH/AA/pyridine/water (85/4/1/10) 0.32 × 700 mm Hypersil ODS or BDS ACN/DMF (90/10) or (3/5) or Kromasil C18 (5) ACN/pyridine (95/5)

0.32 × 400 mm YMC C30 (5)

GC-oven

70–130 (G)

cLC-oven GC-oven

7–90 (G) 30–160 (G)

ACN ACN or ethyl acetate/ACN (10/90)

0.32 × 350–1000 mm Kromasil C18 (5) ACN/MIBK (75/25)

GC-oven GC-oven GC-oven

50–120 (G) 30–130 (G) 30–130 (G)

ACN ACN and ACN/MIBK ethyl acetate/ ACN/TEA (40/50/10)

cLC-oven cLC-oven GC-oven

0.32 × 300 mm Kromasil C18 (3.5) 0.32 × 350–1000 mm Kromasil C18 (5) 0.32 × 350 mm Hypersil ODS (3) 0.32 × 350 mm Hypersil BDS (5) 0.32 × 350 mm Kromasil C18 (5) 0.32 × 280 mm Kromasil C18 (5) 0.32 × 280 mm Hypersil ODS (3) 0.32 × 700 mm Hypersil BDS (5)

6–90 (G+I) 5–90 (G) 28–120 (G)

ACN/water ACN/(TEA+FA) ACN/TEA (95/5)

0.32 × 250 mm Hypersil C18 (3) 0.32 × 500 mm Kromasil C18 (3.5) 0.32 × 400 mm Hypersil ODS (3)

ELSD

ICP-MS

UV-VIS

UV or ELSD

UV-VIS

ELSD

UV ELSD

UV + FT-IR ELSD ELSD

UV ELSD ELSD

0.06

0.06

100

0.06

20 at 10˚C

0.06 at 100˚C

100 at 7˚C 0.05

0.06 0.06 0.50

0.05 100 at 5˚C 0.05

[159]

[168]

[134] [150]

[166] [167] [147]

[164] [165] [148]

[171]

[170]

Fatty amides, [172] glycerylmonoste arate

Retinol, retinoic acids Tetreethyl- and tetramethyl lead

Polymer additives [169]

Irganox 1076 Polymer additives and glyceryl monostearate Hydrocarbon waxes Retinyl esters

Iodixanol Ceramides Hindered amine stabilizers Polymer additives Technical waxes Hindered amine stabilizers

Temperature Effects in Liquid Chromatography 59

c

b

a

MeOH/acetate buffer pH 5 (30/70) ACN/water ACN/water (72/28) MeOH/water ACN/DMF or ACN/N,Ndimethylacetamide ACN/water or MeOH/water ACN/water or MeOH/water MeOH or ACN/water

Mobile Phase(s)

Lab. made Lab. made Lab. made

GC-oven SFC-oven GC-oven GC-oven

30–120 (G) 24–80 (I) 25–150 (I+G) 50–150 (G) 100–150 (I) 100–200 (I) 30–80 (G)

Lab.-made

Heating System

150 (I)

Temperature Range (°°C) (G = Gradient, I = Isothermal)

cold zone at 10˚C using a column switching system columns used: Zorbax C18 (XDB, SB and Extend), Hypersil ODS, XTerra MS 18 and YMC C30

AA = acetic acid FA = formic acid TCB = trichlorobenzene

0.05 × 1500 mm SB-methyl-100 0.05–0.1 × 8000mm octyl polysiloxane 0.08–0.25 × 1000–5000 mm Develosil ODS (3/5/7/10)

0.05 × 2000 mm cyano and octyl polysiloxane 0.18 × 250 mm Zorbax SB ODS (6) 0.2 × 800 mm Phenomenex C18 (5) 0.05 × 1500 mm octyl polysilixane 0.32 ×700 mm Hypersil ODS (3/5)

Dimension and Packing Particle Diameter (µm)

TABLE 2.1 (CONTINUED) Temperature Studies on Capillary Columns

UV UV UV

UV UV UV + MS UV

UV

Detector

0.07 0.005 0.1–0.2

0.06 0.06 0.01 0.06

10

Inj. Vol. (µL)

Test compounds Test compounds p-nitrobenzyl esters of fatty acids

Alkylbenzenes Test compounds Test compounds Polymer additives

Test compounds

Analyte(s)

[123] [122] [175]

[48] [173] [125] [174]

[126]

Ref.

60 Advances in Chromatography

Temperature Effects in Liquid Chromatography

61

Signal intensity

Column 1 55ºC

Column 1 55ºC

Column 2 56ºC

Column 2 55ºC

Column 3 57ºC

Column 3 55ºC

Column 4 55ºC 0

2

12

4 Time (min)

6

13 14 % acetonitrile (a)

8

15

Column 4 55ºC 0

2

10

4 6 Time (min) 11

8

10

12 13 % acetonitrile

14

(b)

FIGURE 2.5 Adjustment of the chromatographic elution profiles of different columns by means of variation of column temperature. (a) The same sample was injected onto four different columns kept at the same temperature in a commercially available enforced air circulation oven with Peltier cooling. Because of differences in surface area, retention of DNA fragments varies between columns. (b) Variation of the individual column temperatures allows harmonization of chromatographic profiles. Columns, 4 × monolithic PS/DVB, 60 × 0.2 mm i.d., mobile phase, (A) 100 mM TEAA at pH 7.0, (B) 100 mM TEAA at pH 7.0, 25% acetonitrile; linear gradient, (a) 45–61% B in 8.0 min, (b) 38–56% B in 10 min; flow rate, 2.0–3.0 μL/min; temperature, (a) 55˚C, (b) 55–57˚C; detection, LIF, emission monitored at 525 nm; injection volume, 1 μL each; sample, equimolar mixture of two FAM-labeled 209bp Y chromosome alleles differing in a single base (168 A > G). From [119].

62

Advances in Chromatography

material is most often fused silica — PEEK shielded fused silica and glass-lined steels have also been employed in some commercial columns, but these are less suited for temperature-gradient programming. The length of the packed columns varies from 100 to 1000 mm, depending on the particle size and the desired resolution. A particle size of 3 or 3.5 μm is most common and the columns are relatively short. When the inner diameter of the column is reduced, the injection volume must be reduced accordingly (the down-scale factor) to maintain high efficiency. Typical injection volumes in capillary LC should be in the range 0.05 to 0.1 μL (see Table 2.1). However, packed capillary columns can be used advantageously for trace determination of analytes in samples of limited sample size [133] by using larger sample volumes that can be introduced either by on-column focusing techniques [134] or by column switching techniques [135, 136]. Injection of large volumes without detrimental effect on the column efficiency can be achieved by solute focusing with a solvent with low elution strength [137], or by temperature [138]. Up to 1.0 mL of sample has been applied utilizing solvent-promoted focusing in a 0.32 mm I.D. column switching system [135, 137] (see Figure 2.6). In reversedAUFS

0.02

3 0.01

1 4 2

0.00

20

40

60

min

FIGURE 2.6 Packed capillary column-switching LC separation of 12 ng of each of the 12MP derivatives of 2,6-TDI (1), HDI (2), 2,4-TDI (3) and MDI (4) from a spiked air sampling filter on a 3 μm Hypersil ODS (0.32 × 250 mm) column. The injection volume was 1.0 mL and the mobile phase was acetonitrile: 10 mM ammonium formate (pH 6.0) (40:60, v/v). A temperature program from 80˚C (37 min) to 95˚C (25 min) using a temperature ramp of 5.0˚C/min was used. From [135].

Temperature Effects in Liquid Chromatography

63

phase systems, the retention is usually lower at low temperatures and up to 500 μL of a sample containing a hydrophobic analyte has been introduced at 5˚C using a column switching system [136]. In this case both the sample solvent and eluting mobile phase consisted of 100% acetonitrile and the analyte Irganox 1076 was completely retained at the 27 mm-long C18 enrichment column at 5˚C. A subsequent temperature program from 5 to 90˚C provided optimal separation of the analyte from the matrix components, which was coextracted from the polymer sample using 100% acetonitrile as extraction solvent. Of the packed capillary columns, silica-based reversed-phase column materials have been most extensively studied (see Table 2.1). Only a limited number of studies have been published using other materials like silica and ion exchange. In an early study, Jinno and Hirata investigated the (low) temperature effect in normal phase micro LC [16], and lately Molander et al. [17] resolved an extruded polymer blend by using temperature-programmed (110 to 170˚C) elution on large-pore silica stationary phase. The separation was obtained according to both size exclusion and normal phase interaction. Isoelectric point (pI) separation of proteins by pH-gradient ion-exchange with temperature control was recently demonstrated on a packed capillary column [19]. The temperature tolerance of the zirconia materials has promoted their use for high temperature separations [15, 79, 139, 140]; however, only one paper on their use in capillary LC has been published [139]. Polybutadiene-modified zirconia packed in 0.1 mm I.D. glass-lined stainless steel columns was found to be stable at 100˚C using acetonitrile/phosphate buffer (pH 7) as mobile phase. On the other hand, in temperature-programming mode (not exceeding 100˚C), a rapid decrease in both column efficiency and retention factors was observed. Even though carbon materials have shown high stability under temperature-programmed high-temperature LC conditions [15], carbon materials have been used mostly at ambient temperature in capillary LC [141–143]. So far, only one study has used the temperature for retention control [144]. Linear van’t Hoff plots were obtained for sulphated glycosamin glycan disaccharides, however, of different sign at acidic (2.2) and alkaline (12) pH. The effect of temperature on retention on the cellulose triacetate enantioselective stationary phase has been studied [145], but so far no separation using the temperature-responsive stationary phases [21–24] have been demonstrated on capillary columns.

2.4.2 TEMPERATURE PROGRAMMING

AND INSTRUMENTATION

The temperature range studied in capillary LC has been limited on the high end mainly by the stability of the packing material but, in some cases, also by the stability of the solutes. Temperatures below 5˚C have seldom been explored due to the lower efficiency at low temperature and by the available ovens (see Table 2.1). Only two commercially available LC ovens with the option for temperature programming exist with temperature ranges of 5 to 150˚C and < 0 to 200˚C, respectively. The former oven (cLC-oven) has been used in a number of publications, as can be seen in Table 2.1, but the drawback is the slow cooling back to 5˚C after performing a run. For

64

Advances in Chromatography

temperature programming, the ovens should provide rapid return to the start of the temperature program; however, the Peltier elements in the larger ovens are too slow and cryo-cooling is required for a rapid return to start. If temperatures below 0˚C are needed (e.g., for focusing of solutes), an oven with two separate temperature zones is advantageous [138]. For temperatures above ambient, GC-ovens are suitable both for isothermal and temperature-programmed elution. Other laboratory-made devices for column temperature control have also been used, as shown in Table 2.1. The repeatability of retention time with temperature programs has been found to be comparable to that obtained with solvent gradient elution. The relative standard deviations (%RSDs) were in the range 0.5 to 1.4 for a 30 to 150˚C (at 1.5 K/min) gradient [146], 0.4 to 1.2 for a 7 to 90˚C (at 3 K/min) gradient [134], and 2 for a 30 to 120˚C (at 5 K/min) gradient [138]. Temperature programming has been used in a number of applications (Table 2.1) to obtain separation in a reasonable time with a good efficiency. Temperatures up to 90˚C have been used on silica-based C18 columns with aqueous mobile phases, even though temperatures above 60˚ are not recommended [15]. When nonaqueous mobile phases are used, higher temperatures may be utilized. Andersen et al. separated hindered amine stabilizers using nonaqueous mobile phases and temperature programming from 30 to 120/130˚C on C18 columns [146–148]. Temperatures up to 200˚C have been used on the open tubular columns by Liu et al., obtaining over 1 mill theoretical plates in 50 minutes using a 20 m-long column [122]. The majority of studies on capillary LC with temperature control have utilized UV detection or evaporative light scattering detection (ELSD) (see Table 2.1). When temperature−gradient programming is used in combination with UV detection, a rising baseline is often observed due to refractive index changes but a rising baseline can also be due to column bleeding [15]. The rising baseline may be corrected mathematically by a commercial software package [149] or by dual wavelength measurements and corrections [77]. The nebulizer of the ELSD has been modified for the low flow rates used in capillary LC to obtain improved limits of detection and increased linearity [150] and baseline drift with temperature−gradient programming is normally not a problem. The laboratory-made electrochemical detector (ECD) for packed capillary columns also showed a rising baseline with temperature programming but the drift could be corrected mathematically [151]. Mass spectrometry (MS), especially with electrospray ionization, benefits from the low flows used in capillary LC and the compatibility between temperature programming in LC and MS was shown more than a decade ago [152]. However, MS has only been applied in a few studies where the temperature has been used actively for retention control in capillary LC [146, 153]. Since temperature programming can be expected to affect the electrospray ionization process less than solvent gradients, temperature gradient programming in LC with MS detection can in many cases be a better choice than solvent gradient elution. Low-volume flow cells for NMR have now become commercially available and the combination of capillary LC and NMR seems very promising compared to conventional-sized systems when the sample amount is limited [154]. Resonance shifts due to changes in solvent composition are avoided by isocratic elution and much better detection limits can be achieved.

Temperature Effects in Liquid Chromatography

65

A small overpressure is necessary to keep the mobile phase in the liquid phase when temperatures above the boiling point of the mobile phase are applied. This is normally done by connecting a fused silica capillary with small I.D. (15 to 30 μm) between the column and the detector when MS or ELSD is used and by using a similar capillary from the detector to waste when UV detection is utilized.

2.4.3 APPLICATIONS Even though most of the studies on temperature effects in capillary columns have utilized test compounds (see Table 2.1) for exploring the separation power of packed capillary column systems, a limited number of papers have also shown the applicability of this methodology for real-life samples [19, 134, 136, 155–159]. Capillary LC with large-volume injection and temperature-programmed elution was used for the determination of Irganox 1076 that was extracted from low-density polyethylene by microwave-aided extraction with acetonitrile [134]. In this case, acetonitrile was used both as injection solvent (100 μL) and as the mobile phase. By utilizing temperature-promoted, large-volume solute focusing, sample volumes up to 500 μL could be used to improve the detection limit to 0.6 ppm [136]. The concentration of retinyl palmitate in Arctic seal liver was determined using a nonaqueous temperature-programmed reversed-phase separation with large-volume oncolumn injection [159]. A method for the simultaneous determination of three selective serotonin reuptake inhibitors and their metabolites in solid phase extracts from plasma, using temperature-programmed aqueous reversed-phase separation with on-column focusing of large injection volumes, provided quantification limits in the concentration range 0.05 to 0.26 μM [157]. The presence of eight triazines in ground water was determined using temperature-programmed elution on a reversed-phase column after large volume injection [158], while chlorophenoxy acid herbicides and their esters were determined in extracts from soil using large-volume injection and temperature-programmed elution on a C18 reversed-phase column [156] (see Figure 2.7). The detection limits reported were between 0.3 and 0.5 μg/g of soil. Low levels of polyethylene glycols (PEG) in extracts from low-density polyethylene could be determined by large-volume injection, temperature-programmed elution on a C18 reversed-phase column using a column switching system [155]. In this case, inverse temperature programming from 68 to 45˚C was used since the retention of PEGs increases with increasing column temperature, a fact explained by conformational changes of PEGs at different temperatures. One special application is the isoelectric point separation of proteins by capillary pH-gradient ion exchange chromatography, where the temperature was used to optimize the separation of skimmed milk proteins [19]. Temperature has been actively used for high-resolution chromatographic determination of nucleic acids on PS-DVB monolithic columns as well as on packed PSDVB columns [160]. In addition to their use in genomic research, the PS-DVB monolithic columns have also been used in proteomics [161]. Huber et al. recently published a study on the effect of temperature and mobile phase additives on peptide and protein analysis on PS-DVB monoliths [162] where they showed that the efficiency improved significantly upon increasing the temperature up to 70˚C. The

66

Advances in Chromatography

2, 4-D

2, 4-D methyl ester

2, 4-DB 2, 4-D butyl ester

b a

0

10 Time, min

20

FIGURE 2.7 LC chromatogram of (a) unspiked soil; (b) soil spiked with 4.0 μg/g of chlorophenoxy acids. Column: LC Packing Hypersil C18 BDS, 3 μm, 150 mm × 0.3 mm I.D. Mobile phase: MeOH/H3PO4 0.8% aqueous solution (56/44, v/v). Flow rate in column: 10 μL/min. Temperature gradient from 20 to 65 ˚C at 9˚C/min. Injection volume: 20 μL. From [156].

selectivity of peptide elution was significantly modulated by the temperature, whereas the effect on protein analysis was only minor. The construction of monolithic capillary arrays, fine-tuned by using the temperature (see also Figure 2.5), allows high-throughput analyses and their use will probably increase in the years to come.

REFERENCES 1. Ooms, B. Temperature control in high performance liquid chromatography. LC-GC, 1996, 14, 306–313. 2. Greibrokk, T. and Andersen, T. High-temperature liquid chromatography. J. Chromatogr. A, 2003, 1000, 743–755. 3. Li, J.W., Hu, Y., and Carr, P.W. Fast separations at elevated temperatures on polybutadiene-coated zirconia reversed-phase material. Anal. Chem., 1997, 69, 3884–3888.

Temperature Effects in Liquid Chromatography

67

4. Guillarme, D., Heinisch, S., and Rocca, J.L. Effect of temperature in reversed phase liquid chromatography. J. Chromatogr. A, 2004, 1052, 39–51. 5. Dolan, J.W. Temperature selectivity in reversed-phase high performance liquid chromatography. J. Chromatogr. A, 2002, 965, 195–205. 6. Greibrokk, T. and Andersen, T. Temperature programming in liquid chromatography. J. Sep. Sci., 2001, 24, 899–909. 7. Sander, L.C. and Wise, S.A. The influence of column temperature on selectivity in reversed-phase liquid chromatography for shape-constrained solutes. J. Sep. Sci., 2001, 24, 910–920. 8. McCalley, D.V. Effect of temperature and flow-rate on analysis of basic compounds in high-performance liquid chromatography using a reversed-phase column. J. Chromatogr. A, 2000, 902, 311–321. 9. Böhm, V. Use of column temperature to optimize carotenoid isomer separation by C30 high performance liquid chromatography. J. Sep. Sci., 2001, 24, 955–959. 10. Szabelski, P., Cavazzini, A., Kaczmarski, K., Liu, X., Van Horne, J., and Guiochon, G. Experimental studies in pressure/temperature dependence of protein adsorption equilibrium in reversed-phase high-performance liquid chromatography. J. Chromatogr. A, 2002, 950, 41–53. 11. Saha, S., Smith, R.M., Lenz, E., and Wilson, I.D. Analysis of ginger extract by highperformance liquid chromatography coupled to nuclear magnetic resonance spectroscopy using superheated deuterium oxide as the mobile phase. J. Chromatogr. A, 2003, 991, 143–150. 12. Lee, D.L., Mant, C.T., and Hodges, R.S. A novel method to measure self-association of small amphipathic molecules. J. Biol. Chem., 2003, 278, 22918–22927. 13. Chen, Y., Mant, C.T., and Hodges, R.S. Temperature selectivity effects in reversed phase liquid chromatography due to conformation differences between helical and non-helical peptides. J. Chromatogr. A, 2003, 1010, 45–61. 14. Borch, T. and Gerlach, R. Use of reversed-phase high-performance liquid chromatography-diode array detection for complete separation of 2,4,6-trinitrotoluene metabolites and EPA Method 8330 explosives: Influence of temperature and ion-pair reagent. J. Chromatogr. A, 2004, 1022, 83–94. 15. Marin, S.J., Jones, B.A., Felix, W.D., and Clark, J. Effect of high-temperature on high-performance liquid chromatography column stability and performance under temperature-programmed conditions. J. Chromatogr. A, 2004, 1030, 255–262. 16. Jinno, K. and Hirata, Y. Investigation of the low temperature effect in normal phase micro high performance liquid chromatography. J. HRC & CC, 1982, 5, 85–90. 17. Molander, P., Greibrokk, T., Iveland, A., and Ommundsen, E. Simultaneous dualmechanism separation of a semicrystalline polymer blend on large-pore silica by temperature-programmed packed capillary liquid chromatography. J. Sep. Sci., 2001, 24, 136–140. 18. Busch, M. and Seubert, A. Influence of column temperature on the ion chromatographic separation of aluminum species. Anal. Chim. Acta, 1999, 399, 223–235. 19. Andersen, T., Pepaj, M., Trones, R., Lundanes, E., and Greibrokk, T. Isoelectric point separation of proteins by capillary pH-gradient ion-exchange chromatography. J. Chromatogr. A, 2004, 1025, 217–226. 20. Wei, Y., Yao, C., Zhao, J., and Geng, X. Influences of the mobile phase composition and temperature on the retention behaviour of aromatic alcohol homologues in hydrophobic interaction chromatography. Chromatographia, 2002, 55, 659–665.

68

Advances in Chromatography 21. Kanazawa, H., Yamamoto, K., Kashiwase, Y., Matsushima, Y., Takai, N., Kikuchi, A., Sakurai, Y., and Okano, T. Analysis of peptides and proteins by temperatureresponsive chromatographic system using N-isopropylacrylamide polymer-modified columns. J. Pharm. Biomed. Anal., 1997, 15, 1545–1550. 22. Go, H., Sudo, Y., Hosoya, T., Ikegami, T., and Tanaka, N. Effects of mobile-phase composition and temperature on the selectivity of poly(N-isopropylacrylamide)bonded silica gel in reversed-phase liquid chromatography. Anal. Chem., 1998, 70, 4086–4093. 23. Song, Y.-X., Wang, J.-Q., Su, Z.-X., and Chen, D.-Y. High-performance liquid chromatography on silica modified with temperature-responsive polymers. Chromatographia, 2001, 54, 208–212. 24. Kurata, K., Shimoyama, T., and Dobashi, A. Enantiomeric separation using temperature-responsive chiral polymers composed of L-valine diamide derivatives in aqueous liquid chromatography. J. Chromatogr. A, 2003, 1012, 47–56. 25. Jiskra, J., Claessens, H.A., and Cramers, C.A. Thermodynamic behaviour in capillary electrochromatography. J. Sep. Sci., 2002, 25, 569–576. 26. Snyder, L.R. Temperature-induced selectivity in separations by reversed-phase liquid chromatography. J. Chromatogr., 1979, 179, 157–172. 27. Grushka, E., Colin, H., and Guiochon, G. Retention behavior of alkylbenzenes as a function of temperature and mobile phase composition in reversed-phase chromatography. J. Chromatogr., 1982, 248, 325–339. 28. Takeuchi, T., Kumaki, M., and Ishii, D. Role of column temperature in open-tubular microcapillary liquid chromatography. J. Chromatogr., 1982, 235, 309–322. 29. Miyabe, K. and Guiochon, G. Thermodynamic interpretation of retention equilibrium in reversed-phase liquid chromatography based on enthalpy-entropy compensation. Anal. Chem., 2002, 74, 5982–5992. 30. Colin, H., Diez-Masa, J.C., Guiochon, G., Czajkowska, T., and Miedziak, I. The role of the temperature in reversed-phase high-performance liquid chromatography using pyrocarbon-containing adsorbents. J. Chromatogr., 1978, 167, 41–65. 31. Knox, J.H. and Vasvari, G. The performance of packings in high-speed liquid chromatography. III. Chemically bonded pellicular materials. J. Chromatogr., 1973, 83, 181–194. 32. Colin, H. and Guiochon, G. Introduction to reversed-phase high-performance liquid chromatography. J. Chromatogr., 1977, 141, 289–312. 33. Colin, H. and Guiochon, G. Comparison of some packings for reversed-phase highperformance liquid — solid chromatography. II. Some theoretical considerations. J. Chromatogr., 1978, 158, 183–205. 34. Horvath, C. and Melander, W. Liquid chromatography with hydrocarbonaceous bonded phases, theory and practice of reversed phase chromatography. J. Chromatogr. Sci., 1977, 15, 393–404. 35. Issaq, H.J. and Jaroniec, M. Enthalpy and entropy effects for homologous solutes in HPLC with alkyl chain bonded phases. J. Liq. Chromatogr., 1989, 12, 2067–2082. 36. Alvarez-Zepeda, A., Barman, B.N., and Matire, D.E. Thermodynamic study of the marked differences between acetonitrile/water and methanol/water mobile-phase systems in reversed-phase liquid chromatography. Anal. Chem., 1992, 64, 1978–1984. 37. Guillaume, Y.C. and Guinchard, C. Retention mechanism of weak polar solutes in reversed phase liquid chromatography. Anal. Chem., 1996, 68, 2869–2873. 38. Guillaume, Y.C. and Guinchard, C. ACN clusters in a water/ACN mixture, with implications for the RPLC weak polar solute retention. Anal. Chem., 1997, 69, 183–189.

Temperature Effects in Liquid Chromatography

69

39. FuAn, W., JianChi, S., YingChen, G., YiHong, W., and Yan, Z. Rule of intersection point between retention of homologues and column temperature in reversed-phase liquid chromatography. Microchem. J., 1995, 52, 200–204. 40. Bai, Q. and Geng, X. The temperature convergence of biopolymers in reversed-phase liquid chromatography. Chromatographia, 2001, 54, 335–338. 41. Chester, T.L. and Coym, J.W. Effect of phase ratio on van ‘t Hoff analysis in reversedphase liquid chromatography, and phase-ratio-independent estimation of transfer enthalpy. J. Chromatogr. A, 2003, 1003, 101–111. 42. Jinno, K., Ohshima, T., and Hirata, Y. Investigation of the low temperature effect in reversed phase micro high performance liquid chromatography. J. High Resolut. Chromatogr., 1982, 5, 621–624. 43. Coym, J.W. and Dorsey, J.G. Reversed-phase retention thermodynamics of pure-water mobile phases at ambient and elevated temperature. J. Chromatogr. A, 2004, 1035, 23–29. 44. Castells, C.B., Gagliardi, L.G., Rafols, C., Roses, M., and Bosch, E. Effect of temperature on the chromatographic retention of ionizable compounds. I. Methanol-water mobile phases. J. Chromatogr. A, 2004, 1042, 23–35. 45. Li, J. Effect of temperature on selectivity in reversed-phase liquid chromatography, a thermodynamic analysis. Anal. Chim. Acta, 1998, 369, 21–37. 46. Buckenmaier, S.M.C., McCalley, D.V., and Euerby, M.R. Rationalisation of unusual changes in efficiency and retention with temperature shown for bases in reversedphase high-performance liquid chromatography at intermediate pH. J. Chromatogr. A, 2004, 1060, 117–126. 47. Bowermaster, J. and McNair, H.M. Temperature programmed microbore HPLC-Part I. J. Chromatogr. Sci., 1984, 22, 165–170. 48. Chen, M.H. and Horvath, C. Temperature programming and gradient elution in reversed-phase chromatography with packed capillary columns. J. Chromatogr. A, 1997, 788, 51–61. 49. Tran, J.V., Molander, P., Greibrokk, T., and Lundanes, E. Temperature effects on retention in reversed phase liquid chromatography. J. Sep. Sci., 2001, 24, 930–940. 50. Zhu, P.L., Snyder, L.R., Dolan, J.W., Djordjevic, N.M., Hill, D.W., Sander, L.C., and Waeghe, T.J. Combined use of temperature and solvent strength in reversed-phase gradient elution. I. Predicting separation as a function of temperature and gradient conditions. J. Chromatogr. A, 1996, 756, 21–39. 51. Zhu, P.L., Dolan, J.W., and Snyder, L.R. Combined use of temperature and solvent strength in reversed-phase gradient elution. II. Comparing selectivity for different samples and systems. J. Chromatogr. A, 1996, 756, 41–50. 52. Zhu, P.L., Dolan, J.W., Snyder, L.R., Hill, D.W., VanHeukelem, L., and Waeghe, T.J. Combined use of temperature and solvent strength in reversed-phase gradient elution. III. Selectivity for ionizable samples as a function of sample type and pH. J. Chromatogr. A, 1996, 756, 51–62. 53. Zhu, P.L., Dolan, J.W., Snyder, L.R., Djordjevic, N.M., Hill, D.W., Lin, J.T., Sander, L.C., and Van Heukelem, L. Combined use of temperature and solvent strength in reversed-phase gradient elution. IV. Selectivity for neutral (non-ionized) samples as a function of sample type and other separation conditions. J. Chromatogr. A, 1996, 756, 63–72.

70

Advances in Chromatography 54. Dolan, J.W., Snyder, L.R., Djordjevic, N.M., Hill, D.W., Sounders, D.L., Van Heukelem, L., and Waeghe, T.J. Simultaneous variation of temperature and gradient steepness for reversed-phase high-performance liquid chromatography method development. I. Application to 14 different samples using computer simulation. J. Chromatogr. A, 1998, 803, 1–31. 55. Wolcott, R.G., Dolan, J.W., and Snyder, L.R. Computer simulation for the convenient optimization of isocratic reversed-phase liquid chromatographic separations by varying temperature and mobile phase strength. J. Chromatogr. A, 2000, 869, 3–25. 56. Neue, U.D. and Mazzeo, J.R. A theoretical study of the optimization of gradients at elevated temperature. J. Sep. Sci., 2001, 24, 921–929. 57. Cole, L.A., Dorsey, J.G., and Dill, K.A. Temperature dependence of retention in reversed-phase liquid chromatography. 2. Mobile-phase considerations. Anal. Chem., 1992, 64, 1324–1327. 58. Kim, H., Gritti, F., and Guiochon, G. Effect of the temperature on the isotherm parameters of phenol in reversed-phase liquid chromatography. J. Chromatogr. A, 2004, 1049, 25–36. 59. Issaq, H.J., Fox, S.D., Lindsey, K., McConnell, J.H., and Weiss, D.E. Effect of temperature on HPLC separations using C1, C4, C8, and C18 alkyl chain bonded silica columns. J. Liq. Chromatogr., 1987, 10, 49–70. 60. Dolan, J.W., Snyder, L.R., Blanc, T., and Van Heukelem, L. Selectivity differences for C18 and C8 reversed-phase columns as a function of temperature and gradient steepness. I. Optimizing selectivity and resolution. J. Chromatogr. A, 2000, 897, 37–50. 61. Ducry, M.W.J., Orendorff, C.J., Pemberton, J.E., and Sander, L.C. Structure-function relationships in high-density octadecylsilane stationary phases by Raman spectroscopy. 1. Effects of temperature, surface coverage, and preparation procedure. Anal. Chem., 2002, 74, 5576–5584. 62. Tchapla, A., Heron, S., Colin, H., and Guiochon, G. Role of temperature in the behavior of a homologous series in reversed phase liquid chromatography. Anal. Chem., 1988, 60, 1443–1448. 63. Yamamoto, F.M., Rokushika, S., and Hatano, H. Comparison of thermodynamic retention behaviors on various C18 columns different in their hydrophobicity. J. Chromatogr. Sci., 1989, 27, 704–709. 64. Sander, L.C. and Field, L.R. Effect of eluent composition on thermodynamic properties in high-performance liquid chromatography. Anal. Chem., 1980, 52, 2009–2013. 65. Cole, L.A. and Dorsey, J.G. Temperature dependence of retention in reversed-phase liquid chromatography. 2. Stationary-phase considerations. Anal. Chem., 1992, 64, 1317–1323. 66. Sentell, K.B., Ryan, N.I., and Henderson, A.N. Temperature and solvation effects on homologous series selectivity in reversed phase liquid chromatography. Anal. Chim. Acta, 1995, 307, 203–215. 67. Chmielowiec, J. and Sawatzky, H. Entropy dominated high performance liquid chromatographic separations of polynuclear aromatic hydrocarbons. Temperature as a separation parameter. J. Chromatogr. Sci., 1979, 17, 245–252. 68. Albert, K. Correlation between chromatographic and physiochemical properties of stationary phases in HPLC: C30 bonded reversed-phase silica. Trends Anal. Chem., 1998, 17, 648–658. 69. Bell, C.M., Sander, L.C., and Wise, S.A. Temperature dependence of carotenoids on C18, C30 and C34 bonded stationary phases. J. Chromatogr. A, 1997, 757, 29–39.

Temperature Effects in Liquid Chromatography

71

70. Pursch, M., Strohschein, S., Händel, H., and Albert, K. Temperature-dependent behaviour of C30 interphases. A solid-state NMR and LC-NMR study. Anal. Chem., 1996, 68, 386–393. 71. Srinivasan, G., Pursch, M., Sander, L.C., and Müller, K. FTIR studies of C30 selfassembled monolayers on silica, titania and zirconia. Langmuir, 2004, 20, 1746–1752. 72. Yang, Y. Stationary phases for high-temperature liquid chromatography. LC-GC Europe, 2003, 16, 37–41. 73. Bullock, J. High-temperature reversed-phase high-performance liquid chromatographic analysis of a synthetic copolymer on a non-porous support. J. Chromatogr. A, 1995, 694, 415–423. 74. Claessens, H.A. and van Straten, M.A. Review on the chemical and thermal stability of stationary phases for reversed-phase liquid chromatography. J. Chromatogr. A, 2004, 1060, 23–41. 75. He, P. and Yang, Y. Studies on the long-term thermal stability of stationary phases in subcritical water chromatography. J. Chromatogr. A, 2003, 989, 55–63. 76. Bolliet, D. and Poole, C.F. Influence of temperature on retention and selectivity in reversed phase liquid chromatography. Analyst, 1998, 123, 295–299. 77. Renn, C.N. and Synovec, R.E. Thermal gradient microbore liquid chromatography with dual-wavelength absorbance detection. Anal. Chem., 1991, 63, 568–574. 78. Li, J.W. and Carr, P.W. Effect of temperature on the thermodynamic properties, kinetic performance, and stability of polybutadiene-coated zirconia. Anal. Chem., 1997, 69, 837–843. 79. Dunlap, C.J., Carr, P.W., McNeff, C.V., and Stoll, D. Zirconia stationary phases for extreme separations. Anal. Chem., 2001, 73, 598A–607A. 80. Li, J.W. and Carr, P.W. Evaluation of temperature effects on selectivity in RPLC separations using polybutadiene-coated zirconia. Anal. Chem., 1997, 69, 2202–2206. 81. Wolcott, R.G., Dolan, J.W., Snyder, L.R., Bakalyar, S.R., Arnold, M.A., and Nichols, J.A. Control of column temperature in reversed-phase liquid chromatography. J. Chromatogr. A, 2000, 869, 211–230. 82. Spearman, L., Smith, R.M., and Dube, S. Monitoring effective column temperature by using shape selectivity and hydrophobicity and the effects of mobile phase temperature. J. Chromatogr. A, 2004, 1060, 147–151. 83. Poppe, H. and Kraak, J.C. Influence of thermal conditions on the efficiency of highperformance liquid chromatographic columns. J. Chromatogr., 1983, 282, 399–412. 84. Thompson, J.D., Brown, J.S., and Carr, P.W. Dependence of thermal mismatch broadening on column diameter in high-speed liquid chromatography at elevated temperatures. Anal. Chem., 2001, 73, 3340–3347. 85. Hesse, G. and Engelhardt, H. Temperatureprogrammierung bei der adsorptionschromatographie von lösungen. J. Chromatogr., 1966, 21, 228–238. 86. Djordjevic, N.M., Fowler, P.W.J., and Houdiere, F. High temperature and temperature programming in high-performance liquid chromatography: Instrumental considerations. J. Microcol. Sep., 1999, 11, 403–413. 87. Molander, P., Olsen, R., Lundanes, E., and Greibrokk, T. The impact of inner diameter on chromatographic performance in temperature gradient liquid chromatography. Analyst, 2003, 128, 1341–1345. 88. Biggs, W.R. and Fetzer, J.C. Thermal gradient liquid chromatography: Application to selective element detection by inductively coupled plasma emission spectrometry. Anal. Chem., 1989, 61, 236–240.

72

Advances in Chromatography

89. Kim, W.-S., Lee, J.Y., Lee, S.K., Park, Y.H., and Lee, D.W. Investigation of retention behavior for herbicides in micro-HPLC with temperature changes. J. Liq. Chromatogr., 2003, 26, 3357–3369. 90. Hu, W., Hasebe, K., and Haddad, P.R. Temperature effects on retention in reversedphase liquid chromatography of nucleosides and their bases using water as the mobile phase. Anal. Commun., 1997, 34, 322–324. 91. Moore, L.K. and Synovec, R.E. Axial thermal gradient microbore liquid chromatography by flow programming. Anal. Chem., 1993, 65, 2663–2670. 92. Sander, L.C., Sharpless, K.E., and Pursch, M. C30 stationary phases for the analysis of food by liquid chromatography. J. Chromatogr. A, 2000, 880, 189–202. 93. Hancock, W.S., Chloupek, R.C., Kirkland, J.J., and Snyder, L.R. Temperature as a variable in reversed-phase high-performance liquid chromatographic separations of peptide and protein samples. I. Optimizing the separation of a growth hormone tryptic digest. J. Chromatogr. A, 1994, 686, 31–43. 94. Chloupek, R.C., Hancock, J.R., Marchylo, B.A., Kirkland, J.J., Boyes, B.E., and Snyder, L.R. Temperature as a variable in reversed-phase high-performance liquid chromatographic separations of peptide and protein samples. II. Selectivity effects observed in the separation of several peptide and protein mixtures. J. Chromatogr. A, 1994, 686, 45–59. 95. Mant, C.T., Chen, Y., and Hodges, R.S. Temperature profiling of polypeptides in reversed-phase liquid chromatography. I. Monitoring of dimerization and unfolding of amphipathic γ-helical peptides. J. Chromatogr. A, 2003, 1009, 29–43. 96. Mant, C.T., Tripet, B., and Hodges, R.S. Temperature profiling of polypeptides in reversed-phase liquid chromatography. II. Monitoring of folding and stability of twostranded γ-helical coilded-coils. J. Chromatogr., 2003, 1009, 45–59. 97. Chen, Y., Mehok, A.R., Mant, C.T., and Hodges, R.S. Optimum concentration of trifluoroacetic acid for reversed-phase liquid chromatography of peptides revisited. J. Chromatogr. A, 2004, 1043, 9–18. 98. Hayakawa, K., Hirano, M., Yoshikawa, K., Katsumata, N., and Tanaka, N. Separation of phenylthiohydantoin-amino acids by temperature-controlled reversed-phase highperformance liquid chromatography. J. Chromatogr. A, 1999, 846, 73–82. 99. Dolfinger, R. and Locke, D.C. Dependence of selectivity on eluent composition and temperature in the HPLC separation of taxanes using fluorinated and hydrocarbon phases. Anal. Chem., 2003, 75, 1355–1364. 100. Issaq, H.J., Xu, H., Chan, K.C., and Dean, M.C. Effect of temperature on the separation of DNA fragments by high-performance liquid chromatography and capillary electrophoresis: A comparative study. J. Chromatogr. B, 2000, 738, 243–248. 101. Yang, Y., Belghazi, M., Lagadec, A., Miller, D.J., and Hawthorne, S.B. Elution of organic solutes from different polarity sorbents using subcritical water. J. Chromatogr. A, 1998, 810, 149–159. 102. Fields, S.M., Ye, C.Q., Zhang, D.D., Branch, B.R., Zhang, X.J., and Okafo, N. Superheated water as eluent in high-temperature high-performance liquid chromatographic separations of steroids on a polymer-coated zirconia column. J. Chromatogr. A, 2001, 913, 197–204. 103. Kondo, T., Yang, Y., and Lamm, L. Separation of polar and non-polar analytes using dimethyl sulfoxide-modified subcritical water. Anal. Chim. Acta, 2002, 460, 185–191. 104. Jones, A. and Yang, Y. Separation of nonpolar analytes using methanol-water mixtures at elevated temperatures. Anal. Chim. Acta, 2003, 485, 51–55.

Temperature Effects in Liquid Chromatography

73

105. Zarzycki, P.K., Wierzbowska, M., and Lamparczyk, H. The influence of temperature on the high performance liquid chromatographic separation of steroids using mobile phases modified by β-cyclodextrin. J. Pharm. Biomed. Anal., 1996, 14, 1305–1311. 106. Karlsson, A. and Charron, C. Reversed-phase ion-pair chromatography at a column temperature below 0˚C using three generations of Hypercarb as solid-phase. J. Chromatogr. A, 1996, 732, 245–253. 107. Tanaka, H., Zhou, X., and Masayoshi, O. Characterization of a novel diol column for high-performance liquid chromatography. J. Chromatogr. A, 2003, 987, 119–125. 108. Haidekker, A. and Huber, C.G. Ion chromatography on chelating stationary phases: Separation of alkali metals. J. Chromatogr. A, 2001, 921, 217–226. 109. Hosoya, K., Kimata, K., Araki, T., Tanaka, N., and Frechet, J.M.J. Temperaturecontrolled high-performance liquid chromatography using a uniformly sized temperature-responsive polymer-based packing material. Anal. Chem., 1995, 67, 1907–1911. 110. Kanazawa, H., Kashiwase, Y., Yamamoto, K., Matsushima, Y., Kikuchi, A., Sakurai, Y., and Okano, T. Temperature-responsive liquid chromatography. 2. Effects of hydrophobic groups in N-isopropylacrylamide copolymer-modified silica. Anal. Chem., 1997, 69, 823–830. 111. Berthod, A., He, B.L., and Beesley, T.E. Temperature and enantioseparation by macrocyclic glycopeptide chiral stationary phases. J. Chromatogr. A, 2004, 1060, 205–214. 112. Kazusaki, M., Shoda, T., Kawabata, H., and Matsukura, H. Influence of mobile phase composition on thermodynamic properties in high-performance liquid chromatography. J. Liq. Chromatogr. R.T., 2001, 24, 141–151. 113. Peyrin, E., Guillaume, Y.C., and Guinchard, C. Interactions between dansyl amino acids and human serum albumin using high-performance liquid chromatography: Mobile-phase pH and temperature considerations. Anal. Chem., 1997, 69, 4979–4984. 114. Takeuchi, T., Watanabe, Y., and Ishii, D. Role of column temperature in micro high performance liquid chromatography. J. High Resolut. Chromatogr., 1981, 4, 300–302. 115. Hirata, Y. and Sumiya, E. Temperature-programmed reversed-phase liquid chromatography with packed fused-silica columns. J. Chromatogr., 1983, 267, 125–131. 116. Bowermaster, J. and McNair, H. Microbore high-performance liquid chromatographic columns: Speed, efficiency, sensitivity and temperature programming. J. Chromatogr., 1983, 279, 431–438. 117. Jinno, K., Phillips, J.B., and Carney, D.P. Temperature-controlled high-speed microcolumn liquid chromatography. Anal. Chem., 1985, 57, 574–576. 118. McNair, H. and Bowermaster, J. HPLC column performance and temperature programming capabilities. J. High Resolut. Chromatogr., 1987, 10, 27–31. 119. Premstaller, A., Xiao, W., Oberacher, H., O’Keefe, M., Stern, D., Willis, T., Huber, C.G., and Oefner, P.J. Temperature-modulated array high-performance liquid chromatography. Genome Res., 2001, 11, 1944–1951. 120. Premstaller, A., Oefner, P.J., Oberacher, H., and Huber, C.G. Capillary array highperformance liquid chromatography of nucleic acids and proteins. Anal. Chem., 2002, 74, 4688–4693. 121. Andersen, T., Nguyen, Q.-N.T., Trones, R., and Greibrokk, T. Studies on the longterm thermal stability of particulate and monolithic poly(styrene-divinylbenzene) capillaries in reversed-phase chromatography. Analyst, 2004, 129, 191–197. 122. Liu, G., Djordjevic, N.M., and Erni, F. High-temperature open-tubular capillary column liquid chromatography. J. Chromatogr., 1992, 592, 239–247.

74

Advances in Chromatography

123. Djordjevic, N.M., Stegehuis, D., Liu, G., and Erni, F. Improved ultraviolet detection in high-temperature open-tubular liquid chromatography. J. Chromatogr., 1993, 629, 135–141. 124. Ryan, K., Djordjevic, N.M., and Erni, F. Temperature programming in open tubular liquid chromatography. J. Liq. Chromatogr. R.T., 1996, 19, 2089–2099. 125. Nyholm, L.M., Sjöberg, P.J.R., and Markides, K.E. High-temperature open tubular liquid chromatography coupled to atmospheric pressure chemical ionisation mass spectrometry. J. Chromatogr. A, 1996, 755, 153–164. 126. Nyholm, L.M. and Markides, K.E. Column preparation for reversed-phase hightemperature open tubular column liquid chromatography. J. Chromatogr. A, 1998, 813, 11–20. 127. Premstaller, A., Oberacher, H., and Huber, C.G. High-performance liquid chromatography-electrospray ionization mass spectrometry of single- and double-stranded nucleic acids using monolithic capillary columns. Anal. Chem., 2000, 72, 4386–4393. 128. Huber, C.G., Premstaller, A., Xiao, W., Oberacher, H., Bonn, G.K., and Oefner, P.J. Mutation detection by capillary denaturing high-performance liquid chromatography using monolithic columns. J. Biochem. Biophys. Methods, 2001, 47, 5–19. 129. Oberacher, H., Oefner, P.J., Hölzl, G., Premstaller, A., Davis, K., and Huber, C.G. Re-sequencing of multiple single nucleotide polymorphisms by liquid chromatography-electrospray ionization mass spectrometry. Nucleic Acids Res., 2002, 30, e67/1–e67/6. 130. Premstaller, A., Oberacher, H., Rickert, A., Huber, C.G., and Oefner, P.J. Multiplex analysis of single-nucleotide extension products on a 16-capillary, denaturing, highperformance liquid chromatography array. Genomics, 2002, 79, 793–798. 131. Oberacher, H., Premstaller, A., and Huber, C.G. Characterization of some physical and chromatographic properties of monolithic poly(styrene-co-divinylbenzene) columns. J. Chromatogr. A, 2004, 1030, 201–208. 132. Dulay, M.T., Quirino, J.P., Bennett, B.D., Kato, M., and Zare, R.N. Photopolymerized sol-gel monoliths for capillary electrochromatography Anal. Chem., 2001, 73, 3921–3926. 133. Vissers, J.P.C., Claessens, H.A., and Cramers, C.A. Microcolumn liquid chromatography: Instrumentation, detection and applications. J. Chromatogr. A, 1997, 779, 1–28. 134. Molander, P., Haugland, K., Hegna, D.R., Ommundsen, E., Lundanes, E., and Greibrokk, T. Determination of low levels of an antioxidant in polyolefins by large volume injection temperature-programmed packed capillary liquid chromatography. J. Chromatogr. A, 1999, 864, 103–109. 135. Molander, P., Thomassen, A., Lundanes, E., Fladseth, G., Thorud, S., Thomassen, Y., and Greibrokk, T. Determination of 1-(2-methoxyphenyl)-piperazine derivatives of airborne diisocyanates by packed capillary liquid chromatography with pre-column large-volume enrichment. J. Sep. Sci., 2001, 24, 947–954. 136. Molander, P., Holm, A., Lundanes, E., Hegna, D.R., Ommundsen, E., and Greibrokk, T. Temperature-promoted large-volume solute enrichment in column-switching miniaturized liquid chromatography: Determination of an antioxidant. Analyst, 2002, 127, 892–897. 137. Holm, A., Molander, P., Lundanes, E., and Greibrokk, T. Determination of rotenone in river water utilizing packed capillary column switching liquid chromatography with UV and time-of-flight mass spectrometric detection. J. Chromatogr. A, 2003, 983, 43–50.

Temperature Effects in Liquid Chromatography

75

138. Holm, A., Molander, P., Lundanes, E., and Greibrokk, T. Novel column oven concept for cold spot large volume enrichment in high throughput temperature gradient capillary liquid chromatography. J. Sep. Sci., 2003, 26, 1147–1153. 139. Andersen, T., Nguyen, Q.-N.T., Trones, R., and Greibrokk, T. Mesoporous polybutadiene-modified zirconia for high-temperature packed capillary liquid chromatography: Column preparation and temperature programming stability. J. Chromatogr. A, 2003, 1018, 7–18. 140. Xiang, Y., Yan, B., Yue, B., McNeff, C.V., Carr, P.W., and Lee, M.L. Elevatedtemperature ultrahigh-pressure liquid chromatography using very small polybutadiene-coated nonporous zirconia particles. J. Chromatogr. A, 2003, 983, 83–89. 141. Holm, A., Solbu, K., Molander, P., Lundanes, E., and Greibrokk, T. Sensitive biomonitoring of phthalate metabolites in human urine using packed capillary column switching liquid chromatography coupled to electrospray ionization ion-trap mass spectrometry. Anal. Bioanal. Chem., 2004, 378, 1762–1768. 142. Törnkvist, A., Sjöberg, P.J.R., Markides, K.E., and Bergquist, J. Analysis of catecholamines and related substances using porous graphitic carbon as separation media in liquid chromatography-tandem mass spectrometry. J. Chromatogr. A, 2004, 801, 323–329. 143. Törnkvist, A., Nilsson, S., Amirkhani, A., Nyholm, L.M., and Nyholm, L. Interference of the electrospray voltage on chromatographic separations using porous graphitic carbon columns. J. Mass Spectrom., 2004, 39, 216–222. 144. Koivisto, P. and Stefansson, M. Retention studies of sulphated glycosaminoglycan disaccharides on porous graphitic carbon capillary columns. Chromatographia, 2003, 57, 37–45. 145. Chen, J., Steenackers, D., and Sandra, P. Micro LC on cellulose triacetate (CTA): Influence of particle diameter and column temperature. J. Microcol. Sep., 1993, 7, 259–263. 146. Andersen, T., Holm, A., Skuland, I.L., Trones, R., and Greibrokk, T. Characterization of complex mixtures of polyglycerol fatty acid esters using temperature and solvent gradients in packed capillary LC. J. Sep. Sci., 2003, 26, 1133–1140. 147. Trones, R., Andersen, T., Greibrokk, T., and Hegna, D.R. Hindered amine stabilizers investigated by the use of packed capillary temperature-programmed liquid chromatography I. Poly((6-((1,1,3,3-tetramethylbutyl)-amino)-1,3,5-triazine-2,4-diyl)(2,2,6, 6-tetramethyl-4-piperidyl)imino)-1,6-hexanediyl((2,2,6,6-tetramethyl-4-piperidyl)imino). J. Chromatogr. A, 2000, 874, 65–71. 148. Trones, R., Andersen, T., Hegna, D.R., and Greibrokk, T. Hindered amine stabilizers investigated by the use of packed capillary temperature programmed liquid chromatography II. Poly-(N-β-hydroxyethyl-2,2,6,6-tetramethyl-4-hydroxypiperidyl succinate). J. Chromatogr. A, 2000, 902, 421–426. 149. Bruheim, I., Molander, P., Theodorsen, M., Ommundsen, E., Lundanes, E., and Greibrokk, T. Temperature-programmed packed-capillary liquid chromatographic separation of polystyrenes. Chromatographia, 2001, 53, S266–S270. 150. Trones, R., Andersen, T., and Greibrokk, T. Improved modification of a laser lightscattering detector for use in packed capillary high temperature liquid chromatography. J. High Resolut. Chromatogr., 1999, 22, 283–286. 151. Nordstrom, O., Molander, P., Greibrokk, T., Blomhoff, R., and Lundanes, E. Evaluation of temperature programming for gradient elution in packed capillary liquid chromatography coupled to electrochemical detection. J. Microcol. Sep., 2001, 13, 179–185.

76

Advances in Chromatography

152. Yoo, J.S., Watson, J.T., and McGuffin, V.L. Temperature-programmed microcolumn liquid chromatography/mass spectrometry. J. Microcol. Sep., 1992, 4, 349–362. 153. Andersen, T., Skuland, I.L., Holm, A., Trones, R., and Greibrokk, T. Temperatureprogrammed packed capillary liquid chromatography coupled to evaporative lightscattering detection and electrospray ionization time-of-flight mass spectrometry for characterization of high-molecular-mass hindered amine light stabilizers. J. Chromatogr. A, 2004, 1029, 49–56. 154. Sandvoss, M., Roberts, A.D., Ismail, I.M., and North, S.E. Direct on-line hyphenation of capillary liquid chromatography to nuclear magnetic resonance spectroscopy: Practical aspects and application to drug metabolite identification. J. Chromatogr. A, 2004, 1028, 259–266. 155. Skuland, I.L., Andersen, T., Trones, R., Eriksen, R.B., and Greibrokk, T. Determination of polyethylene glycol in low-density polyethylene by large volume injection temperature gradient packed capillary liquid chromatography. J. Chromatogr. A, 2003, 1011, 31–36. 156. Rosales-Conrado, N., Leon-Gonzalez, M.E., Perez-Arribas, L.V., and Polo-Diez, L.M. Determination of chlorophenoxy acid herbicides and their esters in soil by capillary high performance liquid chromatography with ultraviolet detection, using large volume injection and temperature gradient. Anal. Chim. Acta, 2002, 470, 147–154. 157. Molander, P., Thomassen, A., Kristoffersen, L., Greibrokk, T., and Lundanes, E. Simultaneous determination of citalopram, fluoxetine, paroxetine and their metabolites in plasma by temperature-programmed packed capillary liquid chromatography with on-column focusing of large injection volumes. J. Chromatogr. B, 2001, 766, 77–87. 158. Leon-Gonzalez, M.E., Perez-Arribas, L.V., Polo-Diez, L.M., Panis, C., and San Andres, M.P. Determination of triazine herbicides by capillary liquid chromatography with on-column focusing and temperature gradient. Anal. Chim. Acta, 2001, 445, 29–34. 159. Molander, P., Thommesen, S.J., Bruheim, I.A., Gundersen, T.E., Trones, R., Greibrokk, T., and Lundanes, E. Temperature-programmed packed capillary liquid chromatography separation with large volume on-column focusing injection of retinyl esters. J. High Resolut. Chromatogr., 1999, 22, 490–494. 160. Oefner, P.J. and Huber, C.G. A decade of high-resolution liquid chromatography of nucleic acids on styrene-divinylbenzene copolymers. J. Chromatogr. B, 2002, 782, 27–55. 161. Walcher, W., Oberacher, H., Troiani, S., Hölzl, G., Oefner, P.J., Zolla, L., and Huber, C.G. Monolithic capillary columns for liquid chromatography — Electrospray ionization mass spectrometry in proteomic and genomic research. J. Chromatogr. B, 2002, 782, 111–125. 162. Walcher, W., Toll, H., Ingendoh, A., and Huber, C.G. Operational variables in highperformance liquid chromatography-electrospray ionization mass spectrometry of peptides and proteins using poly(styrene-divinylbenzene) monoliths. J. Chromatogr. A, 2004, 1053, 107–117. 163. Andersen, T., Molander, P., Trones, R., Hegna, D.R., and Greibrokk, T. Separation of polyethylene glycol oligomers using inverse temperature programming in packed capillary liquid chromatography. J. Chromatogr. A, 2001, 918, 221–226. 164. Molander, P., Theodorsen, M., Lundanes, E., Soerenssen, D.M., and Greibrokk, T. Temperature effects on packed-capillary liquid chromatography of the x-ray contrast agent Iodixanol. J. Chromatogr. Sci., 2000, 38, 157–161.

Temperature Effects in Liquid Chromatography

77

165. Molander, P., Holm, A., Lundanes, E., Greibrokk, T., and Ommundsen, E. Separation of ceramides by sub-ambient temperature-assisted large volume injection in temperature-programmed packed capillary liquid chromatography. J. High. Resolut. Chromatogr., 2000, 23, 653–655. 166. Bruheim, I., Molander, P., Lundanes, E., Greibrokk, T., and Ommundsen, E. Temperature-programmed packed capillary liquid chromatography coupled to Fouriertransform infrared spectroscopy. J. High Resolut. Chromatogr., 2000, 23, 525–530. 167. Molander, P., Ommundsen, E., and Greibrokk, T. Separation of technical waxes by temperature-programmed packed-capillary liquid chromatography. Chromatographia, 2000, 51, 349–352. 168. Molander, P., Ommundsen, E., and Greibrokk, T. High temperature injection in capillary high temperature liquid chromatography. J. Microcol. Sep., 1999, 11, 612–619. 169. Molander, P., Trones, R., Haugland, K., and Greibrokk, T. Aspects and applications of non-aqueous high temperature packed capillary liquid chromatography. Analyst, 1999, 124, 1137–1141. 170. Molander, P., Gundersen, T.E., Haas, C., Greibrokk, T., Blomhoff, R., and Lundanes, E. Determination of retinoids by packed-capillary liquid chromatography with largevolume on-column focusing and temperature optimization. J. Chromatogr. A, 1999, 847, 59–68. 171. Trones, R., Tangen, A., Lund, W., and Greibrokk, T. Packed capillary high-temperature liquid chromatography coupled to inductively coupled plasma mass spectrometry. J. Chromatogr. A, 1999, 835, 105–112. 172. Trones, R., Andersen, T., Hunnes, I., and Greibrokk, T. Modified laser light-scattering detector for use in high temperature micro liquid chromatography. J. Chromatogr. A, 1998, 814, 55–61. 173. Sheng, G., Shen, Y., and Lee, M.L. Elevated temperature liquid chromatograpy using reversed-phase packed capillary columns. J. Microcol. Sep., 1997, 9, 63–72. 174. Trones, R., Iveland, A., and Greibrokk, T. High-temperature liquid-chromatography on packed capillary columns with nonaqueous mobile phases. J. Microcol. Sep., 1995, 7, 505–512. 175. Hirata, Y. and Jinno, K. High resolution liquid chromatography with a packed micro glass capillary column. J. High Resolut. Chromatogr., 1983, 6, 196–199.

3

Lipophilicity Measurements by Liquid Chromatography Simion Gocan Babes-Bolyai University, Cluj-Napoca, Romania

Gabriela Cimpan Sirius Analytical Instruments Ltd., Forest Row, U.K.

John Comer Sirius Analytical Instruments Ltd., Forest Row, U.K.

CONTENTS 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9

Introduction ....................................................................................................80 Lipophilicity: Concept and Definitions .........................................................81 The Uses of Lipophilicity ..............................................................................81 logP, logD, and Other Terms for Expressing Lipophilicity ..........................82 The Effect of Chemical Reactions on Partitioning .......................................83 Octanol and Other Partition Solvents ............................................................87 Calculations Involving LOGP ..........................................................................90 logP Measurement by Shake-Flask ...............................................................91 Lipophilicity Measurement by High-Performance Liquid Chromatography (HPLC)...............................................................................94 3.9.1 HPLC with Octanol-Coated Columns ...............................................96 3.9.2 Other Methods using Octanol-Coated Columns .............................100 3.9.3 ELogP Method .................................................................................100 3.9.4 Reversed-Phase HPLC (RP-HPLC).................................................102 3.9.5 The extrapolation method, logkw .....................................................102 3.9.6 Isocratic logk Values ........................................................................105 3.9.7 The Significance of Slope S ............................................................106 3.9.8 Correction for Ionization .................................................................107 3.9.9 The Influence of Stationary Phase on Lipophilicity Measurements...................................................................................108 3.9.10 Mobile Phase Composition and Lipophilicity.................................112 79

80

Advances in Chromatography

3.9.10.1 Organic Modifier...............................................................112 3.9.10.2 Buffer in the Mobile Phase ..............................................114 3.9.10.3 Ionic Strength.................................................................... 114 3.9.10.4 Masking Agents ................................................................115 3.9.10.5 Association in Water.........................................................116 3.9.10.6 Association in the Organic Phase.....................................116 3.9.10.7 The Effect of Temperature ...............................................116 3.9.10.8 Gradient RP-HPLC...........................................................117 3.9.11 Chromatographic Hydrophobicity Index (CHI) ..............................117 3.9.12 Abraham-Type Equations for RP-HPLC .........................................125 3.10 Lipophilicity Estimation by Thin-Layer Chromatography .........................127 3.10.1 Relationship between RM and Molar Fraction X of the Organic Modifier in the Mobile Phase ............................................128 3.10.2 Influence of pH ................................................................................131 3.10.3 Influence of Mobile Phase Composition on RM0 Value..........................................................................................132 3.10.4 Influence of Stationary Phase on RM0 Value ...................................133 3.10.5 Specific Hydrophobic Surface Area of the Solute ..........................134 3.10.6 Relationship between Intercepts (RM0) and Slopes (a1) ..................135 3.10.7 RM Determination and Correlation with Partition Coefficients, logP .............................................................................143 3.11 Chromatographic Determination of Hydrophilic–Lipophilic Balance........145 3.11.1 Drug Lipophilicity............................................................................146 3.11.2 Miscellaneous...................................................................................148 3.11.3 Studies using Both RP-HPLC and RP-HPTLC ..............................149 3.12 Other Chromatographic Methods ................................................................154 3.12.1 Micellar Liquid Chromatography (MLC)........................................154 3.12.2 Reversed-Phase Ion Pair Chromatography (RP-IPC)......................155 3.12.3 Countercurrent Chromatography (CCC) .........................................156 Acknowledgments..................................................................................................157 Glossary .................................................................................................................157 References..............................................................................................................158

3.1 INTRODUCTION Many methods have been studied in an attempt to find a suitable model for the difficult, classical partition experiment in a shake-flask vessel. Liquid chromatography provided a particular approach due to the possibility of using a liquid mobile phase and a liquid or liquid-like stationary phase. This chapter will review different liquid chromatographic methods for measuring lipophilicity, including liquid-liquid chromatography, reversed-phase liquid chromatography (isocratic, gradient, and micellar methods) and counter current chromatography. This chapter will not review liquid chromatography with stationary phases that mimic the biological environment such as affinity chromatography or immobilized artificial membrane chromatography, as these form a distinct group of methods and

Lipophilicity Measurements by Liquid Chromatography

81

comprehensive reviews have been published recently [1–5]. Neither will it review micellar electrokinetic chromatography (MECK) or microemulsion electrokinetic chromatography (MEECK), as these techniques are based on capillary electrophoresis and details can also be found in a recent review [6].

3.2 LIPOPHILICITY: CONCEPT AND DEFINITIONS Lipophilicity is a general concept that characterizes the affinity of molecules for lipid-like structures. A related term, hydrophobicity, refers to water’s “dislike” of a molecule. The two terms are often used interchangeably in the literature but it has been agreed that hydrophobicity relates to the forces between the nonpolar groups in an aqueous environment. Water’s internal molecular organization excludes nonpolar moieties; therefore, hydrophobicity is only one component of the more general term lipophilicity [7]. For many years, confusion existed between lipophilicity and hydrophobicity and both terms were used to characterize molecules that show a preference for a nonpolar phase. The definitions agreed by IUPAC for these terms are [8]: lipophilic, adj., -ity, n.: Having an affinity for fat and high lipid solubility: a physicochemical property which describes a partitioning equilibrium of solute molecules between water and an immiscible organic solvent, favouring the latter, and which correlates with bioaccumulation. hydrophobic, adj., -ity n.: Describing the character of a molecule or atomic group which is insoluble in water, or resistant to wetting or hydration. Hydrophobic interaction: the tendency of hydrocarbons (or of lipophilic hydrocarbon-like groups in solutes) to form intermolecular aggregates in an aqueous medium, and analogous intramolecular interactions. The name arises from the attribution of the phenomenon to the apparent repulsion between water and hydrocarbons. However, the phenomenon ought to be attributed to the effect of the hydrocarbon-like groups on the water-water interaction. Hydrophobicity contributes towards lipophilicity but the two terms are not synonymous. Hydrophobic interaction is a different process and can be found as a driving force in lipophilicity studies.

3.3 THE USES OF LIPOPHILICITY Since the pioneering work of Overton and Meyer at the beginning of the 20th century [9, 10], when compound partition in olive oil was found to correlate well with biological activity, much work has been done to standardize experimental conditions and to understand molecular influences on lipophilicity. Applications of lipophilicity can now be found in the pharmaceutical industry, in environmental applications, toxicology, and any other field involving a potential biological effect on a living

82

Advances in Chromatography

organism. It has been widely used in quantitative structure-activity relationships (QSAR) and in drug and pesticide design as well as in toxicology studies either in the pharmaceutical or environmental sciences. Lipophilicity has been used in the following areas of study: •

• • • •

• • • •

Estimation of solubility — lipophilicity is used (in conjunction with other parameters) to estimate the aqueous solubility by a range of semi-empirical equations [11] Environmental applications [12] General [13] Pesticide occurrence in the environment [14] Drug discovery in the pharmaceutical industry. When creating a new molecule with potential biological activity, knowledge of its lipophilicity helps to build an understanding of its absorption, distribution, metabolism, excretion, and toxicity (ADMET) [15] Oral absorption [16] Serum albumin binding [17] Permeability [18] CNS penetration [19]

According to EU regulations, lipophilicity (expressed as logPo/w) is a required value for any new commercially available chemical and a liquid chromatographic method for this measurement has been approved [20, 21].

3.4 logP, logD, AND OTHER TERMS FOR EXPRESSING LIPOPHILICITY Lipophilicity is generally expressed by the partition between water and a waterimmiscible solvent. Traditionally, the partition coefficient was defined as the ratio between the equilibrium concentration of a compound (C) in a water-immiscible solvent (s) and its concentration in water (w) [22, 23]. In 1891 Walther H. Nernst was appointed as Extraordinary Professor at Göttingen University. During the same year, he developed the distribution law: If two liquids (or solids) a and b are partially immiscible and if there is a third component i present in both phases that behaves individually as an ideal solute (if it is sufficiently dilute), the ratio of its concentrations (x) is independent of the individual values of x. xia = constant xib

(3.1)

Nernst’s law of distribution is described today by the equation for the partition coefficient of a solute between two immiscible phases:

Lipophilicity Measurements by Liquid Chromatography

P=

[C ]s [C ]w

83

(3.2)

The term C refers to the neutral form of the molecule and the square brackets that enclose it denote that its concentration is at equilibrium. According to IUPAC recommendations from 1993, the partition coefficient Po/w or Ko/w is a “measure of lipophilicity by determination of the equilibrium distribution between octan-1-ol and water, as used in pharmacological studies and in the assessment of environmental fate and transport of organic chemicals.” From the same source, the partition coefficient is the ratio of the distribution of a substance between two phases when the heterogeneous system (of two phases) is in equilibrium; the ratio of concentrations (or, strictly speaking, activities) of the same molecular species in the two phases is constant at constant temperature. The partition coefficients most frequently used in acute toxicology are lipid/water and octan-1-ol/water distributions. From now on in this review the term “octanol” will refer to octan-1-ol unless otherwise stated [8]. Measurable P values are usually in the range 10–2 to 107. As these values extend over several orders of magnitude, they may by expressed logarithmically (as log10) and logP is often used instead of P for practical reasons. If the molecule is ionisable, the pH of the aqueous phase will influence the concentrations of the ionised and neutral forms of the molecule.

D=

∑ [C ] ∑ [C ]

s

(3.3)

w

D is the distribution coefficient of a compound at a defined pH and will take into account both neutral and ionized species present in the aqueous and in the organic phase. Partition coefficients have been widely used to define the lipophilic character of compounds [24] and logP and logD have been extensively included in quantitative structure-activity relationships (QSAR) [25].

3.5 THE EFFECT OF CHEMICAL REACTIONS ON PARTITIONING The dissociation of ionizable compounds in the aqueous phase is the most common reaction that accompanies the partition in an octanol-water system. Two simple cases, a monoprotic acid and monoprotic base, will illustrate how dissociation influences the partition. The ionization of a monoprotic acid in solution may be written as, HA ⇔ H+ + A–

(3.4)

84

Advances in Chromatography

Organic solvent HA

Water

HA

H+ + A−

FIGURE 3.1 Solvent–water system and the species partition for a weak acid.

The neutral species, HA, will be able to partition in the lipophilic solvent, while the ionized species will stay in water. Therefore, the organic solvent will contain only the HA form, while the aqueous phase will contain both HA and A– forms. In this case, the distribution coefficient can be expressed as follows:

D=

[ HA ]0 [ HA ]w + ⎡⎣ A – ⎤⎦w

(3.5)

where the indices “o” and “w” stand for organic solvent and water, respectively, and the square brackets symbolize that the concentrations of species at equilibrium is taken into consideration. The ionization process in the aqueous phase is governed by the ionization constant,

Ka =

⎡⎣ A – ⎤⎦ ⎡⎣ H + ⎤⎦ [ HA ]w w w ⇒ ⎡⎣ A – ⎤⎦ = K a w ⎡⎣ H + ⎦⎤ [ HA ]w w

(3.6)

The distribution coefficient will have the expression: P ⎡⎣ H + ⎤⎦ P ⎡⎣ H + ⎤⎦ P = = D= ≈ Ka Ka ⎛ ⎡⎣ H + ⎤⎦ + K a Ka ⎞ 1 + [ HA ]w ⎜⎜ 1 + ⎡ + ⎤ ⎟⎟ ⎡⎣ H + ⎤⎦ ⎝ ⎣H ⎦ ⎠

[ HA ]0

(3.7)

The logarithm of this approximation is, logD = logP + log[H+] – logKa

(3.8)

logD = logP + pKa – pH

(3.9)

which may be written as:

S

O

NH3+

N

H

XH2+

H H

O

XH0

H

XH

N CH3

±

COO−

NH3+

N

CH3

HO

O

O−

O

H H NH

S

CH3

O

H

X−

NH2

N

N CH3

COO−

−

CH3 X

0

25

50

75

100

0

25

50

75

100

1

1

2

2

3

XH2+

3

XH2+

4

4 6 7 8

9

6

7

8

9 pH (concentration scale)

5

XH0

Distribution of species

pH (concentration scale)

5

XH±

Distribution of species

FIGURE 3.2 The distribution of species for Ampicillin (a zwitterion) and for morphine (an ampholyte).

HO

O

HO

S

CH3

NH

O

CH3

NH+

Morphine (an ampholyte) pKas = 9.26, 8.17

HO

O

HO

Ampicillin (a zwitterion) pKas = 7.14, 2.55

O

NH

H H

COOH

CH3 XH2+

% Species % Species

CH3

10

10

11

11

12

X−

12

−

X

13

13

Lipophilicity Measurements by Liquid Chromatography 85

86

Advances in Chromatography

A similar equation can be derived for monoprotic bases, whose ionization may be written as, B + H+ ⇔ BH+

(3.10)

Considering that the only reaction is the base ionization in water, the organic solvent will contain only the neutral base, B, while the aqueous phase will contain both neutral and ionized base, B and BH+, and the following approximation may be derived: logD = logP – pKa + pH

(3.11)

In many applications concerned with biological activity of different organic compounds (e.g., drug discovery, toxicology studies), it is considered that the distribution coefficient at pH = 7.4, the pH of the blood, is sufficient to give an estimate for the in vivo human absorption of compounds. However, the distribution of species can change dramatically with the pH, as in the gastrointestinal tract where pH can vary from 1 to 8 depending on the fed–fasted state and the part of the tract considered (Figure 3.3 and Figure 3.4). The relationship between logP and logD has been derived for amphoteric compounds as a function of the two dissociation constants, pKa1 and pKa2 [26]: logP = logD + log (1 + 10pH – pKa1 + 10pKa2 – pH)

(3.12)

Figure 3.5 shows the lipophilicity profiles for partitioning of diprotic molecules. The curves were drawn using the “What-if” feature of Sirius’ RefinementPro software for simulated samples with aqueous pKa values of 8 and 5. Although the pKa values of real samples may differ, the shapes and relative positions of the curves will follow the examples below for different classes of molecule. A detailed description of lipophilicity-pH profiles for acids, bases, and ampholytes can be found in [27]. 2 1

LogD

0 −1 −2 −3 −4 −5

0

1

2

3

4

5

6 7 pH

8

9 10 11 12 13

FIGURE 3.3 Lipophilicity profile for morphine, pKas = 9.26, 8.17; logP = 0.90.

Lipophilicity Measurements by Liquid Chromatography

5

87

Desipramine pKa = 10.14 Diclofenac pKa = 3.99

4

Diphenhydramine pKa = 8.26

3 Phenobarbital pKa = 7.43

LogD

2

Triamterene pKa = 3.92

Nifuroxime pKa = 10.56

1

0 −1 −2 0

2

4

6

8

10

12

14

pH

FIGURE 3.4 logD = f(pH) for acids (negative slope) and bases (positive slope).

If a compound is suspected to form dimers in one of the phases, this process has to be taken into consideration when defining logD [28]. Applications in the pharmaceutical industry tend to correlate logP/logD values with drug absorption in vivo. In many cases, a good correlation is obtained if octanol is used for the partition measurements and the logD values measured at pH 5 to 7.4 (physiological pH values) provide more relevant information compared with logP [29]. Although the IUPAC recommended notation for partition coefficient is K, the large majority of publications still use logP and logD to differentiate between the partition of the neutral and charged molecule. For this reason, we will keep the same notations throughout this chapter.

3.6 OCTANOL AND OTHER PARTITION SOLVENTS Octanol was not the first partition solvent used as a model for the in vivo process. Olive oil was one of the first investigated lipophilic solvents but differences between batches and difficult standardization led to the use of other solvents such as paraffin oil. Hansch later proposed the use of octanol and the largest database in the world for partition coefficients is based on the octanol–water system [30–32]. Today, octanol–water partition coefficients (logPo/w) are established as a significant physical property correlated with biological activity. An important reason is the ability of octanol to form clusters in an aqueous environment, thus providing a

10

H2Z = H2Z* Z = Z* logPH2Z > logPZ

H2Z = H2Z* Z = Z* logPH2Z = logPZ

Less octanol, low ionic strength

HX = HX*

Less octanol, low ionic strength

B = B*

Less octanol, low ionic strength

H2A = H2A*

H2Z = H2Z* Z = Z* logPZ > logPH2Z

More octanol, high ionic strength

HX = HX* X = X* logPHX > logPX

Special case-very rare

B = B* BH = BH* logPB = logPBH

Special case-very rare

H2A = H2A* HA = HA* logPH2A = logPHA

Most zwitterionic molecules are hydrophilic, logD values are usually negative

More octanol, high ionic strength

H2X = H2X* HX = HX* logPHX > logPH2X

More octanol, high ionic strength

B = B* BH = BH* logPB > logPBH

More octanol, high ionic strength

2 0 H2A = H2A* −2 HA = HA* −4 logP H2A > logPHA −6 pH −8 2 4 6 8

FIGURE 3.5 Lipophilicity profiles for diprotic molecules.

4: Zwitterion One acid, one base group (acid pKa < basic pKa), HZ is the “neutral” species but is zwitterionic: HZ± pKa equations: H + Z = HZ, H + HZ = H2Z

3: Ordinary ampholyte One acid, one base group (acid pKa > basic pKa), HX is the neutral species pKa equations: H + X = HX, H + HX = H2X

2: Base Two base groups, B is the neutral species pKa equations: B + H = BH, BH + H = BH2

1: Acid Two acid groups, H2A is the neutral species pKa equations: H + A = HA, H + HA = H2A

88 Advances in Chromatography

Lipophilicity Measurements by Liquid Chromatography

89

model for the partition into a real biological membrane. Octanol has an amphiprotic character and, as well as displaying lipophilic interaction, it can form hydrogen bonds. It can therefore interact with with neutral or ionized molecules and with ionpairs [33, 34]. The partition of a solute between water and octanol must take into account the interactions between the solute and the water present in the organic phase, as octanol can dissolve a significant amount of water (2.3 to 2.5 M). Experimental evidence exists that amphiprotic and aprotic solutes can either “drag” water molecules or “exclude” them from the octanol phase. It appears that besides octanol–water clusters containing a variable number of octanol molecules around water, there also exist an octanol–water complex having a highly organized structure with a water molecule in the center and four octanol molecules around it bound by hydrogen bonds. Any interaction between the solute and octanol is likely to disturb this arrangement [35, 36]. Octanol and water can both participate in hydrogen bonding because both have hydroxyl groups and, therefore, a considerable amount of water can dissolve in octanol. The distribution of a compound between octanol and water is mainly dependent on the differences between the interactions of the lipophilic part of the molecule with water and with the organic solvent rather than the interaction of the polar and hydrogen bonding groups [37]. The extent to which the octanol model is the right choice to model human, animal, or plant membranes, as well as environmental distribution of toxins, still has to be established. Nevertheless, a very large number of correlations between the logPo/w value and biological activity have been reported during the last 50 years. Although octanol is commonly used in lipophilicity studies, the partition in other solvents is sometimes a better model for the in vivo process. For example, it is known that hydrogen bonding plays little role in transport across the blood–brain barrier, which is better modelled by alkane–water systems. Many alkanes have been used including hexane, cyclohexane, heptane, isooctane, dodecane, and hexadecane. Other investigated systems include chloroform–water (H-donor) and di-n-butylether (H-acceptor) [38]. A correlation between the partition coefficient in heptane/water and absorption from the cerebrospinal fluid was reported in 1959, where it was noted that the oxidative dealkylation of foreign N-alkylamines by rabbit liver microsomes appeared to be limited to compounds that were lipid soluble, as shown by high chloroform to water partition coefficients at physiological pH. Since the microsomal hydroxylation of aromatic compounds also appeared to be limited to lipid soluble substances, an intracellular fat-like boundary was suggested that separates normally occurring polar substances from the highly nonspecific microsomal enzymes [39]. The choice of four solvents has been proposed for measuring partition coefficients, dependent upon the specific application of the data obtained. These are octanol (amphiprotic), alkanes (neutral), chloroform (proton donor), and propylene glycol dipelargonate (PGDP) (proton acceptor) [40]. Octanol-water systems have been used to model serum protein binding as well as the interaction with membranes formed mainly of proteins. Biological membranes can have very different hydrogen bonding properties ranging from neutral — when

90

Advances in Chromatography

the partition in an alkane may be a better model — to structures comprising mainly proton donor groups, modelled by chloroform [41]. They can also comprise mainly proton acceptor groups as in phospholipid membranes, where the partition in propylene glycol dipelargonate may be a better model [41, 42]. According to Buchwald and Bodor, the main reasons for using octanol-water for lipophilicity estimations are: • • • • • •

Large database of experimental data Octanol has low vapor pressure at room temperature and is therefore suitable for different experimental methods Octanol has low UV absorption and therefore does not interfere in the measurement of solute concentration Octanol retains some of the hydrogen-bonding ability of water and is able to dissolve a significant amount of water There are many methods available that reasonably model the partition in the octanol-water system (e.g., RP-HPLC) There are many correlations between the logP value of a compound and its biological activity and therefore it is reasonable to assume that compounds with certain logP values will be absorbed in vivo [43]

Other systems, like oil-water or hexadecane-water, may be better models for the blood–brain barrier than octanol-water. However, the chromatographic applications are limited and the rest of this chapter will describe methods that model the partition in an octanol-water system.

3.7 CALCULATIONS INVOLVING

LOGP

There are a number of methods for calculating logP from chemical structures and for using logP values in other mathematical models and predictions. These have been reviewed [44] and are summarized below: • • • • •

Group contribution models: Hansch [45, 46], Rekker [47, 48], Fujita [31] Atomic contribution or surface area models [49] Molecular methods [50] Linear solvation energy relationships (LSER) [51] Commercially available software for logP calculation: CLogP [52–54], ACD [55], ADME Boxes [56], KowWin, ALogP, IAlogP, XLOGP [57]

The link between the partition in different water-solvent systems can be achieved by applying the Collander equation or by free energy solvation relationships. In 1951, Collander observed the linear relationship between the partition coefficients obtained in different alkanol-water systems, 1 and 2 [58]: logP1 = a logP2 + b

(3.13)

Lipophilicity Measurements by Liquid Chromatography

91

where a and b are coefficients. The Collander equation looked very promising in the beginning because it allowed a transfer from a measured partition coefficient in a defined partitioning system to another one. In practice, the Collander equation has limited application to solutes with similar hydrogen bonding characteristics [34]. Another attempt to generalize the theory of lipophilicity is seen in the linear solvation energy relationship (LSER) defined by Abraham et al. (see RP-HPLC section). LSER equations have been used to compare published partition coefficients in octanol-water (613 compounds), hexadecane-water (370 compounds), alkanewater (200 compounds) and cyclohexane-water (170 compounds) [59]. The main factors governing the partitioning of a compound in the octanol-water system are its dipolarity/polarizability and hydrogen bond basicity, which orient the compound towards water, and its molecular size, which orients the compound towards octanol. In alkane-water systems, polarizability and all forms of hydrogen bonding orient the compound towards water while molecular size orients the compound towards alkane.

3.8 logP MEASUREMENT BY SHAKE-FLASK The shake-flask method has changed very little over the decades and is still the reference for any new method of measuring lipophilicity. The compound is mixed with two immiscible solvents until equilibrium is reached, then the two phases are separated as well as possible and the solute concentration is measured in both of them. If the concentration measurement is difficult in one of the phases, the compound concentration can be measured in the other phase and the concentration in the unmeasured phase obtained by the difference from the total compound quantity. It is essential to mutually presaturate both phases for a successful experiment. Water is usually the lower layer (e.g., when using octanol or cyclohexane) but sometimes can be the upper layer (e.g., when using chloroform). The subscript “o/w” refers to octanol-water partitioning, as in Po/w. In many cases, the equilibrium is established in 5 to 30 minutes but sometimes a longer time is required. Although, in theory, the equilibration time is infinite, the system is considered as reaching equilibrium when two consecutive measurements from both phases show no detectable concentration change. However, this approach requires care as any sampling from one phase will modify the compound concentration and, therefore, the partition process. Also, the shaking of the two immiscible solvents can produce stable emulsions, making the two phases very difficult to separate. Shake-flask experiments can be prone to serious errors and the published logP data should be reviewed carefully, as up to 1 unit span range is indicated for many compounds with reported logPo/w values in the range 1.2 to 5.6 [20]. Although the need to measure the partition coefficients exists in many industrial areas, the pharmaceutical industry has the largest number of samples. The last decade has seen the development of the high-throughput methods for screening compound libraries. Several automated approaches have been developed since the shake-flask approach was decided to be the reference method in industry for logP/logD measurements. An automated shake-flask method for high-throughput logD measurements in an octanol-water system has been described by Valkó [60]. This micro

92

Advances in Chromatography

shake-flask method is performed in capped vials and the dual phase system is sampled by a needle piercing the vial cap. The method was developed for the measurement of logD values at pH = 7.4. A similar system has been described for samples in 96 well plates, where the entire liquid handling is performed by a robotic system. Samples are taken from the octanol and aqueous phase and are consequently analysed by reversed-phase highperformance liquid chromatography (RP-HPLC). The system was validated for a set of chemically diverse compounds as 1 mg/mL stock solutions in octanol in 96 well plates. Liquid handling was performed by a Beckman Biomek 2000 robot (Beckman Instruments, Fullerton, CA, USA). The aqueous phase was 400 μL phosphate buffer (pH = 7.4) and the organic phase was 200 μL octanol. A quantity of 200 μL of the solute stock solutions was transferred into each of the wells. Different methods for creating the dual-phase system were investigated, all of them leading to the same analyte concentration and the same final volumes. The plates were sealed and placed in a reciprocal shaker for 30 minutes, then centrifuged. Aliquots were taken from both phases by sampling at different heights and analysed by isocratic and gradient RP-HPLC on an ODS column. Similar samples were taken from both phases and analysed in the same RP-HPLC conditions from the solution resulting from the other two procedures, the Eppendorf shake-flask and the manual plate method. Details of the methods are shown in Table 3.1. This automated method was fully validated for logD values in the range 2 to 4 against the logD values obtained from the Eppendorf shake-flask method and against the manual plate method [61]. The system described above can measure 24 to 96 samples per day. The methods were validated for equilibration time and reproducibility within the same method. Very good correlations (r2 = 0.996 to 1.000) were obtained between the data produced by the three methods. A commercially available instrument using automated shake-flask method is manufactured by Analiza Inc. [62] Octanol–buffer solutions at different pH values are preprepared and the sample is injected into the dual-phase system. When the equilibrium is established, the vials are centrifuged for the phase separation. Aliquots are sampled from both phases and analysed by UV, nitrogen detector, or LS-MS. This system can measure up to 200 samples per day with a typical logD range from 3 to 4. A micro shake-flask experiment was reported for derivatives of purine and pyrimidine nucleosides. The compounds were stored as 0.5 mg/mL DMSO stock solutions. For experiments, 20 μL aliquots of the stock solutions were dissolved in 1 mL buffer saturated octanol and then the compound was partitioned between the octanol and 1 mL aqueous phosphate buffer phase (pH = 7.0) in a Mixxor apparatus (Genex Corp., Gaithersburg, MD, USA). Instead of shaking, this apparatus uses piston strokes. The mass transfer after 6 piston strokes is equivalent to 40 shakes in a separatory funnel. The phases were separated by centrifugation and aliquots from both phases were analysed by HPLC on an ODS column and acetonitrile-phosphate buffer (pH 7.0) mobile phase [63]. Compiling the data from personal experience and from the numerous published papers, the drawbacks are summarized here for the shake-flask method:

Lipophilicity Measurements by Liquid Chromatography

93

TABLE 3.1 Experimental Settings for Three Methods for Measuring the Distribution Coefficients, logD Eppendorf Method Method

shake-flask

Sampling Samples (stock solutions)

manual 1 mg/mL analyte in octanol (1.8 mL Eppendorf vials) octanol-phosphate buffer pH 7.4 100, 300, 400 μL stock solution mixed with 400, 200 or 100 μL octanol, then 500 μL aqueous phosphate pH 7.4 was added. The procedure was done with the same volumes, but starting with aqueous solutions for the hydrophilic compounds, then adding octanol. 3

Partition system Experimental

Manual Plate Method modified shakeflask in 96 well plates manual

octanol-phosphate buffer pH 7.4 200, 300 or 400 μL stock solution plus 200, 100 or 0 μL octanol (or phosphate buffer pH 7.4 solutions), then 400 μL of the other phase was added.

Automated Method modified shake-flask in 96 well plates automated 1 mg/mL analyte in octanol (96 well plates) octanol–phosphate buffer pH 7.4 400 μL phosphate buffer (100 mM pH 7.4) were pipetted in each well; 0, 100, 200 μL octanol added to each well, then 400, 200 and 300 μL respectively, analyte stock solution was also added

Repeat experiment/ 3 3 compound Compound partition reciprocal shaker, 5 min reciprocal shaker, 30 min Phase separation centrifugation centrifugation centrifugation Amount of sample 10 μL of organic and 80 10 μL of organic various injected in the μL of aqueous phase and 80 μL of HPLC system aqueous phase Analyte concentration measured by: Isocratic RP-HPLC: 100 mm long column, packed with 2 μm TKS gel Super-ODS (TosoHaas, Stuttgart, Germany); mobile phase: acetonitrile in 25 mM phosphate buffer with 0.2% triethylamine, pH 3.0; acetonitrile concentration adjusted to obtain retention times about 5 min; flow rate 1 mL/min UV detection, 260 nm Fast Gradient RP-HPLC: 50 mm long column packed with 2 μm TKS gel Super-ODS (TosoHaas, Stuttgart, Germany); mobile phase A: 0.07% (v/v) formic acid in acetonitrile/water (95:5) mobile phase B: 0.1% (v/v) formic acid in water The gradient starts with 100% B up to 100 A in 3.5 min and held there for 1 min. The total cycle time was 5.5 min

94

Advances in Chromatography

1. The formation of extremely stable emulsions that cannot be separated can prevent the accurate measurement of the analyte concentration in both phases. Complete separation is desirable as any droplets of octanol in water can contain significant quantities of the investigated compound. Even when octanol/water emulsions are quite stable, the two phases usually separate after 10 minutes of centrifugation. It takes a longer time to separate the phases when using propylene glycol dipelargonate (PGDP) as the partition solvent. 2. Measurement of extreme logP/logD values (e.g., for very hydrophilic or very hydrophobic compounds) requires a disproportionate volume ratio of the aqueous and solvent phase. In this situation, it is very difficult to sample a reasonable amount of solution from the phase with the smaller volume and automatic sampling is almost impossible. 3. The sample concentration must be less than the critical micelle concentration. 4. The sample concentration must be set below the aqueous solubility limit. 5. Some compounds, particularly the lipophilic compounds, can adhere to the surface of the vessel. 6. Some compounds can form micelles in the aqueous phase, act as surfactants that concentrate at the interface between the two liquids, or form foams. These compounds cannot be measured by a shake-flask method or any other method requiring agitation. 7. Compounds must have high purity. Measuring the exact concentration of compound in both phases is crucial. If a separation method is used for the quantification, the method is also applicable for impure compounds. When the volume ratio between the two solvent phases is high, an aliquot may be sampled from the larger volume before a suitable solvent is added to homogenously mix the two formerly immiscible phases, after which an aliquot is sampled from the resulting solution.

3.9 LIPOPHILICITY MEASUREMENT BY HIGHPERFORMANCE LIQUID CHROMATOGRAPHY (HPLC) Chromatography has been intensively studied because of its remarkable capacity for separating compounds from the most complex mixtures. Assuming constant temperature, chromatographic separation is determined by three main properties: the chemical structure of the solute, the physicochemical properties of the mobile phase and of the stationary phase, respectively [64]. Of all chromatographic methods, liquid chromatography was of particular interest for lipophilicity measurement because of the similarity between the solute partition in a chromatographic system, having a liquid mobile phase and a liquid-like stationary phase, and the solute partition in a dual liquid phase environment.

Lipophilicity Measurements by Liquid Chromatography

95

What determines whether compound will spend more time in the mobile phase or in the stationary phase of a chromatographic system? The answer lies in the interactions between the solute and the two phases. These interactions include van der Waals forces, hydrogen bonding, interactions between dipoles, electron donor or acceptor interactions, and solvophobic interactions. Liquid chromatography is a more complex system than gas chromatography (GC), where the intermolecular interactions in the mobile phase are minimized by the gaseous state of the mobile phase. Therefore, creating model equations for the partition in a liquid chromatographic system is a complex task that has to take into account many variables. Reversed-phase liquid chromatography (RP-LC) is the generic name of a range of liquid chromatographic methods in which the stationary phase is less polar than the mobile phase. RP-LC is usually employed for lipophilicity estimations as it is easier to immobilize a nonpolar, viscous phase on the stationary phase than to use it as a liquid. RP-LC has other advantages over the traditional shake-flask method: • • •

• • • •

The method requires a much smaller amount of sample It is simple and rapid It is relatively insensitive to impurities, as it is a separation method — however, the impurity peak can have a larger area than the main compound if the impurity has a better UV absorption than the sample It can be used for a larger lipophilicity scale than is practically possible in shake-flask experiments Other practical problems encountered in shake-flask experiments are eliminated, like the formation of stable emulsions Universal detection methods can be used for the large majority of samples There is a linear relationship between the chromatographic retention parameter and the organic solvent concentration in the mobile phase

Liquid–liquid partition chromatography was pioneered by Martin and Synge, who received a Nobel prize for the separation of amino acids on a silica stationary phase with an aqueous mobile phase [65]. They also provided the first theoretical explanation of chromatographic separation by introducing the concept of theoretical plates, based on the plates used in column fractionation by distillation. Solvophobic interactions are an important part of the retention in RP-LC. Retention in liquid–liquid chromatography is a complex process that must take into account three major equilibria: the solute partition between the two liquid phases, one stationary and the other one mobile; the solute adsorption on the solid support; and the solute absorption at the liquid–liquid interface. Secondary processes such as diffusion into the pores of the solid support and flow patterns through the column are also present and can be a source for retention problems [66]. Thus, the retention is a function of the partition coefficients and the volumes of the two liquid phases. This is reflected by the retention factor (k) in an isocratic system,

96

Advances in Chromatography

k=

CSVS V =K S C M VM VM

(3.14)

where CS and CM are the solute concentration in the stationary and mobile phase, respectively, and VS and VM are the volume of stationary phase and mobile phase, respectively. The value K is the partition coefficient of the solute between the two phases. The retention in a chromatographic system can be expressed by the retention volume (VR) or by the retention time (tR): VR = VM (1 + k)

(3.15)

tR = tM (1 + k)

(3.16)

where tM is the dead time and can be measured for every column as the retention time of a completely unretained compound. Combining Equation 3.15 and Equation 3.16, the retention factor (k) can be expressed as, k=K

VS VR − VM = VM VM

(3.17)

K=

VR – VM VR′ = VS VS

(3.18)

and

Therefore, the partition coefficient between the two phases can be calculated from the retention data and the column parameters. This direct calculation involves errors in estimating the accurate volumes and is much easier to correlate k values with the partition coefficients in a true liquid dual-phase system. The retention in liquid–liquid chromatography is based on the different solvation of the analysed compound in the two phases. By contrast, the retention in reversedphase liquid chromatography is based on the hydrophobic effect caused by the exclusion of the solute by the water molecules in the mobile phase. As a consequence, the compound will orientate with the polar groups towards the aqueous stationary phase and the nonpolar, bulky part of the structure towards the hydrocarbonaceous stationary phase [67, 68].

3.9.1 HPLC

WITH

OCTANOL-COATED COLUMNS

Chromatography with octanol-coated columns is an interesting option for measuring the partition coefficient between an aqueous buffer and octanol. Usually, octanol is physically retained on a silica stationary phase and the mobile phase is an aqueous buffer containing different amounts of methanol or acetonitrile. The method was

Lipophilicity Measurements by Liquid Chromatography

97

described over 25 years ago [69]. McCall showed that logk values for benzene derivatives are in good agreement with logP values (shake-flask), especially for simple hydrocarbons, but significant deviations appear with increasing polarity of compounds [70]. HPLC methods using octanol-coated columns were reported as early as the 1970s. The stationary phase in the column was C18 Corasil, 37 to 50 μm particle size, in a 50 cm long column and 2 mm I.D. The column was coated with octanol following McCall’s method [70]. High purity octanol was injected directly into the column and the mobile phase was aqueous buffer and 25% (v/v) methanol in aqueous buffer (pH 7). The baseline normally stabilized in 30 to 45 minutes. Pyridine Noxide, formamide or simply methanol were used to measure the dead time and the peaks were detected by UV. Flow rate was 1 to 2 mL/min. The correlation with logD measured by the shake-flask method was highly linear for neutral derivatives of N-(-styryl)formamide. However, no linear correlation was seen for compounds with hydrogen-bonding capabilities, leading to the conclusion that methanol present in the mobile phase has a crucial influence [71]. In Mirrlees’ method, silanized kieselguhr was packed in columns (L = 10 to 30 cm, 4.6 mm I.D.) that were then dynamically coated with octanol. The mobile phase was water or pH adjusted buffer saturated with octanol. Pyridine N-oxide or tyrosine (both with logP < 1.5) were added for t0 estimation. Retention times of standards were run for each logP range and elution times up to 2 hours could be measured accurately. Values of logP in the range 0.3 to 3.7 were measured by varying the column length and flow rate. The column could be recoated with octanol after 50 hours of use. A remarkable slope equal to unity was obtained for the correlation between logP values and logarithm of retention times. The method seems to have two limitations. The first concerns compounds with poor solubility that tended to give lower logP values than expected, probably because the elution time was too short to reach equilibrium. The second limitation concerns compounds such as propanolol, where both charged and neutral species were extracted into octanol. This occurred because the pH of the mobile phase (7.4) was rather lower than the pKa of propranolol (9.56). In this case, the logD result obtained was lower than logP for the unionized compound. The authors also reported that a high salt concentration (above 0.1 M) in the buffer used for the mobile phase can influence the resulting logP/logD value. If the coating is not uniform, the resulting peaks will show a degree of splitting, making the choice of the correct retention time difficult. The recommended time for coating and octanol elution is 6 to 24 hours and recoating is normally completed in 2 hours. The flow rate was 0.2 to 8 mL/min and the volume of injected sample was typically 10 μL [69]. Although the chromatographic method using liquid octanol is promising, the authors stated that “it cannot be too strongly emphasized that, on all the available evidence, the only true model for octanol is octanol itself.” Octanol-coated HPLC silica-based columns were used by Slater et al. to measure the partition values for over 20 compounds. The results were compared with the corresponding logP/logD values obtained by potentiometric titration. The correlation between logPHPLC and logPliterature was very good (r = 0.991). Similar results were obtained for the correlation between logPpH-metric and logPliterature, where the intercept

98

Advances in Chromatography

was very close to zero (0.06), the slope close to 1 (0.9928) and a correlation coefficient of r = 0.994 [72]. Diatomaceous earth (Hyflosupercel, Johns Manville Inc., U.S.) has also been used as a support for nonpolar liquid coatings. This porous material is irregularly shaped and has different particle sizes but absorbs nonpolar liquids such as octanol and PGDP for long periods of time. No silanization is required if the material is treated with HCl, washed and dried. Mobile phase degassing is recommended to avoid spiking from the air bubbles trapped in the UV cell but it is not always necessary [34]. A modified HPLC method using octanol-coated columns is now commercially available from Sirius Analytical Instruments Ltd. on the instrument called ProfilerLDA [73]. The columns have a proprietary stationary phase and are dynamically coated with octanol. When the column is equilibrated, the samples are injected and the elution is continuously UV monitored. The software will automatically convert the retention times in logD values at the pH of the mobile phase. Usually, pH = 7.4 is used for the assays as it is the physiological pH value. The pH of the mobile phase can be modified for other physiologically relevant values (e.g., pH = 6.5). The instrument was designed for high-throughput screening of new chemical entities, mostly in the pharmaceutical industry, and can measure one sample in about 15 minutes. Samples are taken directly from 10 mM DMSO stock solutions from 96well plates. Typically, a 5 μL sample is injected into a 20 μL injection loop, filled with methanol, then injected into the column. The effluent is monitored by multi-wavelength UV spectroscopy. To maximize peak height vs. background noise, a maximum of 75 wavelengths are summed together and plotted against a logarithmic time scale. The range of available methods and logD range is shown in Table 3.2. The lifetime of an octanol-coated column is about 300 injections, although up to 600 injections have been reported without a significant change in the stationary phase uniformity. These numbers may seem low when compared with typical HPLC columns, which can withstand thousands of injections before showing the first degradation signs,

TABLE 3.2 Sirius ProfilerLDA

LogD range Column length Flow rate Eluent

Method 1 (low logD range)

Method 2 (medium logD range)

Method 3 (high logD range)

1.0 to 2.0 15 cm 3.5 mL/min 10 mM phosphate buffer; pH 7.4; 0.15 M ionic strength

1.5–3.5 1 cm 12 mL/min 10 mM phosphate buffer; pH 7.4; 0.15 M ionic strength

3.0–4.5 1 cm 12 mL/min 10 mM phosphate buffer containing 30% (v/v) methanol; pH 7.4; 0.15 M ionic strength

All methods use the same column type, coated with octanol; average run is 15 min/sample.

Lipophilicity Measurements by Liquid Chromatography

99

TABLE 3.3 logD7.4 Values Measured on GLpKa and ProfilerLDA

Compound

Predicted logD7.4 Value (ACD) [55]

logD7.4 Measured on ProfilerLDA (HPLC)

logD7.4 Measured on GLpKa (potentiometric)

3-chlorophenol Carazolol Carbendazim Chlorpromazine Desipramine Diclofenac Flurbiprofen Ibuprofen Indomethacin Lidocaine Nifuroxime Nitrazepam Papaverine Propanolol Quinine Tetracaine Tramadol Triamterene

2.42 1.58 1.42 3.36 1.23 0.11 0.92 0.77 –0.07 1.20 0.73 2.84 3.38 1.36 1.72 2.59 0.77 1.27

2.47 1.64 1.51 3.54 2.08 1.15 1.05 1.21 0.95 1.72 1.17 2.07 2.98 1.38 2.27 2.50 0.77 1.08

2.59 1.58 1.89 3.69 1.74 1.11 0.71 1.21 1.02 1.79 1.34 2.31 2.89 1.24 2.31 2.29 0.56 1.21

Compound Type acid base ampholyte base base acid acid acid acid base acid ampholyte base base base base base base

All measurements were performed at the same ionic strength, 0.15 M KCl. Sirius Analytical Instruments Ltd., Forest Row, U.K.

but this liquid-liquid chromatographic system works in a fragile thermodynamic equilibrium and care has been taken to provide a specific flow pattern that extends the column lifetime. The linearity of the relationship between the retention parameter and logD values is periodically checked with standards [74, 75]. Data in Table 3.3 show good correlation between the experimentally measured logD values at a pH of 7.4 by two different methods. ProfierLDA is based on the method described above, while GLpKa is based on a potentiometric titration method of the compound in aqueous solution in the presence of a layer of octanol. The potentimetric method for measuring pKa and logP/logD values is well documented [76–80]. Calibration is required at least once a day for octanol-coated columns to assess the coating uniformity. Column lifetime ranges from 10 to 600 injections, depending on the coating method and instrument used. Very lipophilic compounds have long retention times, often up to a couple of hours. The peaks will be broad and the peak shape nonsatisfactory for a usual HPLC experiment but the retention times can still be assessed with reasonable accuracy and the resulting logP values have a better confidence range than in shake-flask experiments. The lifetime of PGDP coated columns is usually longer that of octanol-coated columns [40, 42].

100

Advances in Chromatography

One of the difficulties of using octanol-coated columns is keeping the octanol in the column — usually by physical forces — and maintaining its initial volume constant during the elution. To the best of our knowledge, there is only one commercially available modified HPLC instrument using immobilized octanol on a solid packing [73]. Another difficulty in using octanol as stationary phase is related to the low number of theoretical plates compared with other HPLC columns and, consequently, a low capacity to separate impurities.

3.9.2 OTHER METHODS

USING

OCTANOL-COATED COLUMNS

One of the first methods for measuring partition coefficients used generator columns with a diluted solution of propylbezene in octanol to cover a silica-based column packing [81]. The octanol–water partition coefficients can be calculated from the measured propylbenzene concentrations in the equilibrated octanol and aqueous phase. The columns can be reused by pumping methanol to remove the organic phase and dried by flowing through a warm stream of nitrogen. The dried column is ready for recoating. The partition coefficient values obtained from the generator column method were used to evaluate six methods for Po/w estimation of highly hydrophobic aromatic hydrocarbons, one a measurement by RP-HPLC and the others computational methods using fragments or molecular descriptors [82]. The solvent-generated technique was designed to overcome the problems of poor modelling of the octanol–water partition by liquid chromatography with a “solid” stationary phase consisting of a long alkyl chain, bonded to a solid support, with similar properties with a liquid phase. The reported chromatographic systems used C8 and C18 chemically bonded silica as stationary phase: LiChrospher Si RP-8 and RP-18 (Merck, Darmstadt, Germany), and Nucleosil Si C8 (Macherey-Nagel, Düren, Germany). The mobile phase was 40% methanol with aqueous buffer (phosphate buffer pH = 2.0 for acids and pH = 7.5 for bases), both saturated with octanol following a special procedure to avoid the formation of stable emulsions and allowing a few drops of octanol on top of the mobile phase reservoir at all times to ensure saturation [83]. The formation of the octanol stationary phase was generated by flowing octanolsaturated phosphate buffer through the column. A group of neutral, acidic and basic aromatic compounds was analyzed by this method. The retention factor (logk) was measured on all columns by RP-HPLC and by the solvent generation method and then compared with logPo/w. The column intercomparison showed that the retention (logk values) were more reproducible when using the solvent generation method than when using traditional RP-HPLC. Also, the estimated logPo/w values were in good agreement with experimental logPo/w values. Other applications of the method are described in [84, 85].

3.9.2 ELogP METHOD Classic C8 and C18 silica-based stationary phases and octanol-enriched methanolwater mobile phases were tested by Minick et al. for lipophilicity measurements [86]. If methanol in the aqueous mobile phase is mixed with a small amount of

Lipophilicity Measurements by Liquid Chromatography

101

octanol, the new system will have similar characteristics with the dual phase octanolwater because octanol is significantly more lipophilic than methanol and should concentrate in the stationary phase, competing with methanol molecules and producing a more octanol-like stationary phase by minimizing hydrogen-bond discrimination. The aqueous part of the mobile phase was buffered with phosphate buffer (0.005 M) or with 4-morpholinepropanesulfonic acid (MOPS, 0.02 M) buffer to pH = 7.4. MOPS is an organic salt and should be less susceptible to precipitation at higher methanol concentration. The nature of the salt used to buffer the mobile phase had no significant influence on the retention of compounds. Amino modifiers such as n-decylamine (0.15%, v/v) or N,N-diethylamine (0.075%, v/v) were added to the deionized/distilled water used for the mobile phase to mask silanophilic interactions with the stationary phase. N-decylamine provided better correlations between logkw and logPo/w than N,N-diethylamine for both C8 and C18 stationary phases. This research work concluded that 0.25% octanol added to the methanol used for the mobile phase is the optimum amount to mimic a true octano–water environment. Higher concentrations were also tested but resulted in octanol separation when methanol was less than 50% (v/v). The method should provide a dynamic coating with octanol of the stationary phase, producing similar results to those obtained with a column coated with liquid octanol by direct injection [69]. Lombardo et al. used Minick’s method with octanol-enriched mobile phase and silica-based columns for the lipophilicity estimation of neutral compounds [15]. In Lombardo’s chromatographic system, the octanol from the mobile phase formed a fine molecular coating on the silica support of the column [87]. In this case, the presence of the amine masking agent was not necessary in the mobile phase. The retention parameter was called ELogP and was compared with logP and logD values, respectively. The ELogP values for neutral compounds [15] and for neutral and basic compounds [88] were reported in Lombardo’s work. The column used in Lombardo’s studies was deactivated, electrostatically coated Supelcosil LC-ABZ. The mobile phase was 15 to 70% (v/v) methanol in 20 mM MOPS buffer, pH = 7.4. The aqueous buffer was saturated with octanol and methanol contained 0.25% (v/v) methanol. A quantity of 0.15% (v/v) n-decylamine was added to the buffer for logD estimations to minimise the interactions between the charged compounds and the stationary phase. Extrapolated logkw values were used for correlations. The difference between the obtained ElogD values and experimental logD values were in the range 0.10 to 0.58 for 90 structurally unrelated drugs. Equation 3.19 describes the correlation between experimental logD values (obtained by other methods or reported in the literature) and extrapolated logkw values for the 90 drugs, covering the lipophilicity range from 1.49 for terbutaline sulphate (experimental logD value was 1.35) to 5.95 for amiodarone (experimental logD value was 6.1) [88]: logD = 0.207 + 1.127 log kw

(3.19)

n = 90, r2 = 0.964 The intercept and the slope are reasonably close to zero and unity, respectively, offering a good correlation between the shake-flask and RP-HPLC experiments. This

102

Advances in Chromatography

method is suitable for accurate lipophilicity estimation of neutral and basic compounds with unrelated structures but it is not recommendable for acids. The authors believe that specific interactions take place between anions and the stationary phase, despite adding different masking agents (some of them being zwitterions or anions) to the mobile phase [15, 88]. Even when great care is taken to model octanol–water partition, a compound can follow different partition patterns in the chromatographic separation and in a shake-flask experiment. The larger the surface between the two phases, the greater the chances that other factors such as surface activity and shape will also influence the partition [89].

3.9.4 REVERSED-PHASE HPLC (RP-HPLC) In reversed-phase high-performance liquid chromatography (RP-HPLC), the stationary phase is nonpolar (or less polar than the mobile phase) and the mobile phase is polar (or more polar than the stationary phase). Octadecylsilane (ODS) is one of the most popular packings used in RP-HPLC and the alkane chains form a molecular “fur” on the silica gel granule. RP-HPLC with a purely aqueous mobile phase is not the best experimental design for separation. The alkane “fur” needs a small amount of organic solvent in the mobile phase to help the alkane chains raise from a flat position and allow the free circulation of the solvent molecules. Methanol and acetonitrile are the most popular solvents used as organic modifiers in the mobile phase for RP-HPLC experiments related to lipophilicity estimations. However, the use of ethanol has also been reported [90]. Any organic compound that has a lipophilic and a hydrophilic part of the molecule will interact with both chromatographic phases with the lipophilic part positioned towards the stationary phase and the hydrophilic groups towards the mobile phase. Many authors believe that the retention in a reversed-phase chromatographic system with a C18 or C8 stationary phase is due to the hydrophobic exclusion of the solute molecule from the structural organization of water molecules in the mobile phase rather than on a specific interaction between the solute and the stationary phase.

3.9.5 THE

EXTRAPOLATION METHOD, LOGKW

The retention factor k, normally expressed as logk (log10k), is used for correlations with logPo/w. The relationship between logk — sometimes called capacity factor — and the fraction of organic modifier in the aqueous mobile phase (ϕ) is one of the most studied in RP-HPLC. The relationship is often linear (Equation 3.20) but, in many cases. there is parabolic dependence, especially at low organic solvent concentration (Equation 3.21): logk = a1 + b1 ϕ

(3.20)

logk = a2 + b2 ϕ + c2 ϕ2

(3.21)

Lipophilicity Measurements by Liquid Chromatography

103

where a1, b1, a2, b2 and c2 are coefficients. logk = logkw + S ϕ

(3.22)

This relationship is not linear for the entire range of organic modifier concentration in the mobile phase but is a good approximation for 0.5 < logk < 1.5 [89]. The extrapolated retention factor value (logkw) shows the retention of a compound in a chromatographic environment where only buffered water is used as a mobile phase. The value of logkw cannot usually be measured directly because of extremely long elution times. Therefore, the solute partition between the stationary phase and water could be a good approximation for the partition between an organic phase (e.g., octanol) and water if there are physicochemical similarities between the organic solvent and the stationary phase. All extrapolation methods rely on the linearity of Equation 3.22, assuming that the linearity is preserved even at low concentrations of organic solvent in the mobile phase. Sometimes, the extrapolation is performed from very high concentrations of organic solvent in the mobile phase (40 to 50%, v/v) without having an idea about how retention changes at lower concentrations. In practice, it is not always possible to measure retention times for very lipophilic compounds at low organic solvent concentration in the mobile phase because of very long elution times and distorted peaks The linearity of Equation 3.22 changes when water exceeds 90% (v/v) in the mobile phase. The logk values of nonpolar compounds measured in methanol-water systems are lower than the values predicted by extending the linear correlation; therefore, any prediction will provide a higher value than the real one. For polar compounds, however, the real logk values are higher than the extrapolated ones and any prediction will underestimate the real logk value [91]. A similar trend in the sign of the error between the measured and the extrapolated logkw values was found for acetonitrile-water mixtures (Figure 3.6). Accurate determination of the dead time (tM or t0) is very important, especially for logk values lower that 1 (e.g., for polar compounds) [92]. Errors in measuring the dead time will affect the k values and, ultimately, the correlation with logP values.

(a)

(b)

FIGURE 3.6 ODS structure in water and in water with methanol.

104

Advances in Chromatography

Most RP-HPLC applications for lipophilicity measurements use ODS stationary phases and methanol–buffer mobile phases. Standard procedures are still to be established as the methods are quite diverse. A compound’s capacity to donate or accept hydrogen bonds influences its partition between the alkyl-bonded stationary phase and the mobile phase. To eliminate this influence, some methods require the extrapolation of logk values to logkw (mobile phase containing 100% water). The extrapolation to 0% organic modifier in the mobile phase is based on the assumption that the linearity of the relationship between the retention factor logk and the concentration of the organic solvent in the aqueous mobile phase is maintained for the entire concentration range, even in the range 0 to 5%. In many cases, it is not possible to measure logkw because of the very high elution times. A linear relationship between logP and logkw for over 50 monosubstituted heteroaromatic series — derivatives of pyridine, pyrazine, prymidine, pyridazine, thiazole, and furane — was reported with a very good correlation coefficient, r2 = 0.95 [93]. According to Yamagami [93], logP values correlate well with logkw values for most substituents but, in some cases, usually for strong hydrogen-bond acceptors, the error in estimating logP is quite high and the obtained logkw value is usually higher than the real one. These conclusions may strongly depend on the type of stationary phase, particularly the amount of free silanol groups [94]. There are many studies reporting the use of logkw for lipophilicity studies. Most of them use methanol–water or methanol–aqueous buffer binary mobile phases with logkw obtained from extrapolations against methanol content [95–100]. The linearity of the relationship between logk and Φ depends on the solvent concentration in the mobile phase. Linearity is maintained for methanol–water mobile phases if the methanol concentration is in the range 30 to 70% (v/v). Outside this range, the relationship can show a pronounced curvature dependent on the retained compound [101]. When acetonitrile is used in the binary mobile phase, the relationship is almost always quadratic [102]. Values of logkw should be the same regardless of the column type and the organic solvent in the mobile phase but often this is not the case [103]. Different organic modifiers used in the mobile phase have different capacities to wet the same stationary phase. Moreover, the same mobile phase will not cover the same type of stationary phases from different manufacturers or from different batches in the same way. As a consequence, the intercept value (logkw) will be dependent on the mobile phase composition and the type of stationary phase used. Very good correlations of logkw to logP have been reported for benzene, furan, benzofuran, 1-Me-pyrrole, and 1-Me-indole derivatives in a chemically bonded, reversed-phase HPLC system with methanol-water mobile phase (r = 0.998 to 1.000). Most of these compounds do not form hydrogen bonds or are weak hydrogenbond acceptors. Pyrrole and indole derivatives are hydrogen-bond donors, so the logkw values have the tendency to underestimate the logP values. The corresponding ester derivatives are strong hydrogen-bond acceptors; therefore, the logkw values will overestimate the corresponding logP values. Good correlations between the isocratic logk50 measured at 50% methanol in the aqueous mobile phase were also obtained for these type of compounds. However, it is well known that the isocratic retention parameters in HPLC usually give good correlations with logP values for

Lipophilicity Measurements by Liquid Chromatography

105

congeneric series of compounds. The correlation is not so good when considering a group of chemically unrelated compounds [98]. A good correlation between isocratic logk and experimental logP was obtained for a noncongeneric but closely related group of acidic heterocycles. The chromatographic measurements were performed on a Waters μBondapak C18 column (10 μm packing particle size) with a mobile phase formed by 90% (v/v) distilled water adjusted to a pH of 3.5 with acetic acid, 5% isopropanol, and 5% methanol [104]. logk = 0.675 logP + 0.236

(3.23)

r = 0.964 It has been suggested that logkw is not a suitable parameter to be used in correlations and contributes to poor prediction performance. The value of logkw seems to correlate better with logP in homologous series of compounds and the correlation is rather poor when considering a chemically diverse set of compounds [6].

3.9.6 ISOCRATIC

LOGK

VALUES

The isocratic retention factor, logk, obtained on C18 silica-bonded phase was investigated in an interlaboratory study to determine its suitability for predicting partition coefficients in octanol-water [105]. The results showed that RP-HPLC is a reliable method to replace the reference method, shake-flask partitioning. However, there are a number of conditions to be met: at least six reference substances should be used, the minimum water content of the eluent should be 25% (v/v) and the extrapolation beyond the calibration range of logPo/w should only be carried out for logPo/w > 6. Many studies report isocratic logk values correlating reasonably well with logP values. Usually, logk50 is taken into consideration, obtained for 50% organic modifier in the mobile phase. Yamagami reported [93] an extremely good correlation between logP and logk50 (aqueous mobile phase containing 50% methanol) for series of lipophilic pyrazines and monosubstituted pyrazines covering six orders of magnitude. Measuring only logk50 values saves time compared with measuring logk at different concentrations of the organic solvent in the mobile phase and extrapolating to a hypothetical value of logkw that would be obtained with 100% water in the mobile phase. Why does 50% water in the mobile phase seem to provide a better correlation with logP values than the extrapolated chromatographic parameter, logkw? One possible answer is that when methanol is the organic modifier, the stationary phase is dynamically coated with methanol and gains octanol-like properties. If this hypothesis is true, the chromatographic system with octadecyl silane as stationary phase and equal parts of methanol and water (buffer) in the mobile phase could be more similar to the octanol-water system than any other methanol concentrations in the mobile phase [93]. Other studies also mention an isocratic logk [106–108].

106

Advances in Chromatography

3.9.7 THE SIGNIFICANCE

OF

SLOPE S

Because many experimental results from literature show that logkw values do not always model octanol–water partition correctly, some researchers have investigated the significance of the slope (S) with a view to using it as a second independent variable together with logkw: logP = aS + b logkw + c

(3.24)

where a, b, and c are coefficients [89]. A chromatographic hydrophobicity index (Φ0) was suggested to better model the octanol-water partition in a RP-HPLC system. The index (Φ0) was defined as the organic phase concentration (methanol or acetonitrile) in the mobile phase required to produce a capacity factor of logk = 0 [109] (see also 3.9.11). In this case, the retention time is double the dead time (tR = 2t0) and the molar fraction of the solute is the same in the mobile and the stationary phase (see Equation 3.15 and Equation 3.16): k=K

VS CV = S S VM C M VM

(3.25)

where CS and CM are the solute concentration in the stationary and mobile phase, respectively, and VS and VM are the volume of stationary and mobile phases. The value K is the partition coefficient of the solute between the stationary phase and the mobile phase at a defined composition. The logarithm of this equation can be expressed as: log k = log K + log

VS = Sϕ + log kw VM

(3.26)

A similar relationship can be written for Kw, the partition coefficient between the stationary phase and a purely aqueous mobile phase: log kw = K w + log

VS VM

(3.27)

Taking into account Equation 3.22, Φ0 can be expressed as: ϕ0 = −

log kw S

(3.28)

The Chromatographic Hydrophobicity Index concept was validated on a large number of compounds structurally related or not: nicotinate derivatives, hydroxyl aromatics, herbicides, substituted aromatics, aromatic hydrocarbon derivatives, 143

Lipophilicity Measurements by Liquid Chromatography

107

unrelated drug molecules, morphine derivatives, tricyclic drug molecules, deoxyuridine derivatives, benzodiazepines, aniline and phenol derivatives, adenosine monophosphate derivatives, β-lactam antibiotics, and barbiturate derivatives. The comprehensive study has proved that Φ0 is characteristic for every compound and depends only on the type of organic modifier, temperature, and pH. It is independent of the type of reversed-phase packing in the chromatographic column, the column dimensions, flow rate, and the mobile phase composition of the measurements. The Φ0 values obtained for acetonitrile and for methanol are in excellent correlation with each other, r = 0.96 for 72 compounds. The correlation between Φ0, ACN and logP for 140 compounds (r = 0.88) was slightly better than the correlation between Φ0, MeOH and logP for 448 compounds (r = 0.787) [109] where ACN is acetonitrile and MeOH is methanol.

3.9.8 CORRECTION

FOR IONIZATION

If compounds are ionized at the mobile phase pH, a correction for the right logkw must be applied to the apparent extrapolated logk value (logkwapp): logkw = logkwapp + log (1 + 10a(pH – pKa))

(3.29)

where a = 1 for monoprotic acids and a = –1 for monoprotic bases. More complex equations can be derived for multiprotic acid and bases [38]. Horvath has reported an equation between the capacity factor of a neutral molecule (kn) and the capacity factor of the same molecule but ionized (ki): Ka ⎡⎣ H + ⎤⎦ Ka 1+ ⎡⎣ H + ⎤⎦

kn + k j k=

(3.30)

where k is the capacity factor at a pH where the molecule is partially ionized and Ka is the ionisation constant [110, 111]. The retention of amphoteric compounds is directly dependent on the mobile phase pH, organic modifier in the mobile phase, and the presence of ion pairs. The chromatographic retention for seven ampholytes (niflumic acid, pyridoxine, norfloxacin, pefloxacin, morphine, nitrazepam, sulfadimine, and 11-amino undecanoic acid) was investigated on a Hypersil 5 ODS, 5-μm packing particle size, from Bioseparation Technique Ltd. (Budapest, Hungary) with 50-50 (v/v) methanol–aqueous phosphate buffer solutions, pH = 3 to 8 [26]. The pH of the aqueous buffer was shifted up 1 unit after mixing with equal parts of methanol. Two different categories were investigated, one involving zwitterionic amphoteric compounds when only the zwitterionic species exist at the pH of the isoelectric point (e.g., 11-amino undecanoic acid) and the other involving true ordinary amphoterics when the solution of the pH of the isoelectric point contains both ionized and

108

Advances in Chromatography

neutral species (e.g., nitrazepam and sulfadimine). The retention of the compounds that have a neutral species in the solution is definitely increased around the pH of the isoelectric point. It is reasonable to assume that the maximum retention of a compound will take place at a pH close to the isoelectric point and that the retention at the pH of the isoelectric point will be dramatically different for the compounds in the second category compared with the first category. The ion pairing process could explain why retention increases in pH regions where only ionized species are expected. The results of this study suggest there are enough similarities between the partition in a chromatographic system C18/methanol–water and the octanol–water dual-phase system to support the use of chromatographic retention parameters to model the partition between octanol and water [26]. A detailed study of ionizable compound retention in HPLC can be found in the literature [112, 113].

3.9.9 THE INFLUENCE MEASUREMENTS

OF

STATIONARY PHASE

ON

LIPOPHILICITY

The stationary phase most widely used in lipophilicity studies by RP-HPLC comprises hydrocarbon chains, usually C8 or C18, bonded on a silica-based support. Depending on the manufacturing process, the resulting stationary phase can be monomeric, where only one bond is formed per silane molecule, or polymeric, where multiple bonds are formed between the chains. The monomeric layer has a brushlike structure, activated by the presence of an organic modifier in the mobile phase, while the polymeric structure is not completely known. Both phases have unreacted, free silanol groups but their accessibility to passing molecules from the mobile phase or from the sample, is higher in the monomeric type. Different C18 silica-based stationary phases were studied with methanol-water eluents to investigate the reliability of the extrapolated logkw value. The variability of obtained logkw values was within 2 to 4%, showing that, if correctly determined, it is a solute property and independent of the stationary phase [114]. However, this result should be taken with caution because the practice shows poor reproducibility between similar types of stationary phases of different origin. One of the main problems of silica-based columns is their poor reproducibility when produced by different manufacturers. Although the structure seems identical, there are large differences in the chromatographic properties [115]. Values of logk obtained on a typical RP-HPLC system do not encode the same factors contributing to logPo/w values (e.g., logk values are influenced by the solute hydrogen-bond acidity while logPo/w values are not) [116]. Also, the alkyl-bonded stationary phase is not a bulk liquid and the solute has to penetrate it vertically, unlike the solvation in octanol [117]. Retention properties for 18 RP-HPLC columns were studied in an attempt to find similarities between the retention and the column type [118]. The correlations between logkw and logP were also investigated, showing that the slopes of the linear relationships are less than 1, ranging from 0.79 to 0.86. End-capping of silica-based stationary phases provide more reliable lipophilic correlation because of the blocking of some of the free silanols [87].

Lipophilicity Measurements by Liquid Chromatography

109

Silica-based stationary phases, whether derivatized or not, are suitable for dynamic coating with octanol and provide a system similar to the octanol-water dual phase. However, they are not suitable for coating with other solvents that may be of interest for correlations with biogical properties. A C18 stationary phase coderivatized with styrene divinylbenzene copolymer was reported and could be suitable for alkane coating [96]. Because of the limited chemical stability of silica-based stationary phases used in RP-HPLC, these phases cannot be used for important biologically active weak bases, such as alkaloids, with pKa values above 7.0. Other stationary phases (nonsilica based) were therefore investigated in an attempt to find correlations between the retention factor and lipophilicity. Alkyl-bonded alumina stationary phases have similar retention mechanisms as alkyl-bonded silica phases but have better stability at high pH. Alumina has a couple of advantages over silica gel — it is stable over a wide pH range and has no active silanol groups. An experimental study investigated the retention of alkaloids on three octadecyl-bonded stationary phases, octadecyl alumina (ODA), octadecyl silica (ODS) and octadecyl-modified polystyrene-divinylbenzene (ACT-1), and also on polybutadiene-coated alumina (PBD) [119]. Mixtures of methanol-buffer were used in all cases with different concentrations of methanol to obtain a retention factor in the range of 4 to 6. The aqueous phase was buffered at pH = 7.4 with phosphate buffers in all cases except ODA columns, where 4-morpholinopropanesulfonate (MOPS) was used because of the column instability in phosphate environment. The ionization of bases can be suppressed on ODA columns by working with high pH values in the mobile phase, up to 11.0. The correlations between logk and calculated logPo/w values for the studied alkaloids have better correlation coefficients for ODA columns (r = 0.953 to 0.980) than for ODS (r = 0.938 to 0.970), PBD (r = 0.881 to 0.985) and ACT-1 columns (r = 0.686 to 0.804). Poly(styrene-divinylbenzene) stationary phases are stable over the pH range 1 to 14 but the correlation with logPo/w for structurally unrelated compounds is not very good [120]. Shrinking and welling has been reported for these stationary phases [121]. Other stationary phases used in lipophilicity studies are: • • •

•

Monolithic silica columns [122, 123] Monolithic columns based on polystyrenes, polymethacrylates and polyacrylamides [124] Octadecylpolyvinyl copolymer — this stationary phase can provide good correlations with logP values but can also strongly retain specific compounds [125] Alkylamide silica (alkyl chains attached to the polar alkylamide moiety) [126, 127]

Good correlations for congeneric compounds are reported in O’Gara et al. and Czajkowska et al. [128, 129] and not so good in Buszewski et al. [130] Nonsilicabased stationary phases would not have the residual silanol groups and will be

110

Advances in Chromatography

sufficiently stable at high pH values. Hydrocarbon-coated alumina, zirconia, titania and ceria are also resistant to high pH values [131–134]. Two polyethylene-coated columns, one silica-based (PECSiO2) and the other zirconia based (PECZrO2), were investigated for their hydrophobic properties by testing 25 simple, structurally diverse organic compounds [135]. The mobile phase was 10% methanol in phosphate buffer, pH = 7.2. In some cases, the capacity factor had to be measured with 80% methanol in the mobile phase and the corresponding logk values for 10% methanol obtained by extrapolation against the concentration of the organic modifier in the mobile phase. The comparison between the capacity factors for the same experimental conditions on the two investigated columns has a correlation coefficient r = 0.976 for the 25 investigated compounds (Equation 3.31): logkPECZrO2 = 0.224 + 0.988 logkPECSiO2

(3.31)

n = 25, r = 0.976 The intercept and the slope are reasonably close to zero and unity, respectively, suggesting that the columns have similar properties with stronger retention on the silica-based column. Although the studied chromatographic systems looked promising, the LSER studies showed that a direct correlation with logPo/w values is not possible in this case. However, a good lipophilicity scale and quantitative-structureretention relationships (QSRR) prediction models can be easily obtained. Poly(butadiene)-coated alumina columns, Spherisorb A5Y (ES Industries, Marlton, NJ, USA) or home-made, were also used for lipophilicity evaluation of basic, neutral and acidic compounds. The mobile phase was methanol–buffer (1:1, v/v) with the pH adjusted to keep the compounds in the neutral form (a pH of 1.65 for acids and up to 11.5 for bases). The relationship between logk and logP for 24 unrelated organic compounds was linear with a good correlation coefficient (r = 0.96) but, more importantly, a high pH could be used in the mobile phase to measure the retention of basic compounds in their neutral form [136, 137]. The chromatographic retention data on the diol stationary phase was found to provide better correlations with logPo/w than RP-HPLC for series of polar polycyclic aromatic compounds (N-PAC) [138]. The study used two columns: an ODS (Phenomenex Prodigy ODS-2) and a Diol type (Nucleosil 7 OH, Macherey Nagel). Experiments with n-hexane as eluent showed no difference in retention times of benzene and nitro-benzene on the ODS column, confirming a negligible amount of residual silanol groups. The mobile phase was a mixture of methanol and water, 35 and 50% (v/v) methanol for the Diol column and 65, 75, and 85% (v/v) methanol for the ODS column. The retention data in the RP-HPLC systems were compared with partition coefficient values obtained from generator column experiments, on Chromosorb W generator columns (dimethyl chlorosilane, treated, and acid washed) and an extractor column packed with polygosil 35-42 octadecyl silica. Octadecyl polyvinylalcohol copolymer (ODP) columns can be used with 100% water in the mobile phase where no masking agent is required and the packing is stable at pH from 1 to 14 [139–141].

Lipophilicity Measurements by Liquid Chromatography

111

The influence of organic modifiers in the formation of the stationary phase in RP-HPLC was investigated for methanol, acetonitrile, and tetrahydrofuran [142]. The formation of the stationary phase for C8 and C18 bonded silica involves the adsorption of molecules from the aqueous mobile phase. The organic modifiers used in the mobile phase can selectively enrich the stationary phase while the residual silanol groups will prefer molecules with hydrogen-bonding capabilities like water, methanol, acetonitrile, and tetrahydrofuran. Methanol was widely used in aqueous mixtures and was particularly studied [143]. The concentration of methanol in the aqueous mobile phase was varied from 0 to 100% and the amount of methanol and water that bonds to the C8 or C18 stationary phase was measured by gas chromatography with flame ionization detection (for methanol) and thermal conductivity detection (for water). The increase of methanol fraction in the mobile phase leads to an increase of the amount of methanol bonded to the stationary phase. The effect is more pronounced for C18 phases than for C8. On solvation with methanol, C8 forms a “brush-like” structure on the surface of the solid packing, allowing direct access of the solvent molecules to the free silanol groups. There are structural differences between the C8 and C18 stationary phases, resulting in C18 chains being closer to each other than C8. As a consequence, a lower methanol concentration (less than 20%, v/v) will not open the C18 chains in the same way it does with C8, so the methanol layer on the stationary phase will be less on C18 than on C8. At high methanol concentrations, C18 will also show a “brush-like” structure, allowing the solvent molecules access to all binding sites in the stationary phase. Due to the stationary liquid phase formed on the C8/C18 silica support, the retention mechanism is governed by partition. What happens when the mobile phase is 100% aqueous in a C8/C18 chromatographic environment? Water molecules have been found on the stationary phase and their presence can be explained either by adsorption on the active sites or by being trapped under the hydrocarbon chains. Either way, the retention mechanism will be governed by adsorption rather than partition [143]. A nonporous, ultra-small (1.5 μm), C18 stationary phase was investigated for its suitability in RP-HPLC lipophilicity studies. The column was NPS RP-18 (Micrascientific, Northbrook, IL) and the mobile phases were binary aqueous mixtures, methanol-phosphoric acid (0.03 M) and acetonitrile-phosphoric acid (0.03 M) with various concentrations of organic modifier including 0% (v/v) [144]. Very good correlations between logk and Φ were found in all cases (r2 = 0.99) for the investigated structurally diverse compounds. The following equations were obtained for logPo/w to logkw correlations: logPo/w = 0.87 + 1.18 logkw,ACN,

r2 = 0.977

(3.32)

logPo/w = 1.11 + 0.90 logkw,MeOH,

r2 = 0.961

(3.33)

where logkw,ACN and logkw,MeOH are the extrapolated values for zero content of acetonitrile and methanol, respectively, in the mobile phase. The slopes in the above equations are a measure of the mobile phase sensitivity to changes in the

112

Advances in Chromatography

hydrophobicity of the studied compounds. Both slopes are close to unity, showing a good correlation between logPo/w and logkw with acetonitrile being more sensitive than methanol to structural changes in the solute. Specially formulated stationary phases allow the use of pure aqueous mobile phases in a reversed-phase mechanism. Trifluorpropyl siloxane, commonly used in gas chromatography, has been applied on nonporous glass microspheres [145]. The benefits of a RP-HPLC column that works with pure water as a mobile phase range from low solvent consumption to the possibility of directly measuring the logkw value without the problems faced by the extrapolation method. The retention mechanism on the trifluorpropyl siloxane stationary phase has been proven to be a reversed phase mechanism, similar to the RP-HPLC environment on silica-C18. The relationship between logk and logPo/w has an excellent correlation coefficient (r = 0.998) for ten hydrophobic simple organic compounds.

3.9.10 MOBILE PHASE COMPOSITION

AND

LIPOPHILICITY

3.9.10.1 Organic Modifier Pure water cannot be used on most chromatographic columns, and thus an organic modifier is usually added. The retention on hydrocarbon stationary phases with water as eluent never correlates with logP and many efforts have been made to find the most suitable organic modifier in the mobile phase [34]. Methanol, acetonitrile, and tetrahydrofuran are popular organic modifiers in RPHPLC. In theory, the extrapolated logkw values should be the same regardless of the organic modifier used in the mobile phase. Significant differences exist in practice, making the extrapolation method unreliable. Also, the linear relationship between the retention factor, logk, and the concentration of organic solvent in the mobile phase, tends to curve at high water concentration. The errors depend on the nature of the organic modifier (see related text on the extrapolation method) and can be as much as 10% for methanol-water mixtures and 20% for acetonitrile-water mobile phases [91]. The hydrocarbon-bonded silica-based stationary phase has a brush-like structure at high concentration of organic solvent in the mobile phase. This structure collapses when high water content mobile phases are used, significantly reducing the active surface area of the stationary phase. This could be one of the reasons why the extrapolation method fails to accurately predict the logkw values. Methanol is by far the most popular organic modifier used in RP-HPLC. Good correlations are obtained between logP and logk/logkw for the large majority of studied congeneric compounds: logP = a logk + b

(3.34)

In many cases, the slope (a) is close to unity and the intercept (b) is close to zero, indicating a solid correlation between the forces involved in the two methods. Methanol has hydrogen-bond donor and acceptor properties, changing the cohesion of the water molecules and influencing the way that a solute will interact with the mobile phase.

Lipophilicity Measurements by Liquid Chromatography

113

Acetonitrile is a weak hydrogen bond acceptor and will usually have a greater impact on the solute solvation than methanol. Acetonitrile also has a high dipole moment and can participate in selective dipole–dipole interaction with certain solutes. The participation of residual silanol groups in the retention process is more pronounced in organic modifiers other than methanol [95, 144]. Raman spectroscopy has shown that acetonitrile molecules self-associate as a function of their concentration in an aqueous environment. Solutes will interact with all acetonitrile aggregates and species that exist in the mobile phase [146]. Often, these interactions are not quantifiable and are not taken into account when predicting retention at different acetonitrile concentrations in the eluent. Tetrahydrofuran (THF) is more hydrophobic than methanol and leads to shorter retention times. It seems that THF drags enough water towards the stationary phase to mask the influence of the free silanol groups, so the addition of a hydrophobic amine (e.g., n-decylamine) as a masking agent is not necessary. THF disrupts the cohesion between water molecules in the mobile phase to a greater extent than methanol and, therefore, may show more selectivity in the retention of some compounds. THF is usually employed in concentrations less than 60% (v/v) in the aqueous mobile phase and the relationship between the retention factor, logk, and the THF concentration is often quadratic, not linear. Studies have investigated the influence of the organic modifier, methanol, acetonitrile or THF on the retention of basic compounds. Significantly worse results were obtained for acetonitrile than for methanol or tetrahydrofuran [147, 148]. The influence of organic solvent modifier on the retention of basic compounds was studied on seven C18 columns from different manufacturers and on one C8 column. Nine basic compounds were studied: pyridine (pKa = 5.17), nicotine (pKa = 7.87), amphetamine (pKa = 9.9), codeine (pKa = 7.95), diphenhydramine (pKa = 9.0), nortriptyline (pKa = 10.0), procainamide (pKa = 9.2), quinine (pKa = 8.52), and 2-[N-methyl-N-(2-pyridyl)-amino]ethanol (PAE) (pKa = 6.8) [149]. All compounds have been analyzed with isoeluotropic mobile phases on all columns. The mobile phases were: methanol–phosphate buffer (0.064 M, pH = 7.0, 65:35, v/v), THFphosphate buffer (0.03 M, pH = 7.0, 25:75, v/v), and acetonitrile–phosphate buffer (0.0375 M, pH = 7.0, 40:40, v/v). The columns contained new stationary phases for RP-HPLC, recommended by the manufacturers for the analysis of basic compounds. This study confirms that the organic modifier plays an important role in the retention of basic compounds, resulting in different peak shapes. Acetonitrile produced the worst results and THF may be a better choice than methanol for basic compounds. The chromatographic behavior also depends on the nature of the analyzed solutes. A high pKa value of the solute introduces an extra difficulty in measuring the retention of the neutral form. An interesting “loading” effect was observed for repeated rapid injection of solutes, sometimes improving the peak shape due to previous activation of the stationary phase from the solute. This effect suggests that it is better to inject single samples rather than mixtures. Nonpolar mobile phases have been used for quantitative-structure-retention relationships (QSRR) for series of chalcone, cis, and trans isomers [150]. Eighteen chromatographic systems (three stationary phases and six mobile phases, as shown in Table 3.4) were investigated. The chromatographic parameters obtained for the

114

Advances in Chromatography

TABLE 3.4 Chromatographic Systems with Nonpolar Mobile Phases for the Investigation of Retention of 36 Chalcones, E and Z Configurations Stationary Phase

Mobile Phase

Zorbax ODS LiChrospher 100 DIOL Spherisorb AP (n-propyl picryl ether)

Heptane containing 0.5% (v/v) of THF, dioxane, ethanol, propanol, octanol, and DMF

systems in Table 3.4 for 36 chalcones were correlated with molecular descriptors and logPo/w values calculated from atomic contributions. The logPo/w calculations will not take into consideration the differences between isomers and provide the same value for both of them. The authors claim that it is possible to accurately evaluate logPo/w values with this method, although this was not the purpose of the study. 3.9.10.2 Buffer in the Mobile Phase Buffer is usually added to the mobile phase to maintain a suitable pH for the separation and to maintain the solutes in a neutral or ionised form. Phosphate buffer is very popular and easy to use for a large range of pH values but has the risk of forming ion-pairs with cations. Instead, a zwitterionic buffer can be used, e.g., morpholinopropane sulfonic acid (MOPS). Knowing the pKa of a molecule, the concentration of species can be calculated at each pH. However, it is difficult to predict retention as a function of pH because there are unexpected interactions between the solute and the free silanol groups or the metal contaminants [151]. Most silica-based chemically bonded columns are not stable at pH values above 8.0 because the silica support starts to dissolve [152]. The degradation process is accelerated if the method used for manufacturing the stationary phase produces large surface areas. Several C18 reversed-phase columns from different manufacturers were studied in detailed ageing tests at high pH. Densely bonded monomeric C18 was found to better protect the silica support against solvation in the high-pH mobile phase and acetonitrile prolongs the column life if used as methanol replacement in the aqueous mobile phase. Also, phosphate buffers accelerate the degradation process when compared to similar conditions with organic buffers [153]. Many chromatographic lipophilicity studies are performed at pH = 7.4 because this is the physiological pH. However, the common silica-based stationary phase is partially ionised at pH values higher than 7.0. If the solutes are also partially ionized at this pH, extra interactions occur in the chromatographic system that are difficult to quantify and the peak shape is often not ideal. 3.9.10.3 Ionic Strength The ionic strength of the mobile phase has a significant influence on retention on silica-based alkyl-bonded stationary phases. Experimental results have demonstrated

Lipophilicity Measurements by Liquid Chromatography

115

that ionized organic acids can be excluded from the pores of the stationary phase [154]. This exclusion results in a shorter migration time through the column compared with the eluent molecules and, consequently, in a negative retention factor, k. While the influence of ionic strength on RP-HPLC has not been widely studied, ionic strength has a significant effect on solvent–water partitioning of ionizable compounds. Solute ions can form ion-pairs with ions from the background electrolyte or buffer in the aqueous phase. These ion-pairs can partition into octanol and other solvents with hydrogen-bonding capacity and the logP of the ion-pair increases with ionic strength of the aqueous phase. However, logP of the neutral species is not significantly affected by ionic strength. 3.9.10.4 Masking Agents The interactions between solutes and the free silanol groups present in the stationary phase packing can lead to significant errors. Masking agents are sometimes added to the mobile phase to minimize these interactions. Examples of masking agents and their volumes in the mobile phase include: 1. n-decylamine, 0.15% (v/v). 2. N,N-dimethyloctylamine, 0.15% (v/v). 3. Diethylamine, very small amounts (can increase the measurable chromatographic parameter to an equivalent logP = 8) [155]. 4. Triethylamine (TEA), propylamine, morpholine, glucosamine, N,N-various substituted ethylethanolamines, hydroxylamine, tetramethylammonium chloride, amino alkane derivatives, imine derivatives (a comprehensive list of masking agents used in the RP-HPLC mobile phase can be found in Kaliszan et al.) [156]. 5. Quaternary piperazine derivatives were applied in capillary electrophoresis for dynamically coating the silica walls but they can have potential applications in RP-HPLC [157, 158]. 5. Imidazolinium tetrafluorate ionic liquids, used for a number of basic drugs in RP-HPLC systems on LiChrospher RP-18 column (Merck, Darmstadt, Germany). The logkw values were calculated from the gradient of the range 95 to 0% (v/v) buffer in methanol-buffer mobile phases. The experiments were run with and without ionic liquids in both solvents. The masking agents influenced the retention of basic drugs on the silica-based stationary phases. The relationship between logP and logkw was significantly better for a series of basic drugs when the imidazolinium tetrafluorate ionic liquids were present in the mobile phase (r improved from 0.83 to 0.92) [156]. Although masking agents are employed to block the free silanol groups, they can introduce unwanted, additional ion-pair equilibria. Also, their presence in the mobile phase causes slow equilibration of the chromatographic system when changing the mobile phase composition and they cannot be used in gradient systems [159].

116

Advances in Chromatography

In many cases, reversed-phase chromatography is a good tool for correlation models between retention parameters and lipophilicity but there are many situations where the method fails to correctly model the partition in shake-flask experiments. This occurs mainly due to the differences between a chromatographic system and a shake-flask experiment: • • • • • •

Differences in hydrogen-bonding interactions Compound and mobile phase interactions with unprotected silanol groups on the stationary phase Pore size effects in packed columns poor reproducibility on different batches of the same column type, even from the same manufacturer Limited pH operating scale for many silica-based columns Long elution times for lipophilic compounds, sometimes 2 to 4 hours

Silica gel is by far the most used stationary phase for RP-HPLC applications in lipophilicity studies. However, silica-based column packings have a major disadvantage, expressed by their surface activity (i.e., the free silanol groups). Masking or controlling these effects is difficult in practice and can influence the chromatographic behavior of basic compounds and of acidic solutes [160]. Free silanol groups exist in all modern columns because the chemically bonded phase will cover less than 60% of them [161] and the effectiveness of the masking agents from the mobile phase is limited, covering about one third [162–164]. 3.9.10.5 Association in Water Solute association in water is difficult, especially due to the cohesion between water molecules. Some carboxylic acids produce dimers in water but this process can be neglected as the dimerization constants are usually less than 1 [165]. 3.9.10.6 Association in the Organic Phase The self-association of many solutes in non-polar solvents is particularly known for carboxylic acids, which have a strong tendency to dimerize at concentrations higher than 10–4 M [166]. Another form of solute self-association in the mobile phase is micelle formation. Compounds with hydrophobic chains and polar or ionizable groups will form micelles in water, reducing the value of partition coefficients. To measure the true logP value, the experiments should be conducted at solute concentrations lower than critical micelle concentration (CMC), which sometimes may prove difficult as CMC can be as low as 10–5 M [166]. 3.9.10.7 The Effect of Temperature It is well known that solubility can change dramatically with temperature, but will logP values change with temperature? It has been estimated that a difference logP = ±0.1 can be expected for ΔT = 10 K. The small effect of temperature on logP values can be explained by the increase of mutual solubility of water and octanol

Lipophilicity Measurements by Liquid Chromatography

117

[24]. Although the temperature has little effect on logP values at room temperature, accurate and reproducible results should be obtained in a thermostated environment. Poole and Poole have analyzed the published literature using RP-HPLC for lipophilicity measurements and concluded that many methods have been limited to studies of congeneric series of compounds, failing to provide similar quality results when structurally unrelated compounds are studied [6]. The chromatographic environment provides a partial model for the octanol–water system and has many properties that cannot be found in a dual phase liquid system, e.g., specific interaction with the stationary phase or flow-through porous media. An important problem arises from the lack of reproducibility between similar column packings from different manufacturers and, sometimes, from the same manufacturer. Silica-based columns are the most popular but many have a restricted pH range (2 to 7.5) due to silica instability in basic solutions. The authors applied the solvation parameter model to determine the individual intermolecular interactions responsible for the partition of the neutral molecules in octanol-water and in RP-HPLC systems, and compared these systems. They found that three chemically bonded phases (Supelcosil LC-ABZ [167, 168], Nucleosil RP18 [103], and Hypersil ODS [103]) provided similar system constants as the octanolwater in a shake-flask experiment when used with methanol–water mixtures as mobile phase. Good models [6] are also provided by Supecosil LC-ABZ dynamically coated with octanol from the mobile phase — 5 to 70% methanol containing 0.25% (v/v) octanol in MOPS buffer (20 mM, pH = 7.4) [15, 88] — and silica-gel columns coated with dipalmitoyl phosphatidylcholine with 10% (v/v) acetonitrile in water as mobile phase [167]. The chromatographic parameters used in correlation with experimental logP/logD values were logkw or logk in methanol-water systems. Published examples are shown in Table 3.5. 3.9.10.8 Gradient RP-HPLC Although isocratic chromatographic data can provide very good correlations with logPo/w, the experimental time can be quite long. Gradient HPLC is used frequently in all liquid chromatographic applications to speed up the elution time or to provide a different mobile phase polarity for the separation of mixtures of compounds with different polarities (e.g., natural products). During a gradient run, any point of the run time is equivalent to a particular organic solvent concentration. It is possible to estimate the eluent composition at any time by knowing the dead volume and the dwell volume of the chromatographic system [89].

3.9.11 CHROMATOGRAPHIC HYDROPHOBICITY INDEX (CHI) The Chromatographic Hydrophobicity Index (CHI) for gradient RP-HPLC was defined as a linear dependence to the retention time (tR) obtained by running a linear acetonitrile-buffer gradient: CHI = A tR + B

(3.35)

logkw, S

logkw, S

logkw, S

n-alkylbenzenes

Alkane derivatives

n-alkanols

logk, logkw

logk

Drugs, various in suppository formulations

Fullerene-C60 derivatives

Aurones, substituted logk and related compounds Dipyridylsulphides logk

logkw, S

Method

Alkanes

Compounds 50–95% acetonitrile-water

Mobile Phase logkw–S

Correlation 0.99–1.00

R2 Comments

Differences in logkw are useful for column standardization 60–95% methanol-water logkw–S 0.99–1.00 Differences in logkw are useful for 50–95% acetonitrile-water column standardization methanol-water logkw–S 0.997–0.998 The factors controlling logkw and S and solute size and hydrogenacetonitrile-water bond basicity 50–95% acetonitrile-water C18, various 0.99–1.00 Differences in logkw are useful for logkw–S manufacturers column standardization Hypersil 5 MOS (BST, 52% acetonitrile in aqueous logk – calculated 0.904–0.973 Correlation improves when Hungary) 0.083 M triethyl ammonium logP (CLogP, considering only subseries or phosphate (TEAP), pH 2.25 Daylight, USA) excluding outliers 30% methanol in 13 mM logk – biological 0.800 Silasorb C18, 7.5 μm, logMIC = 0.548 logk +1.241 (Lachema, Brno, phosphate buffer pH 7.0 MIC = Minimal Inhibitory (antituberculous) Concentration Czechoslovakia) activity 0.963 LiChrospher 100 RP-18 25–75% (v/v) methanol Good correlations for nine logkw– CLogP structurally unrelated drugs (ACD Labs) (Merck, Darmstadt, (range differs for different Germany) drugs) in 5 mM phosphate buffer, pH 7.2 — methanol-water 90:10 (v/v) logk, no other RP-HPLC was able to rank the Jupiter C18 correlation (Phenomenex, Torrance, with 0.1% (v/v) lipophilicity of the compounds; CA, USA); LiChrospher trifluoroacetic acid a comparison with reported RP-8 and RP-18 membrane absorption is in good (Merck, Darmstadt, agreement Germany)

C18, various manufacturers C18, various manufacturers C18, various manufacturers

Column, Packing

TABLE 3.5 Examples of RP-HPLC Correlations logP–logk

[174]

[173]

[172]

[171]

[169]

[170]

[169]

[169]

Ref.

118 Advances in Chromatography

logkw

N-butyl-paminobenzoate

N-hydroxyethyllogk, logkw amide of aryloxyalkylen and pyridine carboxylic acids

ln kw

N-alkylbenzenes

Various C18 columns methanol or acetonitrile in water or phosphate buffer (pH 7.0)

0.97–1.00 0.99–1.00 0.94–0.98

logk vs. Φ logkw vs. CLogP logkw vs. calculated logP (Rekker)

[175]

[175]

Despite the high pH, repeatability [178] and accuracy were good (all standard deviations less than 4%) 0.987–0.999 Detailed data presented [179] 0.989–0.995

0.998 0.99

Measurements performed at 23, [176] 27 and 40˚C — for neutral compounds (worse correlations when all compounds) considered) 0.995–0.999 Good correlation for ln (loge) not [177] only for log data

n-decylamine masking agent is not necessary because of the THF presence in the mobile phase

logkw or logk–logP (calculated Rekker)

logk vs. Φ logk vs. Φ logkw vs. logPo/w

Potential anticancer activity

logkw or logk–logP (calculated Rekker)

C18-SIL-X-5, 5 μm isocratic elution with 60–95% ln k vs. organic solvent column (Perkin Elmer) methanol in water concentration logPo/w vs. logkw Radial Pack Resolve C18 methanol-phosphate buffer logkw–logPo/w (Waters), 10 μm (2:1, v/v) pH 11.2

9 H-xanthene logkw, logk at Lichrosorb RP-18, 10 μm methanol in 0.02 M 3various morpholinopropane containing an sulphonic acid buffer (pH aminoalkanamides organic modifier 7.4) in the presence of nor nitrosoureido concentration decylamine (0.2%) as group masking agent 9 H-tioxanthene logkw, logk at Lichrosorb RP-18, 10 μm THF in 0.02 M 3containing an various morpholinopropane aminoalkanamides organic sulphonic acid buffer (pH or nitrosoureido modifier 7.4) group concentration Heteroaromatics, logk, logkw LiChrosorb RP-18 methanol-water cyclic aza acetonitrile-water compounds

Lipophilicity Measurements by Liquid Chromatography 119

logkw

Peptides

Correlation

0.95

0.995

R2

logkw determined 0.85–0.88 on the three on (1) columns vs. logP 0.91–0.94 of amino acid on (2) 0.93–0.94 side chain on (3)

(1) 20 mM MOPS buffer in logPo/w–logkw water, containing 0.15% (v/v) n-decylamine, pH 7.4 (MOPS = 3-morpholinopropane sulphonic acid) (2) 94.75% (v/v) methanol – 5% diethylamine in water (pH 7.4, adjusted with acetic acid – 0.25% v/v octanol) methanol-water, 80:20 (v/v) logPo/w vs. logk

Mobile Phase

(1) LiChrosorb RP-8, 5 0–40% methanol – 0.05 M μm (Merck); KH2PO4 (pH 7.0) (2) PLRP-S, 8 μm containing poly(styrenedivinyl benzene) (Polymer Laboratories); (3) Polyethylene, 20–40 μm, custom made column

Resolve C18, 5 μm (Waters)

logk

PCBs polychlorinated biphenyls (pesticides)

Column, Packing

C18, C8

Method

Non-congeneric logkw series of ionizable drug molecules

Compounds

TABLE 3.5 (CONTINUED) Examples of RP-HPLC Correlations logP–logk Ref.

[181] PCBs have very high lipophilicity, logP = 5–7, so the retention and accumulation in organisms is of particular concern Polyethylene has a homogenous [182] hydrophobic adsorption surface; additional inhibition of polar peptide-solid phase interaction not required. Polymer-based stationary phases are suitable for lipophilicity determination of peptides.

n-decyl amine added in the [180] mobile phase to minimize the solute-silanol interaction; also extends the upper measurable lipophilicity range to logD = 8.3 Octanol added in the mobile phase to mimic the octanol hydrogen bonding activity

Comments

120 Advances in Chromatography

logkw

logkw

Very lipophilic compounds, logP = 3.54–5.95

C18

Delabond C8 column, 5 μm, (Keystone Sci., Belfonte, PA, USA)

logkw–S logk

methanol in 0.05 M sodium log(1/MIC) vs. acetate buffer (pH 5.0) with logkw antimicrobial or without N,Nactivity (MIC) dimethyloctylamine methanol-water logkw vs. logPo/w

C18, Nucleosil 7 (Dalian 70–100% methanol – water Institute of Chemical Physics, Academic Sinica)

logkw

Teicoplanin antibiotics

Capcell Pak C18 (Shiseido, Japan)

logk, logkw

94.75% (v/v) methanol – 3% logkw–S (v/v) diethylamine pH 7.4; 18–92.5% methanol in 30 mM aqueous MOPS buffer methanol (30, 50, 70%, v/v) logk vs. logPo/w in phosphate buffer pH 7.4 logkw vs. LogP

Phenyl N-methyl and phenyl N,N-dimethylcarbamates Sulphur containing compounds

Hypersil, 3 μm MOPS

logkw, S

Phenones

0.957–0.999 Po/w measured shake flask. [184] Estimation of logPo/w (calc) from chromatographic data: logPo/w (calc) = 1.031 S + 1.682 logkw + 1.262, r = 0.930 The best chromatographic system: 61% methanol – 39% water 0.908 N,N-dimethyloctylamine usually [185] decreases the retention, except the situations when it may form ion pairs [186] 0.97

[183]

Differences in logkw are useful for [169] column standardization

0.987–0.997 50% methanol in the mobile phase provides the best results 0.981 for LogP predictions

0.99–1.00

Lipophilicity Measurements by Liquid Chromatography 121

122

Advances in Chromatography

where A and B are coefficients determined from similar chromatographic separations of suitable standards. CHI values were successfully correlated with logP, meaning that a single gradient run can provide the lipophilicity estimation for a compound. The coefficients A and B are dependent on the experimental conditions (e.g., flow rate, column length, gradient time, column) [187]. The CHI measurement takes 5 minutes and covers a 6 to 7 logP unit range of lipophilicity. CHI values are easy to calculate from retention times and can also be transformed into a logarithmic scale for comparison with logP/logD data [89]. Valkó et al. have shown that CHI values can be treated as linear free-energy related parameters, just like logP or logk values [188]. The authors used 20 different chromatographic systems for a diverse set of 29 compounds. The column packing was a range of silica-based ODS with different endcapping and carbon content from various suppliers; modified packing with diol and nitrile groups; polymer coated alumina-based, polymer-based, immobilized artificial membrane (with phospholipids); and permethylated -cyclodextrin. The mobile phase was 50 mM aqueous ammonium acetate (pH = 7.4) with acetonitrile used as an organic modifier. The fast gradient followed the pattern: 0% acetonitrile for the first 0.5 minutes, then to 100% acetonitrile in 3.5 minutes, maintained at 100% acetonitrile for 1 minute, then quickly to 0% acetonitrile in 0.2 minutes and maintained at this concentration for 2.3 minutes. Ten reference compounds were used for the method calibration and their logk values were plotted against the organic modifier concentration in the mobile phase (Φ). Usually three to five concentrations were used in isocratic HPLC experiments for this plotting and the ratio negative-intercept/slope (–logkw/S) (see definition in related isocratic HPLC text) provided the Φ0 values (the hydrophobicity index) for the reference compounds [109]. A good linear correlation has been shown to exist between the gradient retention times (tR) and the isocratically determined Φ0 values [187, 188]: ϕ0 = A tR + B

(3.36)

The coefficients A and B determined for standards were used to obtain CHI values (Equation 3.35). The correlation between the CHI values for the different ODS columns was quite good (r > 0.991) while the correlation between ODS and other types of stationary phase was not as good, showing that other columns have different selectivities. This method is a high-throughput technique for lipophilicity estimation of new molecules and the CHI values showed significant correlation (r = 0.996) with the molecular descriptors. The retention of tripeptide and dipeptide derivatives was measured on ODS, nitrile, immobilized artificial membranes, and permethylated -cyclodextrin columns with acetonitrile as organic modifier in 50 mM ammonium acetate buffer (pH = 7.4) for the gradient HPLC. The acetonitrile concentration was increased in steps of 5% and the elution of peptides took less than 10 minutes [189]. The fast-gradient method can determine quickly the structural parameters of similar oligopeptide derivatives. CHI values correlate well with molecular descriptors, making it possible to explain the hydrophobicity of molecules in terms of hydrogen-bond basicity and acidity as well as dipolarity/polarizability.

Lipophilicity Measurements by Liquid Chromatography

123

The logD values obtained with the CHI method differ significantly from the logD obtained in an octanol–water system. For neutral molecules, H-donor compounds usually have lower logD values than the logD obtained in the octanol-water system while ionisable molecules show a higher logD value [89]. The CHI lipophilicity scale from 1 to 100 can be used to express the logP scale, from 1 to 5, rather than providing an exact value for logP/logD. Perfluorooctyl stationary phase with 2,2,2-trifluoroethanol (TFE) gradient provides a very different selectivity in terms of the solvation equation compared with ODS stationary phases and acetonitrile or methanol used as organic modifiers in the gradient mobile phase [190]. Trifluoroethanol adsorbs more strongly on the fluorinated stationary phase than on C18 phases; therefore, its strong hydrogen-bond acidity and weak hydrogen-bond basicity will highly influence the retention selectivity [191]. This property was investigated for 60 structurally unrelated compounds on three different types of C18 columns: Luna C18 (Phenomenex), XTerra MS C18 (Waters), and Fluorophase RP (Keystone Scientific). For basic compounds, the aqueous phase was 50 mM ammonium acetate (pH = 10.5) and for acidic or neutral compounds the aqueous phase was prepared from 0.1% phosphoric acid (pH = 2.0). The organic modifiers were acetonitrile, methanol, and TFE. The fast gradient followed a 5 minute pattern starting from 0% organic solvent in the mobile phase for 0.5 minute, increasing linearly to 100% organic solvent in 2.5 minutes, maintained for 0.5 minute, then decreasing to 0% organic solvent. The biggest selectivity difference between the investigated stationary phases was observed when TFE was used in the mobile phase instead of methanol or acetonitrile. Another approach used linear and nonlinear gradient elutions for a standard mixture of aromatic hydrocarbon derivatives on a LiChrospher 100 RP-18 column, 5 μm particle size (Merck, Darmstadt, Germany). The linear gradient elution was applied to an acetonitrile-water mobile phase. The acetonitrile-water ratio changed from 20:80 to 100:0 (v/v) in 40 minutes. Using linear gradients, the exponential regression produced a poor estimation of compounds with low LogPo/w values [192]. The kG parameter was introduced by Krass et al. for the fast gradient elution [193]: kG =

VG − VD − VM VM

(3.37)

where VG is the gradient volume, VM is the dead volume and VD is the gradient “dwell volume” corresponding to the delay in the transport of the mobile phase between the gradient mixer and the top of the column. The authors reported good correlations between kG and logkw, not kw. Snyder et al. have shown that it is possible to obtain an approximate logkw value from a single gradient run and to obtain an exact value from two gradient runs [194]. The equation for the analyte retention time is: tR =

t0 log ( 2.3 k0 b + 1) + t 0 + t D b

(3.38)

124

Advances in Chromatography

where

b=

Vm ΔΦS tG F

(3.39)

and t0 is the column dead time; b is the steepness parameter; k0 is the k-value for 0% organic modifier in the mobile phase; tD is the gradient delay time equal to VD/F; VD is the equipment hold-up or “dwell” volume; ΔΦ is the Φ change during the gradient; Φ is the volume fraction of the strong solvent in the mobile phase; S is an analyte parameter equal to d(logk)/dΦ; VM is the column dead volume; tG is the gradient time; and F is the flow rate. Kaliszan et al. have compared the logkw values of 31 drugs, obtained from isocratic and gradient runs on columns packed with octadecyl bonded silica gel and polybutadiene-coated alumina with methanol or acetonitrile in buffer (mixtures of phosphoric, acetic or boric acids and sodium hydroxide to obtain the required pH) as mobile phase [195]. The logkw values for gradient and isocratic runs were calculated with the DryLab software [196]. Very good linear correlations (r = 0.953 to 0.983) were obtained between logkw (isocratic) and logkw (gradient) on both columns, for methanol and for acetonitrile in the mobile phase. When methanol was the organic modifier in the mobile phase, the correlation was slightly better than for acetonitrile. The linear correlation between the logkw (gradient) values and logP values reported in the literature was not as good (r = 0.804), probably due to the uncertainty of some reported logP values. Kerns et al. have applied a rapid-gradient LC-MS method for the initial profiling of lipophilicity in drug discovery [197]. They measured 70 structurally diverse drugs on a Polaris C18-A column (MetaChem Technologies, Torrance, CA) and a linear mobile phase gradient with solvent A as 100% 10 mM ammonium acetate (adjusted to pH = 7.4 with ammonium hydroxide and acetic acid) and solvent B as 100% acetonitrile. The gradient started with 0% B and after 2.5 minutes reached 95% of B, at 4.0 minutes was 95% B, at 4.1 minutes decreased to 0% B, and was kept to 0% B until the timer reached 5.5 minutes. The experiments were performed at 40˚C and a flow rate of 0.8 mL/min. The identity of all peaks was confirmed by mass spectroscopy (MS). The samples were also run at a pH of 6.9 and 7.9 to test the pH influence on the lipophilicity measurements, which was negligible. The method was calibrated with six compounds having logD values at pH = 7.4 (logD7.4) from 2.00 to 5.50. The experiment has shown that the volume of injected sample has a critical influence on the predicted lipophilicity and therefore the injection volume was set at 5 μL. The method provided a good linear correlation between the predicted logD7.4 values using the described method and the literature logD7.4 values (r = 0.94). However, the method should not be used for compounds with molecular mass less than 200 or retention times less than 1.5 minutes because the logD7.4 values are not well predicted. The reproducibility of the method was very good with r = 0.999.

Lipophilicity Measurements by Liquid Chromatography

125

The relative hydrophobicity and lipophilicity of 63 compounds with known permeability through the blood-brain barrier was measured by partitioning in dextran-poly(ethylene glycol)-buffer (pH = 7.4), octanol-buffer (pH = 7.4) and by RP-HPLC [198]. Very lipophilic compounds with logD > 3 could not be partitioned in the aqueous Dex-PEG two phase system. The experimental results suggested that the logD7.4 values obtained by HPLC correlate better with the compound’s ability to cross the blood-brain barrier than the shake-flask experiments for octanol-water. There was a rather poor correlation (r = 0.918) between the lipophilicity values obtained by octanol-water partitioning and the HPLC experiments. Jandera et al. have introduced the oligomeric gradient capacity obtained by gradient HPLC for the separation of compounds with small differences in polarity or for separating samples containing compounds with a wide polarity range [199]. The type of stationary phase used in gradient RP-HPLC experiments significantly influences the sample polarity range that can be covered by using a gradient of organic solvents in water/buffer. The gradient lipophilic capacity (P1) is based on the retention of standard compounds with a repeat lipophilic structural unit (e.g., methylene) and can help the characterization of polarity range available for gradient separation of different samples. Oligo(ethylene glycol)s were analyzed on ten different C18 stationary phases and 5-minute gradient runs with methanol, ethanol, and acetonitrile as mobile phase.

3.9.12 ABRAHAM-TYPE EQUATIONS

FOR

RP-HPLC

A useful link between chromatography and partitioning is provided by Abraham’s solvation equations. In these equations, solute molecules are characterized by “descriptors” that include size, hydrogen bonding acidity, and hydrogen bonding basicity. These descriptors can be derived from measured data or rapidly estimated from structure by a fragment scheme and used for a qualitative chemical insight into the behavior of molecules in two-phase systems. Abraham has written or coauthored hundreds of papers in this area, making him one of the most widely cited authors in the chemical literature. The Abraham equation is usually written as logSP = c + r R2 + s Π2H + a Σα2H + b Σβ2H + v Vx

(3.40)

where logSP is a property for a series of solutes in a fixed solvent system (e.g., logP, logk); c is a constant (close to zero for a well-derived equation); r, s, a, b, and v are coefficients whose sign and magnitude are related to the solute property; and the remaining terms are solute descriptors. Abraham’s original symbols are widely used in the literature but the alphabetical versions are more convenient for computer keyboard entry and have been used in Absolv software (Sirius Analytical Instruments Ltd., Forest Row, Sussex, U.K.) [116].

126

Advances in Chromatography

Abraham’s Original Symbol

Alphabetical Version of Symbol

R2

E

Π2 H Σα2H Σβ2H Vx

S A B V

Meaning of Descriptor An excess molar refraction that can be obtained for complex solutes by the addition of solvents Solute dipolarity/polarizability The solute overall or effective hydrogen-bond acidity The solute overall hydrogen-bond basicity The McGowan characteristic volume that can be calculated from molecular structure

A useful introduction to Abraham’s method is provided in a recent reference. This paper explains the origins and meanings of the descriptors, shows how descriptors are calculated from software, and explains how the approach may be used to predict the oral absorption of drugs and penetration of the blood-brain barrier [200]. Zissimos et al. described several methods for calculating Abraham descriptors from seven HPLC systems and from measurements of logP in partitions between water and four different partition solvents: octanol, chloroform, cyclohexane, and toluene [201, 202]. Many papers have published equations for predicting properties by Abraham’s method. For example, Yamagami et al. have determined these parameters for water−octanol, water−chloroform, water−octane and water−dibutyl ether systems for pyrazines, pyrimidines, and pyridazines [203]. As well as predicting properties in shake-flask systems, Abraham’s method has been used to interpret the behavior of molecules in chromatographic systems. Luna C18 columns can provide a good model for logP estimation of neutral compounds with no significant hydrogen-bonding acidity when using acetonitrileaqueous buffer (40%, v/v) as mobile phase. The mobile phase pH was adjusted to 7.4 for neutral compounds, 10.5 for bases and 2 for acids. The same chromatographic system becomes unsuitable for compounds acting as hydrogen-bond acids [204]. The Abraham equation was: logk = –0.14 + 0.31 E – 0.58 S – 0.51 A – 1.58 B + 1.79 V

(3.41)

showing reasonable correlation with the octanol−water partition system. The correlation between chromatographic parameters obtained in isocratic and gradient RP-HPLC was studied for 55 unrelated compounds. Gradient RP-HPLC was performed on ODS2-IK5 Inertsil column (Capital HPLC Ltd., Broxburn, Scotland). For the acidic compounds, the mobile phase was a mixture of aqueous 0.1% phosphoric acid and acetonitrile. For the basic compounds, the mobile phase was 50 mM ammonium acetate buffer, adjusted to pH = 9.5 with ammonium hydroxide. The isocratic chromatographic runs were done at various concentrations of acetonitrile in the mobile phase: the gradient runs started with 0% acetonitrile for 1.5 minutes, increasing linearly to 100% acetonitrile during the following 9 minutes, then coming back to 0% acetonitrile in 2 minutes.

Lipophilicity Measurements by Liquid Chromatography

127

The isocratic slope S (Equation 3.22), logkw, logk50 (at 50% acetonitrile in the mobile phase), the isocratic chromatographic hydrophobicity index (ϕ0), and the gradient chromatographic hydrophobicity index, CHI, were investigated via an Abraham-type equation for logP. The regression coefficients were similar in most cases but not identical. All the retention parameters obtained by gradient RP-HPLC are a measure of lipophilicity and are influenced by the same properties, i.e., solute dipolarity/polarizability, solute capacity to act as hydrogen bond donor or acceptor, molecular volume. While an increase of solute polarity or hydrogen bonding capacity will decrease the chromatographic retention parameters, the increase in molecular volume will lead to an increase in lipophilicity. The study demonstrates that the CHI values obtained by rapid-gradient RP-HPLC encode the same information with isocratic logk values obtained at 30 to 70% acetonitrile [205]. A computer program called Absolv has been introduced to the market for calculating Abraham descriptors as well as solute properties. The first version was introduced by Sirius in 2000. It was superseded by a new version created by Pharma Algorithms Inc. (Toronto, Canada) that was introduced as an accessory to their ADME Boxes software in 2004. Although there is very large number of correlations between chromatographic parameters and logPo/w values, it cannot be stated that the octanol-water system and the chromatographic partitioning are comparable [89].

3.10 LIPOPHILICITY ESTIMATION BY THIN-LAYER CHROMATOGRAPHY Thin-layer cromatography (TLC) was the first form of liquid chromatography used for the measurement of partition coefficients. The method is simple and evolved along with other chromatographic methods to a high standard of reproducibility and reliability. High-performance thin-layer chromatography (HPTLC) used with automated sample applicators and automated densitometric detection will provide quantitative and qualitative results having similar accuracy and reproducibility to HPLC. The retention parameter used in TLC is Rf and is defined by the ratio between the distance migrated on the plate by the compound (x) and the distance migrated by the solvent front in the same time (xf) with all distances measured from the start line: Rf =

x xf

(3.42)

Martin and Synge, Nobel Prize winners in 1952 for the invention of partition chromatography, were the first to provide a thermodynamic model for this method [206]. They showed that Rf values are related to P values according to Equation 3.43 [207]:

128

Advances in Chromatography

⎞ ⎛ 1 P = a⎜ − 1⎟ R ⎠ ⎝ f

(3.43)

where a is a constant. This finding was followed by the introduction of the RM value [208], defined in a similar way to the capacity factor (k) in HPLC. This relationship is shown in Equation 3.44 and in subsequent correlations between RM values and logP: RM =

1 −1 Rf

(3.44)

In the late 1960s, Snyder and Soczewinski contributed to the development of the semi-empirical model of adsorption chromatography [209, 210]. In 1965, Boyce and Milborrow published the first correlation between RM values obtained on TLC plates and biological activity using the example of some N-alkyltritylamines [211]. TLC has some advantages over HPLC for the assessment of logP: • •

•

The throughput of TLC is higher than than of HPLC. The impregnated layers used in TLC are stable at extreme pH values. Most packing materials used in HPLC columns, especially the silica based ones, are stable only between pH = 2 and pH = 8. Solvents with strong UV absorbance (e.g., acetone) can be used for reversed-phase experiments in TLC because the solvent is completely removed from the plate before the detection. This result means that the entire range of solvents can be used in TLC, unlike HPLC where the solvent use is limited by the instrumentation.

Lipophilicity estimations by TLC are usually done on silica gel layers, either C8 or C18 derivatized or impregnated with a nonpolar liquid phase, and aqueous organic mobile phases. The experimental setup uses the RM values obtained at a specified mobile phase composition or extrapolated RM0 values for a 100% aqueous phase for the correlations with logPo/w data in a similar way as the correlations in RP-HPLC.

3.10.1 RELATIONSHIP BETWEEN RM AND MOLAR FRACTION X THE ORGANIC MODIFIER IN THE MOBILE PHASE

OF

The literature reports a large number of very good linear correlations between the RM and logP values for neutral and ionic compounds. The correlations between logP and logk (in RP-HPLC) or RM (in reversed-phase TLC, or RP-TLC) are frequently linear for nonionic and ionic compounds. When a compound contains one or more ionizable substituents, the pH and the ionic strength of the eluent will modify the apparent lipophilicity according to the ratio of neutral and ionized species [212].

Lipophilicity Measurements by Liquid Chromatography

129

The influence of mobile phase pH and ionic strength on retention was investigated for phenothiazine drugs in [213, 214]. As mentioned earlier, RM was defined in TLC in a similar way with the retention factor (k) in HPLC. Taking into consideration Equation 3.17 defining k, a similar relationship can be written for RM: RM = K

VS = Kr VM

(3.45)

where r is the phase volume ratio (VS/VM) for the stationary and mobile phase, respectively, and K is the partition coefficient of a compound between the stationary phase and the mobile phase. If the stationary phase is octanol and the mobile phase is water (or aqueous buffer) and if the compound is in the neutral form, K equals Po/w. The term r is a constant for a defined chromatographic system. If the compound is partially ionized under the experimental conditions, K equals D (the distribution coefficient). In this case, the logarithm of Equation 3.45 can be written as: RM = logP + log r

(3.46)

RM = logD + log r

(3.47)

or

Hulshoff and Perrin [213, 214] expressed RM values taking the acidity constant into account: ⎛ P = D ⎜1 + ⎜⎝

⎡⎣ H + ⎤⎦ ⎞ ⎟ K a ⎟⎠

(3.48)

⎛ P = D ⎜1 + ⎜⎝

Ka ⎞ ⎟ ⎡⎣ H + ⎤⎦ ⎟⎠

(3.49)

for a monoprotic base and

for a monoprotic acid. Rearranging the equations, RM can be expressed in Equation 3.50 for a base: RM = logP + log f + log r

(3.50)

where f = Ka/(Ka + [H+]), the fraction of the drug present as free base in mobile phase.

130

Advances in Chromatography

Extensive experimental results show the existence of a linear relationship between the retention parameter (RM) and the concentration of organic modifier (usually methanol) in the mobile phase (Φ) in a similar way with the RP-HPLC experiments: RM = a + b ϕ

(3.51)

Because RM and logP are related, a similar equation can be written for logP: logP = a + b ϕ

(3.52)

where a and b are coefficients. Therefore Equation 3.51 becomes: RM = a + b ϕ + log f + log r

(3.53)

When ϕ equals zero (i.e., the aqueous mobile phase does not contain any solvent), logP = a. Equation 3.11 may also be written: RM = logP + log r + log f + b ϕ

(3.54)

RM – log f = logP + log r + b ϕ

(3.55)

Rearranging:

The linear plotting of RM – log f values vs. ϕ will have the slope b and the intercept logP + log r. When ϕ = 0, RM = logP + log r

(3.56)

RM0 is the value at ϕ = 0 (0% organic modifier in the mobile phase), when the experimental conditions are set to have the compounds only in the nonprotonated form. Almost 50 years ago, Boyce and Milborrow proposed the use of RM values obtained from TLC experiments as a reliable alternative to the classical logP determinations [211]. As in HPLC, methanol and acetonitrile are the most popular organic modifiers for the polar mobile phase and silica gel impregnated with a nonpolar solvent or chemically bonded with C18 were used as a nonpolar stationary phase. Planar chromatography has a great advantage over HPLC: it can cope with solvents that are strong absorbers in UV (e.g., acetone) because the solvents are completely removed from the plate before detection. Very good linear correlations were also obtained between chromatographic systems using acetone as organic modifier in the mobile phase and logPo/w. In binary aqueous organic eluents, the organic solvent has a decisive influence on the overall chromatographic distribution equilibrium reflected in the RM values of the investigated solutes. Generally, there is a good linear correlation between RM and X (Equation 3.57) where X is the fraction of organic

Lipophilicity Measurements by Liquid Chromatography

131

modifier in the aqueous mobile phase. However, deviations from linearity are frequently found when the water concentration in the mobile phase is more than 90%. RM = ao + a1 X = RM0 + a1 X

(3.57)

The RM values extrapolated to zero organic solvent content — X = 0 : (RM)X=0 = a0 = RM0 — depend on the structure of the compounds and are considered to be a good estimation for lipophilicity in the same way that logkw values obtained by RPHPLC are. The intercept values (RM0) can be correlated with the compound lipophilicity and the slope (a1) can be considered a measure of mobile phase strength or, in other words, the mobile phase contribution to the solute retention [12]. The absolute values of the negative slopes (a1) of the TLC Equation 3.57 are dependent on the organic modifier used in the mobile phase and decrease in the order acetone > acetonitrile > methanol [215]. In an ideal experiment using a particular stationary phase, the nature of the organic solvent (methanol, acetonitrile, or acetone) in the mobile phase should have little influence on the extrapolated RM0 values. Experimental results confirm this idea, although there are many exceptions. The slopes and intercepts of the RM – ϕ linear correlations usually have values close to 1 and 0, respectively, showing the correct nature of the correlation. In RP-TLC, very hydrophilic compounds can migrate reliably even at 0% organic solvent in mobile phase. The experimentally measured RM0 values correlate well with the extrapolated RM0 values, showing the validity of the presumption that the real RM0 values follow the linearity between RM and X [216]. To support these findings, the chromatographic behavior of 33 xanthine derivatives was studied by RP-TLC on silica gel GF254 impregnated with silicone DC 200 (350 cS) (Applied Science Labs, State College, PA) at pH = 7 and pH = 9 [217, 218]. A very good correlation was reported between the extrapolated (RM0extrap) and experimental RM0 values (RM0exp): RM0exp = 0.036 (± 0.010) + 0.972 (± 0.013) RM0extrap

(3.58)

n = 33, r = 0.997, SD = 0.043, F = 5523 where n is the number of compounds, r is the correlation coefficient, SD is the standard deviation and F is the statistical parameter for the F-test. The intercept and slope were very close to 0 and 1, respectively, showing the validity of the extrapolated methods. Similar results were reported for six serotonergic ligands for RP-TLC experiments with acetone, acetonitrile, or methanol as organic modifier [219].

3.10.2 INFLUENCE

OF PH

The experimental values (RM0exp) of a series of xanthine derivatives were measured by RP-TLC on silica gel GF254 plates when the mobile phase was a glycine buffer, with pH = 7.0 and pH = 9.0 providing the following relationship [217]:

132

Advances in Chromatography

RMexp pH 7.0 = –0.481 (± 0.057) + 1.079 (± 0.100) RMexp pH 9.0

(3.59)

n = 27, r = 0.907, SD = 0.298, F = 116.7, P < 0.005 A similar study was performed on 14 adenosine derivatives and the relationship between the RMexp values of adenosine derivatives at pH = 7.0 and those at pH = 9.0 are given by Equation 3.60: RMexp pH 7.0 = –0.062 (± 0.092) + 1.029 (± 0.105) RMexp pH 9.0

(3.60)

n = 14, r = 0.943, SD = 0.168, F = 96.15, P < 0.005 The higher standard deviation obtained for xanthine derivatives suggests that not all the compounds were equally affected by the increase in the pH of the chromatographic system. By contrast, the increase in pH did not seem to have a significant influence on the chromatographic migration of the adenosine derivatives. In this case, the intercept and slope have values closer to 0 and 1, respectively. The measured RM values for strong bases by RP-TLC on silica-based stationary phases are significantly affected by the influence of the pH of the mobile phase on the silanol groups. Dross et al. have studied the effect of the mobile phase pH on RM values for 17 monoamine oxidase inhibitory drugs by RP-TLC on paraffinimpregnated silica gel using water-methanol systems containing acetic acid, sodium acetate, and sodium chloride [220, 221]. The influence of solvent pH on RM values was also investigated on a series of 1,2-benzisothiazol-3(2H)-ones. The RM0 values were measured by RP-TLC using acetone, acetonitrile, and methanol as organic solvent moderator [222].

3.10.3 INFLUENCE OF MOBILE PHASE COMPOSITION ON RM0 VALUE Methanol, acetone, and acetonitrile are the most widely used organic solvents in the aqueous phase for RP-TLC experiments. Biagi et al. have used all these solvents in binary aqueous mobile phases at pH = 7.0 for lipophilicity measurements of 15 steroids on silica gel GF254 layers impregnated with silicone oil [215]. The linear relationships between the extrapolated RM0 values for the three organic modifiers have shown very good correlation coefficients, described by Equation 3.61 and Equation 3.62: RM0methanol = 0.116 (±0.061) + 0.947 (±0.040) RM0acetone

(3.61)

n = 15, r = 0.989, SD = 0.077 RM0acetonitrile = –0.014 (±0.048) + 0.966 (±0.032) RM0acetone n = 15, r = 0.993, SD = 0.061

(3.62)

Lipophilicity Measurements by Liquid Chromatography

133

where n is the number of studied compounds, r is the correlation coefficient and SD is the standard deviation. Many reported studies have concluded that the RM0 values obtained from extrapolations based on different organic solvents in the mobile phase are very similar [215, 219, 223, 224]. Alternatively, some results suggest that better correlations with logPo/w are obtained when methanol is used [220]. Values for lipophilicity based on RM0 values of 17 monoamine oxidase inhibitory drugs was determined by RP-TLC [225]. The effect of salt concentration on the reversed-phase retention was investigated by adding ammonium chloride to the mobile phase. Each drug showed predictable retention behavior and the RM values decreased linearly with the increase of methanol concentration in the mobile phase. The presence of ammonium chloride led to a decrease in retention and this effect was more evident at low salt concentration, indicating that when the salt concentration is close to the saturation value, the retention will not change greatly. Also, the influence of ammonium chloride is dependent on the methanol concentration.

3.10.4 INFLUENCE

OF

STATIONARY PHASE

ON

RM0 VALUE

The stationary phase has a crucial effect on the chromatographic behavior of different solutes. The stationary phase used in a lipophilicity study is usually hydrophobic, e.g., cellulose [226] or silica based. The large majority of lipophilicity studies by RP-TLC are performed on silica gel layers, impregnated with hydrophobic, oily liquids, or chemically derivatized with hydrocarbon chains (C8 or C18). The plate impregnation is usually done with octanol [227], liquid paraffin [211, 228], or silicone oil [215, 216, 229]. Paraffin oil and silicone oil were popular attempts to produce a nonpolar stationary phase on a silica gel layer. The chromatographic migration of benzophenones in RP-TLC has shown that the replacement of paraffin oil by silicone oil can cause significant changes in the retention of compounds containing particular functional groups [230]. The compounds have different retention on kieselguhr and silica gel layers coated with silicone oil but the influence of the solid support seems negligible when coating the layer with paraffin oil. These subtle influences on the retention of compounds make it difficult to explain the changes in lipophilicity. Although silica gel is the most widely used stationary phase in RP-TLC experiments for lipophilicity studies, the relationship between the retention on other stationary phases and lipophilicity has been investigated. The chromatographic behavior of some neutral, acidic, and alkaline compounds was studied by RP-TLC using different layers: Kieselgel 60 HF254 (Merck), MN-Aluminium oxide G (Macherey-Nagel) and Cellulose powder MN 300 (Macherey-Nagel), all impregnated with different amounts of paraffin oil. The mobile phase was water, water-methanol (1:1 or 3:1, v/v) or water-acetonitrile (1:1 or 2:1, v/v). The retention of all compounds increased in all experiments with the amount of paraffin oil present in the stationary phase. The lowest RM value was determined on cellulose support. Alkaline compounds showed higher RM values on silica gel layers and acidic compounds on aluminium oxide [231]. The lipophilicity of 21 amino acids was investigated on different stationary phases: silica gel, aluminium oxide, cellulose, Kieselguhr and mixtures of these,

134

Advances in Chromatography

impregnated with paraffin oil and using water as mobile phase, and a comparison of RM values obtained in all systems was presented by Gullner et al. [232] While the nature of the stationary phase has a strong influence on the absolute RM values, very good correlations have been obtained between RM values obtained for the same series of compounds on different stationary phases [217]. For instance, the correlation between the RM values obtained on RP-TLC plates impregnated with silicone and those obtained on C18 RP-HPTLC plates (Whatman KC18F) for 43 of xanthine and adenosine derivatives at pH = 7.0 is given by Equation 3.63: RM0 silicone = –0.272 (± 0.040) + 0.610 (± 0.023) RM0 C18

(3.63)

n = 43, r = 0.972, SD = 0.154 The intercept value, significantly different from zero, is due to the difference in the nature of the stationary phases used. The C18 RP-HPTLC system seems to be more lipophilic than silicone RP-TLC and the extrapolated RM0 values obtained on C18 plates are higher than the experimental silicone RP-TLC values; therefore, the slope is lower than 1. Although there are significant differences between the two systems, the two sets of extrapolated values, RM0 silicone and RM0 C18, have a satisfactory correlation for the studied xanthine and adenosine derivatives [217]. Data about the influence of TLC layers coated with complexating, ion-exchanging, and ion pairing agents on retention, with particular attention paid to the determination of molecular lipophilicity and comparison between impregnated and chemically modified silica gel layers, was studied by Garssini-Strazza et al. [233] The influence of pH on the surface of the TLC layer was also investigated for silica gel. The separation of peptides at a surface pH of 2.0, 4.5, 7.5, and 9.0, using methanol as organic modifier in the concentration range 0 to 90% (v/v), was discussed in relation with the silanophilic effect and the peptide structures by Cserháti and Szögyi [234]. Although silica gel is the general choice for the stationary phase when investigating compound lipophilicity, there are studies showing good correlations on alumina oxide. The lipophilicity of 18 phenols and aniline derivatives was investigated on DC-Alumina-oxide F254 (Merck, Darmstadt, Germany) and Polygram UV254 silica gel plates (Macherey-Nagel, Düren Germany) impregnated with 2.5 and 10% (v/v) paraffin oil in n-hexane and a mixture of methanol-water as mobile phase. The degree of stationary phase impregnation has a greater effect on the compound retention than the support chemical nature [235].

3.10.5 SPECIFIC HYDROPHOBIC SURFACE AREA

OF THE

SOLUTE

The slope (a1) of the linear correlation between RM values and the concentration of organic modifier (Φ) in the mobile phase (Equation 3.57) is related to the specific hydrophobic surface area of the solute. Many studies have shown that the lipophilicity as well as the specific hydrophobic surface of solutes can be determined from the linear relationship between the RM values and the concentration of methanol in the mobile phase [236–241].

Lipophilicity Measurements by Liquid Chromatography

135

The lipophilicity and specific hydrophobic areas of 18 nonsteriodal anti-inflammatory drugs have been determined by RP-TLC for subsequent use in QSAR studies. The chromatographic experiments were performed on Polygram UV-3254-3 plates (Macherey-Nagel, Düren, Germany) impregnated by overnight development in nhexane-paraffin oil (95:5, v/v); water-methanol was used as mobile phase, the methanol concentration ranging from 0 to 35% (v/v) in steps of 5%. The experimental data showed that the hydrophobicity parameters for nonsteroidal anti-inflammatory drugs are highly dependent on the presence of acetic acid, sodium acetate, and sodium chloride in the mobile phase [236]. The drugs fall into two groups, dissociable acidic and alkaline. The influence of salts on lipophilicity and hydrophobic surface area was investigated by RP-TLC for 11 sulfosuccinic acid esters [237]. The experiments were performed in ion-free eluents and in eluents containing sodium chloride, potassium chloride, magnesium chloride, and calcium chloride. It is important to note that the salt influence was greater for specific hydrophobic surface area estimations than for hydrophobicity. An interesting result was reported for the RP-TLC data on the lipophilicity and specific hydrophobic surface area of 31 commercial pesticides with water-methanol mixtures as mobile phase. The RM values decreased linearly with increasing concentration of methanol in the mobile phase. Although the chemical structures of pesticides were extremely different, the lipophilicity and specific hydrophobic surface area of pesticides were highly intercorrelated [238]. Principal component analysis (PCA) showed that when both lipophilicity and specific hydrophobic surface area were taken into consideration, the herbicides, insecticides, and fungicides could not be distinguished, all forming only one class. The differences in the biological activity of the investigated pesticides cannot therefore be attributed only to their lipophilicity or their specific hydrophobic surface area. The determination of lipophilicity and specific hydrophobic surface area by RPTLC following the methods described above was performed on many different classes of compound, such as fused-nitrogen heterocycles [239], anti-hypoxia drugs [242], or bioactive heterocyclic compounds [241].

3.10.6 RELATIONSHIP BETWEEN INTERCEPTS (RM0) AND SLOPES (a1) Many publications use the term “congeneric” to classify a group of structurally related compounds. In RP-TLC the “congenerity” of compounds can be expressed by the linearity between the extrapolated parameters RM0 and the slopes a1. In order to understand the relationship between slopes and intercepts of the TLC equations, the physico-chemical meaning of these two parameters must be discussed. The intercept (RM0) can be considered a measure of the compound partitioning between a polar mobile phase and a nonpolar stationary phase. The value RM0 can sometimes be measured directly but, for the majority of the compounds, RM0 is obtained by extrapolating the retention parameter RM to a theoretical retention that would be obtained with 100% aqueous mobile phase. The slope a1 indicates the rate at which the solubility of solute increases in mobile phase. The correlation between RM0 and

136

Advances in Chromatography

a1 is usually linear for series of congeneric compounds (Equation 3.64) [215, 216, 243]: RM0 = b0 + b1 a1

(3.64)

The congenerity can be broken by the presence of ionizable groups, which can modify the interactions of the components with the nonpolar stationary phase or polar mobile phase. The physicochemical significance of the slope can be expressed: 1. As reflecting the nature of the compound and mainly determined by the interaction between the solute and the mobile phase 2. In terms of the so-called displacement model, which assumes that a monomolecular layer of mobile phase molecules is adsorbed on the stationary phase and that any other molecule from the mobile phase, solvent or compound, can be retained on the stationary phase only by displacing the molecules already absorbed 3. On the concept of hydrophobic surface area availability [216] RM and slope increase when the retention increases, suggesting that both parameters are related to the same physicochemical factors. Table 3.6 shows results for a series of ionizable quinolines studied on silica gel impregnated with silicone, using methanol, acetone, or acetonitrile in the mobile phase. The results for all three systems seem consistent and good linear correlations were obtained between the parameters RM – ϕ and RM0 – a1. However, there are differences between the piperazine ring and the cinoline ring and sometimes not all the members of a chemical series fit the same linear correlation [216, 243]. The relationship between the intercept and the slope of the linear RM – ϕ dependence was reviewed for series of steroids, nitroimidazothiazoles, phenols, triazines, prostaglandins, 5-nitroimidazoles, dermorphins, 4-nitropyrazoles, 1,4dihydropyridines, naphthalenes, and quinolines [216]. The investigations of different series of compounds in different chromatographic environments led to the conclusion that the relationship between the slope and the intercept of the TLC equations (Equation 3.57) is linear and can be considered a fundamental aspect of the lipophilicity estimations by RP-TLC but only for strictly congeneric compounds [216, 219]. A recent study used the parameter C0 (where C0 = RM0/a1) instead of RM0. However, the correlations between logP and C0 for series of thiazoles are not better than between logP and RM0. Different organic modifiers in the mobile phase were also investigated [244]. Another interesting observation is that the slope of TLC equations in different solvent systems correlate well with the solvent strength parameter (E0) of acetone, acetonitrile, and methanol [215]. The solvent strength is defined by the 1/E0 value and is 1.78 for acetone, 1.54 for acetonitrile, and 1.5 for methanol. The ratios between the slopes in different solvents and the ratios between solvent strength values are shown in Table 3.7 [215, 243]. The ratio between the slopes in two different solvent systems is close to the ratio between the 1/E0 values for the corresponding solvents

Antiarrhythmic and antihypertensive, 1-[2 hydroxy- or 1-[2acetoxy-3-(4-aryl-1piperazinyl)pyrrolidin-2one derivatives

Antiarrhythmics

2-amino-3-cyano-4,5diphenylfurane derivatives

2-(aminoacetylamino) thiazole derivatives

Amido esters of ethanolamine derivatives

Compounds Mobile Phase

silica gel 60 F254 silanized precoated plates (Merck, Darmstadt, Germany) RP-18 F254S plates (Merck, Darmstadt, Germany)

acetonitrile (65/35 to 50/50, v/v) in 0.02 N ammonium acetate buffer (pH 7.4) 21.4–78.5% (v/v) acetone, or 14.3% (v/v) methanol in water 0.996–0.974

0.900

C18 bonded silica gel methanol-water 0.985–0.993 (Merck, Darmstadt, Germany) silica gel plates methanol-water (7:3, impregnated with 5% v/v) and 2% aqueous liquid paraffin in light ammonium hydroxide petroleum ether silica gel 60 F254 RP-18 40–80% methanol in 0.990–0.999 (Merck, Darmstadt, water Germany)

Stationary Phase

Linear Correlation RM – ϕ r

0.992

0.990

Linear Correlation RM0 – a1 r

0.991

0.897 for RM at 50% acetonitrile

0.918

0.979

0.929

Linear Correlation RM0 – logP r

Correlation with calculated values logP (SYBYL 6.4 Tripos)

Calculated logP values using Rekker’s fragmental constant method Correlation with calculated values CLogP

RM values range from 0.153 to –0.753

Tab 2

Comments

TABLE 3.6 Examples of RP-TLC Lipophilicity Studies on Different Classes of Compounds (where r = Correlation Coefficient)

[259, 260]

[252, 255]

[258]

[256]

[257]

Ref.

Lipophilicity Measurements by Liquid Chromatography 137

Polygram UV254 plates (MachereyNagel, Düren, Germany)

Stationary Phase

β-adrenoceptor antagonists n-octanol on RP-TLC plates 1-aryl-2-imine-3silica gel RP-18 F254S arylaminocarbonyl (Merck Darmstadt, imidazolidines Germany) Benzamides silica gel 60 F254 silanised precoated plates (Merck, Darmstadt, Germany) 1,2-benzisothiazol-3(2H)- precoated RP-18 F254S ones plates (Merck, Darmstadt, Germany)

Anticancer drugs

Compounds

10–75% (v/v) methanol, ethanol, 1propanol, 2-propanol, actonitrile, dioxane or tetrahydrofurane in water phosphate buffer (pH 7.4) 60–90% (v/v) methanol in 4% aqueous acetic acid acetonitrile (65/35 to 50/50, v/v) in 0.02 N ammonium acetate buffer (pH 7.4) 40–80% acetone in water; 20–60% acetonitrile in water; 40–80% methanol in water (all v/v)

Mobile Phase

Linear Correlation RM0 – a1 r Linear Correlation RM0 – logP r Comments

0.992–0.999 (acetone) 0.987–0.996 (acetonitrile) 0.992–0.999 (methanol)

0.900

0.999–0.984

0.992–0.999 (acetone) 0.987–0.996 (acetonitrile) 0.992–0.999 (methanol)

0.992 (calibration compounds) 0.716 for RM at 50% acetonitrile

Correlation with calculated values CLogP

Experiment conducted at 37˚C logPTLC = 3.870–5.452

lipophilicity mapping for 22 anticancer drugs, using multivariate mathematicalstatistical methods

Linear Correlation RM – ϕ r

TABLE 3.6 (CONTINUED) Examples of RP-TLC Lipophilicity Studies on Different Classes of Compounds (where r = Correlation Coefficient)

[264]

[252, 255]

[32, 262]

[212]

[261]

Ref.

138 Advances in Chromatography

silica gel 60 F254 silanized precoated plates (Merck, Darmstadt, Germany) Chalcones and cyclic silanized silica gel 60 chalcone analogs F254 (Merck, Darmstadt, Germany) activated by washing with methanol, drying and heating at 106˚C for 1 hour Carbonyl derivatives 1-(1- silica gel RP-18 F254S arylimidazolin-2-yl)-3 (Merck Darmstadt, arylureas Germany) Estrane and secoestrane RP-18 silica gel F254 derivatives (Merck, Darmstadt, Germany) 2-hydrazinothiazolic silica C8 plates F254 derivatives (Merck, Darmstadt, Germany)

β-blockers

75–90% (v/v) methanol in water

60–90% (v/v) methanol in 4% aqueous acetic acid methanol-water

acetonitrile (65/35 to 50/50, v/v) in 0.02 N ammonium acetate buffer (pH 7.4) 40–75% (v/v) methanol or acetone in water

0.999

> 0.99

0.992–0.999

> 0.99

0.900

[32, 262]

logPTLC = 3.394–5.193

0.930–0.937

logP values calculated with ACD/logP or with Rekker’s method. Although the chromatographic behavior is ideal, no direct correlation was found with biological activity.

[266]

Calculated logP (Rekker) [263]

[265]

logPo/w measured by shake-flask

0.996

0.992 (calibration compounds) 0.868

[252, 255]

Correlation with calculated values CLogP

0.960 for RM at 50% acetonitrile

Lipophilicity Measurements by Liquid Chromatography 139

Mobile Phase

40–80% (v/v) 0.969–0.998 (methanol) methanol in water; 30–70% (v/v) acetone 0.947-0.999 (acetone) in water

HPTLC plates RP-18 F254S (Merck, Darmstadt, Germany)

Maleic, succinic, and phthalic acid derivatives

> 0.98

methanol-water, methanol-glycine buffer pH 1.95, and methanol-glycine buffer pH 10

slurry of silica gel GF254 aqueous buffer alone, 0.941–0.999 obtained with 0.09 M saturated with HCl for pH 1.2 or with silicone, or mixed 0.36 M NaOH for pH with various amounts 9.0; the silica gel layer of acetone, methanol, was impregnated with or acetonitrile silicon DC 200 (Applied Science, State College, PA, USA)

Stationary Phase

N- and O-substituted silica gel RP-18 F254 alkanoic acids of 1,2plates (Merck, benzisothiazol-3(2H)-one Darmstadt, Germany)

Ionizable quinolines

Compounds

Linear Correlation RM – ϕ r 0.972 (acetonebuffer pH 9.0) 0.943 (methanolbuffer pH 9.0) 0.866 (acetonitrilebuffer pH 9.0) 0.960 (acetone-buffer pH 1.2) significant correlation for structurally different compounds

Linear Correlation RM0 – a1 r Tab 3-4

Comments

0.931 (methanol- logP values calculated buffer pH 1.95) Hyper Chem 5.0 Chem 0.879 (methanol- Plus 2.0 water) 0.893 (methanolbuffer pH 10)

Linear Correlation RM0 – logP r

TABLE 3.6 (CONTINUED) Examples of RP-TLC Lipophilicity Studies on Different Classes of Compounds (where r = Correlation Coefficient)

[268]

[267]

[243]

Ref.

140 Advances in Chromatography

Serotonergic ligands

Phenothiazine

glycine buffer at pH 9.0 0.999–0.829 alone or mixed with various amounts of acetone, methanol or acetonitrile 30–70% (v/v) acetoneglycine buffer pH 9.0

0.900

silica gel 60 F254 silanized precoated plates (Merck, Darmstadt, Germany) Kieselguhr G or Kieselguhr G and cellulose impregnated with oleyl alcohol or liquid paraffin silica gel GF254 impregnated with silicone DC200 (350 CS) (Applied Science, State College, PA, USA) and C18 Whatman KC18F HPTLC plates

Phenothiazines

acetonitrile (65/35 to 50/50, v/v) in 0.02 N ammonium acetate buffer (pH 7.4) methanol-water

RP-18 F254S TLC plates methanol (25–65%, RM0 (95% (Merck Darmstadt, v/v), buffer and acetic confidence) = 1.613–2.508 Germany) acid mixtures

N-substituted amides of piperazine-hydroxybutyric acid

N-benzylamides of -(4RP-18 F254S TLC plates methanol (25–65%, RM0 (95% benzylpiperazine)-(Merck Darmstadt, v/v), buffer and acetic confidence) = 1.055–1.747 hydroxybutyric acid Germany) acid mixtures N-hydroxyethylamides of silica gel RP-8 F254 60–90% (v/v) aryloxyalkylene and methanol in water HPTLC plates pyridine carboxylic acids (Merck, Darmstadt, Germany)

0.928 (acetone) 0.946 (acetonitrile) 0.994 (methanol)

0.9994

0.922 (on silicone) 0.870 (on C18)

0.977

0.832 for RM at 50% acetonitrile

0.874

0.990–0.992

0.893

[213, 214]

[252, 255]

[269]

[179]

[269]

Correlation with [219] calculated values CLogP (QSAR Program Version 1.87, BioByte, Claremont, CA, 1994)

logP values calculated with Pallas (CompuDrug Chemistry Ltd., 1995) logP values calculated with Rekker’s method or from Pomona College database (Claremont, CA, USA) logP values calculated with PALLAS (CompuDrug Chemistry Ltd., 1995) Correlation with calculated values CLogP

Lipophilicity Measurements by Liquid Chromatography 141

Xanthine and adenosine derivatives

Vinpocetine and related compounds

Thiazoles

Compounds

RP-HPTLC on Whatman KC18F plates

silica gel impregnated with silicone

homemade plates with silica gel 60 GF254, (Merck, Drmstadt, Germany) impregnated with paraffin oil silanized silica gel 60 F254 (Merck, Darmstadt, Germany)

Stationary Phase

aqueous buffers alone or mixed with various amounts of acetone; glycine buffer pH 9.0 or sodium acetateVeronal buffer pH 7.0 methanol-phosphate buffer pH 7.0

4–34% (v/v) methanol; 0.999–0.951 4–24% (v/v) acetone; or 1–20% (v/v) dioxane in concentrated ammonia 35–65% (v/v) acetone 0.989–0.997 in 0.1 M NaOH

Mobile Phase

Linear Correlation RM – ϕ r

0.993

Linear Correlation RM0 – a1 r

0.952

0.935

0.997

0.830 (methanol) 0.956 (acetone) 0.945 (dioxane)

Linear Correlation RM0 – logP r

[244]

Ref.

CLogP values calculated with CLOGP Program, Version 3.42, Pomona College, Claremont, CA

The value r for the linear [246] correlation RM0 – a1 was obtained from data in Table 3.1 [48] CLogP values calculated [218] with CLOGP Program, Version 3.42, Pomona College, Claremont, CA

Comments

TABLE 3.6 (CONTINUED) Examples of RP-TLC Lipophilicity Studies on Different Classes of Compounds (where r = Correlation Coefficient)

142 Advances in Chromatography

Lipophilicity Measurements by Liquid Chromatography

143

TABLE 3.7 Ratios between Slopes in Different TLC Systems and Solvent Strength Ratio of Slopes a1 Compounds Steroids Prostaglandins Dermorphins Triazines Naphthalenes and quinoline Quinolones Mean

Ratio of Slopes b1

Acetone/ Acetonitrile

Acetone/ Methanol

1.12 1.07

1.70 1.67 1.36 1.37 1.53

1.52 1.56

0.91 0.83

1.33

0.94

1.13 1.09 ± 0.02

1.39 1.50 ± 0.06

1.22 1.41 ± 0.08

1.15

1/E0 1.70

1.47

1.03

Acetonitrile/ Acetone/ Methanol Acetonitrile

Acetone/ Methanol

Acetonitrile/ Methanol

0.59 0.71 0.82 0.63 0.71

0.65 0.86

0.87 0.89 ± 0.02

0.73 0.70 ± 0.03

0.84 0.76 ± 0.05

0.86

E0 0.59

0.68

0.68

Data from [215]

for all studied chemical series. Similar results were obtained for serotonergic ligands [219]. Also, the ratio between the slopes b1 (Equation 3.64) are in good agreement with the ratio between the E0 values for the same solvents. In particular, the ratios between the mean slopes of the TLC equations in different solvent systems are close to the ratios between the 1/E0 values from the corresponding solvent pairs, with the exception of the acetone system where the slopes seem to be lower than expected on the basis of the 1/E0 values. In fact, the acetone/methanol slope ratio (1.50) is lower than the corresponding reciprocal solvent strength ratio (1.70).

3.10.7 RM DETERMINATION AND CORRELATION COEFFICIENTS, LOGP

WITH

PARTITION

Determination of logP by TLC is based on the linear relationship between the chromatographic retention, RM and the octanol/water partition coefficient (logPo/w), usually determined by shake-flask for a set of reference compounds. Many studies exist in the literature reporting linear correlations between RM or RM0 and logP, either calculated or experimentally measured, for series of congeneric or structurally unrelated compounds. This approach was applied for the lipophilicity study of 17 compounds of 2methyl-4-oxo-3H-quinazoline-3-alkyl-carboxylic acid derivatives. Twelve standard compounds, five quizolones, and seven pyrido[1,2-a]pyrimidines, were selected to investigate the suitability of pyrido-pyrimidines as standards for the chromatographic logP determination of quizolones [245]. The logP values were measured by the shake-flask method at 25˚C in mutually saturated octanol and water phases. The pH was set to the isoelectric points for two compounds while the rest were measured at pH = 7.4. The concentrations of the compounds were measured by UV spectroscopy in aliquots taken from each phase.

144

Advances in Chromatography

Each logP value was obtained as an average of 12 parallel measurements. The RPTLC lipophilicity measurements were performed on silanized precoated plates with UV indicator (Merck, Darmstadt, Germany) and methanol-water (55:45, v/v) as mobile phase. Before applying the sample, the plates were washed in methanol, dried, and heated at 160˚C for one hour. Spot detection was done under a UV lamp or by densitometry. The seven pyrido-pyrimidines were selected as standards because their logP values have a wider range, from 0.21 to 2.08 log units (Equation 3.65): logPshake-flask = 1.18 + 3.77 RM, r = 0.996

(3.65)

The calibration for the complete set of 12 compounds is shown in Equation 3.66: logPshake-flask = 1.00 + 3.36 RM, r = 0.995

(3.66)

Equation 3.66 was used to estimate logP values of ten other compounds from the same series. The standard deviation on the measurement of all logPTLC values was ±0.06 log unit, generating an acceptable confidence range of ±0.05 for the logP values. Another recent publication used calibration methods to characterize lipophilicity of vinpocetine and related compounds by RP-TLC [246]. The RP-TLC experiments were carried out on silanized silica gel 60F254 (Merck, Darmstadt, Germany). The use of methanol was insufficient to move the highly lipophilic compounds from their starting point; therefore, 35 to 65% (v/v) acetone was chosen as the cosolvent. The investigated compounds were bases, so the aqueous part of the eluent was 0.1 M NaOH to maintain them in the unprotonated form. The spots were determined by densitometry at 279 nm. The partition coefficients were measurements by a stir-flask method in saturated octanol-water system at 25˚C. Concentration of compounds was determined in the aqueous phase by UV spectrometry at several values above 250 nm, before and after adding the octanol phase. Equation 3.67 was obtained for five pyrido[1,2-a]pirimidines used for the calibration set. logPshake-flask = 0.422 + 1.157 RM0, r = 0.995

(3.67)

The logPTLC and logPshake-flask values show a good agreement for chlorpromazine (3.54 and 3.75, respectively) and progesterone (5.13 and 5.07, respectively) and these compounds were added to the calibration set (Equation 3.68): logPshake-flask = 0.419 + 1.147 RM0, r= 0. 997

(3.68)

Equation 3.68 was used for the determination of logPTLC values of the eburnone alkaloids with values ranging from 2.94 to 4.85. Dross et al. used RP-18 F254S RP-TLC plates and various mixtures of methanolbuffer at pH = 7.0 with 0.1 M ionic strength in the mobile phase and obtained a reasonable correlation of the chromatographic data with logPo/w for 65 aromatic compounds [220] (Equation 3.69):

Lipophilicity Measurements by Liquid Chromatography

RM0 = 0.151 + 1.008 logPo/w, r = 0.970

145

(3.69)

These values have been analyzed by Abraham et al. [247] using linear free energy relationship analysis (LFER) for 76 compounds and obtaining Equation 3.70: RM0 = 0.259 + 0.239R2 – 0.662π2H – 0.666Σα2H – 3.006Σβ2O + 3.603Vx (3.70) n = 76, r = 0.9891, SD = 0.206 where the solute descriptors are R2, an excess molar refraction; π2H, the dipolarity/polarizability; α2H and β2O, the overall or effective hydrogen-bond acidity and basicity; and Vx, the McGowan characteristic volume (see related text on Abraham equation). The extrapolated RM0 values from aqueous methanol are well correlated with logPo/w, and the ratio of coefficients in the LFER equations for RM0 and logPo/w are very close. However, the correlations between TLC data and logPo/w for amines (especially heterocyclic) must be evaluated carefully before use for logPo/w predictions because of possible secondary equilibrium.

3.11 CHROMATOGRAPHIC DETERMINATION OF −LIPOPHILIC BALANCE HYDROPHILIC− The hydrophilic−lipophilic balance (HLB) is a valuable parameter for the classification of nonionic surfactants. It is possible to measure HLB values by the classic shake-flask method but chromatographic methods can overcome the practical problems of investigating surfactants in an aqueous environment. A number of publications have reported that the chromatographic parameter determined by RP-HPLC on C18 columns [248–250] or RP-TLC [223] extrapolated to 100% aqueous eluent can be considered a good predictor of the HLB value. RP-TLC was used for the determination of the HLB values of some polyethoxylated nonionic surfactants: (i) octylphenol and nonylphenol ethoxylated of increasing molar ratios of ethylene oxide, Triton X® and Triton N® series (from Rohm and Haas); and (ii) polyethoxyl lauryl, stearyl and oleyl ethers, Brij® (from ICI Americas) [251]. The HLB values of these surfactants were taken from literature. The RP-TLC experiments were performed on Nano-Sil-C18-100 UV254 plates (from Aldrich) at a constant temperature of 30˚C. The mobile phase was acetone–water for Triton® and tetrahydrofuran–water for Brij series. The retention was not significantly affected by adsorptive side effects due to free silanol groups. Good linear correlations were obtained for nine Triton products (Equation 3.71 and Equation 3.72) and for five Brij compounds: RM0 = 4.58 (± 0.08) – 0.07 (± 0.01) HLB, r = 0.991

(3.71)

RM0 = 3.49 (± 0.20) – 0.11 (± 0.02) HLB, r = 0.951

(3.72)

146

Advances in Chromatography

3.11.1 DRUG LIPOPHILICITY Mannhold et al. [252] investigated four groups of drugs: antiarhythmics (AA), βblockers (BB), phenothiazines (PT), and benzamides (BA). The calculation of octanol-water logP values was achieved by two approaches, the hydrophobic fragmental constant method of Rekker [253] and the computerized version of the Hansch procedure [254]. Calculated fragmental constants (Σf) and logP values (ClogP) for 48 test compounds were compared with the measured logP values. Differences up to 0.49 between the calculated and measured logP values were considered as acceptable (a), differences greater than 1.0 as unacceptable (b), and differences in between as disputable (c). The comparison of the two calculation procedures led to the following conclusions: 26 Σf values and 26 ClogP values fell in category (a); the Rekker system calculates 12 compounds in category (c) and 10 in category (b); the corresponding numbers in the ClogP system are 11 compounds in category (c) and 11 in (b). The measured logP values indicate significant variation in lipophilicity, AA > BB > BA > PT. Dross et al. studied the same four groups of drugs — 15 AA, 11 BB, 13 PT and 9 BA — by RP-TLC on silica gel 60 F254 silanized precoated plates (Merck, Darmstadt, Germany) and acetonitrile (65:35 to 50:50, v/v) in 0.02 N ammonium acetate buffer (pH = 7.4) as mobile phase [255]. The same authors published in a previous paper the logPobs, ClogP, and Σf values [252]. The logP* values were obtained from the Med Chem Data Base THOR. Equation 3.73 through Equation 3.76 are reported [255] for 48 compounds, except Equation 3.74, which was obtained for 24 compounds: RM0 = 0.682 + 0.214 logPobs,

r = 0.919

(3.73)

RM0 = 0.709 + 0.206 logP*,

r = 0.944

(3.74)

r = 915

(3.75)

RM0 = 0.585 + 0.210 Σf, RM0 = 0.814 + 0.174 ClogP,

r = 0.899

(3.76)

A simple correction for the mobile phase pH and the pKa constant of ionizable compounds could be misleading, especially when acetonitrile is present, because this solvent can significantly change the pH of the aqueous mixture. Repeating the correlations of the extrapolated and precisely pKa-corrected RM0 values for the four sets, the correlation between logP and RM0-corrected values worsens significantly. The authors concluded that the correction for pH for the partition of ionizable compounds is necessary only in the shake-flask procedure, where the complete equilibrium of ionized and unionized forms is fully achieved. Apparently, the correction is not necessary in TLC [255]. The retention of phenotiazines (Rf values) can be predicted using Equation 3.77 [214]:

Lipophilicity Measurements by Liquid Chromatography

Rf =

a0 + a1 ⎡⎣ H + ⎤⎦

147

(3.77)

1 + a0 + a1 ⎡⎣ H + ⎤⎦

where a0 = 1/(P r) and a1 = 1/(P r K). (P is the partition coefficient, r is the phase volume ratio — a constant for a defined chromatographic system — and K is the dissociation constant in methanol–water mixtures and depends on the solution concentration). The biological activity of 2-(aminoacetylamino)thiazole derivatives was correlated with the retention parameters obtained by RP-TLC on silica gel plates impregnated with liquid paraffin (Table 3.6) [256]. The dependent variable is the percent inhibition of carrageenin-induced edema in mice with the i.p. administration of 2.2 × 103 mmol/100 g body weight of the tested compounds (Equation 3.78): log % = –0.793 (± 0.596) – 1.518 (± 0.381) RM

(3.78)

n = 5, r 0.958, SD = 0.048, F = 33.43 The logP values for N-substituted amides of α-piperazine-γ-hydroxybutyric acid (A) and derivatives of N-benzylamides of α-(4-benzylpiperazine)-γ-hydroxybutyric acid (B) were calculated using the PrologP module of the PALLAS system (CompuDrug Chemistry Ltd., 1995). This program predicts the partition coefficient of the neutral species of compounds in octanol-water systems and PrologP has three fragment databases for calculating logP values. The obtained values are: logPCDR, logPATOMIC and logPcombined. The value of logPCDR is obtained with the CDR Database, based on Rekker’s collection of hydrophobic fragmental constants [270]. The value of logPATOMIC is calculated from atomic fragments and the combined logP values are obtained as the weighted average of values from the equation: logPcombined = Σ wi logPin

(3.79)

where wi the weight of database i and logPi the logP value predicted using database i. The logPcombined of the molecule is calculated as the sum of the contributions of fragments and interactions. The correlation between RM and the calculated data is given by Equation 3.80 and Equation 3.81 for amides in series A and B, respectively: RM0 = 0.708 + 0.746 logPcombined,

r = 0.893

(3.80)

RM0 = 0.408 + 0.626 logPcombined,

r = 0.874

(3.81)

The lipophilicity index (RM0) was determined by RP-TLC for - and -quinolinyl sulfides [271]. The silica gel 60 F254 (Merck, Darmstadt, Germany) impregnated with paraffin oil was used as reversed-phase stationary phase and aqueous methanol (60 to 80% in steps of 5%, v/v) as mobile phase. The value of RM0 was extrapolated

148

Advances in Chromatography

for each compound for zero concentration of the methanol in the eluent. The results show that the introduction of a sulfide substituent significantly increases the lipophilicity of the compounds, changing the lipophilicity index RM0 from 0.51 to 0.92. N-hydroxyethylamides of aryloxyalkylene and pyridine carboxylic acids have shown growth-stimulating effect in the Moewus test with Lepidium stativum [272]. The lipophilicities of the compounds were investigated on C8 silica gel plates with methanol-water mixtures as mobile phase (Table 3.6). A very good linearity was obtained for RM – ϕ, RM – a1 and RM0 – logP correlations, showing that the studied compounds are congeneric. The lipophilicity of aniline and 36 ring-substituted aniline derivatives was determined by RP-TLC in the presence of different salts on silica gel impregnated with NaCl, KCl, MgCl2, CaCl2, AlCl3 , and tetramethylammonium hydroxide or with these salts added to the eluent [273]. The plates were impregnated by immersing them in a solution of 5% paraffin oil in n-hexane. The mobile phase was water–methanol (4:1, v/v) with or without the presence of salts. In most cases, the salts decreased the retention power of silica, resulting in increased mobility of aniline derivatives. The monovalent cations had the lowest impact on selectivity while AlCl3 and tetramethylammonium hydroxide had the highest. The correlation between RM values and logPo/w (calculated, Hansch method [65]) was better in all cases in salt-free systems. The lipophilicity of 1,2-benzisothiazol-3(2H)-ones was investigated by RP-TLC but, in this case, the correlation with logPo/w was not relevant as the purpose of the study was to determine a lipophilicity scale [264]. The correlations between the extrapolated RM0 values for the three chromatographic systems are shown in Table 3.6 with slope and intercept close to 1 and 0, respectively, and are described by Equation 3.82 and Equation 3.83: RM0,acetone = 1.114 RM0,acetonitrile – 0.174, r = 0.983

(3.82)

RM0,methanol = 1.210 RM0,acetone – 0.236, r = 0.998

(3.83)

3.11.2 MISCELLANEOUS The measurement of the RM values of a series of triazine herbicides was performed by RP-TLC on silica gel impregnated with silicone [274]. Acetone, methanol, or acetonitrile were used as organic modifier of the mobile phase. The influence of the organic solvent on RM values and composition of the mobile phase was investigated. The RM values of twelve colchicines and eight colchiceinamides were determined using RP-TLC on silica gel impregnated with 5% liquid paraffin [275]. The RM0 values were calculated by extrapolation from the linear range of a plot of RM values vs. the composition of the mobile phase, ϕ. In the colchicine series, substitution at the nitrogen in position C7 decreased the lipophilicity, whereas in the colchiceinamide series, substitution at nitrogen in position C10 increased the lipophilicity. The influence of the substituent groups on RM values was also discussed. Other RP-TLC applications in lipophilicity studies include a new group of 2,4dihydroxythiobenzanilides with fungistatic activity [276]; coumarine derivatives

Lipophilicity Measurements by Liquid Chromatography

149

designed as possible anti-inflammatory/antioxidant agents [277]; only a poor relationship exists between lipophilicity and anti-inflammatory activity; derivatives of the phenyl glycine octal ester [278]; good correlation of RM0-calculated logP but a poor relationship between the lipophilicity and the anti-inflammatory activity; nonionic surfactants [279, 280]; good linear correlation between RM0 values and the specific hydrophobic surface, so these surfactants can be considered as a homologous series of solutes independently of their structural inhomogeneity; 2-cyano-3-methylthio-3-substituted amine-acrylates [281]; and piperazine derivatives [282].

3.11.3 STUDIES

USING

BOTH RP-HPLC

AND

RP-HPTLC

Arylalkanoic acids (arylacetic acids, aryloxobutanoic acids and cinnamic acids) were analysed on two columns: μBondapak (Waters, C18, 30 cm × 3.9 mm I.D.) and 5 μm Spherisorb ODS (custom, 25 cm × 3.9 mm I.D.). The flow rate was 1 mL/min in all cases and the mobile phase was a mixture of phosphate buffer (pH = 3.0) and methanol in various ratios. The correlation coefficient for logP vs. logk50 and logkw varied in the range 0.980 to 0.996 [283]. The slopes (s) above have very different values for the correlations logP vs. RM and logP vs. RM0, suggesting that the slope is dependent on the strength of the organic modifier and on the nature of the solutes (Table 3.8). The slopes (a) of the linear relationships between RM and the concentration of organic modifier in the mobile phase correlate well with logP values. The same situation is true when comparing the chromatographic parameters RM0, logk, RM , or logkw at a defined mobile phase composition with the slope, a. These correlations, with a very good statistical significance, may introduce a new parameter for lipophilicity estimations. The lipophilicity of -carbolines was analyzed by RP-TLC and RP-HPLC and the chromatographic parameters compared with logP values. In the RP-TLC experiments, silica gel G was impregnated with silicone G by developing the plates in 5% silicone solution in diethyl ether, and the mobile phase was a glycine buffer (pH = 13.0, aqueous or mixed with various amounts of acetone) and saturated with silicone. The investigated compounds were assumed nonionized at pH = 13.0. The HPTLC experiments were performed on Whatman KC 18F plates with 45 to 80% (v/v) methanol in phosphate buffer (pH = 7.0). RP-HPLC was done on a μBondapack C18 column (30 cm × 3.9 mm I.D.) (Waters, Milford, MA, USA) and the mobile phase was 20 to 80% (v/v) methanol in phosphate buffer (pH = 7.0) at a flow rate of 1 mL/min. The RM values obtained by TLC correlated well with logP values, with r = 0.952 and an intercept close to zero. Worse correlations were obtained for RM(TLC) vs. RM(HPTLC) with r = 0.793 and for RM(TLC) vs. logk with r = 0.841. Intercept values for the latter equations were quite far from zero, indicating systematic errors. A significant correlation was found between RM, logk or logP and receptor binding affinity. One of the influencing factors could be the same pH (7.0) in all experiments [284]. TLC and RP-HPLC lipophilicity studies for cardiac glycosides and digitalis were reported by Biagi, et al. [285] The TLC experiments were performed on silica gel

Stationary Phase Mobile Phase

r for logP r for logP vs. RM0 vs. RM TLC

r = correlation coefficient s = the slope of RM vs. ϕ a = slope of RM vs. ϕ ϕ = organic modifier concentration in the mobile phase Data from [283]

0.992 0.995 r = 0.996 (s = 0.108) (s = 0.081) for logP vs. slope a (TLC exp) 0.987 (s = 0.968 (s = r = 0.951 0.211) for logP vs. slope 0.137) a (TLC exp) Cinnamic acids acetone in buffer 0.993 (s = 0.989 (s = r = 0.983 0.157) for logP vs. slope (pH 3.4) 0.122) a (TLC exp) 16-Φ substituted estra- silanized silicagel acetone-water r = 0.950–0.988 1,3,5(10)-trienes for RM vs. a Androstane-type silica gel G layer impregnated with silicone oil acetone-water r = 0.995–0.971 steroids for RM vs. a Benzodiazepine Silicoplat F254 impregnated with paraffin oil acetonitrile-phosphate r = 0.951 for RM derivatives buffer pH 4.6 vs. a Dermorphine-related r = 0.921–0.987 silica gel G impregnated with silicone oil methanol-aqueous derivatives for RM vs. a buffer pH 7.0 Dansylamides

silanized Kieselgel 60 F254 plates (E. Merck, acetone in buffer Darmstadt, Germany) impregnated with 5% (pH 3.4) ethereal solution of silicone oil Aryloxobutanoic acids acetone in buffer (pH 3.4)

Aryl acetic acids

Compound

TABLE 3.8 Comparison of Lipophilcity Studies by TLC and HPLC

r = 0.975–0.999 for logkw vs. a r = 0.925–0.997 for logk vs. a

r = 0.971 for logP vs. slope a (HPLC exp)

r = 0.929 for logP vs. slope a (HPLC exp)

r = 0.984 for logP vs. slope a (HPLC exp)

HPLC

150 Advances in Chromatography

Lipophilicity Measurements by Liquid Chromatography

151

G impregnated with silicone (prepared by developing the plates in 5% silicone solution in diethyl ether) and the mobile phase was aqueous buffer (sodium acetateVeronal buffer, pH = 7.2) alone or with various concentrations of acetone presaturated with silicone oil. RP-HPLC was done on μBondapack C18 column (10 μm particle size, 30 cm × 3.9 mm I.D.) (Waters, Milford, MA) and mobile phase was 20 to 70% acetonitrile in water at a flow rate of 1 mL/min. RM0 values were measured and calculated for a range of the most hydrophilic cardiac glycosides. The experimental values showed that the extrapolation procedure from a range of RM values obtained at different mobile phase compositions to 0% organic modifier provides the same value as the experimental RM0 value. The extrapolated RM0 values were obtained for the studied cardiac glycosides from digitalis by extrapolating the RM values determined with acetone concentrations ranging from 0 to 24%, or from around 40 to around 60% depending on the lipophilicity of the compounds. The logP values, RM0 and logk from three laboratories were in good agreement with each other (r = 0.914 to 0.964), showing that the chromatographic retention data was a good lipophilicity index for the studied compounds [285]. A similar study was performed for series of benzodiazepines on LiChrosorb RPC18 column (250 × 4.6 mm I.D.), with 30 to 70% (v/v) acetonitrile in phosphate buffer (pH = 4.8) for the RP-HPLC experiments. Values of logk and logkw were used for lipophilicity correlations and for comparison with the RM values obtained by RP-TLC on Silcoplat F254 plates (Labor-MIM) impregnated with paraffin oil with mobile phase 0 to 32.5% acetonitrile in phosphate buffer (pH = 4.8) [286]. The equation: RM = RM0 + b1 C1 + b2 logC2

(3.84)

was determined for the studied benzodiazepines, where C1 and C2 are acetonitrile and buffer concentrations in the mobile phase and b1 and b2 are constants. The buffer concentrations (C2) were 6, 12, 30, 60, and 90 mM KH2PO4. The correlation coefficients for the above equation were r2 = 0.917 to 0.984 for 18 benzodiazepines. Retention was shown to decrease with increasing acetonitrile and buffer concentrations but the impact of the acetontrile concentration is six to nine times higher than that of the buffer concentration. There is a significant correlation between logkw and RM0, expressed by the equation: logkw = c1 + c2 RM0 + c3 b1 + c4 b2, r2 = 0.7492

(3.85)

where RM0, b1 and b2 are constants and c1 through c4 are equation coefficients. Although the correlation coefficient r2 has a good value, the retention in RP-HPLC still cannot be predicted from the chromatographic parameters in RP-TLC. One explanation for this discrepancy may be that the stationary phase is covered differently in HPLC and TLC experiments, resulting in different responses of the chromatographic system to the changes in the concentration of the organic modifier in the mobile phase. Principal component analysis (PCA) of the data obtained from

152

Advances in Chromatography

the two chromatographic systems showed that the retention mechanisms in RP-TLC and RP-HPLC are similar but not identical. The lipophilicity of aromatic alkoxy and hydroxyl acids was analyzed on RP18 F254 TLC plates (Merck, Darmstadt, Germany) and on Nucleosil C18 columns (5 μm, 200 × 4 mm) by RP-HPLC. The mobile phase was 20 to 45% (v/v) acetonitrile or 40 to 90% (v/v) methanol in 0.1 M phosphate buffer (pH = 2) [287]. The extrapolated retention parameters for 0% organic modifier in the mobile phase, RM0 and logkw, were different in most cases for methanol and acetonitrile. However, a very good correlation was observed between the retention parameters obtained for methanol and those obtained for acetonitrile (ACN), making the conversion of data obtained in one chromatographic system to the other possible: RM0,methanol = 1.115 RM0,ACN + 0.147,

r = 0.913

logkw,methanol = 0.968 logkw,ACN + 0.241,

r = 0.938

(3.86) (3.87)

While in most cases the expected contribution of different substituents (e.g., hydroxyl) was observed, the influence of the methoxy group was not very clear. In RP-HPLC experiments, retention increased or decreased according to the position of the methoxy group in the molecule. However, this effect was not observed in the RP-TLC experiments, suggesting that steric effects are more important in HPLC than in TLC. Taking into account the data reported in the above mentioned work, the following equations have been derived: logkw,methanol = 0.364 + 0.789 RM0,methanol, logkw,ACN = 0.193 + 0.967 RM0,ACN,

r2 = 0.978

r2 = 0.951

(3.88) (3.89)

The good correlations suggest that retention parameters can be predicted for the studied compounds on the RP-HPLC system from the retention in a RP-TLC experiment. Different flavonoid compounds were used to investigate the correlation and repeatability between TLC and HPLC experiments, taking into consideration different TLC layers from different manufacturers, different solvent volume fraction, different organic modifiers in the mobile phase and different acid modifiers with the same solvent [288]. The TLC experiments were performed on three different reversed-phase C18 precoated layers: TLC KC18F (Whatman, Clifton, NJ), TLC precoated plates RP-18 F254S (Merck, Darmstadt, Germany), and HPTLC precoated RP-18W F254S (Merck, Darmstadt, Germany); the mobile phase was methanol, acetonitrile, and THF in binary mixtures with aqueous buffer with pH from 2 to 3 (80 mM citric or acetic acid, 8 mM sodium disodium hydrogenphosphate). The TLC retention data were compared with the HPLC experiments performed on a μBondapack C18 column (Waters, Milford, MA). The correlations logk vs. RM are very good for the RP-HPLC and RP-TLC systems with slopes close to 1 and almost constant and intercept values close to 0

Lipophilicity Measurements by Liquid Chromatography

153

(0.20) when methanol is the organic modifier. No other plate-solvent combination gives the same result; this fact reflects that the TLC system is more suitable to model HPLC experiments than the HPTLC plates for mobile phases consisting of methanolbuffer mixtures. The Δlogk and ΔRM values were also investigated in an attempt to explain different molecule behavior. In the HPLC system with methanol in the mobile phase, the group contributions were independent of the mobile phase composition, column type, and acid modifier. The retention data from TLC and HPTLC experiments agreed with these findings. A general agreement between the HPLC and TLC data was also found for THF and acetonitrile as organic modifiers in the mobile phase. For some substituent contributions, a parallel behavior between Δlogk and ΔRM values was observed while for others, the steric effect was not so clear, probably due to secondary polarity and composition gradient effects from the chromatographic system. The fact that TLC models HPLC experiments well makes it a good method due to the possibility of enhancing the peak capacity through bi-dimensional development. Compounds of 1,2,4-triazoles are known to be biologically active with pronounced antimicrobial activity and these compounds have various applications as herbicides and fungicides [289]. HPLC experiments used LiChrosorb RP-8 column (Merck, Darmstadt, Germany; 5 μm, 150 × 4.6 mm I.D.). The mobile phases were binary mixtures of methanol-water and acetonitrile-water containing 10 to 60% (v/v) organic modifier. Values of logkw values were obtained, then the isocratic chromatographic hydrophobicity index (ϕ0) was calculated as the ratio between the intercept (logkw) and the slope, S: ϕ0 = –logkw/S

(3.90)

The parameter ϕ0 is characteristic for every compound and is independent of the reversed phase column used, column length, flow rate, and the mobile phase composition. The TLC experiments used a stationary phase of silica gel impregnated with paraffin oil. The silica gel powder was mixed with a solution of paraffin oil in diethyl ether (3.5%, v/v), then a fluorescent indicator F254 was added and the layer spread on glass plates. The newly prepared plates were left to dry at room temperature, then used for the experiments. The mobile phase was 20 to 60% methanol in water. A reasonably good correlation was found between the chromatographic parameters obtained by the two methods: logkw = 1.69 RM0 – 0.023, r = 0.938

(3.91)

A good correlation was also obtained between logkw values and calculated logP values (ACD software), with correlation coefficients in the range 0.945 to 0.996 for both solvent systems. The correlations between the isocratic chromatographic index and logP were of the same quality for methanol and acetonitrile in the mobile phase with r = 0.928 to 0.986. Similar research work and good correlations between the chromatographic parameters obtained in RP-TLC and RP-HPLC have been reported

154

Advances in Chromatography

for 2,4-dihydroxythiobenzanilides (potential fungicides) [290] and for N-alkylamides of benzomorpholine-2-carboxylic acid [291]. A comprehensive study has taken into consideration extensive published chromatographic parameters for β-lactam antibiotics in an attempt to find a unified hydrophobicity scale [292]. Although it is convenient to compare the large amount of chromatographic data from literature available for antibiotics with the partition in octanol-water, poor correlations seem to be previously obtained due to the selfassociation of β-lactam antibiotics in hydrophilic media. An independent hydrophobicity scale would be based on a linear relationship between the free energy of partitioning in organic-water systems, ΔGi = αΔGj + β

(3.92)

where ΔGi(j) is the free energy of solute transition from water to organic phase i (or j). The value ΔGi(j) is also proportional with the partition coefficient (logP) and with the retention parameters in chromatographic systems, logk or RM (from different experimental conditions reported in the literature). The slope, intercept, and correlation coefficients were reported for Equation 3.92 when one of the variables was an experimentally determined value (logP, logk, RM) and the other variable was calculated for the hydrophobicity scale of six penicillins. The correlation coefficients were in the range 0.98 to 0.99 in most cases, although some significant outliers are also present [292].

3.12 OTHER CHROMATOGRAPHIC METHODS 3.12.1 MICELLAR LIQUID CHROMATOGRAPHY (MLC) Micellar liquid chromatography (MLC) is a form of liquid chromatography where the mobile phase is enriched in a compound above its critical micellar concentration (CMC). The migration of a compound in an MLC system is based on a competing partitioning between the aqueous mobile phase and the micelles and the aqueous mobile phase and the alkyl-bonded stationary phase [293]. Details about calculating the partition coefficient in MLC can be found in Khaledi and Breyer [294]. A linear relationship exists between the retention factor in MLC and the micelle concentration; therefore, the partition of a compound between the aqueous phase and the micelles can be easily determined. Good correlations between the MLC retention factor (logk) and experimental logP values were reported for various compounds on C8 and phenyl columns with CTAB micellar system [294], for β-adrenolytic drugs with SDS micellar system and n-propanol as organic modifier [295] and for barbiturates with Brij35 as micellar system [296]. Micelle formation as a form of solute self-association is not desirable as it will lower the partition coefficient and experiments should be done at concentrations lower than the critical micelle concentrations. However, micelles can be formed intentionally in the mobile phase as a different media into which the solute could partition.

Lipophilicity Measurements by Liquid Chromatography

155

The micelle–water partition coefficient has been measured with an HPLC system but the peaks obtained can be quite broad [297]. Fluoroquinolones were measured by this method on Bondapack C18 columns with an isocratic mobile phase, methanol-water (1:1, v/v), 25 mM phosphate buffer (pH = 2.75), and 1 mM sodium heptanesulphonate as the micelle formation agent [298]. A Spherisorb C8 column was used for benzene derivatives and polycyclic aromatic hydrocarbons with sodium dodecyl sulphate (SDS) and cetyltrimethylammonium bromide (CTAB) for the micelle formation in the concentration 0.05 to 0.14 M. The micellar phase was modified with 3, 5, or 10% of an alcohol (methanol, n-propanol, and n-butanol were tested). The correlation between k vs. logP seems better than logk vs. logP and both methods should be applied [299]. The column efficiency in MLC is inferior to hydroorganic systems and deteriorates as the hydrophobicity of solutes increases [300, 301]. Although “chromatography” is included in the name, other methods are not forms of liquid chromatography but of capillary electrophoresis. MEKC is a modified capillary electrophoresis method that allows the separation of neutral and ionized compounds due to a micellar phase added to the electrolyte, while MEEKC uses a microemulsion in the electrolyte formed by a water-immiscible solvent (e.g., heptane). Both methods use a dual system for the separation, one electrophoretic (electrolyte in an electric field) and the other chromatographic-like (the solute partition between the aqueous electrolyte bulk and the micelle or microemulsion phase). Lipophilicity estimations are possible in MEKC and MEEKC due to the chromatographic component of the migration process and good correlations have been reported between the retention parameter and logP values. However, these methods will not be discussed in this chapter. Details about these methods can be found in Poole and Poole [6].

3.12.2 REVERSED-PHASE ION PAIR CHROMATOGRAPHY (RP-IPC) The lipophilicity of 21 aminoaryl sulphonic acids was studied by RP-HPLC and RP-IPC. The experiments were performed on reversed-phase packing material, Polygosil-ODS, and Spherisorb-ODS. The mobile phase used in RP-HPLC was 10 mM phosphate buffer, while the mobile phase used in the RP-IPC experiments was methanol-water at different ratios with 4 mM tetrabutylammonium iodide, an ionpairing agent. The pH was roughly 7.0 for all experiments. The correlations between logk values and logPo/w in the presence of the ion-pairing agent were much better (r = 0.995) than those obtained by RP-HPLC (r = 0.911). Some organic modifiers in the aqueous mobile phase in the RP-HPLC experiments could be argued as having improved the correlations, but this situation was not investigated [302]. Similar results were reported for the same ion-pairing agent, tetrabutylammonium iodide, but for a series of phenylamine and naphtylamine sulphonic acids [303]. The method is interesting but has poor reproducibility [298, 303].

156

Advances in Chromatography

3.12.3 COUNTERCURRENT CHROMATOGRAPHY (CCC) Countercurrent chromatography (CCC) is an interesting separation method derived from two classical partition methods, countercurrent distribution (CCD) and liquid chromatography (LC). CCC does not use a solid support and is a form of liquidliquid chromatography. Droplets of a liquid mobile phase are forced to flow through an immiscible liquid stationary phase and a large contact interface is formed between the two phases. Centrifugal CCC brings the benefit of rotation forces, making possible applications in analytical or preparative scale [304–306]. If octanol is the stationary phase and water the mobile phase, the P/logP values are obtained directly from the compound retention and no correlation is used. A measurable logP range from 2.0 to 2.3 can be obtained by modifying the ratio of stationary and mobile phase volumes. The extension of the measurable logP range up to 4.5 is possible by modifying the octanol phase with mixtures of hexane, acetonitrile, cyclohexane, chloroform, or methanol. In this case, a correlation to the true logPo/w must be introduced. Very lipophilic compounds can also be measured by the same method by reversing the role of the phases, using water as a stationary phase and octanol as a mobile phase; however, this method brings larger errors [307]. Centrifugal partition chromatography (CPC) can be used for the determination of logPo/w coefficients if octanol and water are employed as the two immiscible liquid phases. However, for accurate logPo/w estimations, the chromatographic system must be in equilibrium and the chromatographic relationships between the retention factor and the mobile phase volume must be proved true. In many cases, the latter condition has been difficult to demonstrate. CPC uses a direct measurement where the aqueous stationary phase is placed in a coil and droplets of octanol flow through the coil. The compounds are injected and the UV absorbance is detected at the coil exit. The coil rotates at over 700 rpm, the flow rate is usually around 4 mL/min and the base line stabilizes in 30 minutes. In its normal mode, CPC can measure logPo/w in the range 0.5 to 2.5. In its dual mode, the flow direction is changed from descending to ascending after a predetermined time. The compounds are injected in the system as solutions in octanol and are then eluted in octanol without having to spend a long time in the column. The measurable logPo/w range is extended to 4.1 in dual mode CPC, although errors can be introduced due to the change in the flow direction and possible leaks [308]. Cocurrent chromatography was introduced in an attempt to solve the high retention of very lipophilic compounds and their tendency of staying too long in the column, trapped in the stationary phase. In this method, two immiscible mobile phases flow in the same direction, octanol with a small flow and water with a high flow, until the retained compound exits the column. The cocurrent mode is especially useful for lipophilic compounds with logPo/w > 3. The technique appeared promising but it generates a large amount of effluent that cannot be recycled due to mixing with a common solvent for detection, a serious drawback [309]. Although CPC has the potential of measuring true logPo/w, the technique was not widely used due to technical difficulties and ambiguous data interpretation. The counter-current methods are difficult to operate, the instrumentation is difficult to clean, requires skilled operators, uses large quantities of solvents and applies to a

Lipophilicity Measurements by Liquid Chromatography

157

limited logP range. Details about partitioning in octanol or liposomes can be found in a recent publication [310].

ACKNOWLEDGMENTS We are grateful to our colleagues at Sirius who provided support with data and figures and who showed patience during the busy editing time. We also thank Klara Valkó (GlaxoSmithKline, U.K.), Michal Markuszewski (Medical University of Gdansk, Poland) and Rebeca Ruiz (University of Barcelona, Spain) for support. We are grateful to Henk Claessens (Technical University Eindhoven, Netherlands) for kindly guiding the steps through the Ph.D. thesis.

GLOSSARY CCC CPC CTAB D

E0 end-capping ΔG HPLC HPTLC K isocratic isoeluotropic logk logkw logk50 logP

logD logPo/w MEEKC MEKC MLC MOPS

countercurrent chromatography centrifugal partition chromatography cetyltrimethylammonium bromide distribution coefficient, defined as the ratio of concentrations of all species of a solute dissolved in each phase of a system consisting of two separate phases at equilibrium at a particular pH. solvent strength chemical reaction between the free silanol groups and a low reactivity substance free energy high-performance liquid chromatography high-performance thin-layer chromatography partition coefficient between any immiscible phases (equivalent to P for octanol-water systems) same mobile phase composition throughout the experiment mobile phases with different composition but with the same elution power retention (or capacity) factor in chromatography isocratic retention factor when the mobile phase is purely aqueous isocratic retention factor when the mobile phase contains 50% (v/v) organic modifier logarithm of the partition coefficient (P), defined as the ratio of concentrations of a solute dissolved in each phase of a system consisting of two separate phases at equilibrium at a pH where the solute is unionized log10 of D log10 of P microemulsion electrokinetic chromatography micellar electrokinetic chromatography micellar liquid chromatography 4-morpholinepropanesulfonic acid

158

Advances in Chromatography

octanol ODS P Po/w r r2 RM RM0 RP-HPLC RP-IPC RP-HPTLC SDS X ϕ ϕ0

n-octanol octadecylsilane partition coefficient between two immiscible (liquid) phases; refers to the neutral form of the molecule partition coefficient between n-octanol and water (or aqueous buffer) correlation coeffcient square of the correlation coeffcieint (r and r2 were kept in the text according to the published data from the references) retention in thin layer chromatography RM value when the mobile phase is purely aqueous reversed-phase high-performance liquid chromatography reversed-phase ion pair chromatography reversed-phase high-performance thin-layer chromatography sodium dodecyl sulphate fraction of the organic modifier in the mobile phase concentration of the organic modifier in the aqueous mobile phase isocratic hydrophobicity index measured by RP-HPLC

REFERENCES 1. Kaliszan, R. Effect of separation conditions on chromatographic determination of hydrophobicity of acid xenobiotics. J. Chromatogr. B, 1998, 717, 125–134. 2 Valkó, K. Application of high-performance liquid chromatography based on measurements of lipophilicity to model biological distribution. J. Chromatogr A, 2004, 1037, 299–310. 3 Valkó, K., Nunhuck, S., Bevan, C., Abraham M.H. and Reynolds, D.P. Fast gradient HPLC method to determine compounds binding to human serum albumin. Relationships with octanol/water and immobilized artificial membrane lipophilcity. J. Pharm. Sci., 2003, 92, 2236–2248. 4. Abraham, M.H., Gola, J.M.R., Kumarsingh, R., Cometto-Muniz, J.E. and Cain, W.S. Connection between chromatographic data and biological data. J. Chromatogr. B, 2000, 745, 103–115. 5. Mannhold, R., Kubinyi, H. and Timmerman, H., eds. Lipophilicity in drug action and toxicology. Methods and Principles in Medicinal Chemistry, Vol. 4, VCH, Weinheim, 1996. 6. Poole, S.K. and Poole, C.F. Separation methods for estimating octanol-water partition coefficients. J. Chromatogr. B, 2003, 797, 3–19. ˇ V., Testa, B. and van de Waterbeemd, H. Lipophilicity, the empirical tool and 7. Pliska, the fundamental objective: An introduction. Lipophilicty in drug action and toxicology. Methods and Principles in Medicinal Chemistry, Vol. 4, VCH, Weinheim-New York, 1996, 1–5. 8. http://www.iupac.org. 9. Mayer, H. Zur Theorie der Alkoholnarkose I. Welche Eigenschaft der Anaesthetika bedingt ihre narkotische Wirkung? Archiv f. Experimentelle Pathologie un Pharmakologie, 1899, 42, 109–118.

Lipophilicity Measurements by Liquid Chromatography

159

10. Overton, E. Studien ber die Narkose, zugleich ein Beitrag zur allgemeinen Pharmakologie, Gustav Fischer. Jena, Switzerland, 1901. 11. Yalkowsky, S.H. and Banerjee, S. Methods of estimation for organic compounds. Aqueous Solubility, Marcel Dekker Inc., New York, 1991. 12. Cimpan, G. The lipophilicity characterization of some groups of organic compounds by reversed-phase partition chromatography. Ph.D. thesis, 1997. 13. Lyman, W.J., Reehl, W.F. and Rosenblatt, D.H. Handbook of Chemical Property Estimation Methods, McGraw-Hill, New York, Chs. 1, 2, and 4, 1982. 14. Noble, A. Partition coefficients (n-octanol-water) for pesticides. J. Chromatogr., 1993, 642, 3–14. 15. Lombardo, F., Shalaeva, M.Y., Tupper, K.A., Gao, F. and Abraham, M.H. ELogPoct: A tool for lipophilicity estimation in drug discovery. J. Med. Chem., 2000, 43, 2922–2928. 16. Lipinski, C.A., Lombardo, F., Dominy, B.W. and Feeny, P.J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Delivey, 1997, 23, 3–25. 17. Comenarejo, G., Alvarez-Pedraglio, A. and Lavandera, J.-L. Cheminformatic models to predict binding affinities to human serum albumin. J. Med. Chem., 2001, 44, 4370–4378. 18. Camenish, G., Folkers, G. and van de Waterbeemd, H. Review of theoretical passive drug absorption models: Historical background, recent developments and limitations. Pharm. Acta Helv., 1996, 71, 309–327. 19. van de Waterbeemd, H., Smith, D.A. and Jones, B.C. Lipophilicity in PK design: Methyl, ethyl, futile. J. Comp.-Aided Mol. Des., 2001, 15, 273–286. 20. OECD Guidelines for Testing of Chemicals. Vol. 117, Partition coefficient (noctanol/water). High Performance Liquid Chromatography (HPLC) Method, OECD, Paris, 1989. 21. Harnish, M., Möckel, H.J. and Schultze, G. Relationship between logPow, shake-flask values and capacity factors derived from reversed-phase high-performance liquid chromatography for n-alkylbenzenes and some OECD reference substances. J. Chromatogr., 1983, 282, 315–332. 22. Nernst, W. Verteilung eines Stoffes zwischen zwei Lösungsmitteln und zwischen Lösungsmitteln und Dampfraum. Z. Phys. Chem., 1891, 8, 110–139. 23. http://www.nernst.de. 24. Leo, A., Hansch, C. and Elkins, D. Partition coefficients and their uses. Chem. Rev., 1971, 71, 525–616. 25. Hansch, C. and Leo, A. Substituent Constants for Correlation Analysis in Chemistry and Biology. John Wiley & Sons, New York, 1979. 26. Takács-Novák, K., Szász, Gy., Budvári-Bárány, Zs., Józan, M. and Lre, A. Relationship study between reversed phase HPLC retention and octanol/water partition among amphoteric compounds. J. Liq. Chromatogr., 1995, 18, 807–825. 27. Comer, J. and Tam, K. Lipophilicity profiles: Theory and measurement. Pharmacokinetic Optimisation in Drug Research, Wiley-VCH, 2001, 275–304. 28. Strong, L.E., Brummel, C.L., Ryther, R., Radford, J.R. and Pethybridge, A.D. Dimerization of some substituted benzoic acids in aqueous solution from conductance measurements. J. Solution Chemistry, 1988, 17, 1145–1167. 29. Faller, B. and Wohnsland, F. Physicochemical parameters as tools in drug discovery and lead optimization. Pharmacokinetic Optimisation in Drug Research, Wiley-VCH, 2001, 258–273.

160

Advances in Chromatography

30. Hansch, C. and Fujita, T. ρ-σ-π Analysis: A method for the correlation of biological activity and chemical structure. J. Am. Chem. Soc., 1964, 86, 1616–1626. 31. Fujita, T., Iwasa, J. and Hansch, C. A new substituent constant, π, derived from partition coefficients. J. Am. Chem. Soc., 1964, 86, 5175–5180. 32. Hansch, C. and Leo, A. Exploring QSAR-Fundamentals and Applications in Chemistry and Biology. American Chemical Society, Washington, DC, 1995, 97–543. 33. Leo, A., Hansch, C. and Church, C. Comparison of parameters currently used in the study of structure-activity relationships. J. Med. Chem., 1969, 12, 766–771. 34. Taylor, P.C. Hydrophobic Properties of Drugs, Comprehensive Medicinal Chemistry: The Rational Design, Mechanistic Study and Therapeutic Application of Chemical Compounds. Pergamon Press, New York, 1990, 241–294. 35. Tsai, R.-S., Fan, W., El Tayar, N., Carrupt, P.-A., Testa, B. and Kier, L.B. Solutewater interactions in the organic phase of a biphasic system. 1. Structural influence of organic solutes on the “water-dragging” effect. J. Am. Chem. Soc., 1993, 115, 9632–9639. 36. Fan, W., Tsai, R.-S., El Tayar, N., Carrupt, P.-A. and Testa, B. Solute-water interactions in the organic phase of a biphasic system. 2. Effects of organic phase and temperature on the “water-dragging” effect. J. Phys. Chem., 1994, 98, 329–333. 37. Tute, M.S. Lipophilicity: A history. Lipophilicity in drug action and toxicology. Methods and Principles in Medicinal Chemistry, Vol. 4, VCH, Weinheim 1996, 7–26. 38. van de Waterbeemd, H. and Manhold, R. Lipophilicity descriptors for structureproperty correlation studies: Overview of experimental and theoretical methods and a benchmark of logP calculations. Lipophilicity in drug action and toxicology. Methods and Principles in Medicinal Chemistry, Vol. 4, VCH, Weinheim 1996, 401–418. 39. Gaudette, L.E. and Brodie, B.B. Relationship between the lipid solubility of drugs and their oxidation by liver microsomes. Biochem. Pharmacol., 1959, 2, 89–96. 40. Leahy, D.E., Taylor, P.J. and Wait, A.R. Model solvent systems for QSAR. Part I. Propylene glycol depelargonate (PGDP). A new standard solvent for use in partition coefficient determination. Quant. Struct.-Act. Relat., 1989, 8, 17–31. 41. Abraham, M.H., Chada, H.S. and Mitchell, R.C. Hydrogen bonding. 33. Factors that influence the distribution of solutes between blood and brain. J. Pharm. Sci., 1994, 83, 1257–1268. 42. Leahy, D.E., Morris, J.J., Taylor, P.J. and Wait, A.R. Model Solvent system for QSAR. Part 2. Fragment values (F-values) for the critical quartet. J. Chem. Soc. Perkin Trans., 1992, 2, 723–731. 43. Buchwald, P. and Bodor, N. Octanol-water partition: Searching for predictive models. Curr. Med. Chem., 1998, 5, 353–380. 44. Leo, A.J. Calculating logPoct from structures. Chem. Rev., 1993, 93, 1281–1306. 45. Hansch, C., Maloney, P.P., Fujita, T. and Muir, R.M. Correlation of biological activity of phenoxyacetic acids with hammett substituent constants and partition coefficients. Nature, 1962, 194, 178–180. 46. Hansch, C., Muir, R.M., Fujita, T., Maloney, P.P., Geiger, F. and Streich, M. The correlation of biological activity of plant growth regulators and chloromycetin derivatives with hammett constants and partition coefficients. J. Am. Chem. Soc., 1963, 85, 2817–2824. 47. Rekker, R.F. and Manhold, R. Calculation of drug lipophilicity. The Hydrophobic Fragmental Constant Approach, VCH, Weinheim, 1992. 48. Dross, K.P. and Rekker, R.F. Drug lipophilicity in QSAR practice. I. A comparison of experimental with calculative approaches. Quant. Struct.-Act. Relat., 1990, 9, 21–28.

Lipophilicity Measurements by Liquid Chromatography

161

49. Viswanadhan, V.N., Ghose, A.K., Revankar, G.R. and Robins, R.K. Atomic physicochemical parameters for three dimensional structure directed quantitative structureactivity relationships. 4. Additional parameters for hydrophobic and dispersive interactions and their application for an automated superposition of certain naturally occurring nucleoside antibiotics. J. Chem. Inf. Comput. Sci., 1989, 29, 163–172. 50. Bodor, N., Gabanyi, Z. and Wong, C.-K. A new method for the estimation of partition coefficient. J. Am. Chem. Soc., 1989, 111, 3783–3786. 51. Kamlet, M., Abboud, J.-L., Abraham, M. and Taft, R. Linear solvation energy relationships. 23. A comprehensive collection of the solvatochromic parameters, π, α, and β, and some methods for simplifying the generalized solvatochromic equation. J. Org. Chem., 1983, 48, 2877–2887. 52. http://clogp.pomona.edu. 53. http://www.biobyte.com. 54. http://www.daylight.com/daygci/clogp. 55. http://www.acdlabs.com. 56. http://ap-algorithms.com. 57. Caron, G. and Ermondi, G. A comparison of calculated and experimental parameters as sources of structural information: the case of lipophilicity-related descriptors. Mini Rev. Med. Chem., 2003, 3, 821–830. 58. Collander, R. Partition coefficients of alcohols (including ethanol) in octyl alcohol/water and oleyl alcohol/water systems. Acta Chem. Scand., 1951, 5, 774–780. 59. Abraham, M.H., Chadha, H.S., Whiting, G.S. and Mitchell, R.C. Hydrogen bonding. 32. An analysis of water-octanol and water-alkane partitioning and the logP parameter of seiler. J. Pharm. Sci., 1994, 83, 1085–1100. 60. Valkó, K. Separation methods in drug synthesis and purification. Handbook of Analytical Separations, Vol.1, Elsevier, Amsterdam, 2000, 539–542. 61. Hitzel, L., Watt, A.P. and Locker, K.L. An increased throughput method for the determination of partition coefficients. Pharm. Res., 2000, 17, 1389–1395. 62. http://www.analiza.com. 63. Ford, H. Jr., Merski, C.L. and Kelley, J.A. A rapid microscale method for the determination of partition coefficients by HPLC, J. Liq. Chromatogr., 1991, 14, 3365–3386. 64. Kaliszan, R. Quantitative structure-chromatographic retention relationships. Chemical Analysis, Vol. 93, John Wiley & Sons, New York, 1987. 65. Martin, A.J.P. and Synge, R.L.M. A new form of chromatogram employing two liquid phases. 1. A theory of chromatography. 2. Application of the microdetermination of the higher mono-amino acids in proteins. Biochem. J., 1941, 35, 1358–1368. 66. Vailaya, A. Liquid-liquid partition chromatography. Encyclopedia of Chromatography, Marcel Dekker, New York, 2001, 494–497. 67. Pesek, J.J. and Matyska, M.T. Reversed-phase chromatography: Description and applications. Encyclopedia of Chromatography, Marcel Dekker, New York, 2001, 719–722. 68. Snyder, L.R. Theory of chromatography. Chromatography, 5th ed., Elsevier, Amsterdam, 1992. 69. Mirrlees, M.S., Moulton, S.J., Murphy, C.T. and Taylor, P.J. Direct measurement of octanol-water partition coefficients by high pressure liquid chromatography. J. Med. Chem., 1976, 19, 615–619. 70. McCall, J.M.. Liquid-liquid partition coefficients by high pressure liquid chromatography. J. Med. Chem., 1975, 18, 549–552.

162

Advances in Chromatography

71. Unger, S.H., Cook, J.R. and Hollenberg, J.S. Simple procedure for determining octanol-aqueous partition, distribution and ionisation coefficients by reversed-phase high-pressure liquid chromatography. J. Pharm. Sci., 1978, 67, 1364–1367. 72. Slater, B., McCormack, A., Avdeef, A. and Comer, J.E.A. pH-metric logP comparison of partition coefficients determined by HPLC and potentiometric methods to literature values. J. Pharm. Sci., 1994, 83, 1280–1283. 73. http://www.sirius-analytical.com. 74. Comer, J.E.A. High-throughput measurements of logD and pKa in drug bioavailability. Estimation of solubility, permeability, absorption and bioavailability. Methods and Principles in Medicinal Chemistry, Vol. 18, Wiley-VCH, 2003, 21–45. 75. Box, K., Elliott, S. and Cimpan, G. High throughput pKa and logD7.4 measurements. PharmaChem, June 2003, 55–59. 76. Caron, G., Steyaert, G., Pagliara, A., Reymond, F., Crivori, P., Gaillard, P., Carrupt, P.-A., Avdeef, A., Comer, J., Box, K.J., Girault, H.H. and Testa, B. Structure-lipophilicity relationships of neutral and protonated–blockers. Part I. Intra and intermolecular effects in isotropic solvent systems. Helvetica Chimica Acta, 1999, 82, 1211–1222. 77. Comer, J.E.A., Avdeef, A. and Box, K.J. Limits for successful measurement of pKa and logP by pH-metric titration. American Laboratory, 1995, 4, 36C–36I. 78. Comer, J., Chamberlain, K. and Evans, A. Validation of pH-metric technique for measurement of pKa and logPow of ionizable herbicides. SAR and QSAR in Environmental Research, 1995, 3, 307–313. 79. Comer, J.E.A. The acid test, ionization and lipophilicity of drugs, influence on biological activity. Chemistry in Britain, 1994, 30, 983–986. 80. Avdeef, A. and Comer, J.E.A. Measurement of pKa and logP of water-insoluble substances by potentiometric titration. QSAR and Molecular Modelling, ESCOM Science Publishers BV, Leiden, 1993, 386–387. 81. DeVoe, H., Miller, M.M. and Wasik, S.P. Generator columns and high pressure liquid chromatography for determining aqueous solubility and octanol-water partition coefficients of hydrophobic substances. Journal of Research of the National Bureau of Standards, 1981, 86, 361–366. 82. Doucette, W.J. and Andren, A.W. Estimation of octanol/water partition coefficients: Evaluation of six methods for highly hydrophobic aromatic hydrocarbons. Chemosphere, 1988, 17, 345–359. 83. Cichna, M., Markl, P. and Huber, J.F.K. Determination of true octanol-water partition coefficients by means of solvent generated liquid-liquid chromatography. J. Pharm. Biomed. Anal., 1995, 13, 339–351. 84. Huber, J.F.K., Pawlowska, M. and Markl, P. Selection of the solid support — A crucial step for the realization of solvent generated liquid-liquid chromatography. J. Chromatogr., 1990, 500, 257–280. 85. Huber, J.F.K., Meijers, C.A.M. and Hulsman, J.A.R.J. Prediction of partition coefficients in liquid-liquid systems and its experimental verification for steroids by static and chromatographic measurements. Anal. Chem., 1972, 44, 111–116. 86. Minick, D.J., Frenz, J.H., Patrick, M.A. and Brent, D.A. A comprehensive method for determining hydrophobicity constants by reversed phase high-performance liquid chromatography. J. Med. Chem., 1988, 31, 1923–1933. 87. Pagliara, A., Khamis, E., Trinh, A., Carrupt, P.-A., Tsai, R.-S. and Testa, B. Structural properties governing retention mechanisms on RP-HPLC stationary phases used for lipophilicity measurements. J. Liq. Chromatogr., 1995, 18, 1721–1745.

Lipophilicity Measurements by Liquid Chromatography

163

88. Lombardo, F., Shalaeva, M.Y., Tupper, K.A. and Gao, F. ELogP: A tool for lipophilicity determination in drug discovery. 2. Basic and neutral compounds. J. Med. Chem., 2001, 44, 2490–2497. 89. Valkó, K. Application of high-performance liquid chromatography based measurements of lipophilicity to model biological distribution. J. Chromatogr. A, 2004, 1037, 299–310. 90. Ellgehausen, H., Hondt, C.D. and Fuerer, R. Reversed-phase chromatography as a general method for determining octanol/water partition coefficients. Pestic. Sci., 1981, 12, 219–227. Cited in: Nobel, A. Partition coefficients (n-octanol-water) for pesticides. J. Chromatogr., 1993, 642, 3–14. 91. Hsieh, M.M. and Dorsey, J.G. Accurate determination of logkw in reversed-phase liquid chromatography. Implications for quantitative structure-retention relationships. J. Chromatogr., 1993, 631, 63–78. 92. Minik, D.J., Frenz, J.H., Patrick, M.A. and Brent, D.A. A comprehensive method for determining hydrophobicity constants by reversed-phase high performance liquid chromatography. J. Med. Chem., 1988, 31, 1923. 93. Yamagami, C. Recent advances in reversed-phase HPLC techniques to determine lipophilicity. Pharmacokinetic Optimisation in Drug Research, Wiley-VCH, 2001, 383–400. 94. van de Waterbeemd, H., Kansy, M., Wagner, B. and Fischer, H. Lipophilicity measurement by reversed-phase high performance liquid chromatography (RP-HPLC). Lipophilicity in drug action and toxicology. Methods and Principles in Medicinal Chemistry, Vol. 4, VCH, Weinheim 1996, 73–87. 95. Braumann, T. Determination of hydrophobic parameters by reversed-phase liquid chromatography: Theory, experimental techniques, and application in studies on quantitative structure-activity relationships. J. Chromatogr., 1986, 373, 191–225. 96. Lambert, W.J. Modeling oil-water partitioning and membrane permeation using reversed-phase chromatography. J. Chromatogr. A, 1993, 656, 469–484. 97. Yamagami, C., Ogura, T. and Takao, N. Hydrophobicity parameters determined by reversed-phase liquid chromatography. I. Relationship between capacity factors and octanol-water partition coefficients for monosubstituted pyrazines and the related pyridines. J. Chromatogr., 1990, 514, 123–136. 98. Yamagami, C., Yokota, M. and Takao, N. Hydrophobicity parameters determined by reversed-phase liquid chromatography. VIII. Hydrogen-bond effects of ester and amide groups in heteroaromatic compounds on the relationship between the capacity factor and the octanol-water partition coefficient. J. Chromatogr. A, 1994, 662, 49–60. 99. Garst, J.E. Accurate, wide-range high-performance liquid chromatographic method for the estimation of octanol/water partition coefficients. II. Equilibrium in partition coefficient measurements, additivity of substituent constants and correlation of biological data. J. Pharm. Sci., 1994, 73, 1623–1629. 100. Hammers, W.E., Meurs, G.J. and de Ligny, C.L. Correlations between liquid chromatographic capacity ratio data on lichrosorb RP-18 and partition coefficients in the octanol-water system. J. Chromatogr., 1982, 247, 1–13. 101. Valkó, K., Snyder, L.R. and Glajch, J.L. Retention in reversed-phase liquid chromatography as a function of mobile-phase composition. J. Chromatogr. A, 1993, 656, 501–520. 102. Schoenmakers, P.J., Billiet, H.A.H. and de Galan, L. Influence of organic modifiers on the rentention behaviour in reversed-phase liquid chromatography and its consequences for gradient elution. J. Chromatogr., 1979, 185, 179–195.

164

Advances in Chromatography

103. Poole, C.F., Gunatilleka, A.D. and Poole, S.K. In search of a chromatographic model for biopartitioning. Adv. Chromatogr., 2000, 40, 159. 104. Lipinski, C.A., Fiese, E.F. and Korst, R.J. pKa, logP and MedChem CLogP fragment values of acidic heterocyclic potential bioisosters. Quant. Struct.-Act. Relat., 1991, 10, 109–117. 105. Klein, W., Kördel, W., Wei, M. and Poremski, H.J. Updating the OECD test guideline 107 “Partition coefficient n-octanol-water,” OECD laboratory intercomparison test on the HPLC method. Chemosphere, 1988, 17, 361–386. 106. Kucha, M., Rejholec, V., Kraus, E., Miller, V. and Rábek, V. Influence of intramolecular interactions on chromatographic behaviour of arylaliphatic acids. I. Comparison of reversed-phase thin-layer and high-performance liquid chromatography. J. Chromatogr., 1983, 280, 279–288. 107. Bechalany, A., Tsantili-Kakoulidou, A., El Tayar, N. and Testa, B. Measurement of lipophilicity indices by reversed-phase high-performance liquid chromatography: Comparison of two stationary phases and various eluents. J. Chromatogr., 1991, 541, 221–229. 108. Kucha, M., Rejholec, V., Kraus, E., Miller, V. and Rábek, V. Influence of intramolecular interactions on chromatographic behaviour of arylaliphatic acids. I. Comparison of reversed-phase thin-layer and high-performance liquid chromatography. J. Chromatogr., 1983, 280, 279–288. 109. Valkó, K. and Slégel, P. New chromatographic hydrophobicity index (0) based on the slope and intercept of the logk versus organic phase concentration plot. J. Chromatogr., 1993, 631, 49–61. 110. Horvath, C., Melander, W. and Molnar, I. Liquid chromatography of ionogenic substances with nonpolar stationary phases. Anal. Chem., 1977, 49, 142–154. 111. Hanai, T. Structure-retention correlation in liquid chromatography. J. Chromatogr., 1991, 550, 313–324. 112. Espinosa, S., Bosch, E. and Roses, M. Retention of ionisable compounds in highperformance liquid chromatography. XI. Global linear salvation energy relationships for neutral and ionisable compounds. J. Chromatogr. A, 2002, 945, 83–96. 113. Espinosa, S., Bosch, E. and Roses, M. Retention of ionisable compounds in highperformance liquid chromatography. IX. Modelling retention in reversed-phase liquid chromatography as a function of pH and solvent composition with acetonitrile-water mobile phases. J. Chromatogr. A, 2002, 947, 47–58. 114. Braumann, Th., Genieser, H.-G., Lüllmann, C. and Jastorff, B. Determination of hydrophobic parameters by reversed-phase high-performance liquid chromatography. Effect of stationary phase. Chromatographia, 1987, 24, 777–782. 115. van Straten, M.A., Lynch, R.J. and Claessens, H.A. The role of stationary phase in reversed-phase liquid chromatography in the application of solvent optimization procedures. LC-GC Int., 1995, 8, 143–150. 116. Abraham, M.H., Chadha, H.S. and Leo, A.J. Hydrogen bonding 35: Relationship between high performance liquid chromatographic capacity factors and water-octanol partition coefficients. J. Chromatogr. A, 1994, 685, 203–211. 117. Tchapla, A., Colin, H. and Guiochon, G. Linearity of homologous series retention plots in reversed-phase liquid chromatography. Anal. Chem., 1984, 56, 621–625. 118. Kaliszan, R., van Straten, M.A., Markuszewski, M., Cramers, C.A. and Claessens, H.A. Molecular mechanism of retention in reversed-phase high-performance liquid chromatography and classification of modern stationary phases by using quantitative structure–retention relationships. J. Chromatogr. A, 1999, 855, 455–486.

Lipophilicity Measurements by Liquid Chromatography

165

119. Haky, J.E. and Vemulapalli, S. Comparison of octadecyl-bonded alumina and other stationary phases for lipophilicity estimation by high performance liquid chromatography. J. Liq. Chromatogr., 1990, 13, 3111–3131. 120. Kaliszan, R. Quantitative structure-retention relationships applied to reversed-phase high-performance liquid chromatography. J. Chromatogr. A, 1993, 656, 417–435. 121. Bechalany, A., Rothlisberger, T., El Tayar, N. and Testa, B. Comparison of various non-polar stationary phases used for assessing lipophilicity. J. Chromatogr., 1989, 473, 115–124. 122. Tanaka, N., Nagayama, H., Kobayashi, H., Ikegami, T., Hosoya, K., Ishizuka, N., Minakuchi, H., Nakanishi, K., Cabrera, K. and Lubda, D. Monolithic silica columns for HPLC, micro-HPLC, and CEC. J. High Resol. Chromatogr., 2000, 23, 111–116. 123. Cabrera, K., Lubda, D., Eggenweiler, H.-M., Minakuchi, H. and Nakanishi, K. A new monolithic-type HPLC column for fast separations. J. High Resol. Chromatogr., 2000, 23, 93–99. 124. Zou, H., Huang, X., Ye, M. and Luo, Q. Monolithic stationary phases for liquid chromatography and capillary electrochromatography. J. Chromatogr. A, 2002, 954, 5–32. 125. Dawkins, J.V., Gaggot, N., Lloyd, L.L., McConville, J.A. and Warner, F.P. Reversedphase high-performance liquid chromatography with a C18 polyacrylamide-based packing. J. Chromatogr., 1988, 452, 145–156. 126. Buszewski, B., Schmid, J., Albert, K. and Bayer, E. Chemically bonded phases for the reversed-phase high-performance liquid chromatographic separation of basic substances. J. Chromatogr., 1991, 552, 415–427. 127. Feibush, B. and Santasania, C.T. Hydrophilic shielding of hydrophobic, cation- and anion-exchange phases for separation of small analytes: Direct injection of biological fluids onto high-performance liquid chromatographic columns. J. Chromatogr., 1991, 544, 41–49. 128. O’Gara, J.E., Alden, B.A., Walter, T.H., Petersen, J.S., Niederländer, C.L. and Neue, U.D. Simple preparation of a C8 HPLC stationary phase with an internal polar functional group. Anal. Chem., 1995, 67, 3809–3813. 129. Czajkowska, T., Hrabovsky, L., Buszewski, B., Gilpin, R.K. and Jaroniec, M. Comparison of the retention of organic acids on alkyl and alkylamide chemically bonded phases. J. Chromatogr. A, 1995, 691, 217–224. 130. Buszewski, B., Gadzaa-Kopciuch, R.M., Markuszewski, M. and Kaliszan, R. Chemically bonded silica stationary phases: Synthesis, physicochemical characterization, and molecular mechanism of reversed-phase HPLC retention. Anal. Chem., 1997, 69, 3277–3284. 131. Nawrocki, J., Rigney, M.P., McCormick, A. and Carr, P.W. Chemistry of zirconia and its use in chromatography. J. Chromatogr., 1993, 657, 229–282. 132. Blackwell, J.A. and Carr, P.W. Development of an eluotropic series for the chromatography of Lewis bases on zirconium oxide. Anal. Chem., 1992, 64, 863–873. 133. Trüdinger, U., Müller, G. and Unger, K. Porous zirconia and titania as packing materials for high-performance liquid chromatography. J. Chromatogr., 1990, 535, 111–125. 134. Blackwell, J.A. and Carr, P.W. The role of Lewis acid-base processes in ligandexchange chromatography of benzoic acid derivatives on zirconium oxide. Anal. Chem., 1992, 64, 853–862. 135. Nasal, A., Kaliszan, R., Forgács, E., Cserháti, T. and Abraham, M.H. Polyethylenecoated silica and zirconia stationary phases in view of quantitative structure-retention relationships. Chromatographia, 1996, 43, 484–490.

166

Advances in Chromatography

136. Kaliszan, R., Blain, R.W. and Hartwick, R.A. A new HPLC method of hydrophobicity evaluation employing poly(butadiene)-coated alumina columns. Chromatographia, 1988, 25, 5–7. 137. Kaliszan, R., Petrusewicz, J., Blain, R.W. and Hartwick, R.A. Determination of hydrophobicity parameters on polybutadiene-coated alumina and their application in quantitative structure-activity relationship analysis. J. Chromatogr., 1988, 458, 395–404. 138. Helweg, C., Nielsen, T. and Hansen, P.E. Determination of octanol-water partition coefficients of polar polycyclic aromatic compounds (N-PAC) by high-performance liquid chromatography. Chemosphere, 1997, 34, 1673–1684. 139. Altomare, C., Cellamera, S., Carotti, A. and Ferappi, M. Linear solvation energy relationships in reversed-phase liquid chromatography. Examination of Deltabond C8 as stationary phase for measuring lipophilicity parameters. Quant. Struct.-Act. Relat., 1993, 12, 261–268. 140. Vallat, P., Fan, W., El Tayar, N., Carrupt, P.-A. and Testa, B. Solvatochromic analysis of the retention mechanism of two novel stationary phases used for measuring lipophilicity by RP-HPLC. J. Liq. Chromatogr., 1992, 15, 2133–2151. 141. Altomare, C., Tsai, R.-S., El Tayar, N., Testa, B., Carotti, A., Cellamare, S. and De Benedetti, P.G. Determination of lipophilicity and hydrogen-bond donor acidity of bioactive sulphonyl-containing compounds by reversed phase HPLC and centrifugal partition chromatography and their applications to structure-activity relations, J. Pharm, Pharmacol., 1991, 43, 191–197. 142. Yonker, C.R., Zwier, T.A. and Burke, M.F. Investigation of stationary phase formation for RP-18 using various organic modifiers. J. Chromatogr. A, 1982, 241, 269–280. 143. Yonker, C.R., Zwier, T.A. and Burke, M.F. Comparison of stationary phase formation in RP- for methanol-water systems. J. Chromatogr. A, 1982, 241, 257–268. 144. Jenke, D.R. Liquid chromatographic estimation of octanol/water partition coefficients with a high efficiency, nonporous, ultrasmall particle size reverse phase stationary phase. J. Liq. Chromat. & Rel. Technol., 1996, 19, 2227–2245. 145. Foster, M.D. and Synovec, R.E. Reversed phase liquid chromatography of organic hydrocarbons with water as the mobile phase. Anal. Chem., 1996, 68, 2838–2844. 146. Rowlen, K.L. and Harris, J.M. Raman spectroscopic study of solvation structure in acetonitrile/water mixtures. Anal. Chem., 1991, 63, 964–969. 147. Claessens, H.A., Vermeer, E.A. and Cramers, C.A. Comparison of reversed-phase performance for some commercial C-18 HPLC columns, LC-GC Int., 1993, 6, 692. 148. McCalley, D.V. Influence of organic solvent modifier and solvent strength on peak shape of some basic compounds in high-performance liquid chromatography using a reversed-phase column. J. Chromatogr. A, 1995, 708, 185–194. 149. McCalley, D.V. Effect of organic solvent modifier and nature of the solute on the performance of bonded silica reversed-phase columns for the analysis of strongly basic compounds by high-performance liquid chromatography. J. Chromatogr. A, 1996, 738, 169–179. 150. Azzaoui, K. and Morin-Allroy, L. Quantitative structure-retention relationships for the investigation of the retention mechanism in high performance liquid chromatography using apolar eluent with a very low content of polar modifiers. Chromatographia, 1995, 40, 690–696. 151. Lewis, J.A., Lommen, D.C., Raddatz, W.D., Dolan, J.W. and Snyder, L.R. Computer simulation for the prediction of separation as a function of pH for reversed-phase high-performance liquid chromatography. I. Accuracy of a theory-based model. J. Chromatogr., 1992, 592, 183–195.

Lipophilicity Measurements by Liquid Chromatography

167

152. Kirkland, J.J., van Straten, M.A. and Claessens, H.A. High pH mobile phase effects on silica-based reversed-phase high-performance liquid chromatographic columns. J. Chromatogr. A, 1995, 691, 3–19. 153. Claessens, H.A., van Straten, M.A. and Kirkland, J.J. Personal communication, 1996. 154. Knox, J.H., Kaliszan, R. and Kennedy, G.F. Enthalpic exclusion chromatography. J. Chem. Soc. Faraday Symp., 1980, 15, 113. Cited in Lipophilicity and pKa estimates from gradient high-performance liquid chromatography. J. Chromatogr. A, 2002, 965, 117–127. 155. Belsner, K., Pfeifer, M. and Wilfert, B. Reversed-phase high-performance liquid chromatography for evaluating the lipophilicity of pharmaceutical substances with ionization up to logPapp = 8. J. Chromatogr., 1993, 629, 123–134. 156. Kaliszan, R., Marsza, M.P., Markuszewski, M.J., Bczek, T. and Pernak, V. Suppression of deleterious effects of free silanols in liquid chromatography by imidazolinium tetrafluoroborate ionic liquids. J. Chromatogr. A., 2004, 1030, 263–271. 157. Sebastiano, R., Gelfi, C., Righetti, P.G. and Citterio, A. Novel, trifunctional diamine for silica coating in capillary zone electrophoresis. J. Chromatogr. A, 2000, 894, 53–61. 158. Olivieri, E., Sebastiano, R., Gelfi, C., Righetti, P.G. and Citterio, A. Quantitation of protein binding to the capillary wall in acidic, isoelectric buffers and means for minimizing the phenomenon. J. Chromatogr. A, 2000, 894, 273–280. 159. Snyder, L.R., Kirkland, J.J. and Glajch, J.L. Practical HPLC Method Development, Wiley, New York, 1977. 160. Gilroy, J.J., Dolan, J.W. and Snyder, L.R. Column selectivity in reversed-phase liquid chromatography. IV. Type-B alkyl-silica columns. J. Chromatogr. A, 2003, 1000, 757–778. 161. Bayer, E. and Paulus, A. Silanophilic interactions in reversed-phase high-performance liquid chromatography. J. Chromatogr. A, 1987, 400, 1–4. 162. Hansen, S.M., Helboe, P. and Thomson, M. Bare silica, dynamically modified with long-chain quaternary ammonium ions — The technique of choice for more reproducible selectivity in reversed-phase high-performance liquid chromatography. J. Chromatogr., 1991, 544, 53–76. 163. Nawroki, J. and Buszewski, B. Influence of silica surface chemistry and structure on the properties, structure and coverage of alkyl-bonded phases for high-performance liquid chromatography. J. Cromatogr., 1988, 449, 1–24. 164. Tan, L.C. and Carr, P.W. Study of retention in reversed-phase liquid chromatography using linear solvation energy relationships. II. The mobile phase. J. Chromatogr. A, 1998, 799, 1–19. 165. Taylor, P.J. Comprehensive Medicinal Chemistry, in: Rational design, mechanistic study and therapeutic application of chemical compounds. Volume 4, Qualitative drug design, Sammes, P.G., Hansch, C., Taylor, J.B.; Ramsden, C.A. (Eds.), Oxford: Pergamon Press, UK, 1989. 166. Dearden, J.C. and Bresnen, G.M. The measurement of partition coefficients. Quant. Struct.-Act. Relat., 1988, 7, 133–144. 167. Abraham, M.H., Chadha, H.S., Leitao, R.A.E., Mitchell, R.C., Lambert, W.J., Kaliszan, R., Nasal, A. and Haber, P. Determination of solute lipophilicity, as log P(octanol) and log P(alkane) using poly(styrene-divinylbenzene) and immobilised artificial membrane stationary phases in reversed-phase high-performance liquid chromatography. J. Chromatogr. A, 1997, 766, 35–47.

168

Advances in Chromatography

168. Dias, N.C., Nawas, M.I. and Poole, C.F. Evaluation of a reversed-phase column (Supelcosil LC-ABZ) under isocratic and gradient elution conditions for estimating octanol-water partition coefficients. Analyst, 2003, 128, 427–433. 169. Nong, C., Lehtonen, P. and Yukui, Z. The linear S-logkw correlation (LSLC) analysis for homologous series in reversed-phase high performance liquid chromatography (RP-HPLC). Chinese J. Chromatogr., 1995, 13, 373–378. 170. Chen, N., Zhang, Y. and Lu, P. Effects of molecular structure on the logkw index and linear S-logkw correlation in RP-HPLC. J. Chromatogr., 1993, 633, 31–41. 171. Hallagas, B., Patonay, T., Kiss-Szikszai, A., Dobos, Zs., Hollósy, F., Ers, D., Orbi, L., Kéry, Gy. and Idei, M. Comparison of measured and calculated lipophilicity of substituted aurones and related compounds. J. Chromatogr. B, 2004, 801, 229–235. ˇ J., Klimesová, ˇ 172. Klimes, V. and Waisser, K. Chromatographic behaviour of dipyridylsulphides. Relationship between logk values and structure by RP-HPLC. J. Chromatogr., 1992, 595, 334–336. 173. Dimitrova, B., Doytchinova, I. and Zlatkova, M. Reversed-phase high-performance liquid chromatography for evaluating the distribution of pharmaceutical substances in suppository base-phosphate buffer system. J. Pharm. Biomed. Anal., 2000, 23, 955–964. 174. Natalini, B., Capodiferro, V., Mattoli, L., Marinozzi, M., Constantino, G. and Pellicciari, R. Chromatographic separation and evaluation of the lipophilicity by reversedphase high-performance liquid chromatography of fulerene-C60 derivatives. J. Chromatogr. A, 1999, 847, 339–343. 175. Tsantili-Kakoulidou, A., Filippatos, E., Todoulou, O. and Papadaki-Valiraki, A. Use of RP-HPLC in lipophilicity studies of 9 H-xanthene and 9 H-tioxanthene derivatives containing an aminoalkanamide or a nitrosoureido group. Comparison between the capacity factors and calculated octanol-water partition coefficients. J. Chromatogr. A, 1993, 654, 43–52. 176. de Voogt, P., Wegener, J.W.M., Brinkman, U.A.Th. and Govers, H. Retention of neutral and basic heteroaromatic hydrocarbons in RPLC systems and its use in predictive studies. I. Concentration of the organic modifier. The Science of the Total Environment, 1991, 109/110, 69–87. 177. Harnisch, M., Möckel, H.J. and Schulze, G. Relationship between logPo/w shake-flask values and capacity factors derived from reversed-phase high-performance liquid chromatography for n-alkylbenzenes and some OECD reference substances. J. Chromatogr., 1983, 282, 315–332. 178. Grouls, R.J.E., Ackerman, E.W., Korsten, H.H.M., Hellebrekers, L.J. and Breimer, D.D. Partition coefficients (n-octanol/water) of N-butyl-p-aminobenzoate and other local anaesthetics measured by reversed-phase high-performance liquid chromatography. J. Chromatogr. B, 1997, 694, 421–425. 179. Cimpan, G., Irimie, F., Gocan, S. and Claessens, H.A. Role of stationary phase and eluent composition on the determination of logP values of N-hydroxyethylamide of aryloxyalkylen and pyridine carboxylic acids by reversed-phase high-performance liquid chromatography. J. Chromatogr. B, 1998, 714, 247–261. 180. Belsner, K., Pfeifer, M. and Wilffert, B. RP-HPLC for evaluation the lipophilicity of pharmaceutical substances with ionisation up to logPapp = 8. J. Chromatogr., 1993, 629, 123–134. 181. de Kock, A.C. and Lord, D.A. A simple procedure for determining octanol-water partition coefficients using reverse phase high performance liquid chromatography (RPHPLC). Chemosphere, 1987, 16, 133–142.

Lipophilicity Measurements by Liquid Chromatography

169

182. Rothemund, S., Krause, E., Ehrlich, A. and Bienert, M. Determination of peptide hydrophobicity parameters by reversed-phase high-performance liquid chromatography. J. Chromatogr. A, 1994, 661, 77–82. 183. Yamagami, C. and Katashiba, N. Hydrophobicity parameters determined by reversed phase liquid chromatography. XI. Prediction of logP values for phenyl N-methyl and phenyl N,N-dimethylcarbamates. Chem. Pharm. Bull., 1996, 44, 1338–1343. 184. Hong, H., Zhou, D., Han, S., Wang, L., Zhang, Z. and Zou, G. Retention of aromatic sulfur containing compounds on RP-HPLC: Correlation with partition coefficients and molecular connectivity indices. J. Liq. Chrom. & Rel. Technol., 1996, 19, 1617–1629. 185. Altomare, C., Carotti, A., Cellamare, S., Carrieri, A., Ciabatti, R. and Malabarba, A. Lipophilicty of Teicoplanin antibiotics as assessed by reversed phase high performance liquid chromatography, quantitative structure-property and structure-activity relationships. J. Pharm. Pharmacol., 1994, 46, 994–999. 186. Guesdon, M.-P., Dallet, Ph. and Dubost, J.-P. Comparaison des logP et des logKw d’une série de molécules très lipophiles. Bull. Soc. Pharm. Bordeaux, 1994, 134, 83–92. 187. Valkó, K., Bevan, C. and Reynolds, D. Chromatographic hydrophobicity index by fast-gradient RP-HPLC: A high-throughput alternative to log P/log D. Anal. Chem., 1997, 69, 2022–2029. 188. Valkó, K., Plass, M., Bevan, C., Reynolds, D. and Abraham, M.H. Relationships between the chromatographic hydrophobicity indices and solute descriptors obtained by using several reversed-phase, diol, nitrile, cyclodextrin and immobilised artificial membrane-bonded high-performance liquid chromatography columns. J. Chromatogr. A, 1998, 797, 41–55. 189. Plass, M., Valkó, K. and Abraham, M.H. Determination of solute descriptors of tripeptide derivatives based on high-throughput gradient high performance liquid chromatography retention data. J. Chromatogr. A, 1998, 803, 51–60. 190. Du, C.M., Valkó, K., Bevan, C., Reynolds, D. and Abraham, M.H. Characterizing the selectivity of stationary phases and organic modifiers in reversed-phase high-performance liquid chromatographic systems by a general solvation equation using gradient elution. J. Chromatogr. Sci., 2000, 38, 503–511. 191. Valkó, K., Espinosa, S., Du, C.M., Bosch, E., Rosés, M., Bevan, C. and Abraham, M.H. Unique selectivity of perfluorinated stationary phases with 2,2,2-trifluoroethanol as organic mobile phase modifier. J. Cromatogr. A, 2001, 933, 73–81. 192. Kaune, A., Knorrenschild, M. and Kettrup, A. Predicting 1-octanol-water partition coefficient by high-performance liquid chromatography gradient elution. Fresenius J. Anal. Chem., 1995, 352, 303–312. 193. Krass, J.D., Jastorff, B. and Genieser, H.G. Determination of lipophilicity by gradient elution high-performance liquid chromatography. Anal. Chem., 1997, 69, 2575–2581. 194. Snyder, L.R. and Dolan, J.W. Initial experiments in high-performance liquid chromatographic method development. I. Use of a starting gradient run. J. Chromatogr. A, 1996, 721, 3–14. 195. Kaliszan, R., Haber, P., Baczek, T., Siluk, D. and Valkó, K. Lipophilicity and pKa estimates from gradient high-performance liquid chromatography. J. Chromatogr. A, 2002, 965, 117–127. 196. http://www.lcresources.com. 197. Kerns, E.H., Di, L., Petusky, S., Kleintop, T., Huryn, D., McConnel, O. and Carter, G. Pharmaceutical profiling method for lipophilicity and integrity using liquid chromatography-mass spectrometry. J. Chromatogr. B, 2003, 791, 381–388.

170

Advances in Chromatography

198. Gulyaeva, N., Zaslavsky, A., Lechner, P., Chlenov, M., McConnell, O., Chait, A., Kipnis, V. and Zaslavsky, B. Relative hydrophobicity and lipophilicity of drugs measured by aqueous two-phase partitioning, octanol-buffer partitioning and HPLC. A simple model for predicting blood-brain distribution. European J Med. Chem., 2003, 38, 391–396. 199. Jandera, P., Halama, M. and Novotná, K. Stationary phase effects in gradient highperformance liquid chromatography. J. Chromatogr. A, 2004, 1030, 33–41. 200. Abraham, M.H., Ibrahim, A., Zissimos, A.M., Zhao, Y.H., Comer, J. and Reynolds, D.P. Application of hydrogen bonding calculations in property based drug design. Drug Discov. Today, 2002, 7, 1056–1063. 201. Zissimos, A.M., Abraham, M.H., Du, C.M., Valkó, K., Bevan, C., Reynolds, D., Wood, J. and Tam, K.Y. Calculation of Abraham descriptors from experimental data from seven HPLC systems, evaluation of five different methods of calculation. J. Chem. Soc. Perkin Trans. 2, 2002, 12, 2001–2010. 202. Zissimos, A.M., Abraham, M.H., Barker, M.C., Box, K.J. and Tam, K.Y. Calculation of Abraham descriptors from solvent-water partition coefficients in four different systems, evaluation of different methods of calculation. J. Chem. Soc. Perkin Trans. 2, 2002, 3, 470–477. 203. Yamagami, C., Takao, N. and Fujita, T. Hydrophobicity parameter of diazines. III. Relationship of partition coefficients of monosubstituted diazines and pyridines in different partitioning systems. J. Pharm. Sci., 1993, 82, 155–161. 204. Du, C.M., Valkó, K., Bevan, C., Reynolds, D. and Abraham, M.H. Rapid method for estimation octanol-water partition coefficient (logPoct) from isocratic RP-HPLC and a hydrogen bond acidity term (A). J. Liq. Chromatogr. Rel. Technol., 2001, 24, 635–649. 205. Du, C.M., Valko, K., Bevan, C., Reynolds, D. and Abraham, M.H. Rapid gradient RP-HPLC method for lipophilicity determination: A solvation equation based comparison with isocratic methods. Anal. Chem., 1998, 70, 4228–4234. 206. Martin, A.J.P. and Synge, R.L.M. A new form of chromatogram employing two liquid phases. Biochem. J., 1941, 35, 1358–1368. 207. Kowalska, T. and Prus, W. Theory and mechanism of thin-layer chromatography. Encyclopedia of Chromatography, Marcel Dekker, New York, 2003, 821–825. 208. Bate-Smith, E.C. and Westall, R.G. Chromatographic behavior and chemical structure. 1. Some naturally occurring phenolic substances. Biochim. Biophys. Acta, 1950, 4, 427–440. 209. Snyder, L.R. Principles of Adsorption Chromatography, Marcel Dekker, New York, 1968. 210. Soczewinski, E. Solvent composition effects in thin-layer chromatography systems of the type silica gel-electron donor solvent. Anal. Chem., 1969, 41, 179–182. 211. Boyce, C.B.C. and Milborrow, B.V. A simple assessment of partition data for correlating structure and biological activity using thin-layer chromatography. Nature, 1965, 208, 537–539. 212. Jack, D.B., Hawker, J.L., Rooney, L., Beerahee, M., Lobo, J. and Patel, P. Measurment of the distribution coefficients of several classes of drug using reversed-phase thinlayer chromatography. J. Chromatogr. 1988, 452, 257–264. 213. Hulshoff, A. and Perrin, J.H. Chromatographic characterization of phenothiazine drugs by a reversed-phase thin-layer technique. J. Chromatogr. 1976, 129, 249–262. 214. Hulshoff, A. and Perrin, J.H. A reversed-phase thin-layer chromatographic method for the determination of relative partition coefficients of very lipophilic compounds. J. Chromatogr., 1976, 120, 65–79.

Lipophilicity Measurements by Liquid Chromatography

171

215. Biagi, G.L., Barbaro, A.M., Sapone, A. and Recanatini, M. Determination of lipophilicity by means of reversed-phase thin-layer chromatography. II. Influence of the organic modifier on the slope of the thin-layer chromatographic equation. J. Chromatogr. A, 1994, 669, 246–253. 216. Biagi, G.L., Barbaro, A.M. and Sapone, A. Determination of lipophilicity by means of reversed-phase thin-layer chromatography. I. Basic aspects and relationship between slope and intercept of TLC equations. J. Chromatogr. A, 1994, 662, 341–361. 217. Biagi, G.L., Barbaro, A.M., Gamba, M.F. and Guerra, M.C. Partition data of penicillins determined by means of reversed-phase thin-layer chromatography. J. Chromatogr., 1969, 41, 371–379. 218. Biagi, G.L., Guerra, M.C., Barbaro, A.M., Barbieri, S., Recanatini, M., Borea, P.A. and Pietrograde, M.C. Study of the lipophilicity character of xanthine and adenosine derivatives. I. RM and logP values. J. Chromatogr., 1990, 498, 179–190. 219. Biagi, G.L., Barbaro, A.M., Sapone, A., Borea, P.A., Varani, K. and Recanatini, M. Study of lipophilic character of serotonergic ligands. J. Chromatogr. A, 1996, 723, 135–143. 220. Dross, K., Sonntag, C.H. and Mannhold, R. Determination of the hydrophobicity parameter RMw by reversed-phase thin-layer chromatography. J. Chromatogr., 1994, 673, 113–124. 221. Dross, K., Sonntag, C.H. and Mannhold, R. On the precise estimation of RM values in reversed-phase thin-layer chromatography including aspects of pH dependence. J. Chromatogr., 1993, 639, 287–294. 222. Slawik, T. and Kowalski, C. Lipophilicity of a series of 1,2-benzisothiazole-3(2H)ones determined by reversed-phase thin-layer chromatography. J. Chromatogr. A, 2002, 952, 295–299. 223. Barbaro, A.M., Pietrogrande, M.C, Guerra, M., Cantelli, G., Borea, P.A. and Biagi, G.L. Relationship between the chromatographic behaviour of dermophin-related oligopeptides and the composition of the mobile phase in reversed-phase thin-layer chromatography. Comparison of extrapolated RM values. J. Chromatogr. 1984, 287, 259–270. 224. Barbaro, A.M., Guerra, M.C, Biagi, G.L., Pietrogrande, M.C. and Borea, P.A. Influence of the mobile phase composition on the reversed-phase thin layer chromatographic behaviour of a series of prosta glandins. Comparison of extrapolated RM values. J. Chromatogr., 1985, 347, 209–218. 225. Cserháti, T. and Magyar, K. Lipophilicity determination of some monoamine oxidase inhibitors: The effect of methanol and ammonia chloride. J. Chromatogr., 1992, 575, 57–62. 226. Pyka, A.J. The topological indices and RM values of benzoic acid derivatives: a structure-activity investigation. Planar Chromatogr., 1994, 7, 108–116. 227. Dzimiri, N., Fricke, U. and Klaus, W. Influence of derivatization on the lipophilicity and inhibitory actions of cardiac glycosides on myocardial sodium-potassium ATPase. J. Pharmacol., 1987, 91, 31–38. 228. Wilson, J. Ion-pair reversed-phase thin-layer chromatography of organic acids: Reinvestigation of the effects of solvent pH on RF values. J. Chromatogr., 1986, 354, 99–106. 229. Biagi, G.L., Barbaro, A.M., Sapone, A. and Recanatini, M. Determination of lipophilicity by means of reversed-phase thin-layer chromatography. I. Basic aspects and relationship between slope and intercept of TLC equations. J. Chromatogr. A, 1994, 662, 341–361.

172

Advances in Chromatography

230. Bijloo, G.J. and Rekker, R.F. Influence of stationary phase modifications on lipophilicity measurement of benzophenone using reversed-phase thin-layer chromatography. J. Chromatogr., 1986, 351, 122–127. 231. Cserháti, T., Darwish, Y.M. and Matolcsy, G. Effect of support characteristics on the determination of the lipophilicity of some neutral, acidic and alkaline compounds by reversed-phase thin-layer chromatography. J. Chromatogr., 1983, 270, 97–104. 232. Gullner, G., Cserháti, T., Bordas, B. and Szögyi M. Mixed support in the reversedphase thin-layer chromatography of amino acids. J. Liq. Chromatogr., 1986, 9, 1919–1931. 233. Garssini-Strazza, G., Carunchio, B. and Girelli, A.M. Flat-bed chromatography on impregnated layer. J. Chromatogr., 1989, 466, 1–35. 234. Cserháti, T. and Szögyi, M. Determination of the lipophilicity of some peptides. Effect of surface pH of silica. J. Chromatogr., 1990, 520, 249–256. 235. Cserháti, T. and Forgács, E. Effect of TLC support characteristics and coating on the lipophilicity determination of phenols and aniline derivatives. J. Chromatogr. Sci., 2000, 40, 564–568. 236. Forgács, E., Cserháti, T., Kaliszan, R., Haber, P. and Nasal, A. Reversed-phase thinlayer chromatographic determination of the hydrophobicity parameters of nonsteroidal anti-inflammatory drugs. J. Planar Chromatogr., 1998, 11, 383–387. 237. Cserháti, T., Forgács, E., Kiss, G.C. and Augustin, J. Effect of salts on the hydrophobicity parameters of sulfosuccinic acid esters studied by reversed-phase thin-layer chromatography. J. Planar Chromatogr., 1997, 10, 441–446. 238. Darwish, Y., Cserháti, T. and Forgács, E. Relationship between lipophilicity and specific hydrophobic surface area of a non-homologous series of pesticides. J. Planar Chromatogr., 1993, 6, 458–462. 239. Cserháti, T., Forgács, E. and Hajos, G. Determination of the lipophilicity of fusedring nitrogen heterocycles by reversed-phase thin-layer chromatography. J. Planar Chromatogr., 1998, 11, 64–69. 240. Wallerstein, S., Cserháti, T. and Fischer, J. Determination of the lipophilicity of some anti-hypoxia drugs: Comparation of TLC and HPLC methods. Chromatographia, 1993, 35, 275–280. 241. Darwish, Y., Cserháti, T. and Forgács, E. Reversed-phase characteristics of some bioactive heterocyclic compounds. J. Chromatogr., 1994, 668, 485–494. 242. Wallerstein, S., Cserháti, T. and Fischer, J. Determination of the luipophilisity of some anti-hypoxia druds: comparation of TLC and HPLC methods. Chromatographia, 1993, 35, 275–280. 243. Biagi, G.L., Barbaro, A.M. and Recanatini, M. Determination of lipophilicity by means of reversed-phase thin-layer chromatography. III. Study of the TLC equations for a series of ionizable quinolone derivatives. J. Chromatogr. A, 1994, 678, 127–137. 244. Perisic-Janjic, N.U., Dajakovi-Seculi, T.L. and Popov-Pergal, K. Effect of the stationary phase and the mobile-phase modifier on the retention of some thiazoles. Correlation with the lipophilicity of the compounds. J. Planar Chromatogr., 2003, 16, 363–368. 245. Almási, J., Takács-Novák, K., Kökösi, J. and Vámos, J.J. Characterization of potential NMDA and cholecystokinin antagonists. II. Lipophilicity studies on 2-methyl-4-oxo3H-quinazoline-3-alkyl-carboxylic acid derivatives. Int. J. Pharm. 1999, 180, 13–22. 246. Mazák, K., Vámos, J., Nemes, A., Rácz, Á. and Noszál, B. Lipophilicity of vinpocetine and related compounds characterized by reversed-phase thin-layer chromatography. J. Chromatogr. A, 2003, 996, 195–203.

Lipophilicity Measurements by Liquid Chromatography

173

247. Abraham, M.H., Poole, C.F. and Poole, S. Solute effect on reversed-phase thin-layer chromatography. A linear free energy relationship analysis. J. Chromatogr. A, 1996, 749, 201–209. 248. Greenwald, H.L., Kice, E.B., Kelly, M. and Kelly, J. Determination of the distribution of non-ionic surface active agents between water and isooctane. Anal. Chem., 1961, 33, 405–468. 249. Schott, H. Hydrophilic-lipophilic balance and distribution coefficients of non ionic surfactants. J. Pharm. Sci., 1971, 60, 648–649. 250. Hahn, L. and Sucker, H. HLB determination by HPLC. Tenside Surf. Det., 1989, 26, 192–194. 251. Trapani, G., Altomare, C., Franco, M., Latrofa, A. and Liso, G. Determination of hydrophile-lipophile balance of some polyethoxylated non-ionic surfactants by reversed-phase thin-layer chromatography. Int. J. Pharm., 1995, 116, 95–99 252. Mannhold, R., Dross, K.P. and Rekker, R.F. Drug lipophilicity in QSAR practice. I. A comparison of experimental with calculative approaches. Quant. Struct.-Act. Relat., 1990, 9, 21–28. 253. Nys, G.G. and Rekker, R.F. The concept of hydrophobic fragmental constants (fvalues). II. Extension of its applicability to the calculation of lipophilicities of aromatic and heteroaromatic structures. Eur. J. Med. Chem., 1974, 9, 361–375. 254. Leo, A., Jow, P.Y.C., Silipo, C. and Hansch, C. Calculation of hydrophobic constant (logP) from π and f-constant. J. Med. Chem., 1975, 18, 865–868. 255. Dross, K.P., Mannhold, R. and Rekker, R.F. Drug lipophilicity in QSAR practice. II. Aspects of RM determinations, critics of RM-corrections, interrelations with partition coefficients. Quant. Struct.-Act. Relat., 1992, 11, 36–44. 256. Geronikaki, A. and Hadjipavlou-Litina, D.J. Lipophilicity and antiinflamatory activity of some 2-(aminoacetylamino)thiazole derivatives. Pharmazie, 1993, 48, 948–949. 257. Gocan, S., Irimie, F. and Cimpan, G. Prediction of the lipophilicity of some plant growth-stimulating amido esters of ethanolamine using reversed-phase thin-layer chromatography. J. Chromatogr. A, 1994, 675, 282–285. 258. Cimpan, G. and Miclu, V. Normal phase thin layer chromatography and a lipophilicity study by reversed phase thin layer chromatography of some 2-amino-3-cyano-4,5diphenylfurane derivatives. Studia Universitatis Babe-Bolyai, Chemia, 1997, 42, 225–231. 259. Malawska, B., Kulig, K. and Winiewska, M. Determination of the lipophilicity of antiarrhythmic and antihypertensive 1-[2-hydroxy- or 1-[2-acetoxy-3-(4-aryl-1-piperazinyl)propyl]-pyrrolidin-2-one derivative. J. Planar Chromatogr., 2000, 13, 187–190. 260. Kulig, K. and Malawska B. Estimation of the lipophilicity of antiarrhythmic and antihypertensive active 1-substituted pyrrolidin-2-one and pyrrolidine derivatives. Biomed. Chromatogr., 2003, 17, 318–324. 261. Forgács, E. and Cserháti, T. Effect of various organic modifiers on the determination of the hydrophobicity parameters of a non-homologous series of anticancer drugs. J. Chromatogr. A, 1995, 697, 59–69. 262. Matosiuk, D. and Jówiak, K. Determination of lipophilicity of carbonyl derivatives of 2-aminoimidazolines-2 by reversed-phase thin-layer chromatography. Part 1. Lipophilicity of 1-(1-arylimidazolin-2-yl)-3-arylureas and 1-aryl-2-imine-3-arylaminocarbonilimidazolidines. J. Planar Chromatogr., 2000, 13, 52–56. 263. Petrovi, S.M,. Lonar, E. and Pejanovi, V. Corelation between retention 1-octanolwater partition coefficients of some extrane derivatives in reversed-phase thin-layer chromatography. J. Chromatogr. Sci., 2002, 40, 569–574.

174

Advances in Chromatography

264. Sawik, T. and Kowalski, C. Lipophilicity of a series of 1,2-benzisothiazol-3(2H)-ones determined by reversed-phase thin-layer chromatography. J. Chromatogr. A, 2002, 952, 295–299. 265. Tacács-Novák, K., Perjési, P. and Vámos, J. Determination of logP for biologically active chalcones and cyclic chalcone analogs by RPTLC. J. Planar Chromatogr., 2001, 14, 42–46. 266. Cimpan, G., Bota, C., Coman, M., Grinberg, N. and Gocan, S. A lipophilicity study for some 2-hydrazinothiazolic derivatives with antifungal activity by reversed phase thin-layer chromatography. J. Liq. Chrom. & Rel. Technol., 1999, 22, 29–40. 267. Sawik, T. and Paw, B. Lipophilicity of some N- and O-substituted alkanoic acids of 1,2-benzisothiazol-3(2H)-one determined by reversed-phase thin-layer chromatograph. J. Liq. Chrom. & Rel. Technol., 2004, 27, 1043–1055. 268. Gocan, S., Cimpan, G. and Panea, T. Normal phase TLC and reversed phase HPTLC of some plant growth stimulators. J. Planar Chromatogr., 1994, 7, 435–439. 269. Malawska, B. Determination of lipophilicity of some N-substituted amides of γpiperazine-γ-hydroxybutiric acid. J. Planar Chromatogr., 1998, 11, 137–140. 270. Rekker, R.F. and de Kort, H.M. The hydrophobic fragmental constant, an extension to a 1000 data point set. Eur. J. Med. Chem.-Chimica Therapeutica, 1979, 14(6), 479–488. 271. Malankiewicz, M.J. and Pluta, K. TLC Separation of - and -quinolinyl sulfides. J. Planar Chromatogr., 1994, 7, 484–486. 272. Irimie, F. Ph.D. thesis, Babes-Bolyai University, Cluj-Napoca, Romania, 1993. 273. Cserháti, T. and Bordás, B. Determination of the lipophilicity of some aniline derivatives by reversed-phase thin-layer chromatography. The effect of salts. Chromatographia, 1986, 21, 312–316. 274. Biagi, G.L., Barbaro, A.M., Sapone, A. and Recanati, M. Thin-layer chromatography study of the lipophilicity of triazine herbicides influence of different organic modifiers. J. Chromatogr., 1992, 625, 392–396. 275. Glavic, D. RM values of some colchicines and colchiceinamides determined by reversed-phase thin-layer chromatography. J. Chromatogr., 1992, 591, 367–370. 276. Matysiak, J., Niewiadomy, A., Macik-Niewiadomy, G. and Kornillowicz, T. Dependence of fungistatic activity of 2,4-dihydroxythiobenzanilides on the structure and lipophilic nature of the compounds. Eur. J. Med. Chem., 2000, 35, 393–404. 277. Kontogiorgis, C. and Hadjipavlou-Litina, D. Biological evaluation of several coumarin derivatives designed as possible anti-inflammatory/antioxidant agents. J. Enzyme Inhib. Med. Chem., 2003, 18, 63–69. 278. Kontogiorgis, C., Hadjipavlou-Litina, D. and Schulz, E. Antioxidant activity of DL -phenyl-amino acid octyl esters with anti-inflamatory activity. Correlation of the structure with lipophilicity. Arzneimittelforschung, 2001, 51, 485–488. 279. Cserháti, T. Lipophilicity determination of non-homologous series of nonionic surfactants by means of reversed-phase thin-layer chromatography. J. Biochem. Biophys. Methods, 1993, 27, 133–142. 280. Cserháti, T., Forgács, E. and Oros G. Effect of salt concentration and pH on the hydrophobicity parameters of surfactants studied by TLC and spectral mapping technique. J. Biochem. Biophys. Methods, 1999, 38, 1–15. 281. Wang, Q.S., Zhang, L., Yang, H.Z. and Liu, H.Y. Lipophilicity determination of some potential photosystem II inhibitors on reversed-phase thin-layer chromatography. J. Chromatogr. Sci., 1999, 37, 41–44.

Lipophilicity Measurements by Liquid Chromatography

175

282. Kastner, P., Kuchar, M., Kilinmes, J. and Dosedlova, D. Relationship between structure and reversed-phase thin-layer chromatographic lipophilicity parameters in a group of piperazine derivatives. J. Chromatogr. A, 1997, 766, 165–170. 283. Hucha, M., Kraus, E. and Jelínková, V. Influence of mobile phase composition on evaluation of lipophilicity by partition chromatography. J. Chromatogr., 1991, 557, 399–411. 284. Biagi, G.L., Pietrogrande, M.C., Barbaro, A.M., Guerra, M.C., Borea, P.A. and Cantelli Forti, G. Study of the lipophilic character of a series of -carbolines. J. Chromatogr., 1989, 469, 121–126. 285. Biagi, G.L., Barbaro, A.M., Guerra, M.C., Borea, P.A. and Recanatini, M. Lipophilic character of cardiac glycosides. RM values as lipophilicity parameters. J. Chromatogr., 1990, 504, 163–178. 286. Valkó, K., Olajos, S. and Cserháti, T. Prediction of the high-performance liquid chromatographic retention behavior of some benzodiazepine derivatives by thin-layer chromatography. J. Chromatogr., 1990, 499, 361–371. 287. Bidló-Iglóy, M. Reversed-phase thin-layer and high-performance liquid chromatography of aromatic alkoxy and hydroxy acids. J. Chromatogr., 1989, 475, 321–329. 288. Dondi, F., Grassini-Strazza, G., Kahie, Y.D., Lodi, G., Pietrogrande, C., Reschiglian, P. and Bighi, C. Comparison of reversed-phase thin-layer and high performance liquid chromatography of flavonoid compounds. J. Chromatogr., 1989, 462, 205–217. ˇ N.U., Aanski, M.M., Janji, N.J., Lazarevi, M.D. and Dimova, V. Study of the 289. Perisic, lipophlicity of some 1,2,4-Triazole derivatives by RPHPLC and TLC. J. Planar Chromatogr., 2000, 13, 281–284. 290. Róyo, J.K., Matysiak, J., Niewiadomy, A. and Abiska, A. Reversed-phase HPLC and HPTLC characterization of potential fungicides. J. Planar Chromatogr., 2000, 13, 176–181. 291. Bieganowska, M. Chromatographic retention data of N-alkylamides of benzomorpholine-2-carboxylic acid in the study of structure-activity relationships. Chromatographia, 1984, 18, 456–458. ˘ 292. Petrauskas, A.A. and Svedas, V.K. Hydrophobicity of -lactam antibiotics. Explanation and prediction of their behavior in various partitioning systems and reversed-phase chromatography. J. Chromatogr., 1991, 585, 3–34. 293. Kaliszan, R., Nasal, A. and Markuszewski, M.J. New approaches to chromatographic determination of xenobiotics. Anal. Bioanal. Chem., 2003, 377, 803–811. 294. Khaledi, M.G. and Breyer, E.D. Quantitation of hydrophobicity with micellar liquid chromatography. Anal. Chem., 1989, 61, 1040–1047. 295. Detroyer, A., Vander Heyden, Y., Carda-Broch, S., Garciá-Alvarez-Coque, M.C. and Massart, D.L. Quantitative structure-retention and retention-activity relationships of -blocking agents by micellar liquid chromatography. J. Chromatogr. A, 2001, 912, 211–221. 296. Martin-Biosca, Y., Molero-Monfort, M., Sagrado, S., Villanueva-Camanas, R.M. and Medina-Hernández, M.J. Development of predictive retention-activity relationship models of barbiturates by micellar liquid chromatography. Quant. Struct.-Act. Relat., 2000, 19, 247–256. 297. Tanaka, A., Nakamura, K., Nakanishi, I. and Fujiwara, H. A novel and useful descriptor for hydrophobicity, partition coefficient micellar-water, and its application to a QSAR study of antiplatelet agents. J. Med. Chem., 1994, 37, 4563–4566.

176

Advances in Chromatography

298. Martorell, C., Calpena, A.C., Escribano, E., Poblet, J.M. and Freixas, J. Influence of the chromatographic capacity factor (log k) as an index of lipophilicity in the antibacterial activity of a series of 6-fluoroquinolones: Relationship between physicochemical and structural properties and their hydrophobicity. J. Chromatogr. A, 1993, 655, 177–184. 299. Garcia, M.A. and Marina, M.L. Study of the k or log k-logPow correlation for a group of benzene derivatives and polycyclic aromatic hydrocarbons in micellar liquid chromatography. J. Chromatogr. A, 1994, 687, 233–239. 300. Li, X. and Fritz, S. Mixed surfactants as mobile phase additives for the separation of organic compounds by HPLC. Anal. Chem., 1996, 68, 4481–4488. 301. Rodgers, A.H. and Khaledi, M.G. Influence of pH on retention and selectivity in micellar liquid chromatography: Consequences of micellar-induced shifts of ionization constants. Anal. Chem., 1994, 66, 327–334. 302. Zou, H., Zhang, Y., Hong, M. and Lu, P. Comparison of measurements of noctanol/water partition coefficient of organic ions by RP-HPLC and RP-IPC. Progress in Natural Science, 1995, 5, 106–112. 303. Zou, H., Zhang, Y., Hong, M. and Lu, P. Measurement of partition coefficients by reversed phase ion-pair chromatography. J. Chromatogr., 1992, 625, 169–175. 304. Foucault, A. Countercurrent chromatography. Anal. Chem., 1991, 63, 569–579. 305. Marston, A. and Hostettmann, K. Counter-current chromatography as a preparative tool, and perspectives. J. Chromatog. A, 1994, 658, 315–341. 306. http://www.pharma-tech.com. 307. Berthod, A. Centrifugal partition chromatography. Chromatographic Science, Vol. 68, Marcel Dekker, 167–196. 308. Berthod, A., Menges, R. and Armstrong, D. Direct octanol/water partition coefficient determination using co-current chromatography. J. Liq. Chromatogr., 1992, 15, 2769–2785. 309. Gluck, S.J., Martin, E. and Benko, M.H. Centrifugal partition chromatography for the determination of octanol-water partition coefficients, in in Foucault, A.P. (Ed), Centrifugal Partition Chromatography. Chromatographic Science, Vol. 68, Marcel Dekker, 1994, 199–218. 310. Avdeef, A. Absorption and drug development. Solubility, Permeability and Charge State, Wiley Interscience, John Wiley & Sons, New York, 2003.

4

Concepts and Practice of Multidimensional HighPerformance Liquid Chromatography R. Andrew Shalliker University of Western Sydney, Australia

Michael J. Gray University of Western Sydney, Australia

CONTENTS 4.1

4.2 4.3

4.4 4.5 4.6 4.7

Introduction ..................................................................................................178 4.1.1 Selectivity in Liquid Chromatography ............................................179 4.1.2 Solvent Selectivity ...........................................................................179 4.1.3 Stationary Phase Selectivity.............................................................180 4.1.4 Temperature Selectivity ...................................................................182 4.1.5 Mobile Phase pH Selectivity ...........................................................184 One-Dimensional Separations......................................................................184 Multidimensional Separations......................................................................185 4.3.1 Assessment of the Separation Potential ..........................................185 4.3.1.1 SMO Treatment of the Two-Dimensional Approach .......186 Types of Chromatographic Separation Displacements ...............................188 Two-Dimensional Chromatographic Systems .............................................191 Sample Dimensionality ................................................................................192 Orthogonality in Two-Dimensional Liquid Chromatography .....................193 4.7.1 Determination of Orthogonality for Two-Dimensional Separations .......................................................................................195 4.7.2 Statistical Approaches ......................................................................197 4.7.2.1 Correlation Coefficients and Color Maps ........................197 4.7.2.2 Principle Component Analysis .........................................199 4.7.2.3 Information Theory...........................................................200 4.7.2.4 Geometric Approach to Factor Analysis ..........................202 4.7.2.5 Combination Methods ......................................................204

177

178

Advances in Chromatography

4.8

Application of Multidimensional Liquid Chromatography ........................208 4.8.1 Designing Two-Dimensional Separations in Accordance with the Findings of Orthogonality Studies ....................................212 4.8.2 Contour Maps...................................................................................215 4.8.3 Two-Dimensional System Design....................................................215 4.8.4 Dual- and Quad-Column Second-Dimension Designs....................218 4.9 Anomalies in Multidimensional Separations...............................................224 4.10 Conclusion....................................................................................................226 Acknowledgments..................................................................................................228 References..............................................................................................................228

4.1 INTRODUCTION High-performance liquid chromatography (HPLC) is a long established method used for the separation, identification, and determination of chemical components in complex mixtures [1]. This analytical tool is employed for the qualitative identification and quantitative determination of separated species [1, 2] and finds widespread application in almost all areas of science. Many different modes of liquid chromatography exist that depend on the type of liquid and the type of stationary phase. For example, the technique referred to as normal phase liquid chromatography (NPLC) employs a polar stationary phase and a nonpolar mobile phase. The reverse of this process, referred to as reversed-phase liquid chromatography (RPLC), utilizes a nonpolar stationary phase and a polar mobile phase. In both these techniques, the stationary phase is usually a chemically modified solid support with either a polar or nonpolar ligand, depending on the desired mode of separation. Ion exchange liquid chromatography is a technique for separating ionic species where the solid support is chemically modified using an ionic ligand. Separation of charged species depends on the pH and electrolytic strength of the mobile phase. Ion exclusion and ion-paired chromatographies are two other techniques employed for the separation of ionic substances. Size-exclusion chromatography (SEC) separates species according to their molecular size. Molecules traverse a porous stationary phase in a mobile-phase environment that is unfavorable for enthalpic interactions. The molecules are consequently separated according to their entropic exclusion from within the porous network. This technique is also referred to as gel permeation chromatography (GPC). Affinity chromatography is a technique with widespread use in the biosciences. In this technique, the stationary phase is modified with a solute receptor targeted for the retention and hence isolation of a specific solute. Chiral chromatography is employed for the separation of enantiomers. The stationary phase is modified with chiral ligands that preferentially retain one enantiomer; the technique is spatially selective. Chiral mobile phases can also be employed to exploit the “handedness” of the target molecules. These illustrations of the different modes of separation are by no means comprehensive but they do show the vast number of liquid chromatographic processes at the disposal of the chromatographer. In fact, too many different stationary phases exist to list here but a scan through any supplies catalogue will

Concepts and Practice of High-Performance Liquid Chromatography

179

illustrate the vast choices available to the chromatographer. Even then, there are many specialty stationary phases that are not commercially available. As a consequence, by appropriate selection of the most suitable techniques, there is virtually no sample — regardless of the complexity — that cannot be reduced to a simpler form, allowing for analysis of the sample constituents. In this chapter, we are not concerned with mechanisms of retention; rather, our interest lies in selectivity. The aim of this chapter is therefore to investigate processes in which selectivity can be exploited in order to gain separation. One way to increase selectivity and also increase the peak capacity of a liquid chromatographic system is to incorporate more than one separation dimension — that is, utilize the process referred to as multidimensional liquid chromatography. Here, multiple separating columns are coupled together in an automated system in such a manner as to improve the separation potential, allowing complex samples to be resolved into simpler systems. Before we begin our discussion on multidimensional chromatography, we must first discuss single-dimensional systems and understand factors that control selectivity.

4.1.1 SELECTIVITY

IN

LIQUID CHROMATOGRAPHY

Chromatographic separation depends on the selective retention of the stationary phase. The selectivity (α) can be expressed as a relative adjusted retention, referred to as the selectivity or separation factor [3]:

α=

k2 k1

(4.1)

where k1 and k2 are the retention factors of compound 1 and compound 2, respectively. By definition, α is greater than or equal to α unity and when α = 1, there is no separation. An increase in the separation selectivity is a powerful technique for improving resolution because it is only through the adjustment of selectivity that resolution can be increased and separation time can be decreased. Factors that influence the separation selectivity include the chemical composition of the stationary and mobile phases, temperature and mobile phase pH. However, it is often difficult to predict selectivity changes and a trial-and-error approach is often made when determining conditions for a change in α.

4.1.2 SOLVENT SELECTIVITY Solvent selectivity is well-documented and can have an enormous effect when choosing a chromatography system for a certain separation. Retention in RPLC, for example, is described in terms of the free energy change upon the transfer of solute from the mobile phase to the stationary phase. In two-phase partitioning processes of solutes, the enthalpic interactions determine the equilibrium, which is driven by the molecular interactions between the sample solute and the two phases [4]. The equilibria between the solutes and the mobile phase are primarily the result of four types of interactions: dispersion, dipole, hydrogen bonding, and dielectric [3].

180

Advances in Chromatography

Similar interactions occur between the sample and the stationary phase [3] and when different sample components have different equilibria and interactions, chromatographic selectivity is established. The ability of a sample or solvent molecule to interact in all four ways is commonly referred to as the polarity. It is well known that polar solvents dissolve polar compounds and, conversely, nonpolar solvents dissolve nonpolar solutes [3]. The effect of these intermolecular interactions on the retention behavior of a solute undergoing separation in liquid chromatography can be illustrated in the separation of the oligostyrenes, as shown in Figure 4.1. Each peak (labeled 3 to 13) corresponds to a different oligomer with the first peak (n = 3) corresponding to the trimer (i.e., an oligomer containing three configurational repeating units). On a C18 column with a methanol mobile phase, all the polystyrene diastereomers are eluted at the same time for each respective molecular weight. The uniformity and systematic band displacement suggests that all components have been completely separated; however, when an acetonitrile mobile phase is employed for the same separation, some of these bands shoulder, as shown in Figure 4.1b. This shouldering is due to the presence of different tactic isomers [5], which perfectly coelute in the C18methanol system, but when the mobile phase is changed to acetonitrile, the expression of diastereomer separation becomes apparent [6]. Various processes have been used to classify solvent selectivity but perhaps the most widely cited is the solvent-selectivity triangle proposed by Snyder [3]. This classification scheme is based on the gas–liquid distribution constants originally reported by Rohrschneider [7]. In this triangle, solvents are grouped on their ability to donate or accept protons and to induce a dipole moment. Eight distinct groups characterize the solvents in this triangle; solvents yielding similar selectivity are grouped in clusters. Separation selectivity is achieved by substituting one solvent from one group with that of a different solvent from another group. Substitution with solvents within the same selectivity group should result in little change to the separation selectivity. More recently, Kamlet et al. [7a] found that this classification was not entirely correct, since the test compounds used by Rohrschneider [7] could exhibit mixed-interaction tendencies. An alternative solvent-selectivity triangle [8] based on the Kamlet-Taft solvatochromic classification system was developed. Here, the solutes were less influenced by mixed-interaction effects. A comparison of the two solvent-selectivity triangles proposed using the data of Rohrschneider, and the Kamlet-Taft solvatochromic classification system revealed that for aliphatic solvents, the two triangles are, in general, similar with little difference in the relative assignment of different solvents according to their dipolarity, acidity, and basicity. However, the triangle proposed using the Kamlet-Taft system did show some important practical differences, especially in the case of reversed-phase liquid chromatography.

4.1.3 STATIONARY PHASE SELECTIVITY Selectivity can also be adjusted by changing the type of stationary phase. In general, this is less convenient than a change in mobile phase selectivity but it is an essential part of multidimensional liquid chromatographic systems. A change in stationary phase selectivity is also a powerful technique for improving resolution while possibly

Concepts and Practice of High-Performance Liquid Chromatography

200

5 6

160 Intensity (mV)

181

7 4

120 8 n=3

80

9 40

10 11

12

13

0 0

5

10

15

20

25

30

Retention time (min) (a)

120 120

5 Intensity (mV)

6 100 4 Intensity (mV)

n= 3 80

n=6

100

7

80 60 40 20

60

0 8

6

40

7

Retention time (min) 9

20

10

11

0 0

5

10

15

20

25

30

35

40

Retention time (min) (b)

FIGURE 4.1 Chromatogram of an oligomer separation of polystyrene with a molecular weight of 760 daltons. Column C18 Nucleosil 10 μm particle diameter (100 × 4.6 mm). (a) Mobile phase 100% methanol, flow rate 1.0 mL/min. (b) Mobile phase 100% acetonitrile, flow rate 1.0 mL/min.

182

Advances in Chromatography

decreasing the analysis time. An enormous array of stationary phases are available to the chromatographer but even if we limit discussion to reversed phase chromatography, a vast choice still remains, such as C18, C8, C4; a variety of polar endcapped reversed-phases and columns, such as the Synergi (Phenomenex) and the Xterra (Waters); and specialty columns like the Cosmosil range. Stationary phases that are highly specific towards diastereoisomer separations, such as the carbon clad zirconia (CCZ, ZirChrom Separations Inc.) and the HyperCarb (Merck) are also available in reversed-phase mode. Each of these phases can impart the possibility of different selectivity to separations governed by differences in retention mechanisms exhibited by solutes as they interact on the surface. As an example, two reversed-phase surfaces are worth contrasting (C18 and CCZ) that offer very different retention behavior, enabling them to be coupled in multidimensional systems that are RP-RP [9]. The C18 column needs no introduction in reversed-phase LC, as it is the most widely employed of all columns; however, the CCZ is relatively new and does not receive widespread application. The CCZ surface functions as an adsorption surface, commonly more retentive towards compounds with functional groups and is often more selective for the separation of isomers. The improved resolution of isomers results from very high sensitivity of the solid carbon surface to the geometric organization of the solute rather than from differences in column efficiency, as is usually observed for other reversed-phase methods [10, 11]. The separation of cis and trans-stilbene demonstrates the profound affect that solute structure and shape has on the diastereomeric selectivity of the carbon media. Each of the diastereomers has a unique shape and thus presents a different surface of electron density to the carbon phase [11] and this difference gives rise to the enormous selectivity factor observed. The separation of diastereomers of low molecular weight polystyrenes is another example where the selectivity between the C18 and CCZ surfaces is very different. Limited diastereomeric selectivity has been observed on conventional reversed-phase columns, such as the C18, where retention is largely governed by the molecular weight of the polymer. The chromatograms in Figure 4.2 illustrate this effect: In this figure, the separation of the diastereomers of the oligostyrene with four configurational repeating units and a sec-butyl end-group is illustrated. Eight diastereomers are present in this sample since the sec-butyl end-group has a site of stereochemistry. In Figure 4.2a, the separation of the eight diastereomers on the C18 column with an acetonitrile mobile phase all coelute in basically two bands. However, when on the CCZ column (also with an acetonitrile mobile phase, Figure 4.2b), the selectivity increases and seven of the eight isomers are resolved. The separation on the CCZ column is almost independent of molecular weight [9].

4.1.4 TEMPERATURE SELECTIVITY A change in the separation temperature can also change the separation selectivity [12, 13], particularly in ion-exchange and ion-pair chromatography. Little effect is often observed in liquid–liquid or liquid–solid chromatography; however, some interesting results have been reported on CCZ surfaces. When temperature is employed to alter the selectivity, decreasing the solvent strength is usually necessary

Concepts and Practice of High-Performance Liquid Chromatography

183

35 30

Intensity (mV)

25 20 Eight diastereomers 15 10 5 0 0

2

4

6 Retention time (min)

8

10

12

(a)

12

10

Intensity (mV)

8 Eight diastereomers 6

4

2

0 0

2

4

6 Retention time (min)

8

10

12

(b)

FIGURE 4.2 Illustration of the isomeric separation of sec-butyl oligostyrene (n = 4) on (a) a C18 column (Allsphere C18, 250 mm × 4.6 mm, Pd 5 μm) and (b) a carbon clad zirconia column (Self packed, 50 mm × 4.6 mm, Pd 3 μm). Flow rate 1.0 mL/min. Detection 262 nm. Injection volume 20 μL. Isocratic elution in 100% acetonitrile.

184

Advances in Chromatography

following an increase in temperature because an increase in temperature usually results in a decrease in retention.

4.1.5 MOBILE PHASE PH SELECTIVITY The separation selectivity of ionizable acids or bases can also be altered by a change in the pH of the mobile phase. When a change in the pH gives an optimum change in the separation selectivity but retention falls outside the desired range, the solvent strength can be adjusted essentially independently of selectivity by varying the proportions of components in the mobile phase, such as adjusting the salt concentration in the separation.

4.2 ONE-DIMENSIONAL SEPARATIONS One-dimensional liquid chromatography employs a single separation process on one chromatography column. That is, only one mechanism (ideally) is responsible for the retention of the solutes onto the stationary phase. For example, in size-exclusion chromatography, separation is governed by the size of the molecular species. Enthalpic interactions between the solute and the stationary phase surface are minimized [14]. If the mobile phase conditions are such that enthalpic interactions between the solute and the stationary phase are significant, the size exclusion process becomes more complicated than if only a single mechanism were responsible for the separation. Multiple mechanisms often lead to a reduction in the separation potential. Despite the fact that modern chromatographic columns are far superior to those of a decade ago, the actual separation power has only marginally increased because, as chromatographic columns have improved, overlapping components have begun to emerge from bands previously considered as single component peaks. Many samples are far more complex than first believed and the peak capacity of a single dimensional system is simply inadequate to resolve the required number of components. The peak capacity of a separation can be defined as the maximum number of peaks that can fit side by side into the available separation space, where each peak is resolved from neighboring peaks to satisfy some analytical goal [15–17]. Peak capacity provides a measure of the maximum number of components that can be resolved on a chromatographic run [16]. A peak capacity of 40 suggests that 40 single component peaks can fit side-by-side into the allowed retention volume range. However, this ideal situation virtually always exceeds the practical reality of the separations using current technologies [16]. Components are rarely uniformly distributed within the separation process and instead fall randomly along the chromatogram [18]. Davis et al. [18] has shown that single chromatographic separation systems often fail to separate even the simplest of complex samples, largely as a result of the random distribution of peaks throughout the separation. Consequently, component overlap theory (SMO) was developed using Poisson statistics to account for the random band displacement [18–20]. With this theory, Davis et al. predicted that the maximum number of randomly distributed bands that can be separated is approximately 37% of the theoretical limit if the bands were uniformly spaced. Moreover,

Concepts and Practice of High-Performance Liquid Chromatography

185

they showed that the number of single component peaks on the chromatogram would not exceed about 18% of the peak capacity even under the most favorable conditions. Finally, they showed that a chromatogram must be approximately 95% vacant in order to provide a 90% probability that a given component of interest will appear as an isolated peak [18]. The principal conclusion from these theories is that chromatograms of multicomponent complex mixtures contain a large fraction of unresolved or multicomponent peaks, and that complete separation of such mixtures by a single column is inadequate for the resolution of all the components. The statistical limitations on separations for a single chromatographic separation have been interpreted by many as a strong incentive for carrying out a separation incorporating more than one type of chromatographic system. A striking example of the limitations of a single chromatographic separation is shown by the separation of low molecular weight oligostyrenes using reversed-phase chromatography. Typically, a C18 column with a methanol mobile phase will bring about separation of the oligomers [21, 22] (Figure 4.1a) but the diastereomers begin to resolve when the mobile phase is changed to acetonitrile (Figure 4.1b). The resolving power of the (one-dimensional) LC system in this instance is insufficient for the complete separation of all the tactic isomers [23] because the sample components’ retention depends upon two different sample attributes: molecular weight and stereochemistry. These limitations and the desire to resolve increasingly complex mixtures have provided incentive for the development of multidimensional chromatography.

4.3 MULTIDIMENSIONAL SEPARATIONS 4.3.1 ASSESSMENT

OF THE

SEPARATION POTENTIAL

In an effort to overcome the limitation of single-dimension chromatographic separations, a great deal of work has been done on techniques that employ multiple retention processes. Multidimensional separation techniques constitute a powerful class of methods in which different separation steps, based on different selective retention processes, are linked [24]. Linkage of separation steps is usually tailored such that each dimension is selective towards a particular sample attribute and, for the sake of simplicity, this discussion will be limited to two-dimensional (2D) separations. The term “two-dimensional chromatography” has its origin in flat-bed or thin-layer chromatography, where chromatographic development is defined by migration of sample components in one direction followed by migration in a second dimension, perpendicular to the first [25–27]. Coincidently, Guiochon and coworkers [28] in the early 1980s suggested a theoretical two-dimensional liquid chromatographic column that operated in a similar manner to a 2D-TLC plate. Such a system, however, would prove to be difficult to manufacture and operate correctly. A 2D liquid chromatographic system is thus one in which two separate LC columns are linked in successive stages. Two-dimensional separation systems encompass an enormous array of combinations of separation systems. Even within distinct techniques such as liquid

186

Advances in Chromatography

chromatography, the variety in selectivities and even combination order varies greatly. Some common examples to realize the potential of 2D separations include IEC-RPLC [29–31], SEC-RPLC [32], RPLC-NPLC [33, 34], LC-CE [35–37], and LC-GC [38–41]. The power of a 2D separation may be measured by the twodimensional peak capacity. Multidimensional techniques have an intrinsic ability for providing significantly larger peak capacities relative to single step separation systems that, in turn, reduce the probability of component overlap [24]. Theoretically, the peak capacity of a 2D separation is calculated by a multiplicative law in which the peak capacity of a 2D separation is the product of the individual peak capacities of the constituent one-dimensional separations (e.g., NT = N1 × N2) [27, 42–44]. Typically, in a coupled system, maximum resolution and peak capacity (separation space) occur when the retention mechanisms are different. The more divergent the retention mechanisms are in each dimension, the greater the separation potential [45]. Separative potential will decrease as retention correlation between separation dimensions increases until perfect correlation is obtained, whereby the two-dimensional system will have a separative power equivalent to a single-mode chromatographic step [45]. Figure 4.3a shows the theoretical peak capacities for two separate chromatographic separations (one- and two-dimensional) in which the peak capacity is equal to the number of boxes along any one axis. The maximum peak capacity of the twodimensional chromatographic separation (first dimension × second dimension) is represented by the total number of boxes and is equal to the product of the peak capacities of the individual separations. Figure 4.3b shows the component peaks distributed randomly over a two dimensional space that has enough space so as to minimize the overlap of individual peaks (represented as ellipses). Substantial peak overlap occurs along any one axis of the two-dimensional plot, shown in Figure 4.13b. In this case, the chromatogram that would be observed along a single chromatographic dimension becomes crowded with substantial band overlap. Thus, a reduction in peak crowding and hence overlap is realized when the chromatographic separation is expanded to a higher dimensional separation [27, 46]. 4.3.1.1 SMO Treatment of the Two-Dimensional Approach Davis has further extended the statistical work of the SMO theory and broadened its applicability to two-dimensional separations. According to SMO in one dimension, the number of single component peaks is calculated by Equation 4.2 [18]: S = nc αe–2α

(4.2)

The maximum number of single component peaks per peak capacity unit is: (s/nc)max = 1/2 e–1 = 0.1839

(4.3)

where s is the expected number of singlet spots or peaks, nc is the peak capacity and α is a saturation factor (equal to m/nc where m is the statistical approximation

Concepts and Practice of High-Performance Liquid Chromatography

2nd

187

dim

ens

ion

(a)

on

nsi

1st

e dim

(b)

FIGURE 4.3 (a) Two-dimensional peak capacity as determined by the multiplicative rule. (b) The reduction in band overlap in going to two dimensions.

to the number of detectable peaks). In two dimensions, the number of single component peaks is calculated by Equation 4.4 [47]: s = m e–4α

(4.4)

Therefore, the probability that a peak will exist as a single component peak is less than that in a one-dimensional separation (by a factor of e–2α). The number of singlecomponent peaks per unit peak capacity is calculated using Equation 4.5 for a twodimensional separation. s/nc = αe–4α

(4.5)

Davis and coworkers were able to show that by differentiating Equation 4.4 with respect to α and equating the result to zero, the maximum value of s/nc (0.092) occurs α = 1/4. Two-dimensional separations actually perform less efficiently per unit peak capacity than corresponding one-dimensional separations [15, 47]. According to Equation 4.4 and Equation 4.5, the maximum number of single component peaks is approximately 9% of the peak capacity, compared to a one-dimensional separation where the number of single component peaks is estimated as approximately 18%.

188

Advances in Chromatography

This discrepancy occurs in a two-dimensional system because overlap can occur in two dimensions [47] and may seem detrimental to the aim of the separation problem; however, the enormous peak capacity afforded by the coupling of two separate dimensions nullifies this effect. As an example, a one-dimensional system having a peak capacity of approximately 50 may be expected to resolve with an 18% probability and obtain nine single-component peaks that are randomly distributed throughout the chromatogram. By comparison, a two-dimensional system having a peak capacity of 50 and 10 in the first and second dimensions, respectively, could be expected to resolve 45 randomly distributed single component peaks with a 9% probability.

4.4 TYPES OF CHROMATOGRAPHIC SEPARATION DISPLACEMENTS Displacements in a two-dimensional separation can be categorized into two broad groups [48]: (1) simultaneous displacement, where the displacement of a sample component zone across the two dimensions of the 2D field occurs at the same time; and (2) sequential displacement, where movement in one dimension is followed by movement in the second dimension. In two-dimensional column liquid chromatography, only sequential displacements are possible since separation must occur in the first dimension or column followed by the analysis of discrete sections of the first dimension in the second dimension. Separations or displacements observed in serially coupled tandem columns or mixed-phase systems could occur in a manner analogous to simultaneous displacements since retention of components will, on average, be dependent on the composition of mixed stationary phases or length of each connected column. An important class of sequential displacements are discrete displacements. Discrete separations are those in which a small discrete sample is applied to a corner of the 2D separation plane, i.e., injection onto the first dimension. Separation occurs along each subsequent axis, producing discrete elliptical zones (Figure 4.4a) [48]. Discrete one-dimensional displacements underlying the two-dimensional displacements fall into two categories [48], selective (S) and nonselective (N). Selective displacements occur in a liquid chromatographic system when sample components have different retention factors or the selectivity factors (α) observed for the sample components are greater than 1 (Figure 4.4a and Figure 4.4b). Selective displacements are considered to be separative displacements. Nonselective displacements provide equal displacement for all the components of a sample (Figure 4.4c). These S and N displacements can be combined in a number of ways (Table 4.1), corresponding to the displacement along each axis of the 2D plane [48]. Two extreme cases of the S × S separation exist. An S × SI separation occurs where the separation mechanisms are totally different (I = independent). Figure 4.4a shows an S × SI separation in which complete separation of components is achieved. One benefit of two-dimensional systems is that although sample components coelute in any single dimension, the component zone can be well-resolved in two dimensions. When the retention mechanisms are identical or correlated (S × SC), most of

(a)

189

2nd dimension

Concepts and Practice of High-Performance Liquid Chromatography

(b)

2nd dimension

1st dimension

(c)

2nd dimension

1st dimension

1st dimension

FIGURE 4.4 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. Adapted from [48].

the 2D space becomes unavailable for separation and the separation will converge back to a one-dimensional separation, as shown in Figure 4.4b by the alignment of data along the main diagonal [48]. This result is expected since for every unit of retention achieved for a sample component in the first dimension, the same gain is observed in the second dimension. When the separation dimensions are partially

190

Advances in Chromatography

TABLE 4.1 Types of Discrete Displacement Combinations and Their Effect on Two-Dimensional Peak Capacity Displacement Pair

Discrete Separation Peak Capacity

S × SI S × SC S×N N×N

n 2 ~ n 12 n2 ~ n1 n2 ~ n1 n2 = n1

correlated, the 2D space available for the separation of the components is reduced due to the correlation between the two dimensions. The separated components begin to cluster closer to the main diagonal similar to but not as severe as the example illustrated in Figure 4.4b. In this instance, resolution and peak capacity between sample components would be expected to decrease. In some instances, discrete N × S can be considered analogous since no gain in the selectivity factor is observed in the first dimension as no components are separated in the first dimension. For discrete N × S combinations, the column of separated components in the second dimension appear at a uniform separation time in the first dimension (Figure 4.4c). The effectiveness of these processes is shown in Table 4.1. In many instances, both selective and non-selective displacements are observed in one liquid chromatographic dimension as is the case for the two-dimensional analysis of low molecular weight polystyrenes (Figure 4.5). In the separation of a group of 58 oligostyrene isomers with tert-butyl, sec-butyl, and n-butyl end-groups, a selective displacement is observed for groups of diastereoisomers with variations in molecular weight, and within these groups, selectivity differences are observed between oligostyrenes with different end-groups. For each of these groups, a nonselective displacement in the first dimension occurs for the diastereoisomers resolved in the second dimension as there are distinct columns of diastereoisomers apparent on the two-dimensional retention plot (labeled T2 to N5). Therefore, it is often necessary to combine selective and nonselective displacements in order to prevent a chaotic two-dimensional component separation. Continuous separations are not discussed since all combinations of the S and N types yield no increase in separation power compared to a one-dimensional separation and continuous separations are impossible with conventional two-dimensional column hardware. In Table 4.1, the only displacement pair providing any real advantage over a one-dimensional separation is the non-correlated S × SI pair. Therefore, in order to gain the maximum separation power from a two-dimensional system, each individual dimension must be of the selective (S) type. A second criterion for selection of multidimensional combinations should be based on minimal correlation between the two single-dimensional separation processes; that is, they should be as different as possible, thereby utilizing more of the potential space available in a twodimensional separation [48].

Concepts and Practice of High-Performance Liquid Chromatography

191

1.000

0.000 0.000 T2

N2

T3 S3 N3

T4 S4

N4

T5 S5

1.000 N5

FIGURE 4.5 Normalized two-dimensional retention plot in the separation of 58 n = 2 to n = 5 oligostyrenes with tert-butyl, sec-butyl, and n-butyl end-groups.

4.5 TWO-DIMENSIONAL CHROMATOGRAPHIC SYSTEMS Two-dimensional chromatography has been extensively studied over the years as a means of increasing the separation capability of a chromatographic system for complex samples. For a separation system to be considered as multidimensional, the following conditions need to be fulfilled: •

•

The components of a mixture are subject to two or more separation steps or mechanisms, in which the displacements are dependent on different factors [49, 50]; and When two or more components are substantially separated in any single step, they must always remain separated until the completion of the total separation [49, 50].

When two different columns are coupled in series, the separation will generally improve because two separation mechanisms operate independently. However, when columns are coupled serially, coelution of bands initially separated on the first column can occur after passage through the second column [15]. In many instances, serially coupled columns are equivalent to mixed-bed columns of the same length and solutes elute with retention factors equal to the average of the two columns [15, 51, 52]. To overcome this problem, heart-cutting techniques were introduced in which a small fraction of solutes eluting from the first column are transferred to the second column [53–66]. In this way, separation order and resolution are maintained [67]. In most instances, heart cutting can be performed by manual manipulation of the fractions collected from the first dimension followed by reinjection onto the

192

Advances in Chromatography

second dimension [68–70] but this carries with it the obvious disadvantage of manual labor and dilution of the sample. Therefore, from a practical aspect, the automated coupling of two systems serves as a more useful process. In a heart-cutting system, the peak capacity is equivalent to the sum of the peak capacities of the first dimension plus the peak capacity of each heart-cut fraction. In the optimal case, where all peaks are isolated as individual fractions on the first column, the peak capacity of the twodimensional separation is equivalent to the product of the peak capacities of the first and second dimensions. The ideal mode of a multidimensional separation for maximizing chromatographic information is the comprehensive approach, where all components in the eluent from the first column are consecutively transported to and separated in the second dimension [31, 44]. In a comprehensive (automated) two-dimensional chromatographic system, small discrete fractions of mobile phase eluted from any one chromatographic separation are transported at regular intervals to the next dimension throughout the entire chromatographic run [44, 71, 72]. The transfer interval is determined according to the width of the peaks eluting from the previous separation dimension [71, 72]. For this process to work efficiently, the second separation must be much faster than the first separation [71]. Examples of comprehensive two-dimensional systems are numerous and are more commonly seen in gas chromatography (GC × GC) because the second separation step can easily be manipulated to yield very rapid separations [73–75]. This fact, together with cryogenic trapping as developed by Marriot et al. [76], allows the band migration to be slowed and focused prior to placement on the second column [77], allowing more time for the second separation to occur. These factors have allowed GC × GC to develop much more rapidly than LC × LC. A detailed discussion on comprehensive and heart-cutting multidimensional liquid chromatographies follows.

4.6 SAMPLE DIMENSIONALITY For a multidimensional system to be useful for a particular separation, the sample itself must exhibit a degree of sample dimensionality (s). Sample dimensionality can be described as the number of features of the sample that can be utilized for separation purposes. To separate an n-dimensional sample, an n-dimensional chromatographic separation system should be employed [24]. For example, consider the separation of low molecular weight polystyrenes where the sample dimensionality is three. Stereo-irregular polymerizations produce polystyrenes that are distributed in three dimensions. The first dimension can be described in terms of variations in molecular weight. These different molecular weight polystyrenes have a number of tactic isomers as a consequence of the stereochemistry of the configurational repeating units. The second dimension can therefore be described in terms of tacticity. Each of the different tactic isomers also has an enantiomer and thus the sample can be described according to three dimensions. However, unless a chiral column is used to express the enantiomeric aspect of the sample, this third dimension is essentially unnoticed in the chromatographic system. Therefore, the sample is essentially two-dimensional and can be effectively

Concepts and Practice of High-Performance Liquid Chromatography

193

described by the structures illustrated in Figure 4.6. The separation system that should be employed for this separation problem would have one dimension that gives a separation based on molecular weight (e.g., size-exclusion chromatography or reversed-phase chromatography), while the second should separate the oligomer classes according to their tacticity. The oligomeric separations illustrated in Figure 4.1a and Figure 4.1b illustrate the two-dimensional chromatographic behavior of the low molecular weight polystyrenes. In Figure 4.1a, only one sample dimension is expressed in the chromatographic separation and all diastereomers of their respective molecular weights elute as one oligomeric band. Diastereomer selectivity becomes apparent when the mobile phase is changed to acetonitrile (Figure 4.1b). Thus, two sample attributes are being expressed in the separation, albeit with the peak capacity of the single-dimension separation being insufficient to realize the resolution of all isomeric forms. Further, if separation of the components of the two-dimensional polystyrene sample were to be achieved in a one-dimensional system, the resulting band displacements could be disordered, exhibiting a mixed dependency on all sample properties. Generally, if the dimensionality of the sample exceeds that of the system, components of the sample will not resolve systematically. The resulting separation pattern is disordered and could be termed chaotic [24] — clearly, a two-dimensional separation would be more appropriate. If the retention behavior of sample components were dominated by only one sample attribute with all available chromatographic modes, a 2D chromatographic separation would prove useless since the sample can only effectively be specified according to one dimension. In this case, Giddings [24] suggests that a one-dimensional chromatogram will exhibit apparent order despite the fact that sample dimensionality is greater than the system dimensionality. If a two-dimensional separation system is used, the location of sample components in the separation space will be restricted and will cluster towards the main diagonal. This behavior is shown in Figure 4.7.

4.7 ORTHOGONALITY IN TWO-DIMENSIONAL LIQUID CHROMATOGRAPHY Divergence in separation selectivity between combinations of liquid chromatographic systems is known to be a paramount requisite when coupling two liquid chromatographic modes together with the purpose of conducting a two-dimensional separation. Although orthogonality is technically a binary property, in the chromatographic community orthogonality is used as a measure of the difference in selectivity or separation ability of two or more chromatographic steps [78, 79]. In order to resolve the maximum number of sample components, two orthogonal liquid chromatographic systems should be chosen, that is, the peak capacity of the system is maximized [80]. Orthogonality in two-dimensional liquid chromatography can be achieved through selection of systems that maximize the selectivity differences between each system by choosing the most appropriate combinations based on the type of stationary phases [81], different liquid chromatographic modes [31,

C

C

H

H

H

H

C

CH2 CH2CH2CH2CH3

CH2CH2CH2CH3

H

CH2

C

C

H

CH2

C

H

C

CH2

CH2

H

C

H

C

CH2

CH2

C

C

CH2

CH2

CH2

H

H

C H

CH2CH2CH2CH3

H

H

H

H

n=5

CH2CH2CH2CH3

CH2

C

H

C

CH2

CH2

CH2

CH2 H

C

H

CH2 C

C

C H

CH2

C

H

C

CH2

CH2

C

CH2

H

C H

H

FIGURE 4.6 Diastereoisomers of n-butyl polystyrene oligomers with n = 2 to n = 5 styrene repeating units.

CH2CH2CH2CH3

CH2CH2CH2CH3

CH2CH2CH2CH3

CH2CH2CH2CH3

H

CH2

H

CH2

C

H

C CH2

C CH2

C CH2

H

CH2

H

C

CH2

H

H

H CH2

C CH2

CH2

H

C

C

CH2

C

H

H

H

CH2CH2CH2CH3

C CH2

H

H

C CH2

CH2CH2CH2CH3

H

H

CH2

C

CH2

CH2CH2CH2CH3

C

C

H

H

CH2

H

C

H

H

H

CH2

CH2

CH2

CH2

H

H

C

C

H

CH2

C

CH2

CH2

H

C

C

C H

H

n=4

H

H

n=3

CH2

C

H

n=2

H

H

H

H

H

H

H

H

H

H

CH2CH2CH2CH3

CH2

C

CH2

C

CH2

C

CH2

C

CH2

C

H

CH2CH2CH2CH3

CH2

C

CH2

C

CH 2

C

CH2

C

CH2

C

H

H

H

H

H

H

H

H

H

H

H

CH2CH2CH2CH3

CH2

C

CH2

C

CH2

C

CH2

C

CH2

C

H

CH2CH2CH2CH3

CH2

C

CH2

C

CH2

C

CH2

C

CH2

C

H

194 Advances in Chromatography

195

2nd dimension

Concepts and Practice of High-Performance Liquid Chromatography

1st dimension

FIGURE 4.7 Illustration of the separation dimensionality when one sample dimension is weakly expressed. The sample components fall into a narrow band on the 2D chromatographic plot. Reprinted with permission from [24]. Copyright (1995) with permission from Elsevier.

34], mobile phase [82], mobile phase additives, and temperature [12], as discussed in a previous section. Since every sample contains different attributes, the orthogonality of a particular combination of liquid chromatographic systems changes with the sample base. Hence, a particular combination of chromatographic systems can be orthogonal for one sample but highly correlated for a different sample base. Nevertheless, there are some system combinations that are invariably orthogonal, such as size-exclusion/reversed-phase [36, 83], ion exchange/reversed-phase [84], normal-phase/reversed-phase [33], and size-exclusion and liquid chromatography at the critical composition [85]. These combinations are usually orthogonal because the retention mechanisms in each dimension are usually uncorrelated. In order to maximize the two-dimensional separation space or the two-dimensional peak capacity, and consequently optimize the separation of a complex sample, the orthogonality should be assessed. This assessment is most commonly achieved by the evaluation of the selectivity of a large number of chromatographic systems (stationary and mobile phase combinations). The effort may seem like a significant degree of pre-separation work-up but, in the long run, the separation performance gained through some basic system assessment ultimately leads to a higher performance separation. Determination of the orthogonality can be done in a number of ways, as will be discussed below.

4.7.1 DETERMINATION OF ORTHOGONALITY TWO-DIMENSIONAL SEPARATIONS

FOR

Assessment of the orthogonality of two separation systems can be a relatively simple process. In its most basic form, selectivity differences can be determined by assessing changes in elution order for each separation mode [4]. Rocklin et al. [86] investigated the orthogonality of micellar electrokinetic chromatography (MEKC) and capillary electrophoresis (CE). The orthogonality of these techniques was investigated using a five-component mixture of fluorescently labeled compounds. Differences in migration order were evident for the two methods, indicating differences in retention mechanisms [86]. Steuer et al. [2] compared RP-HPLC, supercritical fluid chromatography (SFC) and capillary zone electrophoresis (CZE) for the analysis of drugs. Retention data

196

Advances in Chromatography

1.0

Xa dimension 2

0.8

0.6

0.4

0.2

0.0 0.0

0.2

0.4

0.6 Xa dimension 1

0.8

1.0

FIGURE 4.8 Hypothetical two-dimensional normalized retention plot.

was normalized according to Equation 4.6 [2]. A typical example of the normalized retention plots is illustrated in Figure 4.8. χi =

t1 – t 0 Δt

(4.6)

where t0 is the retention time of the void marker, ti is the retention time of the component of interest and t is the entire retention time range. By comparing normalized two-dimensional retention plots, Steuer found that RP-HPLC and CZE were usually orthogonal, while RP-HPLC and SFC showed some slight correlation. The orthogonality results from the study were later used to determine which separation method should be used in the separation of particular classes of drugs. Valkó and coworkers [87] utilized two-dimensional plots of gradient retention times for a group of 60 solutes to assess the selectivity differences between a Luna C18, Xterra™ C18 and a perfluorinated C6 stationary phase. The least correlation was found between a Luna C18 column with methanol mobile phase and the perfluorinated C6 stationary phase with a trifluoroethanol mobile phase operated in gradient elution mode. Fields et al. [88] applied a similar approach using twodimensional plots of retention factors of testosterone-related compounds relative to testosterone, allowing for the determination of differences in the selectivity of polybutadiene-coated zirconia (PBD-Zr), C18 and PBD-Zr phases using high-temperature water as mobile phases. Neue et al. [89] followed this procedure by comparing relative retention times of various analytes, allowing the characterization of

Concepts and Practice of High-Performance Liquid Chromatography

197

certain phases based on certain chemical descriptors. Neue [89] was able to distinguish between classical C18 packings and polar end-capped phases.

4.7.2 STATISTICAL APPROACHES 4.7.2.1 Correlation Coefficients and Color Maps Van Gysegham et al. [78, 90, 91] recently published a series of works regarding the determination of orthogonal chromatographic systems for the characterization of pharmaceutical drug mixtures. Although the aim was not to couple two chromatographic systems together, the methods they utilized would be valuable in the determination of orthogonal chromatographic systems for multidimensional chromatography. Initially, the orthogonality of 11 chromatographic systems was evaluated using Pearson correlation coefficients of retention factors for all combinations of the chromatographic systems. In this instance, values close to |1| (Figure 4.9a) indicate total correlation while values close to zero indicate total divergence in retention behavior (Figure 4.9b) [78]. Selectivity differences between the 11 chromatographic systems were visualized using two-dimensional plots of the retention factors of the pharmaceutical compounds. Although the Pearson correlation coefficients generally provided a reasonable assessment of the correlation exhibited by two liquid chromatographic systems, in certain cases this measure of correlation failed. Figure 4.9c illustrates the example of a theoretical system in which the majority of compounds are aligned along the main diagonal of the two-dimensional retention plot, indicating a high correlation for these compounds on these chromatographic systems. The correlation coefficient is moderately low due to a reasonably small number of divergent outliers. Van Gysegham et al. [78] concluded that a visual inspection of two-dimensional retention plots was also necessary to determine retention correlation between chromatographic systems. They then applied score plots using principle-component analysis (discussed shortly) and OPTICS color maps in order to assess orthogonality. Color maps based upon hierarchical weighted-average-linked dendrograms could be constructed, as shown in Figure 4.10. In this instance, a classification or differences between chromatographic systems can be achieved by interpreting the differences in the color patterns [78, 92]. However, one limitation was the tendency for the eye to be attracted towards color regions within the map, hence biasing the result. Van Gysegham et al. [78] indicates that OPTICS uses “as much dimensions as needed to explain 99.0 to 99.5% of the variance.” Assessment of the orthogonality of the chromatographic systems can also be made from the cluster analysis dendrogram [89] since the differences between chromatographic systems and groups are based on the linkage pattern. Differences between the chromatographic systems can be visualized based on the length or height of the branching between chromatographic systems or groups of chromatographic systems [89]. The higher two chromatographic systems or groups are linked, the higher their orthogonality. Van Gysegham et al. [90] concluded that the stationary phase was the most significant factor affecting the orthogonality of chromatographic combinations.

198

Advances in Chromatography

30 25

r = 0.951

(a)

k on CS5

20 15 10 5 0

0

5

10

20

15 k on CS4

25

30

35 r = 0.240

30

(b)

k on CS3

25 20 15 10 5 0

0

2

4

6

8

10

12

14

k on CS1 30 25

(c)

k on CSy

20 r = 0.385

15 10 5 0

0

5

10

20 15 k on CSx

25

30

FIGURE 4.9 Retention factors on two systems: (a) CS4-CS5, (b) CS1-CS3 and (c) hypothetical systems, CSx–Csy. r = correlation coefficient. Reprinted with permission from [78]. Copyright (2003) with permission from Elsevier.

0.8

199

WPGMA clustering

0.6 0.4 0.2 0

5 9

7 5

4 11 2 10 6 8 1 3

10

15

20

25

30

WPGMA clustering

Objects

Correlation coefficient 1-|t| (dissimilarities)

Concepts and Practice of High-Performance Liquid Chromatography

Variables

0 0.2 0.4 0.6 0.8 1 Correlation coefficient 1-|t| (dissimilarities)

FIGURE 4.10 Color map based on dendrograms resulting from the hierarchical average linkage technique on retention factors k, for 11 chromatographic systems (abscissa) and 68 substances (ordinate). Reprinted with permission from [78]. Copyright (2003) with permission from Elsevier.

4.7.2.2 Principle Component Analysis Principal component analysis (PCA) has been used in chromatography for the comparison of separation methods, characterization of substances based on chromatographic data, and prediction of retention behavior [93, 94]. The most important application of PCA in the context of the present discussion is in the comparison of separation techniques and comparisons within separation methods. Principal component analysis has been described as a mathematical manipulation of a data matrix where the goal is to represent the variation present in many variables using a small number of factors [95]. The mathematical operations in PCA can be performed using eigenvalue-eigenvector extraction algorithms. Using PCA, each row or column in a data matrix is considered to be a point in a multidimensional space with coordinates defined by the values corresponding to the appropriate number of columns or rows in the data matrix [93]. A major strength of PCA is that it extracts from the data axis eigenvectors that best span the data matrix. The first eigenvector or principal component explains the maximum amount of variation possible in the data set in one direction [93, 95]. The second eigenvector is chosen so as to be orthogonal to the first. The second eigenvector or principal component then describes as much of the remaining variation in data as possible. Subsequent vectors and the corresponding projections of the data can be described in a like manner until all the variation can be explained in terms of the extracted eigenvectors and the coordinates along the vectors [95].

200

Advances in Chromatography

Assessment of chromatographic data and hence system orthogonality requires the construction of a data matrix where chromatographic systems can be located in the columns and autoscaled retention times of analytes in the rows (or vice versa). Depending on the construction of the data matrix, two main types of plots can be obtained, score plots and loading plots [94]. Van Gysegham et al. [78] and Detroyer et al. [94] illustrated the utility of PCA in the comparison of chromatographic systems using score plots and loading plots. Figure 4.11 illustrates a score plot in the comparison of 11 chromatographic systems for the determination of orthogonality. Generally, the greater the distance between chromatographic systems on the score plot (numbered 1 to 11), the higher the orthogonality between chromatographic systems. Detroyer et al. [94] reported the results of a chemometric study on retention times of 83 pharmaceuticals from five different chemical families. They conducted their study on eight different HPLC systems with the application of PCA, cluster analysis and sequential projection pursuit. Primarily, the study aimed to classify these pharmaceutical substances based on the chromatographic data obtained from each separation system. From the score plots obtained and using the first four principle components, this group was able to compare the similarity in the retention of components between the chromatographic systems. Explanations regarding the principal components and chromatographic systems were then proposed. For instance, one principle component was explained in terms of variations in molecular weight while another was attributed to the acid-base behavior of the chromatographic systems [94]. Lurie et al. [96] assessed several separation methods for the qualitative and quantitative determination of anabolic steroids in forensic samples, a qualitative determination employing PCA. In a comparison between MECC, HPLC, and GC, they produced a matrix with the chromatographic methods as the rows and the retention indices of the anabolic steroids relative to testosterone as the columns. PCA analysis was then applied and all three methodologies were found to be orthogonal [96]. 4.7.2.3 Information Theory Information theory (IT) is a chemometric technique that allows the mathematical evaluation of qualitative methods by calculation of the expected or average amount of information obtained from an analysis [97]. The information obtained (i.e., retention times) from a chromatographic analysis is equal to the amount in “bits” of information gained from the reduction in uncertainty after a chromatographic analysis [97, 98]. Information theory has been utilized in chromatographic science for a number of years for the selection of chromatographic columns that yield optimal amounts of information and the identification of components based on retention times [98–100]. Slonecker and coworkers [46] used information theory to calculate the informational orthogonality of a large selection of combinations of separation systems. Information theory is ideal in this instance since the amount of information in “bits” gained from a chromatographic separation combination is sensitive to the degree of correlation between the two chromatographic dimensions [98]. The higher

Concepts and Practice of High-Performance Liquid Chromatography

201

8 +9

6

+7

PC2 (19.12%)

4 2

+18 +6 +10

0

+1

−2 −4

+4

A

+8

+5 +2

−6 −15

−10

−5

5

0

10

PC1 (48.91%) (a) 5 +4

4 3

+5

PC3 (11.66%)

2

+2

+7 +9

1

+11

0

+8

−1

+6

−2

+10

−3 −4 −5 −15

+1 +3 −10

−5

0

5

10

PC1 (48.91%) (b)

FIGURE 4.11 Score plots of the autoscaled k data for 11 chromatographic systems and 68 substances (11 × 68 matrix) on (a) PC1-PC2 and (b) PC1-PC3. Reprinted with permission from [78]. Copyright (2003) with permission from Elsevier.

the correlation between two separation systems, the lower the amount of information obtained [46]. Using information theory for a single chromatographic step the reduction in the uncertainty or entropy (I) of a component having a particular retention time is equal to [101]:

202

Advances in Chromatography

n

I=

∑ – p ( x ) log

2

p ( x)

(4.7)

d =1

Using the method of Slonecker et al. [46], the informational entropy (I) is calculated for each chromatographic step in a two-dimensional separation individually and then for two dimensions. Since retention correlation invariably exists in the majority of coupled chromatographic systems, the two-dimensional information gained is reduced due to mutual information [46]. Then, by calculating the fractional information content (h), the informational similarity (H) can be calculated according to Equation 4.8 and Equation 4.9: h (1, 2 ) = 1 –

I (1; 2 ) I (1, 2 )

(4.8)

1/ 2

H (1, 2 ) = ⎡⎣1 – h 2 (1, 2 ) ⎤⎦

(4.9)

where I(1;2) is the mutual information between the two chromatographic dimensions (1 and 2) and I(1,2) is the total two-dimensional informational entropy. The informational similarity, as described by Slonecker et al. [46], is used as a measure of solute crowding on a normalized [2] two-dimensional retention plot, as shown in Figure 4.8. The retention times are scaled according to the method of Steuer et al. [2] to allow comparison of a diverse range of chromatographic systems. Values of informational similarity range between 0 and 1. A value of 1 represents the highest level of solute crowding and occurs in conjunction with total correlation. Also determined using the method of Slonecker and coworkers is the percentage synentropy, a measure of the retention mechanism equivalency between two chromatographic systems. This value is calculated according to Equation 4.10: ⎛ Ia ⎞ %synentropy = ⎜ ⎟ × 100 ⎝ I (1, 2 ) ⎠

(4.10)

where Ia is the informational entropy of the data aligned (within some arbitrary boundary) along the unit diagonal of a normalized two-dimensional retention plot. Values of the percentage synentropy range between 0 and 100 with a value of 100 indicating that the chromatographic systems in a two-dimensional combination are 100% equivalent. 4.7.2.4 Geometric Approach to Factor Analysis Factor analysis (FA) is a common mathematical tool that can be used to examine a large range of data sets. Liu et al. [80] used factor analysis for the evaluation of the orthogonality and estimation of the two-dimensional peak capacity. Using factor

Concepts and Practice of High-Performance Liquid Chromatography

203

analysis in the evaluation of two-dimensional orthogonality, correlation matrices can be constructed using the scaled retention times of analytes obtained in each chromatographic dimension [80, 102]. The correlation matrix (C) is calculated according to Equation 4.11: ⎛ 1 ⎞ C=⎜ M ′T M ′ ⎝ N – 1 ⎟⎠

(4.11)

where N is the number of scaled retention times, M′ is a matrix of scaled retention times and M′T is the transposed matrix of the matrix of scaled retention times. This computation yields a square correlation matrix in the form of:

C=

1 C21

C12 1

(4.12)

where C12 = C21 and is a measure of the correlation between two sets of retention time data and, ultimately, the orthogonality of a two-dimensional system. Total correlation between chromatographic systems is observed when C21 = 1; totally orthogonal chromatographic systems are observed when C21 = 0. Theoretically, the peak capacity of two orthogonal chromatographic systems is equal to the product of the peak capacities of the constituent dimensions [103]. Correlation between separation dimensions reduces this available separation space to a fraction of the orthogonal case [103]. Using the retention correlation coefficients, Liu et al. [80] considered the correlation coefficients as the cosine of the angle between two vectors. In this instance, the inverse cosine of the correlation matrix (C) would yield the spreading angle (β). Values for the spreading angle range between 0 and 90˚ and a spreading angle of 90˚ occurs when two separation systems yield totally orthogonal retention data. The peak spreading angle β is a relative measure of the theoretical retention space used. A totally orthogonal separation results in a spreading angle of 90˚ while a totally correlated case, in essence a onedimensional separation, yields a peak spreading angle of 0˚. Most two-dimensional separations lie between perfect correlation and perfect orthogonality. Calculation of the peak spreading angle and construction of a geometric plot (Figure 4.12) would show that part of the area in the orthogonal retention space becomes unavailable due to correlation. The gridded area in Figure 4.12 is the separation space of a partially correlated two-dimensional chromatographic system, a fraction of the total area bound by the rectangular surface [80]. Construction of the geometric plot shown in Figure 4.12 is achieved by calculating the angles α, α′ and γ and then the areas designated A and C. The practical two-dimensional peak capacity is then calculated from the theoretical two-dimensional peak capacity using Equation 4.13: NP = NT (A + C)

(4.13)

204

Advances in Chromatography

Effective area Np

Separation step 2

A β γ

αˆ

α

C

Separation step 1

FIGURE 4.12 Effective nonorthogonal, 2D retention space when the peak spreading angle is β. Adapted from [80]. Reprinted with permission from [80]. Copyright (1995) with permission from the American Chemical Society.

where NP is the practical two-dimensional peak capacity and NT is the theoretical two-dimensional peak capacity when the correlation between separation systems is not considered. 4.7.2.5 Combination Methods Gray and coworkers [103, 104] used IT and FA to evaluate the orthogonality and the separation quality of several reversed-phase systems for the separation of oligostyrenes and diastereomers. Using the approach developed by Slonecker and coworkers [46] towards information theory and Liu’s [80] approach to factor analysis, a more complete theoretical assessment could be conducted. The informational similarity (solute crowding) and retention mechanism equivalency could be calculated using IT while the orthogonality (correlation coefficients) and the practical two-dimensional peak capacity (NP) could be determined using the geometric approach to factor analysis. Visual evaluation of the normalized retention plots and geometric plots of the two-dimensional separation space was an enormous benefit in the evaluation of the orthogonality of chromatographic combinations. Successful coupling of two liquid chromatographic systems could be made that contained low to moderate solute crowding, low correlation coefficients, and sufficient practical two-dimensional peak capacity for the separation of a group of oligostyrene isomers when using two-dimensional reversed-phase liquid chromatography [55, 81]. The theoretical performance of two such multidimensional systems for the separation of the diastereomers of oligostyrenes is illustrated below. The sample base contains a mixture of 32 oligostyrenes consisting of chains containing five configurational repeating units and end-groups of either sec-, tert-, or n-butyl. Two multidimensional systems that could be used for the separation of these compounds incorporate in the first dimension a C18 column and in the second a CCZ column. Methanol or acetonitrile could be used in either dimension, but, to limit this

Concepts and Practice of High-Performance Liquid Chromatography

205

discussion, we consider only methanol and acetonitrile in the first dimension with acetonitrile in both systems in the second dimension. Separation in the first dimension that incorporates the C18 column with a methanol mobile phase is highly dependent on the end group. When acetonitrile is employed as the mobile phase, the second sample attribute (stereochemistry) is expressed. Separation in the second dimension (CCZ column) is almost exclusively dependent on the stereochemistry of the oligostyrenes. The chromatogram shown in Figure 4.13 illustrates the separation of the 32-component mixture on the C18 column in (a) the methanol mobile phase and (b) the acetonitrile mobile phase. Figure 4.13c illustrates the separation of the sample mix on the CCZ column with the acetonitrile mobile phase. Clearly, in all three separations, very little resolution of the individual diastereomers is evident. However, if each of the diastereomers are injected individually into each of the three separate chromatographic phases (C18/methanol, C18/acetonitrile and CCZ/acetonitrile) and their respective retention times normalized according to Equation 4.6, the normalized retention plots shown in Figure 4.14 illustrate that there is substantial potential for the separation of these isomers if a two-dimensional system were to be employed. However, assessment of the retention correlation and the degree of orthogonality using IT and FA reveals that a two dimensional system consisting of a C18/methanol first dimension and a CCZ/acetonitrile second dimension would outperform a system that employed C18/acetonitrile in the first dimension. The important attributes that reveal this information are given in Table 4.2. Here, the information similarity of the C18/methanol-CCZ/acetonitrile system is almost half that of the C18/acetonitrile-CCZ/acetonitrile system, while the peakspreading angle (75˚ compared to 42˚) and the useable separation space (90% compared to 56%) is much higher. Further, the system correlation for the C18/methanol-CCZ/acetonitrile is only 0.26, compared with 0.76 for the C18/acetonitrileCCZ/acetonitrile system. Despite this correlation, the maximum number of components that could be expected to be resolved is lower on the C18/methanol-CCZ/acetonitrile, but, because the retention process in the first dimensio, (consisting of the C18/methanol system) expresses only one sample attribute, order in this potential two-dimensional separation is higher. Overall, the success of coupling these two systems would be better than the coupling of the C18/acetonitrile-CCZ/acetonitrile systems. This result will be illustrated in more detail in the following section that details comprehensive 2D separations. Note that the normalized retention plot shown in Figure 4.14a illustrates very nicely the separation order. The more chaotic band displacement in the C18/acetonitrile-CCZ/acetonitrile couple clearly indicates that this system would be more problematic in the transportation of components to the second dimension, especially if one aim of the exercise is isolation in a physical sense with maximized component recovery. As the above example illustrates, multiple approaches give a clear picture as to the orthogonality and theoretical separation quality of combinations of chromatographic systems. The use of multiple visualization methods becomes increasingly important when large numbers of combinations of chromatographic systems are evaluated.

206

Advances in Chromatography

60

Intensity (mV)

50 40 30 20 10 0 8

10

12 Retention time (min)

14

50

Intensity (mV)

40 30 20 10 0 −10 12

13

14

15 16 17 18 19 Retention time (min)

20

21

22

5

10

15 20 25 30 35 Retention time (min)

40

45

50

18 16 14 12

Intensity (mV)

10 8 6 4 2 0 −2 0

FIGURE 4.13 Chromatograms of the 32 component oligostyrene sample: (a) C18 column with 100% methanol mobile phase at 1.0 mL/min. Injection volume 10 μL. (b) C18 column with 100% acetonitrile mobile phase at 1.0 mL/min. Injection volume 10 μL. (c) CCZ column with 100% acetonitrile mobile phase at 2.0 mL/min. Thermostated at 40˚C. Injection volume 10 μL.

Concepts and Practice of High-Performance Liquid Chromatography

207

1.0 0.9

(a)

Xa factor CCZ (acetonitrile)

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0

0.1

0.2

0.3

0.4 0.5 0.6 0.7 Xa factor C18 (methanol)

0.8

0.9

0.0

0.1

0.2

0.3

0.4 0.5 0.6 Xa C18 (acetonitrile)

0.8

0.9

1.0 0.9

(b)

XaCCZ (acetonitrile)

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.7

FIGURE 4.14 Normalized two-dimensional retention plots in the separation of 32 oligostyrene isomers (n = 5) with tert-butyl, sec-butyl, and n-butyl end-groups on (a) a C18 (methanol)/CCZ (acetonitrile) system and (b) a C18 (acetonitrile)/CCZ (acetonitrile) system.

208

Advances in Chromatography

TABLE 4.2 System Attributes for Hypothetical Two-Dimensional Systems Two-Dimensional Systems Attribute

C18-ACN/CCZ-ACN

C18-MeOH/CCZ-ACN

Informational similarity Percentage synentropy Peak spreading angle (β) degrees Theoretical peak capacity (Nt) Practical peak capacity (Np) Correlation (c) Usage Resolved components (/32)

0.92 3.5 41.5 280 156.0 0.75 55.7% 26

0.56 3.3 74.7 60 53.8 0.26 89.7% 26

4.8 APPLICATION OF MULTIDIMENSIONAL LIQUID CHROMATOGRAPHY Comprehensive two-dimensional liquid chromatography implies complete analysis in that the entire first dimension eluent is transported to the second dimension [105]. The application of this separation mode may be undertaken by cutting discrete fractions of the first dimension to the second dimension [84]. Figure 4.15 illustrates this process for the transportation of 24 separate fractions from an ion-exchange first dimension to a reversed-phase second dimension. The size of the cut fractions depends on the speed of the second dimension separations [105], the loading capacity of the second dimension [34], and the sampling rate required to obtain a specific two-dimensional resolution [33]. A limitation of comprehensive analysis lies in the speed of the second separation step. Unlike the gas chromatographic mode, flow velocity in the liquid mode is limited by pressure. Therefore, in order to undertake a comprehensive separation, any analyte transported to the second dimension must elute from the second dimension column prior to components from the next cut fraction; otherwise, a “wraparound” effect (components from the previous cut eluting in the separation space of the following cut) will be observed and a chaotic band displacement will result. Murphy and coworkers [71] claim that in order to achieve maximal two-dimensional resolution, the effective sampling rate of the first dimension requires five samples across a given peak in the first dimension. However, sufficient two-dimensional resolution can be maintained if the sample rate in the first dimension is greater than three samples per peak [71]. They found that an increase in the sampling time of the first dimension results in a reduction in the resolution of the two-dimensional contour plots (Figure 4.16). However, in order to achieve such high sampling rates, the speed of the second dimension separations must be very fast [72], otherwise the wrap-around effect will create chaos [106]. This attribute limits the peak capacity in the second dimension and requires separation be sacrificed for speed. The correlation between both separation dimensions must often increase in order to gain some

Concepts and Practice of High-Performance Liquid Chromatography

209

Number of fractions injected into the 2nd dimension: 0.085

1

2

3 4

5

A

B

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Intensity (AU)

0.065

0.045

0.025

0.005 8

16

C 24

D 32

E 40

F

48 56 Time (min)

64

72

80

88

96

FIGURE 4.15 The chromatogram illustrates the separation of human hemofiltrate on the analytical cation-exchange column in the first dimension after first being subjected to selective enrichment on a cationic RAM. Fractions (24 in total) were continually transferred to the second dimension in 4-minute intervals for subsequent analysis by reversed-phase chromatography. Reprinted with permission from [84]. Copyright (2004) with permission from Elsevier.

degree of predictability in the elution time from the second dimension. Nevertheless, when the wrap-around effect is avoided, the technique has been shown useful in mapping and fingerprinting experiments. In some instances, such as when the wrap-around effect is unavoidable because the retention of some components in the second dimension is long and unpredictable, it may not be feasible to undertake analysis in a manner whereby all the mobile phase is transferred to the second dimension. Under these circumstances, only mobile phase that contains analyte is transported to the second dimension. The first dimension in this instance is usually developed to achieve some sort of group-based separation [55, 107–109] and sample components with a particular attribute are then collectively transported to the second dimension. Although, in certain instances, twodimensional resolution can be sacrificed, the peak capacity of the second dimension can be large because separation time in the second dimension can be extended as there are less cuts being transported to the second dimension. Transportation of entire peaks rather than fractions of peaks to the second dimension are also useful if the aim of the experiment is to separate, isolate, and collect sample components rather than for mapping purposes. Multiple peak sampling in this instance would only increase labor costs and dilute sample components if the aim is to collect separated fractions. Furthermore, since the sample is not diluted as a consequence of the multiple cutting across the band, limitations in sensitivity are less important and quantitation does not require the reconstruction of multiple sections of a particular second-dimension band.

210

Advances in Chromatography

8

1.00 min

1.50 min

0.2 0.4 0.6 0.8

0.2 0.4 0.6 0.8 GPC point

0.2 0.4 0.6 0.8

2.00 min

3.00 min

4.00 min

1.2 1.4 1.6 1.8

1.2 1.4 1.6 1.8 GPC point

1.2 1.4 1.6 1.8

HPLC (min)

0.67 min

18

HPLC (min)

15

25

2.00 2D resolution

Brij 58/78 1.60 PEG 8k/1k 1.20 0.80 0.40

0

1

2 3 Sampling time (min)

4

5

2D resolution

2.00 1.60 Brij 58/78 1.20

PEG 8k/1k

0.80 0.40

0

1

2

3 4 5 6 2D samples per peak

7

8

9

FIGURE 4.16 The effect of sampling time on two-dimensional resolution. The sampling time is noted on each chromatogram and 0.67 minute of each GPC axis is shown for comparison purposes. Amplitude changes due to peak overlap cause a corresponding change in color between chromatograms. Reprinted with permission from [71]. Copyright (1998) with permission from the American Chemical Society.

Whether such a process can be adopted largely depends on the sample. For example, when standards and samples form discrete bands, the later technique can serve as an adequate means of comprehensive analysis. However, if the sample elutes essentially as a continuum but a standard contains discrete markers, it may very well be inappropriate not to sample the entire mobile phase.

Concepts and Practice of High-Performance Liquid Chromatography

211

Regardless of whether whole-band sections or multiple-band sections are transported to the second dimension, a comprehensive two-dimensional separation necessitates that the separation times in the second dimension must equal the sampling times from the first dimension [72]. If they do not, chaotic band displacement results and, as a consequence, the speed of the second dimension must be significantly faster than the first. Moore et al. [36] illustrated the increased separation speeds required in a three-dimensional separation system by incorporating size-exclusion chromatography in the first dimension, reversed-phase chromatography in the second dimension, and capillary zone electrophoresis in the third dimension. In this instance, the separation time of the first dimension was around 500 minutes, the second dimension was approximately 6 minutes, and the third dimension was 2 seconds. To achieve the fast separation times required in the second (or third) dimension to allow for complete two-dimensional analysis, smaller first-dimension columns [109], higher second-dimension flow rates [110], and monolithic columns [111] are commonly being utilized. Venkatramani et al. [82] and Moore et al. [36] illustrated that a portion of the first-dimension chromatogram could be excluded from analysis if each chromatographic peak was sampled at least once, referred to as a discrete sampling approach [36]. Such an approach is interesting, as we live in an information world where collecting the information is often simple but using this information in a meaningful manner is often time-consuming and not worth the end result. Furthermore, this method is a useful screening approach and when anomalies are detected, more detailed studies can be undertaken. For the optimized operation of a two-dimensional system where n fractions are transported from the first dimension to the second dimension, the two-dimensional chromatographic system should have n second-dimension columns to complete the analysis [42, 45], as illustrated in Figure 4.17. This concept should work in practice C2 1 2 Fraction 6

Fraction 5

Fraction 4

Fraction 3

Fraction 2

Fraction 1

C1 Flow diversion device

3

C2

C2

4 5 6

C2

C2

C2

FIGURE 4.17 Hypothetical two-dimensional HPLC system where the number of cut fractions from the first dimension equals the number of second-dimension columns.

212

Advances in Chromatography

when the number of first dimension samples transported does not exceed four or so. The cost and the complexity of the two-dimensional chromatographic system increases with an increasing number of secondary chromatographic columns. Commonly, only one or two second-dimension chromatographic columns are utilized in the construction of a two-dimensional chromatographic system. Köhne and coworkers [109, 112] illustrated the technique of “peak parking,” where first-dimension eluent is transported directly to the second dimension. Once a section is transported to the second dimension, the flow to the first-dimension column is stopped and development of a second-dimension chromatogram is allowed to proceed. In like fashion, the flow to the first dimension is stopped and started continuously until a two-dimensional analysis is completed. It should be noted, however, that “peak parking” could yield undesirable band-broadening effects. In many instances, a comprehensive analysis is unnecessary and only a small section of the first dimension requires analysis in the second dimension. In such instances a technique referred to as “heart-cutting” is employed. With this method, only the relevant section from the first dimension is cut to the second dimension. In this technique, issues such as the wrap-around effect are not encountered and there is often no need for the separation in the second dimension to be fast relative to that of the first dimension. In fact, the resolving power of the second dimension can be exploited if there is no limitation in analysis time [30, 56, 113–116].

4.8.1 DESIGNING TWO-DIMENSIONAL SEPARATIONS IN ACCORDANCE WITH THE FINDINGS OF ORTHOGONALITY STUDIES A mixture of 32 oligostyrene isomers that can be described in terms of three variations in butyl end-groups (tert-butyl, sec-butyl and n-butyl) and stereochemistry (diastereoisomers) is clearly an example of components in a sample that would be impossible to separate in a one-dimensional separation by any means. The onedimensional chromatograms shown in Figure 4.13 verify this fact since, in each onedimensional system, on either the C18 or CCZ columns, all 32 constituents eluted with very little or no resolution. From IT and FA, the C18/methanol-CCZ/acetonitrile system reported in Table 4.2 was predicted to be the optimal two-dimensional combination for the separation of the 32 oligostyrene isomer mixture. The dimensions in this system were almost orthogonal, the system had the lowest solute crowding and utilized the majority of the two-dimensional retention plane. When this system was operated in a heartcutting mode, isomers of each respective end group oligostyrene could be easily resolved following transfer from the C18 dimension to the CCZ dimension. For example, the separation shown in Figure 4.13a illustrates the separation of the 32oligostyrene isomer mixture on a C18 column with a methanol mobile phase. This chromatogram can be divided into three sections corresponding to the resolution of each of the structural isomer classes as indicated by the dashed vertical lines. Retention of the oligostyrenes increased in the order of tert-butyl, sec-butyl, and nbutyl oligostyrenes. There was substantial overlap between the tert- and sec-butyl

Concepts and Practice of High-Performance Liquid Chromatography

213

end-group oligomers but the n-butyl oligomers were well-resolved. Heart-cutting separations of the 32-component oligostyrene mixture are shown in the chromatograms in Figure 4.18. Not all 32 isomers were separated because the resolving power of the CCZ dimension was insufficient to isolate three of the sec-butyl diastereomers and two tert-butyl isomers. In a similar manner to that established for the heart-cutting multidimensional separation of this mixture, a comprehensive separation was undertaken. It is important to note that conducting heart-cutting separations of complex mixtures is highly desirable in order to determine the second-dimension separation times, especially in systems where there is little correlation between each dimension. The normalized two-dimensional retention plane of the two-dimensional separation is shown in Figure 4.14a. In this two-dimensional system, the C18/methanol dimension resolves the oligostyrenes solely on the basis of their end-groups or by structural isomer classes with no stereoisomer resolution. Identical to the heart-cutting separations shown in Figure 4.19, three cuts were made across the C18 dimension, as indicated in Figure 4.18. Completion of the comprehensive two-dimensional separation is shown in Figure 4.19 where, following solute transportation to the second dimension, resolution of the diastereoisomers was achieved. In the second dimension, the dominating aspect of retention is oriented towards molecular shape and thus the diastereoisomers resolve. In total, 27 out of the 32 isomers were resolved in this separation, in agreement with our theoretical predictions (Table 4.2) and achieved in the heart-cutting mode of operation. More than 27 bands were observed in Figure 4.19, due to the duplication of isomers transported as a result of overlap between the tert- and sec-butyl components, as shown in Figure 4.18. (These bands are indicated with an asterisk.) Such an overlap poses some problems for quantitative analysis and, to a certain extent, larger scale preparative separations if such a method is employed. When the first dimension of the two-dimensional system was changed to acetonitrile, the separation was less successful because solute crowding was high, as shown in Table 4.2. Furthermore, the separation dimensions exhibited a higher degree of correlation; that is, there was some isomeric separation in the first dimension that was not apparent when methanol was employed as the mobile phase in this first dimension. This separation resulted in a more disordered first-dimension separation and subsequently an increase in the entire band volume (as shown in Figure 4.13b). The ramifications of this are that more cuts (six) were required to be transported to the second dimension and each cut had a greater degree of heterogeneity. The overlap in the retention windows between each of the tert-butyl, sec-butyl, and n-butyl isomers became substantial, the resulting chromatograms (not illustrated) were complex, and assignment of component identity was not possible. Less than 12 components would have been isolated as pure bands, substantially less that the 26 predicted using factor analysis. An important conclusion drawn from this application is that maintaining order in the chromatographic separation in both dimensions increases the likelihood of obtaining a successful separation. Furthermore, minimizing the degree of solute crowding (also related to the degree of separation order) increases the likelihood of separation success.

−1

0

1

2

3

5

10

15

**

20 25 30 Retention time (min)

*

2nd dimension

35 40

−5

0

5

0

1

2

3

10

10

15

Intensity (mV) 0 5

11

10

1 3

*

15 20 25 30 35 Retention time (min)

*

2nd dimension

12 13 14 Retention time (min)

2

40

1st dimension

45

15

50

16

0

1

2

3

0

10

20

30 40 50 Retention time (min)

2nd dimension

60

70

FIGURE 4.18 Chromatograms of the two-dimensional heart-cutting HPLC separation of 32 oligostyrene isomers (n = 5) with tert-butyl, sec-butyl and n-butyl end-groups on a C18 (methanol)/CCZ (acetonitrile) system. Each numbered section (1–3) from the first dimension was transported to the second dimension. Flow rates: first dimension 1 mL/min, second dimension 2 mL/min. First dimension operated at room temperature, second dimension thermostated at 30˚C.

Intensity (mV)

4

Intensity (mV)

20

Intensity (mV)

25

214 Advances in Chromatography

Concepts and Practice of High-Performance Liquid Chromatography

215

9 8 7

Intensity (mV)

6 5 4 3 2 1 0 −1

0

20

40

60 80 Retention time (min)

100

120

140

FIGURE 4.19 Chromatograms of the comprehensive two-dimensional HPLC separation of 32 oligostyrene isomers (n = 5) with tert-butyl, sec-butyl, and n-butyl endgroups on a C18 (methanol)/CCZ (acetonitrile) system. Flow rates: first dimension 0.1 mL/min, second dimension 2 mL/min. First dimension operated at room temperature, second dimension thermostated at 30°C.

4.8.2 CONTOUR MAPS Presentation of the results of two-dimensional liquid chromatographic separations is commonly done using three-dimensional contour plots that mimic two-dimensional TLC analysis [26]. The x-axis represents the first-dimension separation, the y-axis the second dimension separation, and the z-axis the peak height or peak area of the constructed contour plot. Commonly, a series of second-dimension chromatograms are produced following a two-dimensional separation. The second-dimension chromatographic series is then broken down into a series of high-speed seconddimension chromatograms as shown in Figure 4.20. Through various computations, construction of a three-dimensional contour plot is then produced.

4.8.3 TWO-DIMENSIONAL SYSTEM DESIGN Construction and hyphenation of two chromatographic systems to form a twodimensional liquid chromatographic system is achieved using electronically actuated 4-port, 6-port, 8-port, or 10-port two-position switching valves. Three general or common configuration themes exist for two-dimensional liquid chromatographic systems: those that (1) incorporate two sampling loops and a switching valve, (2)

25.5 −10 0 10 20 30 40 50 60 Detector response (mAU)

26.0

26.5

27.0

−5 25

5

15

25

5

15

25

5

15

25

5

15

25

5

15

25

40 30 35 Retention time (seconds)

Five high-speed secondary chromaiograms

45

25.5

Primary column retention (minutes) 27.5

Detector response (mAU)

25

45 35 Secondary column retention (seconds)

FIGURE 4.20 The process of converting high-speed secondary chromatograms into two-dimensional contour plot is shown above. The chromatogram on the left is a partial plot of an HPLC separation lasting 2 min. The stacked plots in the middle are five high-speed secondary chromatograms resulting from fractions transferred to the secondary column (indicated by error bar). The resulting contour plot is shown to the right. Reprinted with permission from [82]. Cop yright (2003) with permission from the American Chemical Society.

Retention time (minutes)

27.5

HPLC separation on primary column (partial)

216 Advances in Chromatography

Concepts and Practice of High-Performance Liquid Chromatography

217

contain two sample traps and switching valves, and (3) utilize switching valves in conjunction with a dual- or quad-column second dimension. One of the earliest and understated reports of the construction of a two-dimensional chromatographic system is the configuration used by Erni and Frei [25] and shown in Figure 4.21a. This configuration consisted of a specialized 8-port twoposition switching valve with two sample loops that alternately trapped first-dimension eluent. While loop A was loaded, the contents of loop B were injected into the second dimension. A similar design employed by Bushey and Jorgenson [105] is shown in Figure 4.21b and has been utilized by a large number of chromatographers [33, 71, 85, 110, 117–119]. While loop 1 is loaded with first-dimension eluent, the contents of loop 2 are injected onto the second dimension. A limitation of this design is that the contents of loop 1 are reversed flushed onto the second dimension, while the contents of loop 2 are forward flushed into the second dimension. This feature can lead to slight differences in the retention time of certain components when loop sizes become large [83]. Dugo et al. [34] and van der Horst and Schoenmakers [83] also utilized a twoloop system consisting of a 10-port switching valve with the two loops interfaced between the two chromatographic dimensions. Two separate configurations (I and II) are illustrated in Figure 4.22. van der Horst et al. [83] noticed that with the loop sizes used in system I, retention times varied depending on whether the sample was forward flushed from loop 1 or reversed flushed from loop 2. However, when both loops were forward flushed (as in system II) identical retention times were observed for their polystyrene standard. In our studies we routinely employ 6-port two-position switching valves rather than 10-port or 8-port valves primarily because of the versatility that the system design affords. For example, two switching valves are suitable for heart-cutting techniques or [81] four 6-port switching valves in conjunction with two sample loops can be used in a comprehensive analysis if retention in the second dimension is significantly greater than the time between successive cuts from the first dimension (Figure 4.23). In this system design, the contents of the sample loops are isolated from the flow in each separation dimension, an advantage when both loops contain first-dimension eluent and the second-dimension separation of a particular firstdimension section is not completed, although this strategy can only be used when empty first-dimension chromatographic space occurs. In other situations, the 6-port valve combination allows systems to be built for component trapping purposes, where cut fractions can be concentrated on a column and reversed flushed to the second dimension, perhaps for analysis by NMR [120]. Obviously, the size of the sample isolation loops is important for comprehensive 2D separation systems. The loop size is usually prepared to match the volume of the fractions to be transported to the second dimension, or a loop size is chosen to exceed the first dimension volume to be transported to the second dimension. Opiteck and coworkers [119], however, determined the loop volume size by multiplying the first-dimension flow rate by the sum of the second-dimension separation time and equilibration time (in the case of a gradient elution mode in the second dimension).

218

Advances in Chromatography

Column 1

Inj.

. Pos

Det 1

Pump

B

Loop 1

Loop 1

Waste

A Loop 2

Pump

Det 2

. Pos

Column 2 (a)

FIGURE 4.21 Comprehensive two-dimensional liquid chromatograph configurations that incorporate 8-port two-position switching valves. (a) Design suggested by Erni and Frei [25] and (b) Bushey and Jorgenson [105]. Adapted from [25] and [105].

4.8.4 DUAL-

AND

QUAD-COLUMN SECOND-DIMENSION DESIGNS

A number of researchers report the use of two-dimensional chromatographic systems incorporating two or four identical second-dimension chromatographic columns. These setups can but usually do not contain sample loops [31, 32, 72, 84, 107, 108, 121, 122] as the interface between chromatographic dimensions. Dual- and quadcolumn second dimensions give more flexibility than single-column second-dimension configurations since the second-dimension separation times can be twice (or more) as long. In addition Opiteck et al. [32] suggested that the use of loops introduces extra dead time that might increase the second-dimension separation time or, alternatively, reduce the time available for the elution of analytes before the next first-dimension sample is loaded onto the second dimension. Opiteck and others [32, 121] in the analysis of peptides and proteins were among the first to develop a two-dimensional liquid chromatographic system that incorporated a dual-column second dimension. The comprehensive system developed (shown in Figure 4.24a) consisted of a series of size-exclusion columns (first dimension), two 4-port two-position switching valves, a dual second-dimension incorporating two nonporous C18 columns operated in gradient elution mode (run in parallel), plus UV and mass spectrometric detection or a fraction collector. With

Concepts and Practice of High-Performance Liquid Chromatography

219

Pump

Det 1

Column 1

Inj.

Det 2

Waste

Pump

Column 2

Pos. A

Waste (b)

FIGURE 4.21 (Continued.)

this configuration, the eluent from the first separation dimension was diverted alternately to each of the C18 columns in the second dimension, allowing successive band cuts to be loaded onto these columns. Since water was a main component of the first-dimension eluent, sample fractions from the first-dimension would concentrate at the head of the second-dimension reversed-phase columns. This practice focused the sample in a manner similar to that in GC × GC [77] — while the separation was underway on C18(1), a second first-dimension sample would be loaded onto C18(2). At the completion of the separation on C18(1) and equilibration time, a gradient elution of the sample analytes focused onto the head of C18(2) would start. At the same time, another first-dimension fraction would be loaded onto C18(1). This process would continue, allowing the two-dimensional analysis of all sample constituents. Unlike the two-loop design, use of a dualcolumn second dimension would require significantly more computing power to construct contour plots, and so forth. A much simpler design [72, 107, 108, 122] is shown in Figure 4.24b. This design consists of a single 10-port switching valve and two second-dimension columns. Fractions could be alternately loaded onto one of the second-dimension columns while development of the separation on the other second-dimension column occurs. Again, in all cases, band focusing on the head of the second dimension columns was a requisite for this design to be usable, otherwise intolerable band spreading or breakthrough might result.

220

Advances in Chromatography

Pump

Det 1

Column 1

Inj.

Waste Det 2

Loop Loop Column 2

Pump

Waste

Waste (a)

FIGURE 4.22 Comprehensive two-dimensional liquid chromatograph configurations that incorporate 10-port two-position switching valves. (a) Reverse-flush and forward flush configuration and (b) reverse-flush configuration. Adapted from [83].

Tanaka et al. [111] illustrated the use of a dual-column second-dimension 2DHPLC system that incorporated two sample loops. The configuration shown in Figure 4.24c consists of two 6-port switching valves, two loops, a divert valve, two monolithic columns in the second dimension, three pumps, and a first dimension column. Operation of this system could be carried out in a number of ways [111]. In order to allow the system to work comprehensively, each loop could be alternately loaded with first-dimension eluent followed by subsequent injection to one of the seconddimension columns operated at 10.0 mL/min. These second-dimension separations were completed in less than 30 seconds. Potentially, this system could provide more flexibility than a dual-column second-dimension system with no loops since the first loop could be loaded and injected, then the second loop loaded while the first separation developed. If a particular second-dimension separation has not been completed on a particular second-dimension column, extra time is available for subsequent chromatographic development since the separation can still proceed in the second dimension while the loop, attached to that second-dimension column, is being loaded. Wagner et al. [31] and Machtejevas et al. [84] recently illustrated a comprehensive two-dimensional analysis employing four second-dimension columns, as shown in Figure 4.25. Prior to the two-dimensional chromatography, samples were subjected to size fractionation and enrichment using a restricted access media (RAM) column. Elution of retained components on the RAM column onto the first dimension

Concepts and Practice of High-Performance Liquid Chromatography

221

Column 1

Injector

Pump 1

Det 1

Pump 2

Loop 1

Loop 2 Waste

Waste Det 2

Column 2 (b)

FIGURE 4.22 (Continued.)

(TSKgel SP-NPR ion exchange column) was facilitated by in-line back-flushing and the first dimension was operated under a gradient elution mode. The 96-minute firstdimension separation was divided into 24 discrete sections. Each section was subsequently loaded onto one of the four second-dimension nonporous C18 columns, all operating in gradient elution mode. The system contained three gradient pumping systems, one for the first dimension and two for the second dimension. An isocratic pump was also used to regenerate or reequilibrate second-dimension columns. This particular design, however expensive, tolerated longer separation times in the second dimension, allowing for a higher peak capacity to be achieved relative to a dualcolumn second dimension. The quad-column design was needed in the case of Wagner et al. [31] since 4-minute sections of the first dimension were transported to the second dimension and each second-dimension separation required approximately 8 minutes for complete development. In this instance, unless the mobile phase gradient was altered or mobile phase flow rate increased in the second dimension, a dual-column second dimension would not be adequate. The peak capacity of each of the four columns was estimated to be approximately 130. A number of analysts [82, 107] have reported the use of complementary stationary phases in dual-column 2D-HPLC. Venkatramani et al. [82] illustrated the

222

Advances in Chromatography

S1

S2

S3

S4 Injector

P1

C1 D2

Waste D1 Sample Loop 2 V2

V1

S5 Waste P2

Sample Loop 1 V4

V3

C2 D3

Waste

FIGURE 4.23 Comprehensive two-dimensional liquid chromatograph configuration which incorporates four 6-port, two-position switching valves [81].

use of a system incorporating a C18 column in the first dimension and amino and cyano columns, both in the second dimension. First-dimension samples were alternately loaded onto either the cyano or amino columns in the second dimension. Venkatramani et al. [82] suggested that in the separation of a test mixture of amines, acids, and other substances, the C18 with water/TFA-acetonitrile/TFA gradient exhibited a mixed mode of retention based upon hydrophobicity and polarity. The second-dimension mobile phase was the same as the first; however, progressive elution from the first dimension at increased organic content in the mobile phase meant that the mobile phase for each fraction transported to the second dimension was developed in the second dimension with a gradient with increased organic content. Therefore. they [82] suggested that the cyano column interacted with polar functional groups, resulting in a separation based upon functional group differences. The amino column was said to resolve the test mixture based upon hydrophobicity, polarity, and ionic interactions. With this system and using complementary stationary phases in the second dimension, an increased amount of information can be gathered regarding a

Concepts and Practice of High-Performance Liquid Chromatography

HPLC pump #2

HPLC pump #1

Injector

(a)

223

RPLC column

Size exclusion columns

Waste

RPLC column

UV detector

Fraction collector

Pump I (RP)

Ion-exchange

Pump II (RP)

Ten-port switching valve

(b)

Injector 1st-D column 1st-D pump

Switch

Mobile phase 1st-D

(c)

UV detector A

Loop A load, loop B inject Loop A inject, loop B load

1st-D UV detector

2nd-D column A

2nd-D column B

Injector A

Waste

UV detector B

Injector B

Loop A (500 μ)

Waste

Loop B (500 μ)

Pump A

Pump B

Mobile phase 2nd-D

FIGURE 4.24 Comprehensive two-dimensional liquid chromatograph configurations incorporating a dual-column second dimension. (a) Confi guration as utilized by Opiteck et al. [121]. Reprinted with permission from [121]. Copyright (1998) with permission from Elsevier. (b) Configuration utilizing a 10-port, two-position switching valve. Reprinted with permission from [122]. Copyright (2000) with permission from Elsevier. (c) Configuration as utilized by Tanaka et al. [111]. Reprinted with permission from [111]. Cop yright (2004) with permission from the American Chemical Society.

224

Advances in Chromatography

Integral workstation Injector

Gradient pump autosampler control unit Injection valve

Sample loop

RAM Waste

Gradient system 1

Gradient system II

By-pass loop Isocratic pump Waste

IEX

2

RP I

1 10

2

3

9

4

8 5

6

7

RP II

1 10

2

3

9

4

8 5

6

7

Waste UV detector

Fraction collection A for MALDI MS analysis

RP III

1 10 9

3

8

4 5

6

RP IV

7

Waste UV detector

Fraction collection B for MALDI MS analysis

FIGURE 4.25 Schematic representation of the on-line comprehensive two-dimensional HPLC system, including an integrated sample preparation step. Reprinted with permission from [31]. Copyright (2002) with permission from the American Chemical Society.

particular sample. However, since first-dimension eluent is alternately transported to either of the second-dimension columns, two chromatographic runs would be required to comprehensively map the eluent from the first dimension on each column in the second dimension, although if the peaks are sampled at a significantly fast rate in the first dimension, one two-dimensional chromatographic run might suffice.

4.9 ANOMALIES IN MULTIDIMENSIONAL SEPARATIONS As mentioned earlier, orthogonality between chromatographic dimensions is of primary importance in two-dimensional liquid chromatographic systems. Such a requirement places stringent conditions as to the types of stationary phases and liquid chromatography modes that should be employed. Two very different modes may potentially use significantly different mobile phases; hence, incompatibility problems may arise between the solvent plug being transported from the first dimension to the second dimension. Such problems might be due to immiscibility or solvent

Concepts and Practice of High-Performance Liquid Chromatography

225

strength — one obvious example is that of a normal-phase/reversed-phase twodimensional combination. Potentially, a highly organic normal-phase mobile phase would be immiscible with an aqueous reversed-phase mobile phase. Alternatively, size-exclusion chromatography (SEC)/reversed-phase chromatography of certain analytes might be problematic. Mobile phases in size-exclusion modes are chosen such that the absorption effects are minimised or absent. SEC mobile phases are strong or good solvents for the analytes of interest, and injection of a plug of SEC eluent containing analytes from a SEC dimension onto a reversed-phase system or normal-phase system might result in sample breakthrough [123] or total washthrough with no retention on the second-dimension column, especially if the cut volume is large. Jiang et al. [123] illustrated the effect of breakthrough peaks on polymers in relation to solvent strength. Although this study was conducted using a single-column chromatographic system, this setup would obviously be problematic in two-dimensional liquid chromatography where large solvent plugs are injected onto the second dimension. Breakthrough peaks were observed when the sample solvents for a polymethyl methacrylate (PMMA) were strong. Larger breakthrough peaks or multiple breakthrough peaks were observed when the mobile phase starting conditions provided a stronger solvent environment. The size of the breakthrough peaks increased as the injection volume of the PMMA standard increased. The “real” retention time of a polystyrene (PS) standard was also shown to vary as the injection volume for the PS standard was varied. Jiang et al. [123] suggested that the breakthrough phenomenon occurred because polymer molecules at the front and center of a strong solvent plug were caught and would be eluted in a breakthrough peak. At some location in the solvent plug, the composition might vary by dilution with mobile phase. Analytes at the rear of the solvent plug then experience solvent conditions that are below the point of critical composition and rapidly fall behind and are retained on the chromatographic column. This study illustrates that careful consideration should be given to the mobile phase compositions in the first and second dimensions. One method to overcome the effect of solvent plug strength following transfer of a solute in a strong solvent environment from the first dimension to the second dimension is to add a poor solvent to the flow stream via a splitting T-joint after the first dimension and prior to the second dimension [36, 58, 64, 107, 124]. Murahashi [124] illustrated this technique in the analysis of polyaromatic hydrocarbons. Murahashi [124] used a Cosmosil 5PBB column with acetonitrile mobile phase in the first dimension (1.0 mL/min) and two C18 monoliths in the second dimension with acetonitrile mobile phase in the second dimension (16.0 mL/min). A third pump in the system delivered water (3.0 mL/min) via a mixing T-joint to eluent exiting the PBB column (resulting in a 75/25 water/acetonitrile mixture) so that analytes concentrated on the C18 reversed phase columns in the second dimension. A position change of the 10-port valve then changed the mobile phase from 25% acetonitrile/75% water at 4.0 mL/min to 100% acetonitrile at 16 mL/min. This process has the added advantage of focusing the sample at the head of the column in the second dimension. Many examples exist that illustrate this process [32, 72, 108, 122] and the majority include a combination ion-exchange in the first dimension and a

226

Advances in Chromatography

reversed-phase second dimension. Commonly, aqueous buffer mobile phases are extremely poor reversed-phase solvents and so components eluting from an ionexchange column concentrate on the head of the reversed-phase second-dimension column. Only when the ion-exchange solvent plug has eluted from the head of the second-dimension reversed-phase column do concentrated components start to resolve along the reversed-phase column. Other strategies that overcome the problem of solvent miscibility or incompatibility involve the transportation of small solvent plugs to the second dimension. Transportation of large amounts of first-dimension eluent in some instances has resulted in increased peak broadening or peak skewing and poor peak shape when band focusing in the second dimension is not possible [64]. Each second-dimension column and sample analysis is different and requires specific tailoring of the firstdimension sample volume to minimize peak degradation in the second dimension. Obviously, a balance between first-dimension sampling volume, second-dimension separation times, and overall analysis time must be arrived at so as to avoid excessively large analysis times. Köhne et al. [109] suggests that in order to achieve a peak volume/injection volume ratio close to unity, the first-dimension column should be much smaller in diameter compared to the second, although sample dilution can occur in this instance, a problem with lower intensity peaks. Trapping columns, as described previously [29, 30, 66, 113, 114, 125, 126], alleviate many of the problems associated with the transfer of immiscible and strong solvents to the second dimension. These columns can be employed in addition to post-first-dimension eluent modification [125] to allow for a solvent exchange prior to injection onto the second dimension. Viscosity differences between the solvents in the first and second dimension can also result in deleterious chromatographic behavior. When the solvent cut from the first dimension has a viscosity with approximately 10% difference to that of the solvent in the second dimension, the interface between the two solvents becomes unstable and the lower viscosity solvent displaces the higher viscosity solvent in a complex process called viscous fingering. An illustration of this effect is shown in the photograph in Figure 4.26. In this instance, the viscous fingering is severe as the viscosity difference is in excess of 10% (i.e., the mobile phase was 0.56 cP and the injection plug was 0.38 cP). Under these conditions, the fingering process results in substantial band broadening and distorted profiles, which quite clearly renders the two-dimensional separation practically useless. Consequently, care should be taken to not only minimize solvent differences with regard to solvent strength but also differences in solvent viscosity should be carefully considered.

4.10 CONCLUSION Multidimensional separation techniques will no doubt continue to gain favor. The basic instrumental operation of multidimensional systems is now no more complicated than a single dimensional separation. All modern LC systems can operate switching valves that can be programmed to operate continually (in the case of a comprehensive separation) or periodically (as in a heart-cutting system) throughout a separation. Perhaps the limiting aspect of the technique at present is the collection

Concepts and Practice of High-Performance Liquid Chromatography

Viscous fingering

2500

227

Non viscous fingering

Intensity (mV)

2000

1500

1000

500

0 0

2

4 6 Retention time (min)

8

10

FIGURE 4.26 Illustration of viscous fingering and the subsequent effect on band shape. The photographs of viscous fingering were taken inside the chromatography column in a mobile phase (28% dichloromethane, 36% toluene, 36% cyclohexanol) that had the same refractive index as the stationary phase (C18 silica). A red dye (pictured in grayscale) was used to illustrate the flow region of the viscous fingering. The viscosity of the mobile phase in which there was no viscous fingering was 0.38 cP, whereas the viscosity of the mobile phase where viscous fingering was observed was 0.56 cP. In both instances the viscosity of the injection plug was 0.38 cP. Column: C18 Nucleosil (17 mm × 7.2 mm). Mobile phase flow rate 2.0 mL/min, detection at 575 nm.

of two-dimensional data and the representation of the resulting two-dimensional chromatograms. However, many groups have built their own software (including ourselves) to suit their own needs with regard to the collection and display of twodimensional chromatographic information. Perhaps some of these software programs will become commercially available. Regardless of whether this happens, the application of two-dimensional chromatography should not be limited since the data in each dimension can be collected and essentially displayed in separate one-dimensional formats. Contour plots can also be drawn using packages such as Minitab or Origin. Two-dimensional liquid chromatography can be used in many different types of analysis. Applications can be found in complex sample environments, such as environmental analysis and forensic analysis, where chemical signatures can be obtained by expression of the sample components across two different separation stages, or as we have illustrated in this text, the analysis of complex samples such as polymers. Perhaps not as obvious is the fact that two-dimensional chromatography can be employed for the isolation of specific components in complex samples at the preparative level of separation where, under appropriate conditions, the efficiency of

228

Advances in Chromatography

the separation can yield vastly improved specific recovery compared to that of conventional one-dimensional separations [116]. Another area that may find increased use for two-dimensional liquid chromatography is the routine analysis of simple pharmaceuticals. Essentially, two-dimensional chromatography allows two independent modes of analysis to be incorporated into one separation. The displacement of the sample constituents across a two-dimensional plane provides a high degree of surety associated with the identity of the components because of the low probability of two components having exactly the same two-dimensional retention. Furthermore, the process of analysis is also quantitative. Designing a two-dimensional separation is more laborious than optimizing a one-dimensional separation. However, the power of the two-dimensional separation cannot be ignored and, consequently, makes the effort in many instances worthwhile. Our basic approach is to first carry out selectivity studies, either on the real sample or using model compounds. We then employ statistical measures to determine which systems should be coupled. At this point, having a clear objective is helpful; that is, are you interested in complete sample analysis or simply in the analysis or collection of specific components. To a large extent, this objectve will dictate how systems are chosen and ultimately coupled. The application of multidimensional liquid chromatography is a perfect example of the union of several important areas of chromatographic investigation, namely the design of stationary-phase surfaces providing selectivity changes that enable compatible solvent environments to be coupled and the design of new stationaryphase technology, i.e., monoliths that will serve to provide high-speed separations in the second dimension, and also the developments in instrumentation that enable the coupling and automation of system configurations.

ACKNOWLEDGMENTS The authors are grateful for the useful discussions relating to this work with Gary Dennis, Patrick Slonecker, and Alan Sweeney, and Heather Catchpoole for the assistance in the viscous fingering experiment illustrated in Figure 4.26. We would like to acknowledge the receipt of a University of Western Sydney Postgraduate Research Award. Part of this work was supported by a University of Western Sydney Internal Research Award.

REFERENCES 1. Skoog, D.A., West, D.M., and Holler, F.J. Fundamentals of Analytical Chemistry, 7th ed., Saunders College Publishing, 1996. 2. Steuer, W., Grant, I., and Erni, F. Comparison of high-performance liquid chromatography, supercritical fluid chromatography and capillary zone electrophoresis in drug analysis. J. Chromatogr., 1990, 507, 125–140. 3. Snyder, L.R. and Kirkland, J.J. Introduction to Modern Liquid Chromatography, John Wiley & Sons, New York, 1976.

Concepts and Practice of High-Performance Liquid Chromatography

229

4. Turowski, M., Morimoto, T., Kimata, K., Monde, H., Ikegami, T., Hosoya, K., and Tanaka, N. Selectivity of stationary phases in reversed-phase liquid chromatography based on the dispersion interactions. J. Chromatogr. A, 2001, 911, 177–190. 5. Mourey, T.H., Smith, G.A., and Snyder, L.R. Solvent selectivity in the liquid chromatographic separation of polystyrene oligomers on silica. Anal. Chem., 1984, 56, 1773–1777. 6. Gray, M.J., Sweeney, A.P., Dennis, G.R., Wormell, P., and Shalliker, R.A. Selectivity and the separation of isomers of low molecular weight polystyrenes. J. Liq. Chromatogr. Rel. Tech., 2004, 27(18), 2905–2920. 7. Rohrschneider, L. Solvent classification by gas-liquid partition coefficients of selected solutes. Anal. Chem., 1973, 45, 1241–1247. 7a. Kamlet, M.J., Abboud, J.L.M., Abraham, M.H., and Taft, R.W. J. Org. Chem., 1983, 48, 2877. 8. Snyder, L.R., Carr, P.W., and Rutan, S.C. Solvatochromically based solvent-selectivity triangle. J. Chromatogr. A, 1993, 656, 537–547. 9. Sweeney, A.P., Wong, V., and Shalliker, R.A. The separation of diastereoisomers of polystyrene oligomers in reversed phase HPLC. Chromatographia, 2001, 54(1), 24–30. 10. Jackson, P.T., Schure, M.R., Weber, T.P., and Carr, P.W. Intermolecular interactions involved in solute retention on carbon media in reversed-phase high-performance liquid chromatography. Anal. Chem., 1997, 69, 416–425. 11. Jackson, P.T., Kim, T.-Y., and Carr, P.W. Diastereomeric selectivity of carbon-coated zirconia reversed-phase liquid chromatographic media. Anal. Chem., 1997, 69(24), 5011–5017. 12. Bolliet, D. and Poole, C.F. Influence of temperature on retention and selectivity in reversed phase liquid chromatography. Analyst, 1998, 123, 295–299. 13. Sentell, K.B., Ryan, N.I., and Henderson, A.N. Temperature and solvation effects on homologous series selectivity in reversed phase liquid chromatography. Anal. Chim. Acta, 1995, 307, 203–215. 14. Philipsen, H.J.A. Determination of chemical composition distributions in synthetic polymers. J. Chromatogr. A, 2004, 1037, 329–350. 15. de Geus, H.-J., de Boer, J., and Brinkman, U.A.Th. Multidimensionality in gas chromatography. Trends Anal. Chem., 1996, 15(5), 398–408. 16. Giddings, J.C. Use of multiple dimensions in analytical separations, in multidimensional chromatography. Techniques and Applications, Marcel Dekker, 1990. 17. Giddings, J.C. Maximum number of components resolvable by gel filtration and other elution chromatographic methods. Anal. Chem., 1967, 39, 1027–1036. 18. Davis, J.M. and Giddings, J.C. Statistical theory of component overlap in multicomponent chromatograms. Anal. Chem., 1983, 55, 418–424. 19. Davis, J.M. and Giddings, J.C. Statistical method for estimation of number of components from single complex chromatograms: Theory, computer-based testing, and analysis of errors. Anal. Chem., 1985, 57, 2166–2177. 20. Rowe, K. and Davis, J.M. Relaxation of randomness in two-dimensional statistical model of overlap: Theory and verification. Anal. Chem., 1995, 67, 2981–2993. 21. Lewis, J.J., Rogers, L.B., and Pauls, R.E. Solvent selectivity in reversed-phase highperformance liquid chromatographic separation of isomers of polystyrene oligomers. J. Chromatogr., 1983, 264, 339–356. 22. Shalliker, R.A., Kavanagh, P.E., and Russell, I.M. Reversed-phase separations of high-molecular-weight polystyrenes in a dichloromethane — Acetonitrile. J. Chromatogr., 1991, 558, 440–445.

230

Advances in Chromatography

23. Pasch, H., Hiller, W., and Haner, R. Investigation of the tacticity of oligostyrenes by on-line HPLC/1H NMR. Polymer, 1998, 39, 1515–1523. 24. Giddings, J.C. Sample dimensionality: A predictor of order-disorder in component peak distribution in multidimensional separation. J. Chromatogr. A, 1995, 703, 3–15. 25. Erni, F. and Frei, R.W. Two-dimensional column liquid chromatographic technique for resolution of complex mixtures. J. Chromatogr., 1978, 149, 561–569. 26. Zakaria, M., Gonnord, M.-F., and Guiochon, G. Applications of two-dimensional thin-layer chromatography. J. Chromatogr., 1983, 271, 127–192. 27. Bertsch, W. Multidimensional gas chromatography. Multidimensional Chromatography: Techniques and Applications, Marcel Dekker, 1990. 28. Guiochon, G., Beaver, L.A., Gonnord, M.F., Siouffi, A.M., and Zakaria, M. Theoretical investigation of the potentialities of the use of a multidimensional column in chromatography. J. Chromatogr., 1983, 255, 415–437. 29. Chelius, D., Zhang, T., Wang, G., and Shen, R.-F. Global protein identification and quantification technology using two-dimensional liquid chromatography nanospray mass spectrometry. Anal. Chem., 2003, 75(23), 6658–6665. 30. Davis, M.T., Beierle, J., Bures, E.T., McGinley, M.D., Mort, J., Robinson, J.H., Spahr, C.S., Yu, W., Luethy, R., and Patterson, S.D. Automated LC-LC-MS-MS platform using binary ion-exchange and gradient reversed-phase chromatography for improved proteomic analysis. J. Chromatogr. B, 2001, 752, 281–291. 31. Wagner, K., Miliotis, T., Marko-Varga, G., Bischoff, R., and Unger, K.K. An automated on-line multidimensional HPLC system for protein and peptide mapping with integrated sample preparation. Anal. Chem., 2002, 74, 809–820. 32. Opiteck, G.J., Jorgenson, J.W., and Anderegg, R.J. Two-dimensional SEC/RPLC coupled to mass spectrometry for the analysis of peptides. Anal. Chem., 1997, 69, 2283–2291. 33. Murphy, R.E., Schure, M.R., and Foley, J.P. One- and two-dimensional chromatographic analysis of alcohol ethoxylates. Anal. Chem., 1998, 70, 4353–4360. 34. Dugo, P., Favoino, O., Luppino, R., Dugo, G., and Mondello, L. Comprehensive twodimensional normal-phase (adsorption)-reversed-phase liquid chromatography. Anal. Chem., 2004, 76, 2525–2530. 35. Bushey, M.M. and Jorgenson, J.W. Automated instrumentation for comprehensive two-dimensional high performance liquid chromatography/capillary zone electrophoresis. Anal. Chem., 1990, 62, 978–984. 36. Moore, A.W., Jr. and Jorgenson, J.W. Comprehensive three-dimensional separation of peptides using size exclusion chromatography/reversed phase liquid chromatography/optically gated capillary zone electrophoresis. Anal. Chem., 1995, 67, 3456–3463. 37. Moore, A.W., Jr. and Jorgenson, J.W. Rapid comprehensive two-dimensional separations of peptides via RPLC — Optically gated capillary zone electrophoresis. Anal. Chem., 1995, 67, 3448–3455. 38. Blomberg, J., Mes, E.P.C., Schoenmakers, P.J., and van der Does, J.J.B. Characterization of complex hydrocarbon mixtures using on-line coupling of size-exclusion chromatography and normal-phase liquid chromatography to high resolution gas chromatography. J. High Resol. Chromatogr., 1997, 20, 125–130. 39. Häkkinen, V.M.A., Virolainen, M.M., and Riekkola, M.-L. New on-line HPLC-GC coupling system using a 10-port valve interface. HRC & CC, 1988, 11, 214–216. 40. Barcarolo, R. Coupled LC-GC: A new method for the on-line analysis of organochlorine pesticide residues in fat. HRC, 1990, 13, 465–469.

Concepts and Practice of High-Performance Liquid Chromatography

231

41. Senoráns, F.J., Reglero, G., and Herraiz, M. Use of a programmed temperature injector for on-line reversed-phase liquid chromatography-capillary gas chromatography. J. Chromatogr. Sci., 1995, 33, 446–450. 42. Liu, Z. and Lee, M.L. Comprehensive two-dimensional separations using microcolumns. J. Microcolumn Separations, 2000, 12(4), 241–254. 43. Phillips, J.B. and Beens, J. Comprehensive two-dimensional gas chromatography: A hyphenated method with strong coupling between the two dimensions. J. Chromatogr. A, 1999, 856, 331–347. 44. Holland, L.A. and Jorgenson, J.W. Separation of nanoliter samples of biological amines by a comprehensive two-dimensional microcolumn liquid chromatography system. Anal. Chem., 1995, 67, 3275–3283. 45. Venkatramani, C.J., Xu, J., and Phillips, J.B. Separation orthogonality in temperatureprogrammed comprehensive two-dimensional gas chromatography. Anal. Chem., 1996, 68, 1486–1492. 46. Slonecker, P.J., Li, X., Ridgway, T.H., and Dorsey, J.G. Informational orthogonality of two-dimensional chromatographic separations. Anal. Chem., 1996, 68, 682–689. 47. Davis, J.M. Statistical theory of spot overlap in two-dimensional separations. Anal. Chem., 1991, 63, 2141–2152. 48. Giddings, J.C. Two-dimensional separations: concept and promise. Anal. Chem., 1984, 56(12), 1258A–1264A. 49. Blomberg, J., Schoenmakers, P.J., Beens, J., and Tijssen, R. Comprehensive twodimensional gas chromatography (GC × GC) and its applicability to the characterization of complex (petrochemical) mixtures. J. High Resol. Chromatogr., 1997, 20, 539–544. 50. Giddings, J.C. Concepts and comparisons in multidimensional separation. HRC, 1987, 10(5), 319–323. 51. Mao, Y. and Carr, P.W. Adjusting selectivity in liquid chromatography by use of the thermally tuned tandem column concept. Anal. Chem., 2000, 72, 110–118. 52. Akard, M. and Sacks, R. Pressure-tunable selectivity for high-speed gas chromatography. Anal. Chem., 1994, 66, 3036–3041. 53. Wang, K., Blain, R.W., and Szuna, A.J. Multidimensional narrow bore liquid chromatography analysis of Ro 24-0238 in human plasma. J. Pharm. Biomed. Anal., 1994, 12(1), 105–110. 54. Andersson, T., Hyötyläinen, T., and Riekkola, M.-L. Analysis of phenols in pyrolysis oils by gel permeation chromatography and multidimensional liquid chromatography. J. Chromatogr. A, 2000, 896, 343–349. 55. Gray, M.J., Dennis, G.R., Slonecker, P.J., and Shalliker, R.A. Separation of oligostyrene isomers in a complex mixture using two-dimensional heart-cutting reversedphased liquid chromatography. J. Chromatogr. A, 2004, 1028, 247–257. 56. Wong, V., Sweeney, A.P., and Shalliker, R.A. Using analytical multidimensional isocratic HPLC methods of separation to isolate active constituents in natural products. J. Sep. Sci., 2004, 27, 47–52. 57. Pascoe, R., Foley, J.P., and Gusev, A.I. Reduction in matrix-related signal suppression effects in electrospray ionization mass spectrometry using on-line two-dimensional liquid chromatography. Anal. Chem., 2001, 73, 6014–6023. 58. Trathnigg, B., Rappel, C., Raml, R., and Gorbunov, A. Liquid exclusion-adsorption chromatography, a new technique for isocratic separation of nonionic surfactants. V. Two-dimensional separation of fatty acid polyglycol ethers. J. Chromatogr. A, 2002, 953, 89–99.

232

Advances in Chromatography

59. Agbaria, R., Ford, H., Jr., Kelley, J.A., Xie, F., Politi, P., Grem, J.L., Cooney, D.A., Marquez, V.E., Allegra, C.J., and Johns, D.G. Measurement of cyclopentenyl cytosine 5-triphosphate in vitro and in vivo by multidimensional high performance liquid chromatography. Anal. Biochem., 1993, 213, 90–96. 60. Cass, Q.B., Gomes, R.F., Calafatti, S.A., and Pedrazolli, J., Jr. Determination of amoxycillin in human plasma by direct injection and coupled-column high performance liquid chromatography. J. Chromatogr. A, 2003, 987, 235–241. 61. Kindberg, C.G., Slavik, M., Riley, C.M., and Stobaugh, J.F. Analysis of 5-flourouracil in plasma by precolumn derivatization with 4-bromomethyl-7-methoxycoumarin, followed by multi-dimensional high-performance liquid chromatography. J. Chromatogr., 1989, 473, 431–444. 62. Kindberg, C.G., Slavik, M., Riley, C.M., and Stobaugh, J.F. High-performance liquid chromatography of 5-fluorouracil after derivatization with 4-bromomethyl-7-methoxycoumarin. Characterization of the derivative and the use of column switching for the improvement of resolution and the enhancement of sensitivity. J. Pharm. Biomed. Anal., 1989, 7(4), 459–469. 63. Matsui, T., Tamaya, K., Kawasaki, T., and Osajima, Y. Determination of angiotensin metabolites in human plasma by fluorometric high-performance liquid chromatography using heart-cut column-switching technique. J. Chromatogr. B, 1999, 729, 89–95. 64. Veigl, E., Posch, W., Lindner, W., and Tritthart, P. Selective and sensitive analysis of 1-nitropyrene in diesel exhaust particulate extract by multidimensional HPLC. Chromatographia, 1994, 38(3), 199–206. 65. Murahashi, T. Determination of mutagenic 3-nitrobenzanthrone in diesel exhaust particulate matter by three-dimensional high-performance liquid chromatography. Analyst, 2003, 128, 42–45. 66. Serena, P., Tapparo, A., and Bombi, G.G. Direct determination of t,t-muconic acid in human urine by two-dimensional liquid chromatography. Analyst, 2000, 125, 689–692. 67. Vissers, J.P.C., van Soest, R.E.J., Chervet, J.-P., and Cramers, C.A. Two-dimensional capillary liquid chromatography based on microfractionation. J. Microcolumn Separations, 1999, 11(4), 277–286. 68. Glöckner, G., van den Berg, J.H.M., Meijerink, N.L.J., Scholte, T.G., and Koningsveld, R. Two-dimensional analysis of copolymers by size-exclusion chromatography and gradient-elution reversed-phase precipitation chromatography. J. Chromatogr., 1984, 317, 615–624. 69. Glöckner, G. and Barth, H.G. Use of high performance liquid chromatography for the characterization of synthetic copolymers. J. Chromatogr., 1990, 499, 645–654. 70. Augustein, M. and Stickler, M. Gradient high performance liquid chromatography of polymers. 1. Characterization of the products obtained by grafting methyl methacrylate onto ethylene-propene-diene rubber by SEC-HPLC cross fractionation using evaporative light scatting detection. Makromol. Chem., 1990, 191, 415–428. 71. Murphy, R.E., Schure, M.R., and Foley, J.P. Effect of sampling rate on resolution in comprehensive two-dimensional liquid chromatography. Anal. Chem., 1998, 70, 1585–1594. 72. Unger, K.K., Racaiyte, K., Wagner, K., Milliotis, T., Edholm, L.E., Bischoff, R., and Marko-Varga, G. Is multidimensional high performance liquid chromatography (HPLC) an alternative in protein analysis to 2D gel electrophoresis? J. High Resol. Chromatogr., 2000, 23(3), 259–365.

Concepts and Practice of High-Performance Liquid Chromatography

233

73. Phillips, J.B., Gaines, R.B., Blomberg, J., van der Wielen, F.W.M., Dimandja, J.-M., Green, V., Granger, J., Patterson, D., Racovalis, L., de Geus, H.-J., de Boer, J., Haglund, P., Lipsky, J., Sinha, V., and Ledford, E.B., Jr. A robust thermal modulator for comprehensive two-dimensional gas chromatography. J. High Resol. Chromatogr., 1999, 22(1), 3–10. 74. Ragunathan, N., Krock, K.A., and Wilkins, C.L. Multidimensional gas chromatography with parallel cryogenic traps. Anal. Chem., 1993, 65, 1012–1016. 75. Ledford, E.B., Jr. and Billesbach, C. Jet-cooled thermal modulator for comprehensive multidimensional gas chromatography. J. High Resol. Chromatogr., 2000, 23(3), 202–204. 76. Marriott, P.J., Kinghorn, R.M., Ong, R., Morrison, P., Haglund, P., and Harju, M. Comparison of thermal sweeper and cryogenic modulator technology for comprehensive gas chromatography. J. High Resol. Chromatogr., 2000, 23(3), 253–258. 77. Kinghorn, R.M., Marriott, P.J., and Dawes, P.A. Design and implementation of comprehensive gas chromatography with cryogenic modulation. J. High Resol. Chromatogr., 2000, 23(3), 245–252. 78. Van Gysegham, E., Van Hemelryck, S., Daszykowski, M., Questier, F., Massart, D.L., and Vander Heyden, Y. Determining orthogonal chromatographic systems prior to the development of methods to characterise impurities in drug substances. J. Chromatogr. A, 2003, 988, 77–93. 79. Guttman, A., Varoglu, M., and Khandurina, J. Multidimensional separations in the pharmaceutical arena. Drug Discovery Today, 2004, 9(3), 136–144. 80. Liu, Z., Patterson, D.G., and Lee, M.L. Geometric approach to factor analysis for the estimation of orthogonality and practical peak capacity in comprehensive two-dimensional separations. Anal. Chem., 1995, 67, 3840–3845. 81. Gray, M.J., Dennis, G.R., Slonecker, P.J., and Shalliker, R.A. Comprehensive twodimensional separations of complex mixtures using reversed-phase reversed-phase liquid chromatography. J. Chromatogr. A, 2004, 1041, 101–110. 82. Venkatramani, C.J. and Zelechonok, Y. An automated orthogonal two-dimensional liquid chromatograph. Anal. Chem., 2003, 75, 3484–3494. 83. van der Horst, A. and Schoenmakers, P.J. Comprehensive two-dimensional liquid chromatography of polymers. J. Chromatogr. A, 2003, 1000, 693–709. 84. Machtejevas, E., John, H., Wagner, K., Ständker, L., Marko-Varga, G., Forssmann, W.-G., Bischoff, R., and Unger, K.K. Automated multi-dimensional liquid chromatography: sample preparation and identification of peptides from human blood filtrate. J. Chromatogr. B, 2004, 803, 121–130. 85. Adrian, J., Esser, E., Hellman, G., and Pasch, H. Two-dimensional chromatography of complex polymers. Part 1. Analysis of a Graft copolymer by two-dimensional chromatography with on-line FTIR detection. Polymer, 2000, 41, 2439–2449. 86. Rocklin, R.D., Ramsey, R.S., and Ramsey, J.M. A microfabricated fluidic device for performing two-dimensional liquid phase separations. Anal. Chem., 2000, 72, 5244–5249. 87. Valkó, K., Espinosa, S., Du, C.M., Bosch, E., Rosés, M., Bevan, C., and Abraham, M.H. Unique selectivity of perfluorinated stationary phases with 2,2,2-trifluoroethanol as organic mobile phase modifier. J. Chromatogr. A, 2001, 933, 73–81. 88. Fields, S.M., Ye, C.Q., Zhang, D.D., Branch, B.R., Zhang, X.J., and Okafo, N. Superheated water as eluent in high-temperature high performance liquid chromatographic separations of steroids on a polymer-coated zirconia column. J. Chromatogr. A, 2001, 913, 197–204.

234

Advances in Chromatography

89. Neue, U.D., Alden, B.A., and Walter, T.H. Universal procedure for the assessment of the reproducibility and the classification of silica-based reversed phase packings. II. Classification of reversed-phase packings. J. Chromatogr. A, 1999, 849, 101–116. 90. Van Gysegham, E., Crosiers, I., Gourvénec, S., Massart, D.L., and Vander Heyden, Y. Determining orthogonal and similar chromatographic systems from the injection of mixtures in liquid chromatography-diode array detection and the interpretation of correlation coefficients color maps. J. Chromatogr. A, 2004, 1026, 117–128. 91. Van Gysegham, E., Jimidar, M., Sneyers, R., Redlich, D., Verhoeven, E., Massart, D.L., and Vander Heyden, Y. Selection of reversed phase liquid chromatographic columns with diverse selectivity towards the potential separation of impurities in drugs. J. Chromatogr. A, 2004, 1042, 69–80. 92. Smoliñski, A., Walczak, B., and Einax, J.W. Hierarchical clustering extended with visual complements of environmental data set. Chemometr. Intell. Lab. Syst., 2002, 64, 45–54. 93. Lochmüller, C.H., Breiner, S.J., Reese, C.E., and Koel, M.N. Characterization and prediction of retention behavior in reversed-phase chromatography using factor analytical modeling. Anal. Chem., 1989, 61, 367–375. 94. Detroyer, A., Schoonjans, V., Questier, F., Vander Heyden, Y., Borosy, A.P., Guo, Q., and Massart, D.L. Exploratory chemometric analysis of the classification of pharmaceutical substances based on chromatographic data. J. Chromatogr. A, 2000, 897, 23–36. 95. Beebe, K.R., Pell, R.J., and Seascholtz, M.B. Chemometrics: A Practical Guide, John Wiley & Sons, 1998. 96. Lurie, I.S., Sperling, A.R., and Meyers, R.P. The determination of anabolic steroids by MECC, gradient HPLC, and capillary GC. J. For. Sci., 1994, 39(1), 74–85. 97. Massart, D.L., Vandeginste, B.G.M., Buydens, L.M.C., Jong, S.D., Lewi, P.J., and Smeyers-Verbeke, J. Handbook of chemometrics and qualimetrics: Part A. Data Handling in Science and Technology, Elsevier Science, 1997. 98. Huber, J.F.K., Kenndler, E., Reich, G., Hack, W., and Wolf, J. Optimal selection of gas chromatographic columns for the analytical control of chemical warfare agents by application of information theory to retention data. Anal. Chem., 1993, 65, 2903–2906. 99. Dupuis, F. and Dijkstra, A. Application of information theory to analytical chemistry. Identification by retrieval of gas chromatographic retention indices. Anal. Chem., 1975, 47(3), 379–383. 100. Eskes, A., Dupuis, F., Dijkstra, A., De Clercq, H., and Massart, D.S. Application of information theory and numerical taxonomy to the selection of gas-liquid chromatography stationary phases. Anal. Chem., 1975, 47(13), 2168–2174. 101. Massart, D.L., Vandeginste, B.G.M., Deming, S.N., Michotte, Y., and Kaufman, L. Chemometrics: A textbook. Data Handling in Science and Technology, Elsevier Science, 1988. 102. Reese, C.E. and Lochmüller, C.H. Introduction to Factor Analysis, 2001. 103. Gray, M., Dennis, G.R., Wormell, P., Shalliker, R.A., and Slonecker, P. Two-dimensional reversed-phase-reversed-phase separations. Isomeric separations incorporating C18 and carbon clad zirconia stationary phases. J. Chromatogr. A, 2002, 975, 285–297. 104. Gray, M.J., Dennis, G.R., Slonecker, P.J., and Shalliker, R.A. Evaluation of the twodimensional reversed-phase-reversed-phase separations of low-molecular mass polystyrenes. J. Chromatogr. A, 2003, 1015, 89–98.

Concepts and Practice of High-Performance Liquid Chromatography

235

105. Bushey, M.M. and Jorgenson, J.W. Automated instrumentation for comprehensive two-dimensional high-performance liquid chromatography of proteins. Anal. Chem., 1990, 62, 161–167. 106. Gray, M.J., Sweeney, A.P., Dennis, G.R., Slonecker, P.J., and Shalliker, R.A. Comprehensive coupled reversed-phase reversed-phase separations of a complex isomeric mixture. Analyst, 2003, 128, 598–604. 107. Haefliger, O.P. Universal two-dimensional HPLC technique for the chemical analysis of complex sufactant mixtures. Anal. Chem., 2003, 75, 371–378. 108. Lui, H., Berger, S.J., Chakraborty, A.B., Plumb, R.S., and Cohen, S.A. Multidimensional chromatography coupled to electrospray ionization time-of-flight mass spectrometry as an alternative to two-dimensional gels for the identification and analysis of complex mixtures of intact proteins. J. Chromatogr. B, 2002, 782, 267–289. 109. Köhne, A.P. and Welsch, T. Coupling of a microbore column with a column packed with non-porous particles for fast comprehensive two-dimensional high-performance liquid chromatography. J. Chromatogr. A, 1999, 845, 463–469. 110. Siewing, A., Schierholz, J., Braun, D., Hellmann, G., and Pasch, H. Two-dimensional chromatography of complex polymers. 2a. Analysis of the grafting reaction of methyl methacrylate onto EPDM. Macromol. Chem. Phys., 2001, 202, 2890–2894. 111. Tanaka, N., Kimura, H., Tokuda, D., Hosoya, K., Ikegami, T., Ishizuka, N., Minakuchi, H., Nakanishi, K., Shintani, Y., Furuno, M., and Cabrera, K. Simple and comprehensive two-dimensional reversed-phase HPLC using monolithic silica columns. Anal. Chem., 2004, 76, 1273–1281. 112. Köhne, A.P., Dornberger, U. and Welsch, T. Two-dimensional high performance liquid chromatography for the separation of complex mixtures of explosives and their byproducts. Chromatographia, 1998, 48, 9–16. 113. Toriba, A., Chetiyanukornkul, T., Kizu, R., and Hayakawa, K. Quantification of 2hydroxyfluorene in human urine by column-switching high performance liquid chromatography with fluorescence detection. Analyst, 2003, 128, 605–610. 114. Stopher, D.A. and Gage, R. Determination of a new antifungal agent, Voriconazole, by multidimensional high-performance liquid chromatography with direct plasma injection onto a size-exclusion column. J. Chromatogr. B, 1997, 691, 441–448. 115. Szuna, A.J., Mulligan, T.E., Mico, B.A., and Blain, R.W. Determination of Ro-237637 in dog plasma by multidimensional ion-exchange-reversed-phase high-performance liquid chromatography with ultraviolet detection. J. Chromatogr., 1993, 616, 297–303. 116. Wong, V. and Shalliker, R.A. Isolation of the active constituents in natural materials by “heart-cutting” isocratic reversed-phase two-dimensional liquid chromatography. J. Chromatogr. A, 2004, 1036(1), 15–24. 117. Chen, X., Kong, L., Su, X., Fu, H., Ni, J., Zhao, R., and Zou, H. Separation and identification of compounds in rhizoma chuanxiong by comprehensive two-dimensional liquid chromatography coupled to mass spectrometry. J. Chromatogr. A, 2004, 1040, 169–178. 118. Adrian, J., Braun, D., and Pasch, H. New analytical methods for epoxy resins. 2a. Two-dimensional chromatography. Die Angewandte Makromolekulare Chemie, 1999, 267, 82–88. 119. Opiteck, G.J., Lewis, K.C., and Jorgenson, J.W., Comprehensive on-line LC/LC/MS of proteins. Anal. Chem., 1997, 69, 1518–1524. 120. Sweeney, A.P. and Shalliker, R.A. Development of a two-dimensional liquid chromatography system with trapping and sample enrichment capabilities. J. Chromatogr. A, 2002, 968, 41–52.

236

Advances in Chromatography

121. Opiteck, G.J., Ramirez, S.M., Jorgenson, J.W., and Moseley, M.A., III. Comprehensive two-dimensional high-performance liquid chromatography for the isolation of overexpressed proteins and proteome mapping. Anal. Biochem., 1998, 258, 349–361. 122. Wagner, K., Racaityte, K., Unger, K.K., Miliotis, T., Edholm, L.E., Bischoff, R., and Marko-Varga, G. Protein mapping by two-dimensional high-performance liquid chromatography. J. Chromatogr. A, 2000, 893, 293–305. 123. Jiang, X., van der Horst, A., and Schoenmakers, P.J. Breakthrough of polymers in interactive liquid chromatography. J. Chromatogr. A, 2002, 982, 55–68. 124. Murahashi, T. Comprehensive two-dimensional high-performance liquid chromatography for the separation of polycyclic aromatic hydrocarbons. Analyst, 2003, 128, 611–615. 125. Trathnigg, B. and Rappel, C. Liquid exclusion-adsorption chromatography, a new technique for isocratic separation of nonionic surfactants. IV. Two-dimensional separation of fatty alcohol ethoxylates with focusing of fractions. J. Chromatogr. A, 2002, 952, 149–163. 126. Kristensen, B.K., Askerlund, P., Bykova, N.V., Egsgaard, H., and Møller, I.M. Identification of oxidised proteins in the matrix of rice leaf mitochondria by immunoprecipitation and two-dimensional liquid chromatography-tandem mass spectrometry. Phytochemistry, 2004, 65(12), 1839–1851.

5

High-Performance Liquid Chromatography in the Pharmaceutical Industry: Application, Validation, and Regulatory Issues Under the PAT Framework Mauricio Dantus Merck and Company, Rahway, New Jersey

CONTENTS 5.1 5.2

Introduction ..................................................................................................238 Process Analytical Initiative ........................................................................238 5.2.1 Multivariate Data Acquisition and Process Analysis Tools ............240 5.2.2 Process Analytical Tools ..................................................................240 5.2.3 Process and Endpoint Monitoring and Control Tools.....................242 5.2.4 Continuous Improvement and Knowledge Management Tools ......243 5.3 Method Validation ........................................................................................243 5.3.1 Precision ...........................................................................................245 5.3.2 Accuracy...........................................................................................246 5.3.3 Limit of Detection............................................................................246 5.3.4 Limit of Quantitation .......................................................................246 5.3.5 Specificity.........................................................................................247 5.3.6 Range................................................................................................247 5.3.7 Linearity ...........................................................................................247 5.3.8 System Suitability ............................................................................248 5.4 Instrument Qualification ..............................................................................249 5.4.1 Installation Qualification (IQ)..........................................................249 5.4.2 Operational Qualification (OQ) .......................................................250 5.4.3 Performance Qualification (PQ) ......................................................250 5.5 Electronic Records and Signatures ..............................................................250 5.6 Concluding Remarks....................................................................................252 References..............................................................................................................252 237

238

Advances in Chromatography

5.1 INTRODUCTION Chromatographic analyzers are the most widely employed instrumentation in the area of process analysis [1]. This usage is also the case in the pharmaceutical industry, where liquid chromatography is most often the technique of choice for the analysis of substances associated with the manufacture of drug substances and drug products [2]. Historically, pharmaceutical manufacturing has generally consisted of batch processes where analytical tools, including chromatography methods, were used to test a sample of the end product to verify and ensure its quality. The level of quality in a process can also be achieved using on-line process monitoring and control techniques that can improve the efficiency and quality of the product. However, these techniques have not been widely used in the pharmaceutical industry due to the regulatory uncertainty “which may result from the perception that our existing regulatory system is rigid and unfavorable to the introduction of new technologies.” [3] To facilitate the introduction of new and efficient technologies, the Food and Drug Administration (FDA) launched in 2001 the Process Analytical Technology (PAT) initiative. The FDA’s current thinking on PAT was published in a draft guidance PAT — A Framework for Innovative Pharmaceutical Manufacturing and Quality Assurance [3]. According to this guidance, PAT is considered to be a “system for designing, analyzing, and controlling manufacturing through timely measurements (i.e., during processing) of critical quality and performance attributes of raw and in-process materials and processes with the goal of ensuring final product quality.” [3] Although the use of process analytical tools is not a new approach, many of the techniques have been used only on a limited basis by a very small percentage of the pharmaceutical industry [4]. Most of the literature addressing the application of PAT tools in the pharmaceutical industry does not focus on HPLC applications; however, there are a few examples currently available in the literature that serve to illustrate the potential benefit of HPLC as a PAT tool [2, 5–11]. To better understand the benefit of HPLC as a PAT tool, this article presents an overview of the concept of PAT, including a review of the use of chemometrics or multivariate data acquisition and analysis tools to process the information gathered during the manufacturing operation. The use of HPLC techniques in the pharmaceutical industry requires that the instrument and method used have been validated; hence, a review of the regulatory requirements is provided following the PAT discussion. Finally, a review of the Electronic Records and Signatures regulations is included since these regulations apply both to the computer aspect of the HPLC system and to the chemometric component, since the use of PAT tools to acquire information on the process can result in the generation of electronic records.

5.2 PROCESS ANALYTICAL INITIATIVE In August 2002, the Food and Drug Administration launched a new initiative entitled Pharmaceutical cGMPs for the 21st Century: A Risk Based Approach. Some of the main goals of this program are (a) ensure that the manufacture of pharmaceuticals

High-Performance Liquid Chromatography in the Pharmaceutical Industry

239

incorporates the most up-to-date concepts of risk management and quality systems; (b) encourage the use of the latest scientific advantages and technologies; (c) encourage innovation in the pharmaceutical industry by managing the agency’s risk-based approach; and (d) ensure that regulatory review, compliance, and inspection policies are based on state-of-the-art pharmaceutical science [3, 12]. Since 2002, the FDA has published several reports on the progress of their riskbased approach initiative. The final report was just issued in September 2004 [12] and presents the achievements of various groups that were part of this initiative. As part of the agency’s approach to science-based regulation of quality, the PAT initiative is a collaborative effort between the FDA and the pharmaceutical industry to encourage and facilitate the implementation of new technologies. As previously described, PAT is considered by the FDA to be a system for analyzing and controlling the manufacturing process by continuously monitoring its performance. The goal of PAT is to be able to understand and control the manufacturing process by focusing on relevant multi-factorial relationships among materials, processes, and environmental variables. Hence, the PAT focuses on the principles of building quality into the product and process [4]. The concept of continuously monitoring the performance of a process is not new in the chemical industry [13, 14]. However, in the case of the pharmaceutical industry this is a novel approach, since in the pharmaceutical processes samples are generally taken at the end of the manufacturing cycle to ensure the quality of the final product. The concept of PAT goes beyond the primary process measurements (e.g., temperature and pressure) which are currently being monitored in the pharmaceutical process. The PAT extends to secondary measures of process performance and batch quality. These secondary measures can be more complex techniques, and their development and application might not justify using these techniques. Some of the general advantages of using PAT include [3, 4, 13, 14]: • • •

• • • • •

Reduce production cycle times by using on-line, in-line, or at-line process measurements. Consider the possibility of real time release. Increase automation to improve operator safety, reduce human error, and reduce the need for operator interactions with the process. This can be a valuable advantage in the case of sterile operations. Facilitate continuous processing to improve efficiency and manage variability. Lower costs. Increase efficiency and consistency between batches. Enhance process understanding, resulting in a reduction of rejects and reprocessing. Facilitate continuous processing to improve efficiency and manage variability. In the case of HPLC applications, this advantage applies whether HPLC is used on-line or off-line.

In addition, the requirement to manufacture multiple conformance for drug products and active pharmaceutical ingredients subject to premarket approval might

240

Advances in Chromatography

be eliminated, as indicated in the recent FDA guidance, Process Validation Requirements for Drug Products and Active Pharmaceutical Ingredients Subject to PreMarket Approval [15]: Advanced pharmaceutical science and engineering principles and manufacturing control technologies can provide a high level of process understanding and control capability. Use of these advanced principles and control technologies can provide a high assurance of quality by continuously monitoring, evaluating, and adjusting every batch using validated in-process measurements, tests, controls, and process endpoints. For manufacturing processes developed and controlled in such a manner, it may not be necessary for a firm to manufacture multiple conformance batches prior to initial distribution.

There are several tools within the pharmaceutical industry that enable a scientific risk-based management of the development and manufacturing processes. Within the PAT framework, these tools provide the means to acquire and analyze process information to increase process understanding, to achieve control of the process, and facilitate process improvement efforts. The FDA categorized these tools as follows [3]: (a) multivariate data-acquisition and analysis tools; (b) process analytical chemistry tools; (c) process and endpoint monitoring and control tools; and (d) continuous improvement and knowledge management tools.

5.2.1 MULTIVARIATE DATA ACQUISITION TOOLS

AND

PROCESS ANALYSIS

The multivariate data acquisition and process analysis tools focus on the selection of optimal conditions of the manufacturing process by the identification of the variables that have an impact on the process and the existing relationships between these variables. Some of the tools that can be used for this effort include multivariate mathematical approaches and statistical design of experiments [16] that, by building an experiment matrix and changing the values of the variables, allow the identification of those variables that have an impact on the process and the relationships between these variables. Some of the typical experimental design techniques include full factorial, fractional factorial, and Placket-Burman designs. Other approaches include response surface methodologies that develop a simple polynomial expression to replace the complex functional relationships between the process variables [16, 17]. The information acquired can be used to build a process model, increasing the understanding of the process and identifying the variables that could be critical to the quality of the product, as well as identifying the potential failure modes that could have a negative impact on the product.

5.2.2 PROCESS ANALYTICAL TOOLS The process analytical tools used to measure the various process measurements have evolved over time. Initially, process measurements consisted of pH, pressure, and temperature. Technology available today provides the possibility of measuring

High-Performance Liquid Chromatography in the Pharmaceutical Industry

241

Computer

Process parameters (e.g. flows)

Process equipment

Auto-sampling

HPLC

FIGURE 5.1 Typical HPLC system arrangement.

composition and physical attributes. The measurements can be classified into five different groups [3]: 1. Off-line in a laboratory 2. At-line in the production area (close to the manufacturing process) 3. On-line, where the measurement system is connected to the process via a diverted sample stream (the sample can be discarded or returned to the process after measurement) 4. In-line, where process stream may be disturbed 5. Noninvasive, where there is no contact with the material and the process stream is not disturbed Off-line and at-line measurements, for the purposes of the PAT framework include time delays for analysis, statistical sampling errors, and physical sampling errors. The most desirable technology is the noninvasive sensor, where real-time measurements are obtained without disturbing the process stream. Among the techniques that have gained prominence are near infrared (NIR), Raman and chemical imaging technologies [14, 18]. HPLC techniques are not considered noninvasive and are includedf under the first three groups of instruments. However, depending on the application, HPLC techniques providing on-line measurements could be a better analytical technique to monitor and control the manufacturing process. Examples are currently available in the literature that serve to illustrate the potential benefit of HPLC as a PAT tool [2, 5–11]. The typical system arrangement given by these examples is shown in Figure 5.1. In this arrangement, the data obtained from the HPLC is fed to a computer together with other process variables in order to monitor the state of the process and actively manipulate it by changing the process parameter values.

242

Advances in Chromatography

Several options are available for taking measurements; however it is important to realize that the key emphasis of PAT is not so much how to collect data or what kind of instrumentation is used. The key emphasis is on what data should be collected, what is done with the data, and what conclusions can be made using the data [4].

5.2.3 PROCESS TOOLS

AND

ENDPOINT MONITORING

AND

CONTROL

The use of real-time or near real-time analysis and measurement tools can generate large amounts of data. The question then becomes what information can be extracted from the data collected. The data can be analyzed with the help of chemometrics tools, where chemometrics is defined as the use of multivariate data analysis and mathematical tools to extract information from chemical data [14]; in doing so, chemometrics provides real-time information from data [19]. The use of multivariate analysis tools serves three purposes: First, the tools monitor the state of the processes. Second, the tools use a mathematical model to manipulate the process based on the data analyzed. Third, the data and mathematical models generated can be used to optimize the process. A review by Gabrielsson, Lindberg, and Lundstedt [20] cites over 100 references applying multivariate methods to experimental design, process control and optimization in the chemical industry. A review of chemometric analysis and chromatography is given by Synovec et al. [21] The use of statistical process control tools relies on rigorous statistical principles to measure a given attribute. Traditionally, industries used single-variable process control to monitor the state of the process. However, in complex processes such as those encountered in the pharmaceutical industry, the use of multivariate statistical process control tools could be more appropriate. In an example given by Kourti [22], the use of single-variable process control tools can clearly obtain an incorrect conclusion. Furthermore, single-variable statistical process control tools can be useful for analyzing processes with a couple of variables but, for processes with a large number of variables, they may not be practical. A large number of multivariate statistical tools can be used with some of them including: principal component analysis (PCA), which provides the means to get an overview of large quantities of data, detect trends, groups, and outliers [20, 23]; multiway principle component analysis (MPCA), that takes into account the order in which the data was collected [4, 24]; partial least squares (PLS) [20, 25]; and multiway partial least squares [24]. Although there exist many applications of chemometrics to chromatographic data, most of the research has focused on spectroscopic data [21]. The main reason for this emphasis is that chromatographic data exhibit uncertainty due to variations in instrument performance, an uncertainty observed as a shift in retention times of chromatographic peaks between chromatograms. Several publications try to address this issue using PCA [26] and least squares regression [27]. The information obtained from the use of multivariate statistical tools can be used together with a mathematical algorithm to control and manipulate the process. Once the critical material and process attributes have been identified and the

High-Performance Liquid Chromatography in the Pharmaceutical Industry

243

mathematical relationships have been developed, optimization tools can be utilized to improve the performance of the process. A review of various optimization methods is given by Wehrens and Buydens [17].

5.2.4 CONTINUOUS IMPROVEMENT MANAGEMENT TOOLS

AND

KNOWLEDGE

The analysis of the data collected during the history of the product can justify proposals for post-approval changes submitted to the FDA. The data can be stored and managed with electronic databases and, under the PAT framework, large amounts of data will be generated and the need to validate these databases must be evaluated according to the intended use of the information. During evaluation, the implications of the Code of Federal Regulations (CFR) Title 21 Part 11 regulations [28] need to be considered regarding the requirements of managing electronic records.

5.3 METHOD VALIDATION Whether the analytical methods used within the PAT framework are used for inprocess controls or real-time release, these methods must be validated. In the pharmaceutical industry, analytical method validation is conducted to ensure that the methods used to analyze raw materials, in-process samples or finished product give consistent results in order to guarantee the safety and purity of the final drug product. The regulatory requirements to validate analytical methods for drug products in the U.S. are included in the following sections in the Code of Federal Regulations [28]: Section 211.165(e): The accuracy, sensitivity, specificity, and reproducibility of test methods employed by the firm shall be established and documented. Such validation and documentation may be accomplished in accordance with 211.194(a)(2). Section 211.166(a)(3): The written program shall be followed and shall include…reliable, meaningful and specific test methods. Section 211.194(a)(2): A statement of each method used in testing of the sample. The statement shall include the location of data that establish the methods used in the testing of the sample that meet proper standards of accuracy and reliability as applied to the product tested. The suitability of all testing methods shall be verified under actual conditions of use. In the case of active pharmaceutical ingredients (API), the requirements for validating analytical methods are specified in the International Conference on Harmonisation (ICH) Guideline Q7A [29]: Section 12.80: Analytical methods should be validated unless the method employed is included in the relevant pharmacopoeia or other recognized standard reference. The suitability of all testing methods used should nonetheless be verified under actual conditions of use and documented.

244

Advances in Chromatography

Section 12.81: Methods should be validated to include consideration of characteristics included within the ICH guidelines on validation of analytical methods. The degree of analytical validation performed should reflect the purpose of the analysis and the stage of the API production process. These requirements also apply to chromatographic methods which, in the pharmaceutical industry, can be used to identify, quantify, or report impurities in drugs. The amount of validation required might vary depending on the stage of development of a product. In a similar manner, the amount of information required by the regulatory agencies varies according to the stage of development. A review of the various requirements according to the stage of development is given elsewhere [30]; however, once the product reaches the New Drug Application (NDA) stage, full validation of the chromatographic methods is expected. Several guidelines are useful for pharmaceutical applications. The ICH Guideline Q2A [31] defines the validation requirements and provides guidance as to which requirements are expected for each of the four categories of tests: identification, impurity quantitative test, impurity limit test, and assay (see Table 5.1). ICH Guideline Q2B [32] provides guidance on how to perform the method validation. The FDA published a draft Guidance for Industry on Analytical Procedures and Method Validation [33] that provides guidance on the requirements for method validation. The definition of each parameter and recommendations on their implementation are given in the FDA’s guidance on Validation of Chromatographic Methods [34]. The typical validation parameters in the ICH guidelines are the same as in the FDA with one exception — robustness [35–37]. Data obtained from robustness studies is not usually submitted to the FDA but is recommended to be included as part of method validation [34]. Additional criteria not listed in Table 5.1 that can be used during method validation include ruggedness [35, 38–40] and solution stability [34, 39, 41]. The acceptance criteria for each parameter should be established prior to starting method validation activities and should be appropriate for the intended use of the analytical method. However, since neither ICH nor FDA provides guidance on acceptance criteria, these criteria are generally determined by the pharmaceutical

TABLE 5.1 Typical Method Validation Requirements Based on ICH Q2A [31] Method Validation Requirement

Identification Test

Impurity Quantitative Test

Precision Accuracy Limit of detection Limit of quantitation Specificity Range Linearity

— — — — X — —

X X may be needed X X X X

Impurity Limit Test

Assay

— — X — X — —

X X — — X X X

High-Performance Liquid Chromatography in the Pharmaceutical Industry

245

TABLE 5.2 Recommended Values and Examples of Acceptance Criteria Method Validation Requirement

ICH Recommendations [32]

Acceptance Criteria (examples)

Precision

Minimum of 9 determinations covering the specified range of the procedure (3 concentrations with 3 replicates of each concentration).

Accuracy

Minimum of 9 determinations over a minimum of 3 concentration levels covering the specified range.

Typically use the relative standard deviation (RSD). The larger the RSD value the less precise the method. For examples see [41–47]. For number of samples typically used see [32, 34, 41–43, 48–50] Mean recovery should be 100 ± 2% at each concentration over the range of 80–120% [41]. For number of aliquots typically used see [34, 41, 44, 47, 49–51]

Limit of detection

Signal-to-noise ratio between 3:1 and 2:1 is generally considered acceptable. Signal-to-noise ratio of 10:1 is generally considered acceptable [32]. — Range of 80–120% of test concentration for assay of drug substance or finished product; or 70–130% of test concentration unless a wider or more appropriate range based on the nature of the dosage form is justified. —

Limit of quantitation

Specificity Range

Linearity

Other criteria is available [42, 52]

[41, 42, 53, 54] [38, 41, 42, 55–58]

[38, 41, 42, 47, 50, 56, 59]

company. Examples of acceptance criteria found in the literature and ICH recommendations are given in Table 5.2. It is important to clarify that in the case of process measurements, the method validation elements used must be tailored and prioritized with knowledge of the actual process that the method is supposed to control. For example, robustness, which as stated before is not usually submitted to the FDA, is a validation element that is of fundamental importance when methods are used in process control.

5.3.1 PRECISION Precision demonstrates the ability of an analytical method to produce consistent results. Precision is the degree of reproducibility among a set of individual results, including any internal variations inherent in an analytical method. Precision does not ensure that a result is the correct value — what precision provides is the extent

246

Advances in Chromatography

of agreement between multiple results. Precision can be divided into three categories: repeatability, intermediate precision, and reproducibility. Repeatability demonstrates the precision of the instrument itself and is obtained when the analysis is performed using the same instrumental operating conditions over a short period of time. Intermediate precision evaluates the data generated over an extended period of time. The intent is to run the analysis under different operating conditions (e.g., different equipment, different operator, different supplier of reagents). Intermediate precision ensures that an analytical method is precise despite random events affecting operating conditions within the same laboratory. In contrast to intermediate precision, reproducibility analyzes the performance of an analytical method in more than one laboratory.

5.3.2 ACCURACY Accuracy is the closeness of agreement between the accepted value, either as a conventional true value or a reference value, and the value found by analysis. Accuracy can be determined by analyzing a sample of known concentration and comparing the results to the true value [41, 49, 60, 61]. Accuracy can also be determined by comparing the results to an existing well-characterized method whose accuracy is defined [32, 41, 62]. A third method to determine accuracy is to use spiking/recovery studies [41, 48, 49] where the sample is analyzed prior to spiking and then reanalyzed after the addition of a known amount of analyte. The difference between these measurements is called marginal or surrogate recovery and methods with marginal recovery values significantly different than 1 are not considered accurate [32].

5.3.3 LIMIT

OF

DETECTION

The limit of detection (LOD) is the lowest concentration of analyte in a sample that can be detected but not necessarily quantitated under the experimental conditions of the method. In practice, this value is the lowest concentration of analyte that can be distinguished from the blank with a stated degree of confidence [42]. The typical methods recommended by ICH to determine LOD are [32]: 1. Signal-to-noise ratio, where the LOD can be expressed as a concentration at a specified signal-to-noise ratio obtained from samples spiked with analyte 2. Standard deviation of the response and the slope of the calibration curves at levels approximating the LOD

5.3.4 LIMIT

OF

QUANTITATION

The limit of quantitation (LOQ) is the lowest concentration of analyte in a sample that can be quantitated and is the lowest amount of analyte in a sample that satisfies all other method validation requirements. Similarly to LOD, the typical methods recommended by ICH to determine LOQ are [32]:

High-Performance Liquid Chromatography in the Pharmaceutical Industry

247

1. Signal-to-noise ratio, where the LOQ can be expressed as a concentration at a specified signal-to-noise ratio obtained from samples spiked with analyte 2. Standard deviation of the response and the slope of the calibration curves at levels approximating the LOQ

5.3.5 SPECIFICITY Specificity is the ability to assess unequivocally the analyte in the presence of components that can be present in the sample matrix. In the pharmaceutical industry, a sample matrix can include impurities, degradation products, excess raw materials, or excipients. The chromatographic method should be specific and sensitive as required for all known relevant degradation products or impurities. Other potential sample components can be generated by exposing the analyte to stress conditions (e.g., temperature, acid, light, base, oxidant, humidity) [41]. Several tests can be used to determine if a method is specific. ICH recommends that the response of the analyte in a solution containing all impurities, degradation products, or excess raw materials is compared with the response of the solution containing only the analyte, accomplished by spiking the analyte with appropriate levels of impurities, degradation products, or excess raw materials. If impurity or degradation standards are not available, specificity can be demonstrated by comparing the results of the sample containing the impurities or degradation products with a well-characterized procedure, such as those found in compendia or other validated methods [32].

5.3.6 RANGE Range is the interval between and including the upper and lower limits for which the analytical method demonstrates that it satisfies all other method validation requirements. The range can be defined as the set of sample concentrations where the sensitivity of the method is essentially constant [38]. The range is established by confirming that the analytical method provides an acceptable linearity, accuracy, and precision at the extremes of the specified range of the method and should reflect the purpose of the analytical method. “The range selected for validation should not be unrealistically wide, as this may lead to rejection of a method which is really quite suitable for the intended purpose.” [42] For this reason, a balance exists between practical restrictions and theoretical possibilities.

5.3.7 LINEARITY Linearity is the ability of an analytical method to obtain results that are, by a welldefined mathematical transformation, directly or indirectly proportional to the concentration of analyte in samples. Linearity should be demonstrated within the given range and be appropriate for the intended use of the method. Linearity is usually expressed in terms of the square root of the variance (correlation coefficient) around the slope of the regression line and demonstrates that the relationship between the result and sample concentration is continuous and reproducible.

248

Advances in Chromatography

Linearity is typically demonstrated by performing duplicate sample preparations over the specified range. As mentioned previously, the specified range should reflect the intended use of the analytical method. ICH recommends a range from 80 to 120% of test concentration for assays of drug substance or finished product, or from 70 to 130% of test concentration unless a wider or more appropriate range based on the nature of the dosage form is justified [32].

5.3.8 SYSTEM SUITABILITY The system suitability tests for chromatographic methods are based on the concept that the equipment, electronics, analytical operations, and samples to be analyzed constitute an integral system that can be evaluated as such [32]. Through a series of tests and parameters, the system suitability ensures that the total system is functioning adequately and assists in providing assurance of the reliability of the results [34]. If the criteria used to determine the suitability of a system were chosen correctly, the criteria will fail to be met just before the system will generate data that is not acceptable. Hence, the system suitability tests should always be conducted before starting an analysis. Depending upon the length of a run or importance of the sample results, system suitability can be performed during and following the analysis. Typical system suitability parameters for chromatographic methods are listed below: •

•

•

•

•

Capacity factor (k): The capacity factor (k) is a measure of where the peak of interest is located with respect to the void volume. The peak should be well-resolved from other peaks and the void volume, generally k > 2 [34]. Precision/injection repeatability: Expressed as RSD. the precision/injection repeatability indicates the performance of the system at the time the sample is analyzed. Replicate injections of a standard preparation are compared to ensure that requirements for precision are met [34]. One recommendation is to use five replicate injections of the analyte to calculate the RSD if the requirement is less than 2.0% or six injections if the RSD requirement is greater than 2.0% [40]. Relative retention: The relative retention () is a measure of the relative location of two peaks. This parameter is not essential as long as the resolution is stated [34]. Resolution: The resolution is a measure of how well two peaks are separated. The resolution should be greater than 2 between the peak of interest and the closest eluting potential interfering peak [34]. While achieving a resolution greater than 2 may not always be possible, resolving peaks to some degree should be possible, perhaps by changing method parameters. The acceptable resolution for a method should be evaluated on a case-bycase basis. Tailing factor (T): The tailing factor is a measure of peak symmetry and, for perfectly symmetrical peaks, the value is 1. As peak asymmetry

High-Performance Liquid Chromatography in the Pharmaceutical Industry

•

249

increases, precision becomes less reliable. A value of T < 2 is recommended [34]. Theoretical plates (N): The number of theoretical plates is a measure of column efficiency, that is, how many peaks can be located per unit runtime of the chromatogram. The theoretical plate counts depend on elution time but, in general, should be greater than 2000 [34].

The number of system suitability tests required for a particular method will depend on the purpose of the test method. The FDA recommends the following minimum suitability tests [34]: •

•

For dissolution or release-profile tests using external standard methods (impurity quantitative test or major component test): capacity factor, tailing factor, and RSD are minimum recommended system suitability tests. For acceptance, release, stability, or impurities/degradation methods using external or internal methods (impurity quantitative test or major component test): capacity factor, tailing factor, RSD, and resolution are minimum recommended system suitability tests.

5.4 INSTRUMENT QUALIFICATION In the PAT framework, the instruments used as analytical tools must be qualified to provide documented evidence that the equipment is capable of consistently operating within the established design limits and is acceptable for its intended use. According to the FDA, “equipment used in manufacturing, processing…of a drug product shall be of appropriate design…and suitably located to facilitate operations for its intended use” (CFR 211.63) [28]. In addition, as indicated in the ICH Guideline Q7A, “appropriate qualification of analytical equipment should be considered before starting validation of analytical methods” (Section 12.82) [29]. Therefore, analytical instrumentation that is used to test raw materials, in-process samples and final drug products should fully comply with the FDA equipment requirements. The intent of qualification is to ensure that the instrument is operating properly, thus ensuring robust and reliable performance. In other words, qualification provides assurance to the validity of the data generated by an analytical instrument [63, 64]. The qualification process accomplished via a documented procedure can be divided into three phases. These phases are executed in the following order: installation qualification (IQ), operational qualification (OQ), and performance qualification (PQ). There is currently a developing guidance from the ASTM Committee E 55 [3] that is embraced by the FDA that is taking into consideration the attributes of online validation and will provide recommendations regarding the qualification practices for online instruments.

5.4.1 INSTALLATION QUALIFICATION (IQ) Installation qualification (IQ) is verification that the instrument installed in the laboratory meets the design and installation specifications. Typical information

250

Advances in Chromatography

included as part of the IQ is manufacturer; model and serial number; dimensions (physical description) of the instrument; materials of construction in contact with mobile or stationary phase; utility specifications (such as electrical requirements); assembly check; initial calibration records; spare parts list; and location of supporting documentation (such as operating manuals). The majority of these items should be specified in the design or by the instrument manufacturer. As part of the IQ, the design specifications are compared with the actual results obtained from the instrument installed in the laboratory.

5.4.2 OPERATIONAL QUALIFICATION (OQ) Operational qualification (OQ) is verification that the instrument operates within the full range established by the design specifications and includes tests for each function of the instrument. Ideally, OQ should simulate actual operating conditions. OQ tests should be repeated a sufficient number of times to assure reliable results. “OQ verifies key aspects of instrument performance without the aspects of any contributory effects that could be introduced by a method.” [54] OQ testing should include verification that all alarms and interlocks are functional; test control and functionality of unit operation; and verification that the instrument operates correctly over the entire specified range. The purpose of the activities listed above is to ensure that the instrument is capable of meeting design specifications for accuracy, linearity and precision [54].

5.4.3 PERFORMANCE QUALIFICATION (PQ) Performance qualification (PQ) is verification that the ongoing performance of the instrument remains within the designed specifications for its actual use and demonstrates that the instrument performs appropriately for routine use. The PQ can consist of analyzing known standards and evaluating the results vs. system suitability requirements (see text on system suitability for details). PQ testing should be based on good science and the intended use of the instrument. Theoretically, a distinction exists between OQ and PQ. OQ verifies that all of the individual units (parts or modules) of an instrument operate properly while PQ verifies that the entire instrument as a system performs properly. In practice, aspects of OQ and PQ may be evaluated together, particularly when the test can be more easily executed at the system level. A useful Guidance on Equipment Qualification of Analytical Instruments: High Performance Liquid Chromatography (HPLC) [64] can be used to determine the parameters to be evaluated during OQ and PQ. Additional sources addressing the qualification of instruments are also available [63, 65–69].

5.5 ELECTRONIC RECORDS AND SIGNATURES In the PAT framework, the data generated using the analytical tools can be considered an electronic record and is regulated under the CFR Part 11 regulations [28]. The scope of Part 11 regulations is to provide criteria under which the FDA would accept

High-Performance Liquid Chromatography in the Pharmaceutical Industry

251

the use of electronic records and signatures, including considering them equivalent to paper records and handwritten signatures. According to the Part 11 regulations, an electronic record means any “combination of text, graphics, data, audio, pictorial, or other information representation in digital form that is created, modified, maintained, archived, retrieved, or distributed by a computer system.” [28] Based on this definition, electronic data generated by a chromatographic system to meet predicate rules must be compliant with the Part 11 regulations. Since its publication in 1997, the cost and technological limitations associated with Part 11 regulations have prompted significant discussions between the regulated industry and the FDA regarding the interpretation and implementation of the regulations. To address industry concerns, the FDA published numerous guidances on the subject. In August 2003, the FDA issued the Guidance for Industry: Part 11, Electronic Records; Electronic Signatures — Scope and Application [70]. The purpose of this guidance was to define more clearly the intent of the regulations, in particular to provide a clear definition of the records covered under the rule. The guidance also includes specific considerations regarding audit trail, legacy systems, copies of records and record retention. The FDA is currently revising the Part 11 regulations and an amendment will be issued for public comment in 2005 [12]. Part 11 regulations also address the requirement to validate the computer systems used to generate and manage the data. The discussion of instrument qualification in the literature generally does not include the need to validate the computer aspect of the instrument (software and computer hardware), which should be considered an important part of the qualification package. The need to validate computer systems in the regulated industry has been an FDA requirement (e.g., cGMP for finished pharmaceuticals, 21 CFR Part 211) since 1978 and a Guide to Inspection of Computerized Systems in Drug Processing [71] has been available since 1983. However, the role of computer validation has had a significant impact for the FDA regulated industry since the Part 11 regulations became effective in 1997. The decision of which systems to validate, as suggested by the FDA, should consider the impact on the ability to meet predicate rule requirements and “based on a justified and documented risk assessment and determination of the potential of the system to affect product quality and safety, and record integrity.” [70] The validation of the instrument’s computer system includes both the hardware and software. The computer system can be included in the instrument’s qualification package, as described in the previous section. The intent of software validation, similar to the instrument qualification, is to confirm “by examination and provision of objective evidence that software specification conforms to user needs and intended uses, and that the particular requirements implemented through the software can be consistently fulfilled.” [72] The concept of software validation in the PAT framework also applies to the algorithm used to manipulate the process. Several guidances are available that address the requirements of software validation: General Principles of Software Validation; Final Guidance for Industry and FDA Staff [73], Guidance for Industry: Computerized Systems Used in Clinical Trials [72] and Good Automated Manufacturing Practice (GAMP) Guide for Validation of Automated Systems [74]. The FDA’s General Principles of Software

252

Advances in Chromatography

Validation; Final Guidance for Industry and FDA Staff [73] applies specifically to the validation of medical device software or the validation of software used to design, develop, or manufacture medical devices. However, the information included in this guidance provides a comprehensive discussion on the activities involved in the validation of software and can be a valuable tool and starting point for a computer software initiative for other FDA regulated industries. Other references are also available [66].

5.6 CONCLUDING REMARKS The PAT initiative that was launched by the FDA as part of the new initiative entitled Pharmaceutical cGMPs for the 21st Century: A Risk-Based Approach provides the incentives for the pharmaceutical industry to adopt new and emerging technologies. This approach changes the relationship between the agency and the industry and represents a focus on science and engineering as tools used in controlling the quality of the pharmaceutical product. As PAT tools are implemented, substantial benefits will be obtained by the pharmaceutical sector.

REFERENCES 1. Workman, J., Koch, M. and Veltkamp, D.J. Process analytical chemistry. Analytical Chemistry, 2003, 75(12), 2859–2876. 2. Schweitz, L., Fransson, M., Karlsson, L., Torstensson, A. and Johansson, E. On-line process control of gradient elution liquid chromatography. Analytical Chemistry, 2004, 76(16), 4875–4880. 3. FDA. Guidance for Industry: PAT — A Framework for Innovative Pharmaceutical Manufacturing and Quality Assurance. Food and Drug Administration, Washington, DC, 2004. 4. Balboni, M.L. Process analytical technology: Concepts and principles. Pharmaceutical Technology, October 2003, 54–66. 5. Turner, C., Gregory, M.E. and Thornhill, N.F. Closed-loop control of fed-batch cultures of recombinant Eschericia colu using on-line HPLC. Biotechnology and Bioengeneering, 1994, 44(7), 819–829. 6. Hsieh, F.Y.L., Tong, X., Wachs, T., Ganem, B. and Henion, J. Kinetic monitoring of enzymatic reactions in real time by quantitative high performance liquid chromatography-mass spectrometry. Analytical Biochemistry, 1995, 229, 20–25. 7. Liu, Y., Wang, F. and Lee, W. On-line monitoring and controlling system for fermentation processes. Biochemical Engineering Journal, 2001, 7, 17–25. 8. Larson, T.M., Gawlitzek, M., Evans, H., Albers, U. and Garcia, J. Chemometric evaluation of on-line high-pressure liquid chromatography in mammalian cell cultures: Analysis of amino acids and glucose. Biotechnology and Bioengeneering, 2002, 77(5), 553–563. 9. Dantan, N., Frenzel, W. and Küppers, S. Flow injection analysis coupled with HPLC and CE for monitoring chemical production processes. Chromatographia, 2001, 54(3/4), 187–190. 10. Cooley, R.E. and Stevenson, C.E. On-line HPLC as a process monitor in biotechnology. Process Control and Quality, 1992, 2, 43–53.

High-Performance Liquid Chromatography in the Pharmaceutical Industry

253

11. Doyle, M.J. and Newton, B.J. Chromatography with on-line HPLC and ion chromatography for process control. Chromatography and Separation Technology, 2002, 22, 9–12. 12. FDA. Pharmaceutical cGMPs for the 21st Century — A Risk-Based Approach: Final Report. Food and Drug Administration, Washington, DC, 2004. 13. Afnan, A.M. PAT: What’s in a name? Journal of Process Analytical Technology, 2004, 1(1), 8–9. 14. Yu, L.X., Lionberger, R.A., Raw, A.S., D’Costa, R., Wu, H. and Hussain, A.S. Applications of process analytical technology to crystallization processes. Advanced Drug Delivery Reviews, 2004, 56(3), 349–369. 15. FDA. Process Validation Requirements for Drug Products and Active Pharmaceutical Ingredients Subject to Pre-Market Approval. Food and Drug Administration, Washington, DC, 2004. 16. Box, G.E.P., Hunter, W.G. and Hunter, S.J. Statistics for Experimenters. An Introduction to Design, Data Analyisis, and Model Building, John Wiley & Sons, New York, 1978. 17. Wehrens, R. and Buydens, L.M.C. Classical and nonclassical optimization methods. Encyclopedia of Analytical Chemistry, John Wiley & Sons, Chichester, 9678–9689. 18. Ciurczak, E.W. The process analytical technologies initiative: What is it, and where does spectroscopy come in? Spectroscopy, 2003, 18(9), 20–21. 19. Workman, J. The state of multivariate thinking for scientists in industry: 1980–2000. Chemometrics and Intelligent Laboratory Systems, 2002, 60, 13–23. 20. Gabrielsson, J., Lindberg, N. and Lundstedt, T. Multivariate methods in pharmaceutical applications. Journal of Chemometrics, 2002, 16, 141–160. 21. Synovec, R.E., Prazen, K.J., Fraga, C.G. and Bruckner, C.A. Chemometric analysis of comprehensive two-dimensional separations. Advances in Chromatography, Marcel Dekker, New York, 2003. 22. Kourti, T. Process analytical technology and multivariate statistical process control: Wellness index of product and process. Part 1. Journal of Process Analytical Technology, 2004, 1(1), 13–19. 23. Joe Qin, S. Statistical process monitoring: Basics and beyond. Journal of Chemometrics, 2003, 17(8-9), 480–502. 24. Normikos, P. and MacGregor, J. Multiway partial least squares in monitoring batch processes. Chemometrics and Intelligent Laboratory Systems, 1995, 30, 97–108. 25. Hubert, M. and Vanden Branden, K. Robust methods for partial least squares regression. Journal of Chemometrics, 2003, 17(10), 537–549. 26. Malmquist, G. and Danielsson, R. Alignment of chromatographic profiles for principal component analysis: A prerequisite for fingerprinting methods. Journal of Chromatography A, 1994, 687(1), 71–88. 27. Bahowick, T.J. and Synovec, R.E. Correlation of quantitative analysis precision to retention time precision and chromatographic resolution for rapid, short-column analysis. Analytical Chemistry, 1995, 67(3). 28. FDA. Code of Federal Regulations (21 CFR). Food and Drug Administration, Washington, DC, 1997. 29. ICH. Good Manufacturing Practice Guide for Active Pharmaceutical Ingredients: Q7A. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2000. 30. Dantus, M.M. and Wells, M.L. Regulatory issues in chromatographic analysis in the pharmaceutical industry. Journal of Liquid Chromatography & Related Technologies, 2004, 27(7-9), 1413–1442.

254

Advances in Chromatography

31. ICH. Validation of Analytical Procedures: Methodology Q2A. International Conference of Harmonisation of Technical Requirments for Registration of Pharmaceutical for Human Use, 1994. 32. ICH. Validation of Analytical Procedures: Methodology Q2B. International Conference of Harmonisation of Technical Requirements for Registration of Pharmaceutical for Human Use, 1996. 33. FDA. Guidance for Industry: Analytical Procedures and Method Validation. Chemistry, Manufacturing, and Controls Documentation. Draft Guidance: Validation of Chromatographic Methods. Food and Drug Administration, Washington, DC, 2000. 34. FDA. Reviewer Guidance: Validation of Chromatographic Methods. Food and Drug Administration, Washington, DC, 1994. 35. Barwick, V.J. Review of sources of uncertainty in gas chromatography and high performance liquid chromatography. LGC/VAM, 1998. 36. Virrlichie, J.L. and Ayache, A. A ruggedness test model and its application for HPLC method validation. STP Pharma Practiques, 1995, 5(1), 49–60. 37. Hokanson, G.C. A life cycle approach to the validation of analytical methods during pharmaceutical product development. Part II. Changes and the need for additional validation. Pharmaceutical Technology, October 1994, 92–100. 38. Bruce, P., Minkkinen, P. and Riekkola, M.L. Practical method validation: Validation sufficient for an analysis method. Mickrochimica Acta, 1998, 128(1-2), 93–106. 39. Jenke, D.R. Chromatographic method validation: a review of current practices and procedures. III. Ruggedness, revalidation and system suitability. Journal of Liquid Chromatography & Related Technologies, 1996, 19(12), 1873–1891. 40. USP. United States Pharmacopeia: USP 26. United States Pharmacopeial Conventions, Washington, DC, 2003. 41. Shabir, G.A. Validation of high-performance liquid chromatography methods for pharmaceutical analysis: Understanding the differences and similarities between validation requirements of the US Food and Drug Administration, the US Pharmacopeia and the International Conference on Harmonization. Journal of Chromatography A, 2003, 987, 57–66. 42. Jenke, D.R. Chromatographic method validation: A review of current practices and procedures. II. Guidelines for primary validation parameters. Journal of Liquid Chromatography & Related Technologies, 1996, 19(5), 737–757. 43. Buick, A.R., Doig, M.V., Jeal, S.C., Land, G.S. and McDowall, R.D. Method validation in the bioanalytical laboratory. Journal of Pharmaceutical and Biomedical Analysis, 1990, 8(8-12), 629–637. 44. Shah, V.P., Midha, K.K., Dighe, S., McGilveray, I.J., Skelly, J.P., Yacobi, A., Layloff, T., Viswanathan, C.T., Cook, C.E., McDowall, R.D., Pittman, K.A. and Spector, S. Analytical methods validation: Bioavailability, bioequivalence and pharmacokinetic studies. European Journal of Drug Metabolism and Pharmacokinetics, 1991, 16(4), 249–255. 45. Ziakova, A. and Brandsteterová, E. Validation of HPLC determination of phenolic acids present in some Lamiaceae family plants. Journal of Liquid Chromatography & Related Technologies, 2003, 26(3), 443–453. 46. Bressolle, F., Bromet-Petit, M. and Audran, M. Validation of liquid chromatographic and gas chromatographic methods: Application to pharmacokinetics. Journal of Chromatography B, 1996, 686, 3–10. 47. Gupta, A. and Mydral, P.B. On-line high-performance liquid chromatography method for analyte quantitation from pressurized metered dose inhalers. Journal of Chromatography A, 2004, 1033(1), 101–106.

High-Performance Liquid Chromatography in the Pharmaceutical Industry

255

48. Badea, I., Ciutaru, D., Lazar, L., Nicolescu, D. and Tudose, A. Rapid HPLC method for the determination of paclitaxel in pharmaceutical forms without separation. Journal of Pharmaceutical and Biomedical Analysis, 2004, 34, 501–50. 49. Ferretti, R., Gallinella, B., La Torre, F. and Turchetto, L. Validated chiral highperformance liquid chromatographic method for the determination of trans-(-)-paroxetine and its enantiomer in bulk and pharmaceutical formulations. Journal of Chromatography B, 1998, 710, 157–164. 50. Galli, V. and Barbas, C. High-performance liquid chromatographic analysis of dextromethorphan, guaifenesin and benzoate in a cough syrup for stability testing. Journal of Chromatography A, 2004, 1048(2), 207–211. 51. Clarke, G.S. The validation of analytical methods for drug substances and drug products in UK pharmaceutical laboratories. Journal of Pharmaceutical and Biomedical Analysis, 1994, 12(5), 643–652. 52. Rabouan, S., Olivier, J.C., Guillemin, H. and Barthes, D. Validation of HPLC Analysis of aspartate and glutamate neurotransmitters following o-phtaldialdehyde-mercaptoethanol derivatization. Journal of Liquid Chromatography & Related Technologies, 2003, 26(11), 1797–1808. 53. Needham, S. Bioanalytical method validation: Example, HPLC/MS/MS bioanalysis of an anti-nerve gas agent drug for Homeland Security. American Pharmaceutical Review, 2003. 54. Grisanti, V. and Zachowski, E. Operational and performance qualification. LCGC North America, 2002, 20(4), 356–362. 55. Castillo, M.A. and Castells, R.C. Initial evaluation of quantitative performance of chromatographic methods using replicates at multiple concentrations. Journal of Chromatography A, 2001, 921, 121–133. 56. Crowther, J.B. Validation of Pharmaceutical Methods. Handbook of Modern Pharmaceutical Analysis, Academic Press, San Diego, 2001, 415–443. 57. Fabre, H. and Altria, K.D. Key points for validating CE methods, particularly in pharmaceutical analysis. LCGC, 2001, 19(5), 498–505. 58. Thompson, M., Ellison, S.L. and Wood, R. Harmonized guidelines for single-laboratory validation of methods of analysis (IUPAC Technical Report). Pure Applied Chemistry, 2002, 74(5), 835–855. 59. Bakshi, M. and Singh, S. Development of validated stability-indicating assay methods — Critical review. Journal of Pharmaceutical and Biomedical Analysis, 2002, 28, 1011–1040. 60. Qi, M., Chen, R., Cong, R. and Wang, P. A validated method for simultaneous determination of piperacillin sodium and sulbactam sodium in sterile powder for injection by HPLC. Journal of Liquid Chromatography & Related Technologies, 2003, 26(4), 665–676. 61. Mouatassim-Souali, A., Tamisier-Karolak, S.L., Perdiz, D., Cargouet, M. and Levi, Y. Validation of a quantitative assay using GC/MS for trace determination of free and conjugated estrogens in environmental water samples. Journal of Separation Science, 2003, 26, 105–111. 62. Jimidar, M., Hartmann, C., Cousement, N., and Massaty, D.L. Determination of nitrate and nitrite in vegetables by capillary electrophoresis with indirect detection. Journal of Chromatography A, 1995, 706, 479–492. 63. Burgess, C., Jones, D.G. and McDowall, R.D. Equipment qualification for demonstrating the fitness for purpose of analytical instrumentation. Analyst, 1998, 123, 1879–1966.

256

Advances in Chromatography

64. Guidance on Equipment Qualification of Analytical Instruments: High Performance Liquid Chromatography (HPLC). LGC/VAM Instrumentation Working Group, 1998. 65. Rathore, A.S., Kennedy, R.M., O’Donnell, J.K., Bomberis, I. and Kaltenbrunner, O. Qualification of a chromatographic column — Why and how to do it. Biopharm International, 2003(March), 30–40. 66. Huber, L. Validation and Qualification in Analytical Laboratory. Interpharm Press, Buffalo Grove, IL, 1999. 67. Huber, L. and Welebob, L. Selecting parameters and limits for equipment operational qualification. Accreditation and Quality Assurance, 1997, 2(7), 316–322. 68. Zanetti, M. and Huber, L. Validating automated analytical instruments to meet regulatory and quality standard regulations. Mickrochimica Acta, 1996, 123(1-4), 23–31. 69. Hall, G. and Dolan, J.W. Performance qualification of LC systems. LCGC North America, 2002, 20(9), 842–848. 70. FDA. Guidance for Industry: Part 11, Electronic Records; Electronic Signatures — Scope and Application. Food and Drug Administration, Washington, DC, 2003. 71. FDA. Guide to Inspection of Computerized Systems in Drug Processing. Food and Drug Administration, Washington, DC, 1983. 72. FDA. Guidance for Industry: Computerized Systems Used in Clinical Trials. Food and Drug Administration, Washington, DC, 1999. 73. FDA. General Principles of Software Validation; Final Guidance for Industry and FDA Staff. Food and Drug Administration, Washington, DC, 2002. 74. GAMP. The Good Automated Manufacturing Practice (GAMP) Guide for Validation of Automated Systems, GAMP 4. ISPE/GAMP Forum, 2001.

6

The Use of Polysaccharide Phases in the Separation of Enantiomers Rodger W. Stringham Chiral Technologies, Inc., West Chester, Pennsylvania

CONTENTS 6.1 6.2

Introduction ..................................................................................................257 Mechanisms of Selectivity...........................................................................260 6.2.1 Effect of Side Chain Substitutions ..................................................262 6.2.2 Spectroscopic and Thermal Studies.................................................263 6.2.3 Analyte Structure Effects.................................................................264 6.2.4 Molecular Modeling.........................................................................265 6.2.5 Mobile Phase Effects .......................................................................265 6.3 Modes of Separation ....................................................................................267 6.3.1 Normal-Phase Mode ........................................................................267 6.3.2 Polar Organic Mode.........................................................................268 6.3.3 Reversed-Phase Mode ......................................................................274 6.3.4 Supercritical Fluid Chromatography ...............................................275 6.4 Mobile Phase Additives ...............................................................................278 6.5 Conclusions ..................................................................................................283 References..............................................................................................................284

6.1 INTRODUCTION Derivatized polysaccharides coated on silica particles comprise the most widely used chiral stationary phases (CSP). According to a recent survey by Francotte [1], approximately 90% of published analytical separations of racemic compounds are accomplished on these CSPs. This survey also indicated that the number of separations on polysaccharide CSPs is increasing while separations on other CSPs are declining. Francotte’s survey includes a wide variety of journals, not just those specific for separations and analytical applications. This approach includes many

257

258

Advances in Chromatography

separations developed by nonanalytical chemists and will tend to favor CSPs already on hand in these laboratories. As such, there may be some skewing toward polysaccharide phases as separations will be reported as only being accomplished on the first CSP that provides a useful separation. Publications in analytical journals will skew away from well-established polysaccharide CSPs toward newer phases and surveys that focus on these journals will yield different results. Mechanistic studies will also skew toward more defined CSPs. None of these surveys include the large number of proprietary compounds separated but never published. Despite the limitations of literature surveys, it is fair to say that derivatized polysaccharides are by far the most widely used CSPs in the separation of enantiomers. The coated CSPs were developed by Okamoto [2–6] and, as described by Okamoto, the developed synthesis consists of two main steps. In the first step, either cellulose or amylose is derivatized with an excess of isocyanate to yield a triscarbamate derivative (Figure 6.1). Isocyanates can be generated from the corresponding amine and phosgene. The derivatization reaction is quite versatile in that it can be applied to a range of polysaccharides and can use a wide variety of isocyanates to yield a large number of potential CSPs. Acid chlorides can be used in place of the isocyanate to yield tris-ester derivatives that also give useful CSPs. The second step is the coating process, where the tris-carbamate chiral polymer is deposited onto macroporous silica. This step is not as straightforward as the derivatization reaction, as it requires solubilization of the chiral polymer, adsorption onto the silica, and removal of solvent. Correct crystalline structure was reported [7] as crucial to the ability of derivatized polysaccharides to generate useful separations. Solubilization destroys any inherent crystallinity and the challenge becomes to precipitate the polymer onto the silica in a homogenous tertiary conformation. The coating process is not trivial. The solvent selected for coating will affect performance [8, 9], as will procedures used to prepare the silica, solvent evaporation temperatures and pressures, and washing and drying procedures. Coating procedures have been painstakingly developed and a number of cellulose and amylose CSPs are manufactured by Daicel Chemical Industries, Ltd. The structures of the CSPs produced by Daicel are shown in Figure 6.2. The names of these phases are registered trademarks of Daicel.

OH −

OCONHR R N C O

O O HO

–O

OH

O

Isocyanate

RHNOCO

O O OCONHR

n Amylose or Cellulose

Chiral polymer

n

FIGURE 6.1 Schematic of the derivatization step in the synthesis of polysaccharide CSPs.

The Use of Polysaccharide Phases in the Separation of Enantiomers

259

O O CHIRALCEL OJ

OR O

–O

O

R=

OH

RO

N H CHIRALCEL OD

OR

N H

O n

Cellulose

CHIRALCEL OB

CHIRALCEL OG Cl

O

O CHIRALCEL OK

N H CHIRALCEL OF O

O CHIRALCEL OA

N H CHIRALCEL OC

OR –O

RO

O OR

Amylose

R=

H N

O

O n

N H CHIRALPAK AD

O CHIRALPAK AS

FIGURE 6.2 Structures of commercially available polysaccharide CSPs.

These CSPs were selected for commercialization based on results of performance testing for a group of racemic compounds. Selectivity values of probe compounds were summed and the CSPs with the highest totals were developed. This approach was intended to provide CSPs with the broadest application range and has apparently been successful. According to Francotte’s survey [1], 39% of compounds are separated on CHIRALPAK AD, 30% on CHIRALCEL OD, 12% on CHIRALCEL OJ, and 8% on CHIRALPAK AS. CHIRALCEL OC and OA were early commercial products that are now rarely used; CHIRALCEL OG, OF, and OK are still used as they often separate compounds not separable by the main four CSPs. A series of review articles [10–14] from Okamoto and Yashima describe the wide variety of compounds resolved on polysaccharide CSPs. Providing such surveys has become increasingly difficult as the application of these CSPs continues to expand. A major contributing factor to this expansion is that in contrast to other CSPs, dramatic changes in separations can result from changes in mobile phase. Even without commercialization of additional CSPs, development of new mobile phase combinations expand the application range of existing polysaccharide CSPs.

260

Advances in Chromatography

π−

π+ H−

H+

H+

−H

π− π+ H− H+

H+ −H

FIGURE 6.3 Typical three-point interaction schematic representation of chiral selectivity on Pirkle-type CSP. The more-retained enantiomer experiences three simultaneous interactions with the stationary phase while the orientation of the other enantiomer greatly weakens the strength of one of the interactions.

6.2 MECHANISMS OF SELECTIVITY Despite the extensive use of polysaccharide CSPs, these phases are poorly understood. Mechanisms have been proposed and widely accepted for various other CSPs. Figure 6.3 is a simplistic version of the type of representation typically invoked to explain chiral recognition for Pirkle-type CSPs. In this planar depiction, the moreretained enantiomer complex is held together by three points of interaction and the lesser-retained by two points. In Pirkle-type complexes, two interactions are often hydrogen bonding while the third is a putative π-π interaction between aromatic groups. Both hydrogen bonding groups commonly exist on one arm of the chiral center as drawn, although the three interaction sites can arise from three arms off the center. Hydrogen bonding interactions are typically stronger than the π-π interaction and orient the binding so that each enantiomer experiences these interactions. Selectivity is then determined by differences in the weaker π-π interaction. The ππ interaction still exists for the lesser-retained enantiomer but is considerably weaker due to the distance imposed by the unfavorable chiral orientation of this enantiomer. The classic three-point interaction as depicted in Figure 6.3 predicts a practical limit to selectivity. Considering only enthalpic contributions: α=

ΔH AX + ΔH BY + ΔH CZ ΔH CZ = 1+ ΔH AX + ΔH BY ΔH AX + ΔH BY

(6.1)

In cases where both HAX and HBY interactions arise from one arm, at best HCZ approaches HAX + HBY (or else it would be common to both interactions and Equation 6.1 reduces to the same form) and selectivity should have a maximum value of 2. If the interactions arise from three chiral arms, this maximum value reduces even further: ΔHCZ ≤ ΔHAX ~ ΔHBY → ΔHCZ ≤ (ΔHAX + ΔHBY)/2; α → 1.5. Typically, HCZ is considerably weaker than HAX + HBY . Nonspecific interactions and entropic

The Use of Polysaccharide Phases in the Separation of Enantiomers

261

contributions further reduce this predicted maximum selectivity. Reviewing a large number of separations, it was noted that the majority of analytical chiral separations were achieved with selectivities of 2 or less. Separations with larger selectivity values have been observed on Pirkle-type CSPs which have three-dimensional structures giving rise to clefts and potential steric effects [15]. The role of solvent modifiers in chiral separations on Pirkle-type CSP is to disrupt the interactions responsible for retention. In the elution process, both analyte and modifier are considered to be in equilibrium between the stationary and mobile phases. Increasing the level of modifier in the mobile phase will shift the equilibrium of the analyte toward the mobile phase and retention will decrease. The shift will be similar for each enantiomer and selectivity will not change with modifier level. Various alcohols used as modifiers exert similar effects on this equilibrium and selectivity does not vary greatly between modifiers. Retention is typically a function of modifier polarity. Exceptions to these generalizations require invocation of steric effects [16]. The mechanisms invoked for enantioselectivity observed with inclusion complex CSPs such as cyclodextrin and crown ethers tend to be strictly based on steric effects rather than attractive interactions. Proposed mechanisms for both Pirkle-type CSPs and inclusion-type CSPs are fairly well-accepted, as their initial use was based on the a priori expectations that these mechanisms would be effective in enantiomer separations. By contrast, the efficacy of polysaccharide CSPs was more an evolutionary process, ultimately traceable back to paper chromatography. As such, the understanding of the mechanism of enantioselectivity has developed slowly. Polysaccharide CSPs appear to combine both attractive interactions and steric effects. The review articles referenced above [10–14] do an excellent job summarizing efforts to understand the mechanisms contributing to the enantioselectivity of these CSPs. Attempts to understand the mechanisms behind the broad applicability of polysaccharide CSPs include several approaches: observation of chromatographic effects resulting from variations in the stationary phase and also from variations in the structure of probe molecules; spectroscopic measurements of the CSP and of CSPprobe combinations; attempts at molecular modeling to determine relevant contributions to enantioselectivity; and observations of mobile phase effects relevant to proposed mechanisms. Each saccharide in the polymer contains five chiral centers. Derivatization of the three hydroxyl groups to esters or carbamates provides hydrogen bonding opportunities near these three chiral centers. Could binding at these side chains give rise to the observed enantioselectivity in a manner similar to the representation of Figure 6.3? Derivatized polysaccharide CSPs evolved from observations on the utility of microcrystalline cellulose triacetate in chiral separations [17]. When the acetylation reaction was applied to crystalline cellulose, a useful stationary phase resulted. If the cellulose was solubilized prior to acetylation and subsequently crystallized, a much less useful CSP resulted. If the stationary phase prepared from crystalline cellulose was subsequently dissolved and recrystallized, the inferior CSP also resulted. When solubilized cellulose triacetate was coated onto silica, dramatic changes in chromatography were observed [18, 19] that were ascribed to polymeric conformation differences [20]. A useful CSP resulted even though the derivatized

262

Advances in Chromatography

cellulose was characterized as amorphous rather than crystalline [21]. These early findings indicate that the enantioselectivity exhibited by polysaccharide stationary phases derive from the crystallinity or higher-order state of the polymer rather than from chirality within the saccharide unit itself.

6.2.1 EFFECT

OF

SIDE CHAIN SUBSTITUTIONS

Okamoto et al. [3] describe the preparation of a series of cellulose-based phenylcarbamate CSPs with a variety of aromatic side chains. One of the goals of this study was to discern the chromatographic consequences of the electronic effects of the various ring substitutions in anticipation that aromatic side chains in these CSPs participate in π-π interactions, as invoked for Pirkle-type CSPs. Ring substitutions did affect the chromatography but confusing conclusions were drawn that the best enantioselectivity resulted when the substitutions were either electron-withdrawing or electron-donating groups. Much of the problem arose from the attempt to generalize results across ten different chromatographic probes. More interesting are the results of achiral chromatography and nonchromatographic measurements made on these CSPs. Proton NMR experiments showed that ring substitutions affected the acidity of the carbamate NH proton. The acidity of this proton was further correlated to retention times of achiral probes. The electronic effects responsible for changes in the acidity of this proton would also affect the hydrogen bond accepting ability of the carbonyl oxygen in the carbamate linkage. Retention of first eluting enantiomers of the various chiral probes followed this trend. Chromatography of different substituted achiral benzenes showed very little resolution on a phenylcarbamate CSP, indicating that π-π interactions are not discriminating and are unlikely to contribute to the enantioselectivity of these CSPs. Nitrobenzene showed increased retention, likely due to its ability to form an additional hydrogen bond with another carbamate linkage. IR spectroscopy and observations of crystallinity under a polarizing microscope suggested a correlation between sterically regular structure and utility as a chiral stationary phase. The preparation of amylose-based phenylcarbamate CSPs [5] showing dramatically different selectivity also points to the importance of tertiary structure rather than the structure of the saccharide monomers. Amylose and cellulose CSPs differ only in the linkage between monomeric units but are expected to have much different tertiary structures. The effect of derivatization and adsorption onto silica on the native helical structure is not known but it is likely that amylose and cellulose derivatives differ. Proposed structures inferred from x-ray analysis and molecular modeling [11] suggest different helical structure and further that carbamate linkages that offer hydrogen bonding opportunities exist in a curved groove between the aromatic side chains. Kaida and Okamoto [22] synthesized a small series of benzyl carbamates of cellulose and amylose. The benzyl carbamates add a carbon atom between the aromatic ring and the saccharide. Several of the CSPs also incorporate an additional chiral center in the side chain. In a series where the substitutions on this carbon atom ranged from H through CH3, C2H5, CH(CH3)2 to benzyl, only the methyl and ethyl derivatives gave useful cellulose CSPs. Peaks on the other CSPs were broad,

The Use of Polysaccharide Phases in the Separation of Enantiomers

263

interpreted as arising from multiple binding sites with a less-defined tertiary structure. The conclusion was that if substitutions at this carbon atom were too small or too large, the “higher-order structure” was disturbed. Circular dichroism (CD) spectra of these polymers cast onto quartz were obtained. Intense peaks indicative of higher-order structure were observed only for the methyl and ethyl derivatives. Similar results were obtained for amylose derivatives, both in terms of chromatographic utility and CD spectra. Separations were different between amylose and cellulose derived CSPs, as were the CD spectra. This group also synthesized a series of phenylcarbamate derivatives of cellulose and amylose substituted at position 4 with different halogens [23]. Differences in enantioseparation were observed despite similarities in spectroscopic measurements. NMR, IR, and CD results all suggested similar higher-order structure and hydrogen bonding abilities for the CSPs. Selectivity differences were then ascribed to steric factors related to the size of the halogen. Small probes did not show as much selectivity difference between CSPs as did bulkier probes. When invoking steric factors, it would be predicted that as the size of the halogen increased, the access of the probes to the hydrogen binding sites responsible for retention would decrease. This would result in shorter retention times and likely lower selectivity. The examples given [23] exhibit a more interesting behavior in that the retention of the second eluting enantiomer varies. This indicates that steric effects are exerted on either the departure of the probe from the CSP or on access of eluant to binding sites.

6.2.2 SPECTROSCOPIC

AND

THERMAL STUDIES

NMR spectroscopic studies of the interactions between chiral probes and polysaccharide CSPs have been hampered by the need to solubilize the polymer while maintaining chiral selectivity. Yashima et al. [24] found that a trimethylsilylated phenylcarbamate was soluble in chloroform and that selective binding occurred in this solvent. One enantiomer of the trans-stilbene oxide probe showed NMR peak splitting of a proton when exposed to chiral polymer while the other enantiomer did not show this effect. The peak splitting was attributed to involvement of the proton in hydrogen bonding with the polymer as corroborated by chromatographic retention order. The methyl protons of isopropanol were also shifted and split when exposed to polymer in this NMR study, the implications of which will be discussed below. Later work reported by this group [25] extended these findings to other soluble polymers and other probe molecules. Proton NMR results continued to show involvement of hydrogen bonding. 13C NMR results indicated either stacking between analyte and CSP aromatic groups or a shielding effect. The proximity of aromatic groups could arise from orientation imposed by hydrogen bonding in the groove described above. Results obtained with β-naphthol indicated that the rings bearing the hydroxyl groups are oriented deeper in the binding groove than the distal aromatic carbons. Relaxation studies allowed the determination of binding dynamics that were compared to thermodynamic results generated chromatographically. The authors noted that selectivity predicted from NMR was higher than that observed in LC, which they attributed to solvent differences and effects of silica. This observation should serve as a cautionary note when extrapolating results from NMR studies of

264

Advances in Chromatography

solubilized polymer to chromatographic behavior on CSPs. Solid-state NMR was applied to CHIRALPAK AD [26] but this study did not examine CSP-analyte interactions nor could it provide much information about the tertiary structure of the CSP. Interesting observations were noted regarding association of mobile phase components with the CSP that will be discussed below. Researchers at Merck [27] used differential scanning calorimetry (DSC) and IR to examine the effect of temperature on a cellulosic CSP. This work was prompted by the observation of unexpected changes in chromatographic behavior as a function of temperature. From thermodynamic relationships, the logs of retention and enantioselectivity are expected to increase linearly with inverse absolute temperature. These authors noted a discontinuity around 18˚C and DSC indicated that the polymer experienced an exothermic transition around this temperature. The sharpness of this transition was affected by the solvent and was completely reversible. Similar observations were repeated by this group [28] with a different probe on an amylosic CSP. The temperature-induced transition is not always completely reversible [29]. Wang et al. observed an increase in selectivity on a CHIRALPAK AD as a result of heating that was maintained upon cooling. They attributed this effect to a change in polymer conformation. This change was dependent on the polar component of the mobile phase, occurring with ethanol but not isopropanol. The polymer conformation could be reverted by flushing with a hexane/isopropanol mixture. Solid-state NMR studies of this transition [30] did not reveal major changes in conformation of the polymer backbone. Instead, changes in incorporation of ethanol and even the lowlevel acidic additive were indicative of the enantioselectivity shift.

6.2.3 ANALYTE STRUCTURE EFFECTS Attempts to discern the mechanisms behind the ability of polysaccharide CSPs to separate a broad range of compounds have been made by observing the effects of varying the structure of compounds themselves. Many researchers have proffered suggested mechanisms based on the different chromatographic behavior between two or three compounds. Valid comparisons require more data. Wainer and Alembik [31] tested a series of aromatic amides on CHIRALCEL OB and found that selectivity was related to the Hammett constant of the para substituent on the probe’s aromatic ring. The Hammett constant will affect the ability of the aromatic ring to participate in π-π interactions with the CSP. These conclusions are somewhat confounded by the effects the substituent also has on the hydrogen bonding ability of the amides of the probes. Chain length in the alkyl side chains of the probes was roughly correlated with enantioselectivity. This same group [32] extended their studies to a series of aromatic alcohols. Without hydrogen bonding groups attached to the aromatic ring, as was the case in the first study, no correlation exists between the Hammett constant and enantioselectivity. Camilleri et al. [33] attempted to relate the chromatography of 12 benzimidazole sulphoxide analogs on CHIRALPAK AD to physiochemical properties of the analogs. After testing 254 molecular descriptors, these authors identified nine molecular properties as being associated with enantioselectivity. These descriptors reflected electronic effects that would relate to hydrogen bonding and steric factors. Booth

The Use of Polysaccharide Phases in the Separation of Enantiomers

265

and Wainer [34] applied multiparameter regression analysis between various molecular descriptors and the chromatographic results of 28 α-alkylarylcarboxylic acids on CHIRALPAK AD. They found that hydrogen bonding ability was most responsible for retention with lesser contribution of aromaticity. Unfortunately, enantioselectivity was too low to allow further conclusions. Combining multivariate regression with neural networks, this group [35] was able to identify charge transfer ability and steric or geometric factors as being important to enantioselectivity. This approach iteratively eliminates unimportant molecular descriptors.

6.2.4 MOLECULAR MODELING Several groups have attempted to apply molecular modeling approaches to polysaccharide CSPs [13, 14, 25, 27, 28, 34, 36]. While the interaction between chiral analytes and polysaccharide CSPs is too complex to expect to use current molecular modeling techniques for predictions of potential separations, better understanding of separation mechanisms may possibly result. When molecular modeling is used to generate energy-minimized structures of these CSPs, a consistent picture begins to emerge. Using data generated by chromatography, NMR and x-ray diffraction to guide modeling, Yashima et al. [25] derived a left-handed 3/2 helical structure for a cellulose CSP. Hydrogen bonding sites were contained in a chiral groove with CSP side chains directed outward. Energy minimization of an analyte in contact with the modeled CSP indicated orientation of hydrogen bonding groups into the groove at distances consistent with the formation of such bonds. In later work [36], this group indicated that different CSP side chains impart different twists relative to the axis of the helix. Applying a multitude of docking strategies forced the conclusion that enantioselectivity only results when analyte binding involves insertion into the chiral groove. An additional result of interest from modeling studies was the observation of Booth and Wainer [34] that analytes can form simultaneous hydrogen bonds with more than one carbamate linkage. While molecular modeling directed by chromatographic results and NMR and x-ray data can yield a somewhat self-fulfilling prophesy, all the results point toward a consistent mechanism of enantioselectivity on polysaccharide CSPs. Tertiary structure of the polymer is critically important to separation. This structure appears to be helical with a chiral groove presenting opportunities for hydrogen bonding, flanked by side chains that likely present steric restriction to the interaction sites. Different side chains impact the strength of hydrogen bonding with carbamate linkages, impart changes to the helical structure and present different steric restrictions. The π-π interactions expected between aromatic analytes and CSP side chains seem to be less important than originally expected.

6.2.5 MOBILE PHASE EFFECTS These conclusions do not consider the impact of the mobile phase. A large percentage of the early chromatographic investigations were accomplished with a hexane/isopropanol mobile phase, at least partly due to the fact that early investigations from Okamoto’s group were focused on development rather than application of the

266

Advances in Chromatography

polysaccharide CSPs. Expectations at the time were that the effect of the alcohol modifier would be similar to that observed for Pirkle-type CSPs — more polar alcohols would give shorter retention with little effect on enantioselectivity. Gaffney et al. [37] examined the effect a series of alcohol modifiers had on the chromatography of three probes on CHIRALCEL OB. In general, retention was lower when primary alcohols were used as modifiers than when secondary alcohols were used. The primary alcohols are expected to be more polar and better at disrupting hydrogen bonds, and these authors also noted that retention increased with the length of the alkyl chains of the modifiers. While longer alkyl chains decrease polarity, the results suggest a steric effect as well. Most interesting was the observation of a reversal of elution order between alcohol modifiers. Elution order reversal arising from use of different modifiers has been reported by others as well [38–42]. Such dramatic mobile phase effects demand consideration in any discussion of separation mechanisms. Gaffney et al. [37] attributed the observed effects to a shifting between two binding sites on the CSP. Balmer et al. [39] also invoke two binding sites to explain reversal of elution orders resulting from changes in modifier level, changes in water content of the modifier, and temperature. In subsequent work, this group [40] attributed dramatic changes in selectivity resulting from mixtures of modifiers to conformational changes in the polymer. This explanation has become the common rationalization for unexpected mobile phase effects observed on polysaccharide CSPs. Demonstrating conformational changes arising from changes in mobile phase modifier is very difficult. Helical structures are inferred from x-ray crystallography, which precludes solvent effects, and NMR is not sensitive enough to detect minor changes in helical structure. The sharpness of the exothermic transition experienced by CHIRALPAK AD observed by O’Brien et al. [27] was affected by adsorbed mobile phase composition, indicating a difference in polymer conformation. While NMR may not be sensitive enough to detect minor changes in helical structure, it does reveal useful information regarding solvent in the presence of CSP. In an early proton NMR study, Yashima et al. [24] found that alcohol modifier peaks were split in the presence of CSP, indicating their residence in a chiral environment. Solid-state NMR corroborated the incorporation of alcohol modifiers into the CSP [26]. This latter report noted that incorporation was gradual at low levels of modifier but above 5% by volume, no additional incorporation was observed. This effect means that the CSP undergoes a transition in modifier incorporation at low levels and explains the common observation of unstable separations in such mobile phases. Wang et al. [41, 42] also report reversal of elution order resulting from modifier changes, which they attribute to changes in the “steric environment.” This term encompasses both changes in helical structure and modifier molecules incorporated into chiral binding sites. The residence of alcohol modifiers of different shape and size will affect the shape of the binding site. Per NMR studies, bulkier modifiers spend more time in the CSP. If adsorbed alcohol molecules restrict access of analyte to binding sites, this would explain the occasional observation of lower retention with bulky, less polar modifiers. Adsorbed modifier would also change the shape of the binding pocket, yielding different selectivity.

The Use of Polysaccharide Phases in the Separation of Enantiomers

267

While the encompassing term “steric environment” fares well in explaining unexpected effects of modifier changes, it implies that the effects would be more general than are observed. For most applications, the effects of modifier are as expected from experience with Pirkle-type CSPs. A change in the shape of the CSP should affect more analytes than the few observed. Recent work [16] with a Pirkletype CSP expected to have steric hindrance to binding suggested that the bulk of the mobile phase modifier could be important to enantioselectivity due to its impact on the ability of the modifier to penetrate the complex between analyte and CSP. This effect should also be an important factor with polysaccharide CSPs. A good indication of this effect is that when selectivity is larger for a bulky modifier, the retention of the second eluting enantiomer is much longer than for the smaller modifier while that of the first enantiomer is relatively unchanged. Conversely, when selectivity is larger for a smaller modifier, retention of the first peak should be much less than for the bulky modifier while that of the second peak is comparable. While the effects of mobile phase on polysaccharide CSPs are not well understood, modifiers appear to affect hydrogen bonding, tertiary structure and steric factors involved in chiral recognition. The combination of a variety of polysaccharide CSPs with the ability to dramatically alter performance via mobile phase changes makes understanding separations accomplished on these phases very difficult. This combination also offers broad opportunities for manipulation of chiral separations.

6.3 MODES OF SEPARATION 6.3.1 NORMAL-PHASE MODE Most separations on polysaccharide CSPs are accomplished with normal-phase mobile phases. These mobile phases consist of mixtures of nonpolar alkane and a polar alcohol modifier. Many other polar modifiers typically used in achiral normal phase separations cannot be used with these CSPs as they solubilize the polymer, destroying the column. Retention is controlled by varying the proportion of the alcohol in the mobile phase. Increasing the alcohol content decreases retention. Typically, enantioselectivity does not change with alcohol level, although there are exceptions. As noted above, different alcohol modifiers can give dramatic changes in selectivity. While explanations of these effects may be credible, prediction of this a priori effect is not possible. Relating modifier properties to their effects does not appear possible. Application of linear solvation energy relationship (LSER) modeling to separations on a CHIRALPAK AS column with a variety of modifiers failed to identify modifier properties correlated to their effects [43]. This approach attempts to relate molecular properties of a large series of modifiers to their effects on chromatography. It was found to work quite well for a series of phenylalanine analogs for a Pirkle-type CSP. As a general approach, screening polysaccharide CSPs with hexane/ethanol and with hexane/isopropanol mobile phases detects most feasible separations. Based solely on anecdotal observation, when a separation is better with the isopropanol mobile phase, substitution with a bulkier modifier may be beneficial. Conversely, a separation that is better with ethanol may benefit from addition of methanol.

268

Advances in Chromatography

6.3.2 POLAR ORGANIC MODE Increasing the proportion of alcohol modifier in a normal-phase mobile phase is expected to decrease retention and collapse resolution. Observations of useful separations exist with neat alcohol mobile phases [44–51]. Neat alcohol mobile phases, in combination with neat acetonitrile, are collectively referred to as polar organic mobile phases. The occurrence of these separations is counterintuitive, as increases in eluent strength should preclude retention. This result leads to the presumption that there is a different recognition mechanism involved with polar organic mobile phases [49, 50]. Using polar organic mobile phases offers several advantages for preparative applications of chiral chromatography, including enhanced solubility and simplicity of operation and solvent recovery [52]. Chankvetadze et al. [47] screened 20 chiral molecules on CHIRALPAK AD, CHIRALCEL OJ, and CHIRALCEL OD using neat methanol and acetonitrile mobile phases. They obtained baseline resolution for 14 compounds with a resolution of at least 2.0 for 11 of the 20 probes. No structural features appeared to correlate with success. When a structurally diverse set of 82 compounds was screened on AD, AS, OJ, and OD columns using either a 1:1 ethanol:methanol or neat acetonitrile mobile phase, 25 showed selectivity values of 1.5 or greater, giving a success rate of about 30%.* The alcohol mixture was used to detect separations that would occur with ethanol or methanol with the intent to subsequently test the individual alcohols. Results of promising separations are given in the tables below. All of these compounds also gave excellent selectivity with hexane-based mobile phases if elution was observed. The compounds that separated in polar organic mode were structurally diverse and no common elements could be identified in this group. The AD gave the most separations (19) followed by OJ (12), OD (7) and AS (4). Several showed separation on more than one column or with more than one mobile phase. The alcohol mixture was the more successful mobile phase, yielding 31 separations to 12 for the acetonitrile. Most compounds were unretained with acetonitrile and showed no selectivity. Some compounds did not elute and selectivity was not detected. Many polar organic selectivity values were dramatic, including an 8.19 observed for diperidon on OD with acetonitrile. Good selectivity did not always yield high resolution, as some compounds showed distorted peak shapes and some retention times were too short. The errors in determining short retention times could give rise to erroneously high and variable selectivity values. The minimum selectivity value for inclusion in Table 6.1 through Table 6.4 was based on preparative chromatography applications. Several separations with lower values were suitable for analytical purposes. Other factors are also important in selecting mobile phases for preparative chromatography, i.e., low retention is an asset as it speeds the separation. Simple, single-component mobile phases are also desirable for preparative use, along with low viscosity. Acetonitrile is thus a favored mobile phase, followed by neat methanol. Ethanol can be used but is not preferred due to its viscosity. * Data presented in this section are unpublished, generated by the author and others at Chiral Technologies, Inc.

The Use of Polysaccharide Phases in the Separation of Enantiomers

269

TABLE 6.1 Compounds Separating (α ≥ 1.50) on CHIRALPAK AD Using Polar Organic Mobile Phases Sample

m.p.

t’1

t’2

α

Rs

Aminoglutethimide 1-benzoyl-2-t-butyl-3-methyl-4-imidazolidinone 4-benzoyloxy-2-azetidinone 2,3-O-benzylidene DL threitol

M/E ACN ACN M/E ACN M/E M/E ACN M/E ACN M/E M/E M/E M/E ACN M/E M/E M/E M/E

15.5 5.8 5.4 4.1 4.6 6.0 9.1 5.2 4.2 7.8 5.7 5.2 4.4 10.0 8.9 5.1 3.8 5.3 5.1

40.3 8.6 6.9 5.7 6.0 8.0 25.0 6.5 6.9 20.3 9.8 6.6 5.3 46.0 28.2 7.6 4.6 8.0 6.8

2.98 2.01 1.67 2.42 1.94 1.64 3.63 1.62 3.20 3.64 2.49 1.61 1.60 6.14 4.25 2.14 1.95 2.20 1.82

6.70 4.52 4.52 4.09 2.61 4.30 9.40 1.20 6.69 12.45 6.49 4.15 1.16 11.3 7.0 5.96 2.05 5.05 4.42

BOC-2-t-butyl-3-methyl-4-imidazolidinone Chlormezanone α-2,4-dichlorophenyl-1H-imidazole-1-ethanol Diperidon γ-(4-fluorophenyl)-γ-butyrolactone 2-imino--phenyl-3-thiazolidine ethanol α-methyl-α-phenyl succinimide Mianserin Norephedrine Propafenone Tetrahydropalmatine

TABLE 6.2 Compounds Separating (α ≥ 1.50) on CHIRALPAK AS Using Polar Organic Mobile Phases Sample

m.p.

t’1

t’2

α

Rs

Aminoglutethimide

M/E ACN M/E ACN

4.8 4.3 5.7 4.3

6.2 7.1 8.8 6.6

1.77 3.14 2.17 2.70

1.90 6.34 2.78 5.79

Chlormezanone α-methyl-α-phenyl succinimide

Several of the compounds observed to separate on OJ with the mixed alcohol mobile phase also separate with hexane-ethanol mobile phases. This result allowed testing of the presumption of different retention mechanisms for normal phase and polar organic separations. Four compounds were chromatographed at increasing ethanol content, from 10% in hexane to neat ethanol. Results for benzoin ethyl ether are shown in Figure 6.4. Increasing ethanol content relative to hexane gives an initial decrease in retention of both enantiomers for all probes. The retention decrease levels out around 50%

270

Advances in Chromatography

TABLE 6.3 Compounds Separating (α ≥ 1.50) on CHIRALCEL OD using Polar Organic Mobile Phases Sample Diperidon Homatropine Indapamide Laudanosine Tetrahydropalmatine

m.p.

t’1

t’2

α

Rs

M/E ACN M/E M/E ACN M/E ACN

6.0 11.3 4.4 5.0 3.6 5.6 5.5

14.5 70.9 5.2 7.7 4.1 7.2 6.8

3.86 8.14 1.63 2.39 1.75 1.61 1.50

12.10 17.90 3.31 6.81 2.30 4.17 3.29

TABLE 6.4 Compounds Separating (α ≥ 1.50) on CHIRALCEL OJ using Polar Organic Mobile Phases Sample Aminoglutethimide Benzoin ethyl ether 2-bromo-1-indanol Chlophedianol 1-(4-chlorobenzhydryl)-piperazine EEDQ Indapamide Laudanosine α-methyl-α-phenyl succinimide Mianserin Nomefenesine

m.p.

t’1

t’2

α

Rs

M/E M/E ACN M/E ACN M/E M/E M/E M/E M/E M/E M/E

6.5 5.7 4.2 4.6 4.0 8.0 5.2 4.9 5.2 6.0 5.3 5.8

12.3 7.4 6.0 5.4 5.1 12.0 7.3 6.2 6.4 13.6 6.9 7.7

2.67 1.64 2.52 1.50 2.23 1.79 1.93 1.70 1.57 3.55 1.68 1.65

9.50 4.10 6.58 3.12 3.85 1.94 6.37 2.76 3.07 6.30 4.57 4.92

ethanol. Selectivity changes only slightly over the titration range. The lack of a step change in selectivity indicates a consistent recognition mechanism in both the hexane-based and polar organic modes of separation. These results are consistent with those reported by Miller et al. [52]. Increasing levels of modifier affect the partitioning equilibrium between stationary and mobile phases but not the nature of the recognition complex. Chankvetadze et al. [49] observed increased retention resulting from the addition of water to neat ethanol or 2-propanol. Two of six probe sulfonic acids showed increased retention time and selectivity with water, although retentions times could be overlong. Increases above 15% by volume were often detrimental. The effect of adding water to methanol separations from the tables above was evaluated. In most cases, water addition resulted in increased retention. Effects on selectivity were

The Use of Polysaccharide Phases in the Separation of Enantiomers

271

Benzoin ethyl ether 3.00 k1 2.50

k2

2.00

alpha

1.50 1.00 0.50 0.00 0

20

40

60

80

100

% Ethanol (b)

mAU

(a)

4.87

1200 1000

6.21

800 600 400 200 0 0 mAU

1

2

3

4

5

6

7

8

9

min

9

min

(b)

1200

6.16

1000

8.26

800 600 400 200 0 0

1

2

3

4

5

6

7

8

(b)

FIGURE 6.4 (a) Plot of the effect of ethanol level titrations on retention and selectivity of benzoin ethyl ether on CHIRALCEL OJ. (b) Chromatograms of benzoin ethyl ether on CHIRACEL OJ with A = neat ethanol and B = 20/80 ethanol/hexane mobile phases.

mixed and rarely dramatic. Resolution tended to improve with water addition due to retention increases (Figure 6.5). Being more polar than neat alcohol, the addition of water would be expected to decrease retention in normal phase. Results are more consistent with a reversed-phase mechanism, where the addition of water weakens the mobile phase and retention time increases. Although its low volatility could

272

Advances in Chromatography

DAD1 A, Sig=210,4 Ref=360,100 (19JUN03A\003–0201.D) mAU

3.705

600 4.251 500 400 300 200 100 0 0

2.5

5

7.5

10

12.5

15

17.5

min

10

12.5

15

17.5

min

(a)

DAD1 A, Sig=210,4 Ref=360,20 (P215JY03\006–0602.D) mAU 160

4.822

140 6.183

120 100 80 60 40 20 0 0

2.5

5

7.5

(b)

DAD1 A, Sig=210,4 Ref=360,20 (P21JY03\006–0601.D) mAU 120

9.503

100 80

14.305

60 40 20 0 0

2.5

5

7.5

10

12.5

15

17.5

min

(c)

FIGURE 6.5 Chromatogram of 2-bromo-1-indanol on CHIRALCEL OJ using A = neat methanol; B = methanol + 10% water; and C = methanol + 25% water.

The Use of Polysaccharide Phases in the Separation of Enantiomers

273

preclude its use in preparative applications, addition of water appears to give some control over retention and could be useful in the development of analytical methods. The effect of water addition to acetonitrile was tested for compounds from the tables above. Retention was decreased in nearly all instances and effects on selectivity and resolution were mixed. For compounds showing long retention with neat acetonitrile, the addition of water was an effective means of reducing retention. The effect of adding water to neat acetonitrile is opposite that observed with alcohol mobile phases. Decreased retention could arise from disruption of hydrogen bonds only weakly affected by acetonitrile, in which case the addition of alcohols should also decrease retention. Examination of results in Table 6.5 reveals that adding alcohol to acetonitrile mobile phases decreased retention. Increased selectivity resulted when compounds had k′1 values greater than ≥ 0.5 with neat acetonitrile. Alcohol modifier decreased retention of the first eluting enantiomer to a greater extent than the second enantiomer. This effect was more pronounced with isopropanol. Alcohol modifier is disrupting hydrogen bonds responsible for retention and the less-retained enantiomer likely forms a looser complex with the CSP than the more-retained. Alcohol modifier will be more effective in penetrating the first enantiomer-CSP complex, speeding its elution. The size of isopropanol could affect its ability to cause elution. A corollary of this explanation is that at a high enough modifier concentration, the effect on the second eluting enantiomer should be equivalent to that on the first enantiomer and selectivity will decline. This assertion was tested by adding different

TABLE 6.5 Effect of Addition of Alcohols to Acetonitrile Polar Organic Mobile Phase Compound

CSP

Mobile Phase

α-methyl-α-phenyl succinimide

AD

α-2,4-dichlorophenyl1H-imidazole-1-ethanol

Diperidon

t1

t2

α

Rs

ACN

6.96

19.85

4.26

8.78

AD AD AD

ACN + MeOH ACN + IPA ACN

4.86 4.62 7.86

12.62 17.41 20.63

5.17 8.90 3.63

17.52 22.37 12.36

AD AD OD OD OD

ACN ACN ACN ACN ACN

4.34 5.00 9.34 5.36 5.89

8.75 12.24 53.96 18.54 28.37

4.29 4.62 8.04 6.58 8.78

14.16 15.49 18.70 15.22 16.13

+ MeOH + IPA + MeOH + IPA

ACN: neat acetonitrile mobile phase + MeOH: 10% methanol added, by volume + IPA: 10% isopropanol added, by volume

274

Advances in Chromatography

TABLE 6.6 Effect of Varying Levels of Isopropanol in Acetonitrile Separations Compound

CSP

ACN/IPA

Diperidon

OD

α-methyl-α-phenyl succinimide

AD

100/0 95/5 90/10 85/15 80/20 75/25 100/0 90/10 85/15 80/20

t1 (min) 10.4 7.6 6.2 5.5 5.0 4.6 7.0 4.9 4.5 4.2

t2 (min)

α

Rs

61.9 41.1 29.1 22.5 18.3 15.4 19.9 17.9 13.0 10.2

7.93 8.26 8.11 7.91 7.68 7.53 4.26 7.98 6.81 5.85

17.15 16.38 15.25 14.13 13.05 11.97 8.78 15.87 14.33 12.38

amount of isopropanol to acetonitrile for diperidon and α-methyl-α-phenyl succinimide separations. Results are shown in Table 6.6. Addition of isopropanol decreases retention of diperidon enantiomers unequally until a level of 15%. Selectivity increases to a maximum at 5% isopropanol and decreases with further modifier addition. Selectivity for α-methyl-α-phenyl succinimide reaches its maximum at 10% isopropanol.

6.3.3 REVERSED-PHASE MODE Interest in reversed-phase mobile phases for chiral separations arises mostly from the need to analyze biological samples. An aqueous mobile phase would simplify sample preparation, enhance solubility, and ease interfacing with mass spectrum detection. Some resistance to normal-phase application also exists for new practitioners of chiral chromatography. This derives from concerns over perceptions of the finicky nature of normal-phase separations, which dictates dedication of instruments and careful control of water levels in mobile phases. Since polysaccharide CSPs are polymer coated, interactions with silica surface are minimized and separations are not sensitive to issues associated with early practice of normal phase chromatography. The mechanism for enantioselectivity of polysaccharide CSPs almost certainly involves hydrogen bonding with analytes being separated. Use of water in the mobile phase would be expected to completely disrupt these bonds, destroying retention and selectivity. The results from polar organic mobile phases reveal that hydrogen bonding still occurs even with neat alcohol; thus, retention with aqueous mobile phases should not be unexpected. Ishikawa and Shibata [53] used a mobile phase consisting of 40% acetonitrile against various aqueous buffers to separate a variety of compounds on a CHIRALCEL OD column. They concluded that low-pH buffers worked best for acidic compounds and high-pH buffers were best for basic compounds. Neutral compounds were not affected by pH. Perchloric and phosphoric acids were better than acetic acid, and bases were better separated when chaotropic

The Use of Polysaccharide Phases in the Separation of Enantiomers

275

anions were used in the buffer, with hexafluorophosphate (PF6) working the best. These authors reported that methanol did not work as well as acetonitrile and that the column proved stable to the acidic mobile phases. Basic mobile phases resulted in deterioration consistent with damage to the underlying silica. Tang et al. [54] concluded that ethanol gave higher selectivity than methanol for the separation of FMOC-protected amino acids on CHIRALCEL OJ-R, and both alcohols were superior to acetonitrile. Tachibana and Ohnishi [55] published a comprehensive review of almost 300 reversed-phase separations on polysaccharide CSPs. Their recommendations were to test both acetonitrile and alcohol organics. PF6 remained the preferred ion-pair agent with a note to use high concentrations carefully. Precipitation of the salt in highly organic mobile phase combination is a distinct and destructive possibility. They suggested the use of 20 mM borate buffer (pH = 9) as an alternative for basic analytes. With this buffer, the polysaccharide CSPs adapted for reversed-phase application proved stable. These authors note that about 70% of the compounds separating in reversed phase also separate in normal phase and that reversal of elution order occurs in only about 10% of the caes. This suggests that mechanism determining enantioselectivity is the same in normal phase and reversed phase. Perrin et al. [56] described the development of a reversed phase screening approach. Using three polysaccharide CSPs and two mobile phases, they found successful separations of 89% of 37 pharmaceutical compounds tested. While this estimate could be high, it is similar to that experienced with normal mobile phases.

6.3.4 SUPERCRITICAL FLUID CHROMATOGRAPHY Interest in supercritical fluid chromatography (SFC) has shown a resurgence primarily due to its application to chiral separations with packed columns. Recent reviews [57–60] are available describing this application. A supercritical fluid is one that above a critical temperature and pressure will exist in a supercritical state. In this state, viscosity approaches that of a gas while solvent strength is closer to that of a typical liquid. A lower-viscosity mobile phase can be pumped through a column at a higher flow rate and diffusion is faster; faster diffusion gives higher column efficiency. Solvent strength means that analytes need not be volatile as in gas chromatography. By far the most commonly used fluid is liquid carbon dioxide due to its accessible critical region and its low cost, reactivity, and toxicity. The strength of liquid CO2 is relatively low and an alcohol modifier is typically added to speed elution. Early discussions focused overmuch on the effect of alcohol addition on the supercriticality of the resultant mobile phase. Sassiat et al. [61] measured diffusion coefficients in supercritical and subcritical mixtures of CO2 and methanol. They found no discontinuity between subcritical and supercritical mixtures, and estimate that diffusion in the subcritical mixtures was four times as fast as in liquid methanol. Chester [62] notes that misapplication of the definition of the critical point leads to a false concept of distinct regions of supercritical and subcritical states. A wide range of pressuretemperature combinations can be used in chromatography as long as regions near the critical point are avoided. This transition area is to be avoided, as the mobile

276

Advances in Chromatography

phase can separate into liquid and gas phases. The term SFC will apply here to separations accomplished with carbon dioxide-based mobile phases. The earliest report [63] of the use of SFC for chiral separations used a Pirkletype column. These authors noted that separations obtained were very fast with excellent resolution. This group subsequently applied the technique to CHIRALCEL OB [64]. While they reported better SFC separations than HPLC, resolution was disappointing due to lower than expected column efficiency. They attributed this to the need for very low alcohol levels in the mobile phase. As in HPLC, incorporation of modifier into the CSP polymer is incomplete at such low levels. With higher levels of modifier, performance was found to be much improved [65]. In SFC, the additional complication exists that both methanol and CO2 will adsorb from the mobile phase onto the stationary phase [66]. Near the critical temperature, adsorption of mobile phase components will also include analytes. This adsorption is nonspecific and can destroy expected resolution [67]. As long as this critical region is avoided, nonspecific adsorption is not problematic. Consideration of separation mechanisms in SFC should consider the stationary phase as coated with mobile phase components. With the introduction of commercial instrumentation designed for SFC with packed columns, the use of polysaccharide CSPs in SFC conditions increased dramatically. The most commonly noted advantage of SFC in chiral separations is speed. Macaudiere et al. [65] coined the term “resolution per unit time” to quantify this advantage. Early work focused on the effect of different parameters on resolution, which complicated matters. Resolution is the product of enantioselectivity, retention, and column efficiency, all of which can be affected by changes in the chromatographic conditions. Studying the effects on these individual traits is more informative than studying resolution. Parameters available for manipulation in SFC include flow rate, back pressure setting, modifier amount, and choice of modifier. While early comparisons between SFC and HPLC were flawed by the use of less than optimal HPLC separations, achieving separations in SFC in 5 minutes that required significantly longer in HPLC [68] were common. The speed of SFC arises from the ability to use higher flow rates without sacrificing column efficiency and without generating excessive pressure drop. Use of flow rates much above 1 mL/min in HPLC generates pressure drops above the recommended level for polysaccharide CSPs, especially when using those based on higher efficiency 5-μm particles. Due to the low viscosity of SFC mobile phases, flow rates of at least 5 mL/min can be used with these columns without damage. Maximum flow rate is more often limited by pressure limitations of the instrument than by pressure drop across the column. The speed of analysis can also be instrument-limited as some separations require less time than the autosampler requires to make subsequent injections. Increased flow rate has little effect on enantioselectivity [69] or retention factor. At their optimal flow rates, LC and SFC should show the same column efficiency but at commonly used flow rates, SFC does give higher efficiency. The loss in efficiency at high flow rates is minimal. Figure 6.6a shows a 7-minute SFC separation of flurbiprofen on a CHIRALPAK AD-H using 15% isopropanol flowing at 2 mL/min. The same separation accomplished in 1.3 minutes at 10 mL/min is shown in 6.6b. Figure 6.6c shows an overlay of these chromatograms with the time axis

The Use of Polysaccharide Phases in the Separation of Enantiomers

277

10

m Volts

m Volts

10 5

0 0 0

2

4

6

8

10

0

Minutes

0.4

0.8

1.2

1.6

2.0

Minutes

m Volts

10

0

0

2

4

6

8

10

Minutes

FIGURE 6.6 SFC separation of flurbiprofen on CHIRALPAK AD-H using 15% isopropanol at (a) 2 mL/min and (b) 10 mL/min. (c) An overlay of these chromatograms with the time axis normalized per flow rate.

normalized per flow rate. Very little deterioration is found in the separation at the higher flow rate. Varying back pressure in SFC was historically the primary means of controlling retention. Increased pressure gave increased density and increased eluting strength. With the addition of alcohol modifier to control elution, the effect of pressure becomes much less important and this variable is rarely used to alter a chiral separation. Back pressure also has little effect on column efficiency or selectivity. The more common means of altering retention is to vary modifier level. As with normal-phase HPLC, retention decreases sharply with initial increases in alcohol level but decreases more gradually above 20%. Modifier level does not affect column efficiency very much and efficiency decreases slightly as modifier level increases. Enantioselectivity typically does not vary with modifier level, although exceptions are not unusual [70]. Changing modifiers has little effect on efficiency but can have dramatic effects on retention and selectivity [69]. Chen et al. [71] noted an instance where retention with methanol modifier was much longer than with a comparable level of isopropanol. In most cases, retention is less with methanol, consistent with normal-phase behavior and its polarity. This result is not rare and is typically attributed to a change in the tertiary structure of the CSP. The elution order of BOC-phenylalanine ester

278

Advances in Chromatography

enantiomers reverses between methanol and isopropanol modifiers on CHIRALPAK AD-H. In our experience, selectivity observed for ethanol is intermediate between that given by methanol and isopropanol. The use of methanol in a normal phase separation represents an advantage for SFC over HPLC. Other modifiers used include acetonitrile, larger alcohols, and mixtures of alcohol and methylene chloride [72]. SFC offers other advantages in addition to the oft-noted speed of separation; method development takes less time as well. The increased column efficiency can improve a separation that is marginal in HPLC to one that is useful [73, 74]. Column equilibration is also faster in SFC [73–76], allowing faster evaluation of different mobile phases. Polysaccharide CSPs show long-term stability to SFC mobile phases [73, 76] and could be even more stable to various solvents [72] than in HPLC. We have found that CHIRALCEL OF and OG are stable to methanol and acetonitrile under SFC conditions but not HPLC. Neat methyl acetate, tetrahydrofuran and methylene chloride have been used at 20% level for short times in SFC on CHIRALPAK AD-H without damage. SFC also seems to be more forgiving than HPLC to various injection solvents. Figure 6.7 shows the effect of injecting methylene chloride under SFC conditions. In HPLC, the column would be destroyed within the first two or three such injections. There was a slight decrease in performance after the first injection but the column proved stable thereafter. Another advantage of SFC is that the pressure drop across the column is considerably less than in HPLC. Not only does this allow the use of faster flow rates, but connecting columns in series is possible. Berger and Wilson [77] connected 11 achiral columns in series and generated more than 250,000 theoretical plates. Phinney et al. [78] reported the coupling of chiral and achiral columns in SFC to effect the separation of diastereomers. Although it was not noted at the time, early experiments reported [69, 79] the separation of acidic analytes without the use of acidic additive. Acidic additives are required to elute such analytes from polysaccharide CSPs under HPLC conditions. The ability to separate acids without acidic additive is a great benefit to preparative applications as the presence of acid in alcohol-containing mobile phases leads to esterification of separated enantiomers.

6.4 MOBILE PHASE ADDITIVES Use of mobile phase additives is fairly common with polysaccharide CSPs. Acid analytes do not elute with HPLC mobile phases that do not contain acidic additives [80]. Amine additives are commonly added to mobile phases used to separate basic compounds on polysaccharide-based chiral stationary phases. Typically, these additives are included as a default or in hopes of improving peak shapes [81–86]. The expectation of improved peaks shapes is based on a putative interaction between the primary or secondary amine groups of analytes and silanols on the silica support beneath the adsorbed polysaccharide polymer. Addition of amines rarely gives dramatic improvements in peak shape, as nonideal peak shapes in chiral chromatography are more due to kinetic effects than to silanol interactions. Interaction with silanols through the adsorbed polymer layer is minimal. Using amine additives as a default

0

1

2

3

4

5

6

7

8

9

10

FIGURE 6.7 Overlay of TSO chromatograms after 10, 20, 30, 40, and 50 injections (25 μL) of CH2Cl2 on a CHIRALPAK AD column. Mobile phase is 20% isopropanol at 2 mL/min.

−50

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

850

900

950

1,000

The Use of Polysaccharide Phases in the Separation of Enantiomers 279

280

Advances in Chromatography

seems based on a desire for standardization of conditions [87, 88] rather than on actual need. Routine use of additives to improve column efficiency ignores their effect on retention and selectivity. Recent work [89–92] shows that acidic and basic mobile phase additives used to improve peak shapes in chiral HPLC can also affect enantioselectivity. The effect of various acid additives on separations of acidic phenylalanine analogs was tested on a CHIRALPAK® AD® column [89]. There was a large difference in resolution, efficiency, retention, and selectivity among the additives. Stronger acids such as alkyl sulfonic acids gave better shaped peaks and better resolution than commonly used trifluoroacetic acid. Retention proved to be a function of the additive pKa and hydrophobicity. The strength of the acid is apparently important in normal-phase mode to completely suppress ionization. Ion-pair formation is evidenced by differences in retention for alkyl sulfonic acid additives with comparable pKa values. Longer chain sulfonic acids appear to give less polar and less retained ion-pairs. Selectivity was also improved with the sulfonic acids. Enhanced selectivity likely arises from the suppression of nonspecific retention. Nonspecific retention masks true selectivity and its minimization should allow observed enantioselectivity to increase. Separations of phenylalanine analogs with both free amine and acid functionalities were altered by the inclusion of amine additives [90]. General observations included that primary amines were more effective additives than commonly used secondary and tertiary amines. In many cases, additives gave slight increases in selectivity through a larger decrease in retention of the first eluting enantiomer than of the second. Decreased retention is viewed as arising from competition for binding opportunities between the amine additive and the analytes. More interesting were the observations of increased retention in response to inclusion of amine additives. These cases often gave dramatic increases in selectivity. The size and shape of the additive strongly influenced the retention increase, leading to the suggestion that the amine was preventing access of modifier seeking to displace tightly bound enantiomer. Unfortunately, interpretation of these amine effects was somewhat clouded by the requirement for acidic additives in the mobile phase to elute acidic analytes from polysaccharide chiral stationary phases. Ethanesulfonic acid (ESA) was found to give dramatic effects on the chromatography of amino acid esters [91]. Figure 6.8 shows the example of tyrosine methyl ester. This amino acid ester does not separate on a CHIRALPAK AD column with a hexane/ethanol mobile phase. Addition of 0.1% ESA sharply decreases the retention of the D enantiomer. This is consistent with the concept that the additive competes for binding sites decreasing retention. Retention of the L enantiomer, however, increases dramatically. Upon removal of ESA from the mobile phase, the separation was maintained. The same effect could be generated by loading additive via a series of injections. These observations suggest that ESA binds to the CSP and likely offers an additional hydrogen binding site for the L enantiomer. Amine analytes other than amino acid esters show increased retention with ESA-treated CSP, although the increased enantioselectivity is rarely as dramatic. Subsequent NMR attempts to localize the acid binding site within the CSP failed. Stoichiometry calculations indicated that the effect saturates at about one ESA molecule per three

The Use of Polysaccharide Phases in the Separation of Enantiomers

281

1400 1200 D 1000

mAU

L 800

(a)

D

600 L 400

(b)

D+L

200 (c) 0 0

50

100 min

FIGURE 6.8 Chromatogram of tyrosine methyl ester on CHIRALPAK AD using a hexane/ethanol (85/15) mobile phase with C = no additive; B = 0.1% ESA; A = after elimination of additive from mobile phase.

saccharide units. These observations do not support the original concept and it appears more likely that the acid resides in the chiral groove but is not localized. Its effect could be to decrease local pH, leading to enhanced binding of the amine analyte. The memory effect for ESA persisted at least 55 hours after removal from the mobile phase. The effect could be eliminated by flushing with isopropanol/water (80/20). Numerous anecdotal reports exist of an amine “memory effect” where a column exposed to amine additives does not behave as it did prior to exposure. A common observation is that a compound can separate on a used column but not on a new column. Separation is obtained on the new column when additives are incorporated in the mobile phase. Ye et al. [91] demonstrated a memory effect for acidic additives on CHIRALPAK AD enabled by a dramatic change in selectivity with ethanesulfonic acid added to the mobile phase. The effect of n-butylamine on the separation of β-blockers was less dramatic but appeared to be quite persistent. One of the major challenges in demonstrating a memory effect is selecting compounds that clearly show the effects of the additive. Slight improvements in peak shapes tend to be subject to interpretation, leading to a need for more dramatic changes in response to additive. Screening experiments of a large number (50) of amine-containing compounds revealed that only nine compounds showed clear benefit from the addition of diethylamine (DEA) [93], suggesting that default incorporation of DEA into the mobile phase for basic compounds is unwarranted. Four of

282

Advances in Chromatography

the compounds that benefited from DEA were tertiary amines. Beyond that, no consistent structural aspect of compounds was apparent that benefited from additive use — not all tertiary amines benefit from DEA use. CHIRALPAK AD-H showed a persistent memory for DEA when hexane-based mobile phases were used. Polar organic mobile phases showed only a short additive memory. The memory effect was also short-lived on CHIRALCEL OJ-H. Ye et al. [90] suggested that flushing with acidic mobile phase could remove amine additives from the CSP. When this was evaluated, not only did flushing with trifluoroacetic acid-containing mobile phase remove the memory of DEA from the CSP, it gave separations that differed from those obtained prior to exposure to amine. Incorporation of DEA back into the mobile phase restored the separations previously obtained with DEA but memory was not established. The history of polysaccharide CSPs to acidic and amine additives can give unexpected results and dedicated columns are strongly recommended for additive use. As stated above, acidic additives are not required to elute acid analytes in SFC. This is usually attributed to the “acidic” nature of carbon dioxide. A protic modifier is required, and inclusion of an amine additive prevents elution of acidic analytes. These results corroborate an acid-base equilibrium in SFC mobile phases whereby the acidity of carbon dioxide is sufficient to transfer a proton from the alcohol modifier to the acidic analyte. An amine additive is basic enough to prevent this transfer. Amine additives have been used in SFC occasionally with the intent of improving peak shape [68, 69, 75, 94] of amine analytes. The common interpretation is that amine additives mask silanols that contribute to nonspecific retention of such amines. Diminishing nonspecific interactions would decrease retention but should also increase observed selectivity. Amine additives would also be expected to compete with amine analytes for specific binding sites, giving decreased retention but mixed effects on selectivity. This is the typical observation for a broad range of amine analytes [94]. Admittedly, amine additives have not been examined in depth in SFC possibly due to the relative lack of success of the technique with amine analytes. Amines often fail to elute or give peaks so distorted that optimization is not attempted. Ye et al. [95] found that use of cyclic primary amine additives increased retention of amine analytes in SFC, as noted above for HPLC. Increased retention was often accompanied by an increase in enantioselectivity and resolution but the effect required higher levels of additive (> 0.5%) than in HPLC. Recent work [96] tested the effect of ESA in SFC. It was found that incorporation of 0.1% ESA in both modifier and sample diluent gave excellent separation of many amine compounds that previously gave broad or absent peaks in SFC (Figure 6.9). This approach was found to work for a wide variety of basic compounds, including amino acids, amino acid esters, β-blockers and primary, secondary, and tertiary amines on both AD-H and OD-H columns. There was no memory effect. These findings suggest that the acid additive forms a salt ion-pair with the amine analyte, which is then separated intact in SFC.

The Use of Polysaccharide Phases in the Separation of Enantiomers

283

3.14

440 420 400 380 360 340 320 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 0

1

2

3

4

5

6

(a)

3.01

450 400

3.72

350 300 250 200 150 100 50 0 1

2

3

4

5

6

(b)

FIGURE 6.9 SFC chromatogram of chloramphetamine hydrochloride salt on a CHIRALPAK AD-H column using A = 20% ethanol and B = 20% ethanol containing 0.1% ESA modifier. Sample dissolved in modifier.

6.5 CONCLUSIONS The intent of this chapter was to provide an update on current understanding of polysaccharide CSPs in terms of their mechanism and the effect of various mobile phases on their performance. Decisions were made limiting the scope of this chapter, as it was not intended to be an application review. With the growth of applications on these CSPs, generating a comprehensive review that is not immediately outdated may not even be possible. An effort was made to cite helpful reviews that should point the reader to additional information, if desired.

284

Advances in Chromatography

Method development screening approaches that were not described here as testing an array of columns will maximize chances for success. The described effects of various mobile phases strongly indicate evaluation of hexane/isopropanol, hexane/ethanol, methanol, and acetonitrile mobile phases. The means used to accomplish this breadth of screening is a decision based on compromises between equipment availability and time demands. A consistent conclusion is that screening across a range of CSPs and mobile phases will save time in the long run. Without screening, considerable time could be wasted attempting to optimize a partial separation and additives should be used with care. Preparative applications were not discussed, although the majority of preparative chiral separations are accomplished on these phases. Immobilized polysaccharide stationary phases have been recently introduced as commercial products. Preliminary work has yielded intriguing results but there is not enough yet known to allow informed discussion of these CSPs.

REFERENCES 1. Francotte, E. “Enantioselective Liquid Chromatography: A Mature Separation Technology?” Presentation at ISCD-14, Hamburg, Germany 2002. 2. Okamoto, Y., Kawashima, M., and Hatada, K. Useful chiral packing materials for high-performance liquid chromatographic resolution of enantiomers: Phenylcarbamates of polysaccharide coated on silica gel. J. Am. Chem. Soc., 1984, 106, 5357–5359. 3. Okamoto, Y., Kawashima, M., and Hatada, K. Chromatographic resolution. XI. Controlled chiral recognition of cellulose triphenylcarbamate derivatives supported on silica gel. J. Chromatogr., 1986, 363, 173–186. 4. Okamoto, Y., Aburatani, R., and Hatada, K. Chromatographic chiral resolution. XIV. Cellulose tribenzoate derivatives as chiral stationary phases for high-performance liquid chromatography. J. Chromatogr., 1987, 389, 95–102. 5. Okamoto, Y., Aburatani, R., Fukumoto, T., and Hatada, K. Useful chiral stationary phases for HPLC. Amylose tris(3,5-dimethylphenylcarbamate) and tris(3,5-dichlorophenylcarbamate) supported on silica gel. Chem. Letters 1987, 1987, 1857–1860. 6. Okamoto, Y., Kaida, Y., Hayashida, H., and Hatada, K. Tris(1-phenylethylcarbamate)s of cellulose and amylose as useful chiral stationary phases for chromatographic optical resolution. Chem. Letters 1990, 1990, 909–912. 7. Francotte, E., Wolf, R.M., and Lohmann, D. Chromatographic resolution of racemates on chiral stationary phases. I. Influence of the supramolecular structure of cellulose triacetate. J. Chromatogr., 1985, 347, 25–37. 8. Yashima, E., Sahavattanapong, P., and Okamoto, Y. HPLC enantioseparation on cellulose tris(3,5-dimethylphenylcarbamate) as a chiral stationary phase: Influence of pore size of silica gel, coating amount, coating solvent, and column temperature on chiral discrimination. Chirality, 1996, 8, 446–451. 9. Francotte, E. and Zhang, T. Supramolecular effects in the chiral discrimination of meta-methylbenzoyl cellulose in high-performance liquid chromatography. J. Chromatogr., 1995, 718, 257–266. 10. Okamoto, Y. and Kaida, Y. Resolution by high-performance liquid chromatography using polysaccharide carbamates and benzoates as chiral stationary phases. J. Chromatogr. A, 1994, 666, 403–419.

The Use of Polysaccharide Phases in the Separation of Enantiomers

285

11. Yashima, E. and Okamoto, Y. Chiral discrimination on polysaccharide derivatives. Bull. Chem. Soc. Jpn., 1995, 68, 3289–3307. 12. Yashima, E., Yamamoto, C., and Okamoto, Y. Polysaccharide-based chiral LC columns. SYNLETT 1997, 1997, 344–360. 13. Okamoto, Y. and Yashima, E. Polysaccharide derivatives for chromatographic separation of enantiomers. Angew. Chem. Int. Ed., 1998, 37, 1020–1043. 14. Yashima, E. Polysaccharide-based chiral stationary phases for high-performance liquid chromatographic enantioseparation. J. Chromatogr. A, 2001, 906, 102–125. 15. Pirkle, W.H., Gan, K.Z., and Brice, L.J. The enhancement of enantioselectivity by halogen substituents. Tetra. Assymetry, 1996, 7, 2813–2816. 16. Blackwell, J.A., Waltermire, R.E., and Stringham, R.W. The role of modifiers in high selectivity chiral separations on Pirkle-type columns. Enantiomer, 2001, 6, 353–367. 17. Hesse, G. and Hagel, R. Complete separation of a racemic mixture by elution chromatography on cellulose triacetate. Chromatographia, 1973, 6, 227–280. 18. Ichida, A., Shibata, T., Okamoto, I., Yuki, Y., Namikoshi, H., and Toga, Y. Chromatographia, 1984, 19, 280. 19. Shibata, T., Okamoto, I., and Ishii, K. J. Liq. Chromatogr., 1983, 9, 313. 20. Okamoto, Y., Kawashima, M., Yamamoto, K., and Hatada, K. Chem. Lett. 1984, 1984, 739. 21. Shibata, T., Sei, T., Nishimura, H., and Deguchi, K. Chromatographia, 1987, 24, 552. 22. Kaida, Y. and Okamoto, Y. Optical resolution by high-performance liquid chromatography on benzylcarbamates of cellulose and amylose. J. Chromatogr., 1993, 641, 267–278. 23. Yashima, E., Kasashima, E., and Okamoto, Y. Enantioseparation on 4-halogen-substituted phenylcarbamates of amylose as chiral stationary phases for high-performance liquid chromatography. Chirality, 1997, 9, 63–68. 24. Yashima, E., Yamada, M., and Okamoto, Y. An NMR study of chiral recognition relevant to the liquid chromatographic separation of enantiomers by a cellulose derivative. Chem. Lett. 1994, 1994, 579–582. 25. Yashima, E., Yamamoto, C., and Okamoto, Y. NMR studies of chiral discrimination relevant to the liquid chromatographic enantioseparation by a cellulose phenylcarbamate. J. Am. Chem. Soc., 1996, 118, 4036–4048. 26. Wenslow, R.M. and Wang, T. Solid-state NMR characterization of amylose tris(3,5dimethylphenylcarbamate) chiral stationary phase structure as a function of mobile phase composition. Anal. Chem., 2001, 73, 4190–4195. 27. O’Brien, T., Crocker, L., Thompson, R., Thompson, K., Toma, P.H., Conlon, D.A., Feibush, B., Moeder, C., Bicker, G., and Grinberg, N. Mechanistic aspects of chiral discrimination on modified cellulose. Anal. Chem., 1997, 69, 1999–2007. 28. Bereznitski, Y., LoBrutto, R., Variankaval, N., Thompson, R., Thompson, K., Sajonz, P., Crocker, L.S., Kowal, J., Cai, D., Journet, M., Wang, T., Wyvratt, J., and Grinberg, N. Mechanistic aspects of chiral discrimination on an amylose tris (3,5-dimethylphenyl)carbamate. Enantiomer, 2002, 7, 305–315. 29. Wang, F., O’Brien, T., Dowling, T., Bicker, G., and Wyvratt, J. Unusual effect of column temperature on chromatographic enantioseparation of dihydropyrimidinone acid and methyl ester on amylose chiral stationary phase. J. Chromatogr. A, 2002, 958, 69–77.

286

Advances in Chromatography

30. Wang, F., Wenslow, R.M., Jr., Dowling, T.M., Mueller, K.T., Santos, I., and Wyvratt, J. Characterization of a thermally induced irreversible conformational transition of amylose tris(3,5-dimethylphenylcarbamate) chiral stationary phase in enantioseparation of dihydropyrimidinone acid by quasi-equilibrated liquid chromatography and solid-state NMR. Anal. Chem., 2003, 75, 5877–5885. 31. Wainer, I.W. and Alembik, M.C. Resolution of enantiomeric amides on a cellulosebased chiral stationary phase. Steric and electronic effects. J. Chromatogr., 1986, 358, 85–93. 32. Wainer, I.W., Stiffin, R.M., and Shibata, T. Resolution of enantiomeric aromatic alcohols on a cellulose tribenzoate high-performance liquid chromatography chiral stationary phase. A proposed chiral recognition mechanism. J. Chromatogr., 1987, 411, 139–151. 33. Camilleri, P., Livingstone, D.J., Murphy, J.A., and Manallack, D.T. Chiral chromatography and multivariate quantitative structure-property realationships of benzimidazole sulphoxides. J. Computer-Aided Molec. Des., 1993, 7, 61–69. 34. Booth, T.D. and Wainer, I.W. Investigation of the enantioselective separations of alkylarylcarboxylic acids on an amylose tris(3,5-dimethylphenylcarbamate) chiral stationary phase using quantitative structure-enantioselective retention relationships. Identification of a conformationally driven chiral recognition mechanism. J. Chromatogr. A, 1996, 737, 157–169. 35. Booth, T.D., Azzaoui, K., and Wainer, I.W. Prediction of chiral chromatographic separations using combined multivariate regression and neural networks. Anal. Chem., 1997, 69, 3879–3883. 36. Yamamoto, C., Yashima, E., and Okamoto, Y. Computational studies on chiral discrimination mechanism of phenylcarbamate derivatives of cellulose. Bull. Chem. Soc. Jpn., 1999, 72, 1815–1825. 37. Gaffney, M.H., Stiffin, R.M., and Wainer, I.W. The effect of alcoholic mobile phase modifiers on retention and stereoselectivity on a commercially available cellulosebased HPLC chiral stationary phase: An unexpected reversal in enantiomeric elution order. Chromatographia, 1989, 27, 15–18. 38. Okamoto, M. and Nakazawa, H. Reversal of elution order during direct enantiomeric separation of pyriproxyfen on a cellulose-based chiral stationary phase. J. Chromatogr., 1991, 588, 177–180. 39. Balmer, K., Lagerstrom, P.-O., and Persson, B.-A. Reversed retention order and other stereoselective effects in the separation of amino alcohols on Chiracel OD. J. Chromatogr., 1992, 592, 331–337. 40. Balmer, K., Persson, B.-A., and Lagerstrom, P.-O. Stereoselective effects in the separation of enantiomers of omeprazole and other substituted benzimidazoles on different chiral stationary phases. J. Chromatogr. A, 1994, 660, 269–273. 41. Wang, T. and Chen, Y.W. Application and comparison of derivatized cellulose and amylose chiral stationary phases for the separation of enantiomers of pharmaceutical compounds by high-performance liquid chromatography. J. Chromatogr. A, 1999, 855, 411–421. 42. Wang, T., Chen, Y.W., and Vailaya, A. Enantiomeric separation of some pharmaceutical intermediates and reversal of elution orders by high-performance liquid chromatography using cellulose and amylose tris(3,5-dimethylphenylcarbamate) derivatives as stationary phases. J. Chromatogr. A, 2000, 902, 345–355. 43. Blackwell, J.A., Stringham, R.W., Xiang, D., and Waltermire, R.E. Empirical relationship between chiral selectivity and mobile phase modifier properties. J. Chromatogr. A, 1999, 852, 383–394.

The Use of Polysaccharide Phases in the Separation of Enantiomers

287

44. Krstulovic, A.M., Rossey, G., Porziemsky, J.-P., Long, D., and Chekroun, I. Direct determination of the enantiomeric purity of (5S)-3-isopropyl-5-p-toluenesulphonyloxymethyloxazolidin-2-one on a cellulose-based chiral stationary phase. J. Chromatogr., 1987, 411, 461–465. 45. Schulte, M., Ditz, R., Devant, R.M., Kinkel, J.N., and Charton, F. Comparison of the specific productivity of different chiral stationary phases used for simulated movingbed chromatography. J. Chromatogr. A, 1997, 769, 93–100. 46. Weinz, C., Blaschke, G., and Schiebel, H.-M. Investigation of the stereoselective in vitro biotransformation of glutethimide by high-performance liquid chromatography and capillary electrophoresis. J. Chromatogr. B, 1997, 690, 233–242. 47. Chankvetadze, B., Yamamoto, C., and Okamoto, Y. Enantioseparation of selected chiral sulfoxides using polysaccharide-type chiral stationary phases and polar organic, polar aqueous-organic and normal-phase eluents. J. Chromatogr. A, 2001, 922, 127–137. 48. Tachibana, K. and Ohnishi, A. Reversed-phase liquid chromatographic separation of enantiomers on polysaccharide type chiral stationary phases. J. Chromatogr. A, 2001, 906, 127–154. 49. Chankvetadze, B., Kartozia, I., Yamamoto, C., and Okamoto, Y. Comparative enantioseparation of selected chiral drugs on four different polysaccharide-type chiral stationary phases using polar organic mobile phases. J. Pharm. Biomed. Anal., 2002, 27, 467–478. 50. Cass, Q.B. and Batigalhia, F. Enantiomeric resolution of a series of chiral sulfoxides by high-performance liquid chromatography on polysaccharide-based columns with multimodal elution. J. Chromatogr. A, 2003, 987, 445–452. 51. Sousa, E.P., Tiritan, M.E., Oliveira, R.V., Afonso, C.M.M., Cass, Q.B., and Pinto, M.M.M. Enantiomeric resolution of kielcorin derivatives by HPLC on polysaccharide stationary phases using multimodal elution. Chirality, 2004, 16, 279–285. 52. Miller, L., Orihuela, C., Fronek, R., and Murphy, J. Preparative chromatographic resolution of enantiomers using polar organic solvents with polysaccharide chiral stationary phases. J. Chromatogr. A, 1999, 865, 211–226. 53. Ishikawa, A. and Shibata, T. Cellulosic chiral stationary phase under reversed-phase condition. J. Liq. Chromatogr., 1993, 16, 859–878. 54. Tang, L., Xiang, D. and Blackwell, J.A. A comparative study of factors affecting chiral separations under reversed-phase conditions using cellulosic and Pirkle-type chiral selectors. Enantiomer, 2000, 5, 345–355. 55. Tachibana, K. and Ohnishi, A. Reversed-phase liquid chromatographic separation of enantiomers on polysaccharide type chiral stationary phases. J. Chromatogr. A, 2001, 906, 127–154. 56. Perrin, C., Matthijs, N., Mangelings, D., Granier-Loyaux, C., Maftouh, M., Massart, D.L., and Vander Heyden, Y. Screening approach for chiral separation of pharmaceuticals. Part II. Reversed-phase liquid chromatography. J. Chromatogr. A, 2002, 966, 119–134. 57. Kot, A., Sandra, P., and Venema, A. Sub- and supercritical fluid chromatography on packed columns: A versatile tool for the enatioselective separation of basic and acidic drugs. J. Chromatogr. Sci., 1994, 32, 439–448. 58. Petersson, P. and Markides, K.E. Review. Chiral separations performed by supercritical fluid chromatography. J. Chromatogr. A, 1994, 666, 381–394. 59. Williams, K.L. and Sander, L.C. Review. Enantiomer separations on chiral stationary phases in supercritical fluid chromatography. J. Chromatogr. A, 1997, 785, 149–158.

288

Advances in Chromatography

60. Terfloth, G. Review. Enantioseparations in super- and subcritical fluid chromatography. J. Chromatogr. A, 2001, 906, 301–307. 61. Sassiat, P.R., Mourier, P., Caude, M.H., and Rosset, R.H. Measurement of diffusion coefficients in supercritical carbon dioxide and correlation with the equation of Wilke and Chang. Anal. Chem., 1987, 59, 1164–1170. 62. Chester, T.L. Chromatography from the mobile phase perspective. Anal. Chem., 1997, 69, 165A–169A. 63. Mourier, P.A., Eliot, E., Caude, M.H., Rosset, R.H., and Tambute, A.G. Supercritical and subcritical fluid chromatography on a chiral stationary phase for the resolution of phosphine oxide enantiomers. Anal. Chem., 1985, 57, 2819–2823. 64. Macaudiere, P., Caude, M., Rosset, R., and Tambute, A. Chiral resolution of a series of 3-thienylcyclohexylglycolic acids by liquid or subcritical fluid chromatography. A mechanistic study. J. Chromatogr., 1988, 450, 255–269. 65. Macaudiere, P., Caude, M., Rosset, R., and Tambute, A. Chiral resolutions in SFC: Mechanisms and applications with various stationary phases. J. Chromatogr. Sci., 1989, 27, 583–591. 66. Strubinger, J.R., Song, H., and Parhcer, J.F. High-pressure phase distribution isotherms for supercritical fluid chromatographic systems. 2. Binary isotherms of carbon dioxide and methanol. Anal. Chem., 1991, 63, 104–108. 67. Stringham, R.W. and Blackwell, J.A. Factors that control successful entropically driven chiral separations in SFC and HPLC. Anal. Chem., 1997, 69, 1414–1420. 68. Biermanns, P., Miller, C., Lyon, V., and Wilson, W. Chiral resolution of -blockers by packed column supercritical fluid chromatography. LC-GC, 1993, 11, 744–747. 69. Kot, A., Sandra, P., and Venema, A. Sub- and supercritical fluid chromatography on packed columns: A versatile tool for the enantioselective separation of basic and acidic drugs. J. Chromatogr. Sci., 1994, 32, 439–448. 70. Chen, L., Thompson, R., Grinberg, N., Johnson, B., and Ellison, D. Enantioseparation of a protease inhibitor, indinavir, by subcritical fluid chromatography. J. Chromatogr. A, 1996, 752, 101–109. 71. Chen, L., Thompson, R.A., Johnson, B.D., and Wyvratt, J.M. Enatiomeric separation of a thiazolbenzenesulfonamide compound using packed-column subcritical fluid chromatography. Chirality, 2002, 14, 393–399. 72. Loughlin, T., Thompson, R., Bicker, G., Tway, P., and Grinberg, N. Use of subcritical fluid chromatography for the separation of enantiomers using packed cellulose based stationary phase. Chirality, 1996, 8, 157–162. 73. Lynam, K.G. and Nicolas, E.C. Chiral HPLC versus chiral SFC: Evaluation of longterm stability and selectivity of Chiralcel OD using various eluents. J. Pharmaceut. & Biomed. Analysis, 1993, 11, 1197–1206. 74. Stringham, R.W., Lynam, K.G., and Grasso, C.C. Application of subcritical fluid chromatography to rapid chiral method development. Anal. Chem., 1994, 66, 1949–1954. 75. Anton, K., Eppinger, J., Frederiksen, L., Francotte, E., Berger, T.A., and Wilson, W.H. Chiral separations by packed-column super- and subcritical fluid chromatography. J. Chromatogr. A, 1994, 666, 395–401. 76. Medvedovici, A., Sandra, P., Toribio, L., and David, F. Chiral packed column subcritical fluid chromatography on polysaccharide and macrocyclic antibiotic chiral stationary phases. J. Chromatogr. A, 1997, 785, 159–171. 77. Berger, T.A. and Wilson, W.H. Packed column supercritical fluid chromatography with 220,000 plates. Anal. Chem., 1993, 65, 1451–1455.

The Use of Polysaccharide Phases in the Separation of Enantiomers

289

78. Phinney, K.W., Sander, L.C., and Wise, S.A. Coupled achiral/chiral column techniques in subcritical fluid chromatography for the separation of chiral and achiral compounds. Anal. Chem., 1998, 70, 2331–2335. 79. Wilson, W.H. Direct enantiomeric resolution of ibuprofen and flurbiprofen by packed column SFC. Chirality, 1994, 6, 216–219. 80. Okamoto, Y., Aburatani, R., Kaida, Y., and Hatada, K. Chem. Lett., 1988, 1125. 81. Okamoto, Y., Kawashima, M., Aburatani, R., Hatada, K., Nishiyama, T., and Masuda, M. Optical resolution of -blockers by HPLC on cellulose triphenylcarbamate derivatives. Chem Lett. 1986, 1986, 1237–1240. 82. Aboul-Enein, H.Y. and Islam, M.R. Direct separation and optimization of timolol enantiomers on a cellulose tris-3,5-dimethylphenylcarbamate high-performance chiral stationary phase. J. Chromatogr., 1990, 511, 109–114. 83. Okamato, Y., Aburatani, R., and Hatada, K. Chromatographic resolution. XXI. Direct optical resolution of abscisic acid by high-performance liquid chromatography on cellulose tris(3,5-dimethylphenylcarbamate). J. Chromatogr., 1988, 448, 454–455. 84. Krstulovic, A.M., Fouchet, M.H., Burke, J.T., Gillet, G., and Durand, A. Direct enantiomeric separation of betaxolol with applications to analysis of bulk drug and biological samples. J. Chromatogr., 1988, 452, 477–483. 85. Tang, Y. Significance of mobile phase composition in enantioseparation of chiral drugs by HPLC on a cellulose-based chiral stationary phase. Chirality, 1996, 8, 136–142. 86. Tang, Y., Zielinski, W., and Bigott, H. Separation of nicotine and nornicotine enantiomers via normal phase HPLC on derivatized cellulose chiral stationary phases. Chirality, 1998, 10, 364–369. 87. Perrin, C., Vu, V.A., Matthijs, N., Maftouh, M., Massart, D.L., and Vander Heyden, Y. Screening approach for chiral separation of pharmaceuticals. Part I. Normal-phase liquid chromatography. J. Chromatogr. A, 2002, 947, 69–83. 88. Luz de la Puente, M., White, C.T., Rivera-Sagredo, A., Reilly, J., Burton, K., and Harvey, G. Impact of normal-phase gradient elution in chiral chromatography: A novel, robust, efficient and rapid chiral screening procedure. J Chromatogr A, 2003, 983, 101–114. 89. Ye, Y.K. and Stringham, R.W. Effect of mobile phase acidic additives on enantioselectivity for phenylalanine analogs. J. Chromatogr. A, 2001, 927, 47–52. 90. Ye, Y.K. and Stringham, R.W. Effect of mobile phase amine additives on enantioselectivity for phenylalanine analogs. J. Chromatogr. A, 2001, 927, 53–60. 91. Ye, Y.K., Lord, B., and Stringham, R.W. Memory effect of mobile phase additives in chiral separations on a Chiralpak AD column. J. Chromatogr. A, 2002, 945, 139–146. 92. Ye, Y.K., Lord, B., Yin, L., and Stringham, R.W. Enantioseparation of amino acids on a polysaccharide-based chiral stationary phase. J. Chromatogr. A, 2002, 945, 147–159. 93. Stringham, R.W., Lynam, K.G., and Lord, B.S. Memory effect of diethylamine mobile phase additive on chiral separations on polysaccharide stationary phases. Chirality, 2004, 16, 493–498. 94. Phinney, K.W. and Sander, L.C. Additive concentration effects on enantioselective separations in supercritical fluid chromatography. Chirality, 2003, 15, 287–294. 95. Ye, Y.K., Lynam, K.G., and Stringham, R.W. Effect of amine mobile phase additives on chiral subcritical fluid chromatography using polysaccharide stationary phases. J. Chromatogr. A, 2004, 1041, 211– 217.

290

Advances in Chromatography

96. Stringham, R.W. Chiral separation of amines in subcritical fluid chromatography using polysaccharide stationary phases and acidic additives. J. Chromatogr. A, 2005, 1070, 163–170.

7

Chaotropic Effects in RP-HPLC R. LoBrutto Novartis Pharmaceutical Corporation, East Hanover, N.J.

Y.V. Kazakevich Seton Hall University, South Orange, N.J.

CONTENTS 7.1 7.2 7.3 7.4 7.5 7.6

Introduction ..................................................................................................291 Chaotropic Effect .........................................................................................292 Chaotropic Model ........................................................................................295 Effect of Different Counteranions ...............................................................297 Retention of the Counteranions ...................................................................298 Effect of the Counterion Type and Concentration on Peak Efficiency and Asymmetry...........................................................................301 7.7 Applications in the Pharmaceutical Industry ..............................................304 7.8 Concluding Remarks....................................................................................312 References..............................................................................................................313

7.1 INTRODUCTION Most of the samples analyzed in modern reversed-phase HPLC include ionic or ionizable analytes. While the mechanism of HPLC retention in reversed-phase mode is considered to be competitive hydrophobic interactions of the analyte and organic eluent modifier with the stationary phase, the retention of ionic or ionizable analytes is significantly influenced by other factors such as mobile phase pH, salt type, and salt concentration. Ionic interactions are the strongest among all types of molecular interactions employed in HPLC and even minor variations in reversed-phase HPLC conditions (e.g., concentration of counterions, pH) usually lead to dramatic changes in the retention of ionic analytes. Introduction of ionic interactions in reversed-phase HPLC has long been used to the chromatographer’s advantage as a mode to selectively control the retention of ionic compounds and is called ion-pair chromatography (IPC). This mode is especially useful when the retention of an ionic analyte with the lack of hydrophobic

291

292

Advances in Chromatography

moieties needs to be increased. The usual approach is the addition of the amphiphilic ions in the mobile phase such as octyl sulfonic acid and other long chain alkyl acidic modifiers. The increase of the retention of ionic analytes has a dualistic nature: 1. Amphiphilic ions themselves could get adsorbed on the surface of the reversed-phase adsorbent, thus introducing an ionic component in the analyte surface interactions 2. Analyte forms an ion-pair with an amphiphilic ion, which increases its apparent hydrophobicity and increases its hydrophobic interaction with the stationary phase Theory and retention mechanism of ion-pair chromatography have been extensively studied (Ståhlberg, Vigh, Horvath, and others) in the past three decades and several principally different theories have been proposed [1–6]. Both mechanisms outlined above essentially lead to very similar retention dependencies and unambiguous interpretation of chromatographic results is practically impossible since the retention dependencies of the counterion concentration in the eluent are similar in both cases. Moreover, both mechanisms probably coexist and only nonchromatographic experiments can yield information on the magnitude of the predominate process [7]. Application of amphiphilic ions for the alteration of the retention of ionic analytes is considered a last resort measure, since the adsorption of these ions on the surface of the stationary phase usually permanently converts the reversed-phase column into essentially an ion-exchange column. Application of small (usually inorganic) liophillic ions is preferable and their influence on the analyte retention is more subtle. Chromatography performed with these anions is usually more reproducible and the columns employed can be washed with water/organic eluents and reverted back to the native state to be used for other applications. Study of the effect of lipophilic ions on the retention of ionic analytes in reversedphase HPLC has lead to the development of yet another possible theory of their influence on the chromatographic retention of basic compounds [8, 9]. Ionic analytes in water-organic mixtures are solvated. Solvation shell suppresses the analyte’s ability for hydrophobic interactions with the stationary phase, thus effectively decreasing the analtye’s retention. Controlled disruption of the solvation shell allows for control of the analyte retention. Presence of the counterions in close proximity to the ionic solvated analyte leads to the disruption of the analyte solvation shell. This effect is known as chaotropic control for the retention of ionic compounds in reversed-phase chromatography.

7.2 CHAOTROPIC EFFECT The counteranion of an acid interacts with positively charged basic analytes and can form an ion-associated complex. This interaction alters the retention of the basic analyte due to the changes in charge density, polarizability, and solvation. Counteranions that have a less-localized charge, high polarizability, and lower degree of hydration show a significant effect on the retention of protonated basic analytes and are known as chaotropic ions. Chaotropic ions change the structure of water in the

Chaotropic Effects in RP-HPLC

293

(a)

(b)

0

0

5

5

10

10

Total ClO4− (mM)

pH

Rt. (min)

47

1.4

1.4

41

1.5

20

1.8

14

2.0

7

2.2

5.2

Total ClO4− (mM)

pH

100

2.0

89

2.0

79

2.0

70

2.0

55

2.0

Rt. (min) 9.9

7.7

FIGURE 7.1 Variation of the retention of basic analyte (pKa 7.5) with (a) mobile phase pH and (b) counteranion concentration.

direction of greater disorder; therefore, the solvation shell of the basic analytes can be disrupted due to ion interaction with the chaotropic anions. As the concentration of the counteranion increases, the solvation of the protonated basic analyte decreases. The primary sheath of water molecules around the basic analytes is disrupted and this decreases the solvation of the basic analyte. The decrease in the analyte solvation increases the analyte hydrophobicity and leads to increased interaction with the hydrophobic stationary phase and increased retention for the basic analytes. The chaotropic effect discussed above is dependent on the concentration of the free counteranion and not the concentration of the protons in solution, suggesting that change in retention of the protonated basic analyte can be observed with the increase in concentration of the counteranion by the addition of a salt at a constant pH, as shown in Figure 7.1 for a pharmaceutical compound containing an aromatic amine with a pKa of 5. In the example above, the retention of pharmaceutical analyte X was first altered by a decrease of mobile phase pH (Figure 7.1a), and in the second case (Figure 7.1b) the pH was maintained constant and the concentration of counteranion was increased via addition of its sodium salt. The resulting effect on the retention of

294

Advances in Chromatography

2.8 2.6

pH = 2.0 pH = 1.8

2.4 2.2

pH = 2.2

pH = 2.0

2

pH = 2.0 pH = 2.0

k

pH = 1.8

1.8

pH = 2.0

1.6 pH = 2.2 1.4 1.2 1 0

20

40 60 Conc. ClO4− (mM)

Variable pH

HClO4

Variable pH

KH2PO4 adj. w/HClO4

pH = 2.0

80

100

KH2PO4 adj. w/HClO4 and NaClO4

FIGURE 7.2 Retention of lutidine as a function of ClO4 counterion concentration with variable pH (circles), fixed pH (triangles) and variable pH with phosphate buffer (squares).

basic analyte is strikingly similar if both dependencies are plotted against the concentration of free counteranions of ClO4, as shown in Figure 7.2. Disruption of the basic analyte solvation shell should be possible with practically any counteranion employed, and the degree of this disruption will be dependent on the “chaotropic nature” of the anion used. Chaotropic activity of counteranions has been established according to their ability to destabilize or bring disorder (chaos) to the structure of water [10, 11]. Even a very low counteranion concentration in the mobile phase will cause significant initial disruption of the solvation shell, thus leading to the significant increase of the analyte retention while in the high concentration region, a type of a saturation effect is observed (Figure 7.3). Logically, at high counteranion concentration, when all solvation shells are fully disrupted, any further increase of the counteranion concentration should not cause any additional retention increase. As was shown above, chaotropic effect is related to the influence of the counteranion of the acidic modifier on the analyte solvation and is independent on the mobile phase pH, as far as complete protonation of the basic analyte is achieved. Analyte interaction with counteranion causes a disruption of the analyte solvation shell, thus affecting its hydrophobicity. Increase of the analyte hydrophobicity results in a corresponding increase of retention. This process shows a “saturation” limit, when counteranion concentration is high enough to effectively disrupt the solvation of all analyte molecules. A further increase of counteranion concentration does not produce any noticeable effect on the analyte retention.

Chaotropic Effects in RP-HPLC

295

3,4 dimethylpyridine

1.4

1

1.2 2

1

1. Perchlorate Chromatographic conditions Column: 15 cm × 0.46 cm Zorbax XDB-C18 Eluent: 90% Aqueous/10% MeCN 2. Trifluoroacetate Aqueous: 1. Water + xHClO4 pH = 1–3 pH = 1–3 2. Water + yTFA, 3. Water + zH3PO4 pH = 1.6–3 3. Dihydrogen phosphate Flow rate: 1 ml/min 3

k

0.8 0.6 0.4 0.2 0 0

20

40

60

80

Conc. counteranion (mM)

FIGURE 7.3 Influence of different counterions on the retention of 3,4-dimethylpyridine.

7.3 CHAOTROPIC MODEL If the counteranion concentration is low, some analyte molecules have a disrupted solvation shell and some do not, due to the limited amount of counteranions present at any instant within the mobile phase. If we assume an existence of the equilibrium between solvated and desolvated analyte molecules and counteranions, this mechanism could be described mathematically [9]. The assumptions for this model are: 1. Analyte concentration in the system is low enough that analyte–analyte interactions could be considered nonexistent. 2. The chromatographic system is in thermodynamic equilibrium. The analyte solvation–desolvation equilibrium inside the column could be written in the following form: BS+ + A – ⇔ B + … A –

(7.1)

where BS+ is a solvated basic analyte, A– is a counteranion and B+…A– is the desolvated ion-associated complex. The total amount of analyte injected is [B], analyte in its solvated form is [B + S] and analyte in its desolvated form is denoted as [B+…A–], indicating its interaction with counteranions. The equilibrium constant of the Equation 7.1 is ⎡ B + … A – ⎤⎦ K=⎣ + ⎡⎣ BS ⎤⎦ ⎡⎣ A – ⎤⎦

(7.2)

296

Advances in Chromatography

Total analyte amount is equal to the sum of the solvated and desolvated forms of analyte,

[ B ] = ⎡⎣ BS+ ⎤⎦ + ⎡⎣ B+ … A – ⎤⎦

(7.3)

The fraction of solvated analyte could be expressed as ⎡ BS+ ⎤ θ= ⎣ ⎦ [ B]

(7.4)

The fraction of the desolvated analyte in the mobile phase could be expressed as ⎡⎣ B + … A – ⎤⎦ 1– θ = [ B]

(7.5)

Substituting Equation 7.4 and Equation 7.5 in Equation 7.2, we can write an expression for the equilibrium constant: K=

1– θ θ ⋅ ⎡⎣ A – ⎤⎦

(7.6)

Solving Equation 7.6 for θ (solvated fraction) we get θ=

1 K ⎡⎣ A – ⎤⎦ + 1

(7.7)

Equation 7.7 shows that the solvated fraction of the analyte is dependent on the counteranion concentration and the desolvation equilibrium parameter. Completely solvated analyte has a retention factor (even if it is equal to 0), which we denote as ks, while the corresponding retention factor for desolvated form we denote as kus. Assuming that solvation–desolvation equilibrium is fast, we can express the overall retention factor of injected analyte as a sum of the retention factor of solvated form multiplied by the solvated fraction (θ) and the retention factor of the desolvated form multiplied by the desolvated fraction (1 – θ), or k = ks · θ + kus · (1 – θ)

(7.8)

Substituting θ in Equation 7.8, from Equation 7.7 we get ⎛ ⎞ ⎛ ⎞ 1 1 k = ks ⎜ – k ⎟ ⎜ ⎟ + kus us ⎜⎝ K ⎡ A – ⎤ + 1 ⎟⎠ ⎜⎝ K ⎡ A – ⎤ + 1 ⎠⎟ ⎣ ⎦ ⎣ ⎦

(7.9)

Chaotropic Effects in RP-HPLC

297

1.2 kus = 1.13

Desolvated

Retention factor

1 0.8 0.6 0.4

ks = 0.313

Solvated

0.2 0 0

20 40 60 Counteranion concentration (mMole/L)

80

FIGURE 7.4 Experimental dependence of the retention of basic analyte on the counterion concentration (points) and corresponding theoretical curve for this effect calculated using Equation (7.10). Reprinted with permission from [9].

and the final form can be rewritten as: k=

ks – kus + kus K ⋅ ⎡⎣ A – ⎤⎦ + 1

(7.10)

This equation has three parameters: ks is a “limiting” retention factor for solvated analyte, kus is a “limiting” retention factor for desolvated analyte and K is a desolvation parameter [9]. The function described by the equation 7.10 is shown in Figure 7.4.

7.4 EFFECT OF DIFFERENT COUNTERANIONS The chaotropic theory was shown to be applicable in many cases where small inorganic ions were used for the alteration of the retention of basic pharmaceutical compounds [12–15]. Equation 7.10 essentially attributes the upper retention limit for completely desolvated analyte to the hydrophobic properties of the analyte alone. In other words, there could be a significantly different concentration needed when different counterions are employed in the eluent for complete desolvation of the analyte. Therefore, the resulting analyte hydrophobicity, and thus retention characteristics of analyte in completely desolvated form, should be essentially independent of the type of counteranion employed. Alternately, experimental results show that the use of different counterions leads to the different retention limits of completely desolvated analyte (Figure 7.5 clearly illustrates this effect). This discrepancy essentially suggests the existence of other processes with similar equilibrium and similar effects on the overall retention of protonated basic analytes.

298

Advances in Chromatography

4.5 4 3.5

BF4−

3

H2PO4−

2.5 k

CF3COO− PF6−

2

ClO4−

1.5 1 0.5 0 0

20

40 60 Conc. A−(mM)

80

100

FIGURE 7.5 Retention factor variations for acebutolol analyzed with different chaotropic agents. Reprinted with permission from [15].

7.5 RETENTION OF THE COUNTERANIONS Three distinct processes can be envisioned in the effect of chaotropic ions on the retention of basic analytes: 1. Classic ion-pairing: Formation of essentially neutral ion-pairs and their retention according to the reversed-phase mechanism. 2. Chaotropic: Counteranions disrupt the analyte solvation shell, thus increasing its apparent hydrophobicity and retention. 3. Lipophilic: Counteranions are adsorbed on the surface of the stationary phase, thus introducing an electrostatic component into the general hydrophobic analyte retention mechanism. In their recent papers, Griti and Guiochon are essentially advocating the domination of the first process [16–18]. They are explaining the counterion effect on the basis of the formation of a neutral ionic complex followed by its adsorption on the hydrophobic stationary phase. Similarity in adsorption, behavior of anionic and cationic species is interpreted as a confirmation of their adsorption in the form of neutral complexes. The retention of ionic components on reversed-phase columns is essentially regarded as ion-pair chromatography, which has been extensively developed by Horvath [19] and Sokolowski [20, 21] in the form of stochiometric adsorption of ionic species and by Ståhlberg in the form of adsorption of ions and formation of an electrical double layer [22]. The adsorption of amphiphilic ions was also experimentally confirmed while the actual interaction of the small lipophilic ions with hydrophobic stationary phase in reversed phase conditions was found only recently [23]. Most probably all three mechanisms exist while one of them is dominant, depending upon the eluent type, composition, and adsorbent surface properties.

Chaotropic Effects in RP-HPLC

299

For acetonitrile–water systems, acetonitrile forms a thick adsorbed layer on the surface of the hydrophobic bonded phase while methanol adsorption from water formed a classical monomolecular adsorbed layer [24]. The thick adsorbed layer of acetonitrile provides a suitable media for the adsorption of lipophilic ions on the stationary phase, adding an electrostatic component to the retention mechanism, while monomolecular adsorption of methanol should not significantly affect adsorption of ions. The study of the retention of chaotropic anions, BF4, perchlorate and PF6 was performed using acetonitrile/water eluents on alkyl- and phenyl-type phases with LC-MS detection (electrospray, negative ion mode) [23]. At all mobile phase conditions with acetonitrile/water, the PF6 ion exhibits the greatest retention and is the most lipophilic ion in the Hoffmeister series. This ion has the highest degree of charge delocalization and highest polarizability, which facilitates its possible dispersive (or van der Waals) interactions. These properties allow this ion to interact with acetonitrile molecules having significant -electron density. Other ions have similar properties but their ability for dispersive interactions is lower then PF6. At acetonitrile concentrations up to 20% (v/v) acetonitrile, all ions exhibit a maximum retention. General dependence of the analyte retention on the eluent composition in reversed-phase HPLC shows an exponential decay with the increase of the organic modifier concentration. This effect is usually described in the following form: ln(k′) = a + xb

(7.11)

where k′ is a retention factor, x is the eluent composition, and a and b are constants. This relationship has a thermodynamic background, since, in the partitioning retention model, the retention factor is proportional to the distribution equilibrium constant which, in turn, is an exponent of the excessive free Gibbs energy of the analyte in the chromatographic system. Excessive free Gibbs energy is the difference of the analyte potential in the stationary phase and its potential in the eluent, only true if retention is a result of a single process on the adsorbed surface (e.g., partitioning or adsorption). Alternately, if the retention mechanism is complex retention, dependencies will not adhere to Equation 7.11. The thick acetonitrile layer adsorbed on the bonded phase surface acts as a pseudo-stationary phase, thus making retention in acetonitrile/water systems as a superposition of two processes — partitioning into the acetonitrile layer and adsorption on the surface of the bonded phase. Based on the model described in the literature [24], analyte retention could be represented in the following form VR(cel ) = V0 + (Kp(cel ) – 1) Vads. + SKHKp(cel )

(7.12)

where VR(cel ) is the retention of analyte ions as a function of the eluent composition; V0 is the void volume; Kp(cel ) is the equilibrium constant for the distribution of the analyte ions between the eluent and adsorbed layer, Vads.; S is the adsorbent surface area; and KH is the adsorption equilibrium constant for analyte ions adsorption from neat acetonitrile on the corresponding stationary phase.

300

Advances in Chromatography

In a recent publication [23], we derive a semi-empirical equation that allows for the description of the retention of counterions in reversed-phase conditions. The overall expression for the description of the retention dependencies of analyte ions vs. eluent composition will have only four unknowns and allow numerical approximation of experimental retention data (shown as a function of the mole fraction of organic eluent component): ⎛ ΔGMeCN – x ΔGel . ⎞ VR ( x ) = V0 – Vads. ( x ) + A ⋅ exp ⎜ ⎟⎠ ⋅ (Vads. ( x ) + SK H ) (7.13) ⎝ RT

10 5

0

0.5

−5 MeCN mole fraction

1

Adsorption (norm. total)

Adsorption, (uMol/m2)

Essentially, Equation 7.13 describes the retention volume of the analyte as a sum of the mobile phase volume (V0 – Vads., assuming that the adsorbed acetonitrile layer is stagnant) and an energetic term that describes analyte partitioning into the adsorbed layer and its adsorption on the stationary phase surface. Volume of the adsorbed layer on top of the bonded phase is also a function of the acetonitrile concentration in the mobile phase (Figure 7.6). Coefficient ΔGel in Equation 7.13 is the energetic span of partitioning constant in the whole concentration region and reflects the excessive interactions of studied ions with water and acetonitrile as well as structural organization of molecules. The suggested phenomenological model describes the retention of PF6 ions on different reversed-phase columns very well. Average deviation of calculated values from experimentally measured values is on the level of 1%, which confirms that indeed a superposition of several processes govern the retention of liophilic ions in acetonitrile-water systems. Experimental values along with the theoretical curves are shown in Figure 7.7. The multilayered character of acetonitrile adsorption creates a pseudo-stationary phase of significant volume on the surface that acts as a suitable phase for the ion accumulation. In the low-organic concentration region — from 0 to 30% (v/v) of acetonitrile — ions studied show significant deviation from the ideal retention behavior (decrease in ion retention with increase in acetonitrile composition) due to the formation of the acetonitrile layer and significant adsorption of the chaotropic anions was observed, creating an electrostatic potential on the surface that provides 1

0.5

0

0.5 MeCN mole fraction

1

FIGURE 7.6 Acetonitrile excess adsorption isotherm from water on Zorbax Eclipse C8 adsorbent (left); normalized filling of adsorbed layer (right).

Chaotropic Effects in RP-HPLC

301

9 8

50 mM

Retention volume, mL

7 6

20 mM

5 10 mM 4 3

2 mM

2 0 mM 1 0 0

0.05

0.1 MeCN mole fraction

0.15

0.2

FIGURE 7.7 Experimental (points) and mathematical model (lines) dependencies of PF6 retention on Allure-PFP (perfluorinated propyl-phenyl phase) column versus the acetonitrile composition (shown in molar fractions) at different ionic strength (0, 2, 10, 20 and 50 mM adjusted with NH4Cl). Reprinted with permission from [23].

an additional retentive force for the enhancement of the retention of protonated basic analytes. At high concentration of organic content in the mobile phase, the retention of counteranions is decreased. At the same time, the dielectric constant of the mobile phase decreases with the increase of organic content, which favors the formation of the ionic pairs of basic analyte with the counteranions in the mobile phase; thus, this would also lead to the increase of the basic analyte retention. Overall, lipophilic ions (usually small ions capable of dispersive interactions) provide a useful means for selective alteration of the retention of basic analytes. Influence of these ions on the column properties is fully reversible and equilibration requires minimal time. Alternately, the mechanism of their effect is very complex and is dependent on the type of organic modifier used and on the concentration applied. Theoretical description and mathematical modeling of this process is a subject for further studies.

7.6 EFFECT OF THE COUNTERION TYPE AND CONCENTRATION ON PEAK EFFICIENCY AND ASYMMETRY Theoretically, a column can generate a certain maximum number of theoretical plates at the optimum flow rate. This number should be independent of the type of the

Advances in Chromatography

N (h/2)

302

13000 12000 11000 10000 9000 8000 7000 6000 5000 4000 3000

Tailing factor

0

3.4 3.2 3 2.8 2.6 2.4 2.2 2 1.8 1.6 1.4 1.2 1 0.8

0

20 30 BF4− (mM) (a)

10

5

10

15

20 25 30 BF4− (mM)

4-Nitrophenol

Dorzolamide HCl

2-Nitrophenol

Compound X

40

35

40

50

45

50

Timolol maleate

(b)

FIGURE 7.8 Effect of tetrafluoroborate concentration on analyte apparent efficiency and tailing factor. Column: Zorbax Eclipse XDB-C8, mobile phase: 0.1% (v/v) phosphoric acid + xBF4 [1 to 50 mM]: acetonitrile, ophthalmic compounds (10% acetonitrile), phenols (25% acetonitrile), (a) N(h/2) vs. tetrafluoroborate concentration. (b) Tailing factor vs. tetrafluoroborate concentration. Reprinted with permission from [12].

analyte and mobile phase. In reality, any secondary processes, energetic surface heterogeneity, or restrictions in sorption-desorption kinetics in the column will result in the specific decrease of the efficiency for a particular compound. Increasing the chaotropic counteranion concentration of perchlorate, hexafluorophosphate, and tetrafluoroborate in the mobile phase for basic compounds studied led to an increase in the apparent efficiency of the system until the maximum plate number for the column is achieved. In Figure 7.8a, the efficiency for the three basic ophthalmic drug compounds increases relatively fast when the concentration of counteranion BF4 was increased from 1 to 10 mM; the efficiency of the basic compounds then increases slowly until it achieves the maximum column efficiency (phenols, neutral markers). Also, with an increase of BF4 counteranion concentration,

Chaotropic Effects in RP-HPLC

303

the tailing factor of basic compounds decreases and approaches the tailing factor of the neutral analytes, phenolic compounds (Figure 7.8b). The PF6 counteranion has had the greatest effect on the improvement of the peak asymmetry at low concentrations compared to other chaotropic additives. At the highest concentration of counteranions (PF6, ClO4, BF4), the number of plates for most of the basic compounds studied was similar to that of the neutral markers. In contrast, the neutral markers (phenols) showed no significant changes in retention and efficiency with increased counteranion concentration. One of the origins of peak tailing in chromatography can be attributed to energetic surface heterogeneity with overloading of highly energetic adsorption sites [25–30]. Moreover, possible ion-exchange types of interactions with these sites could lead to slow sorption–desorption of solute molecules from the strong sites compared to the weak sites, leading to a further increase in band tailing [31, 32]. McCalley and others also showed that basic analyte sample loading can also have an effect on peak efficiency [25, 33, 34]; thus, a decrease in sample load has led to the improvement in the efficiency of basic compounds. However, injecting large sample sizes is sometimes necessary to enable the detection of small impurities with a consequent increase in basic analyte tailing factor and a decrease in peak symmetry. Chaotropic additives can be added to the mobile phase to suppress secondary interactions with the stationary phase. The adsorption of chaotropic counteranions in the adsorbed organic phase on top of the bonded phase can add an electrostatic component to the retention as well as suppressing some undesired secondary interactions leading to peak tailing of protonated basic compounds. The following trend in increase of basic analyte retention factor and decrease of tailing factor was found: PF6 > ClO4 ~ BF4 > H2PO4 [12]. Figure 7.9 shows an overlay of chromatograms of labetalol with different loads from 1 to 50 μg using a 10-mM dihydrogen phosphate mobile phase. These overlays reveal a typical pattern where the peak tails for different analyte loads coincide, indicating a so-called “thermodynamic overload” that occurs when analyte concentration exceeds the linear region on the adsorption isotherm, and this isotherm curvature inevitably leads to right angled peaks [35–38]. The greater the counteranion concentrations, the higher the adsorption capacity for this type of interaction and the straighter the analyte isotherm, resulting in a shorter tail. Excessive electrostatic interactions are relatively weak in the presence of a significant amount of counteranions in the solution and lead to the relatively low initial isotherm slope. Electrostatic interactions are relatively long-distance interactions, which would explain relatively high adsorption capacity and the nonexponential shape of the peak tail. With an increase in counteranion concentration at all analyte loadings, an increase in peak efficiency and decrease in peak tailing can be achieved [12]. Increasing the load of basic analytes in order to increase analyte sensitivity can lead to a decrease in apparent peak efficiency and an increase in peak tailing. However, if an analysis must be performed at a relatively high sample load, the addition of a chaotropic additive can be employed to increase the apparent peak efficiency and symmetry. Much higher loading capacities could be obtained by

Advances in Chromatography

Abs.

304

e.

67.9 mM ClO4−

d.

40.8 mM ClO4−

c.

27.2 mM ClO4− 6.8 mM ClO4−

b. 10 mM H2PO4−

a. 0

1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 Time (min)

FIGURE 7.9 Chromatographic overlays of Labetalol analyzed at different analyte concentrations using increasing mobile phase concentration of perchlorate anion. Chromatographic conditions: Column: Zorbax Eclipse XDB-C8, Analyte load: 3.3, 6.5, 31.2 μg, (a) 75:0.1 % (v/v) H3PO4: 25% acetonitrile, (b) 75:0.05 % (v/v) HClO4: 25% acetonitrile, (c) 75:0.2 % (v/v) HClO4: 25% acetonitrile, (d) 75:0.4 % (v/v) HClO4: 25% acetonitrile, (e) 75:0.5 % (v/v) HClO4: 25% acetonitrile. Reprinted with permission from [12].

operating columns with these mobile phase additives without substantial deterioration in efficiency.

7.7 APPLICATIONS IN THE PHARMACEUTICAL INDUSTRY Since approximately 80% of drugs include basic functional groups, HPLC behavior of basic compounds has attracted significant interest [39]. Reversed-phase HPLC separation of organic bases of different pKa values is of particular importance in the pharmaceutical industry [40, 41]. The chromatographic analysis of basic compounds is generally recommended to be carried at 2 pH units less than the analyte pKa. However, at these conditions the elution of protonated basic compounds could be close to the void volume. Another option might be to analyze these compounds in their neutral form — a mobile phase pH 2 units above the analyte pKa [42]. Note that going to higher pH might not be feasible due to the pH stability limit of the packing material or long analysis times obtained for the basic analyte in its neutral form. The advantages of employing chaotropic mobile phase additives at a pH where the basic analyte is in its fully protonated form provides the chromatographer an additional approach to adjust basic analyte retention and chromatographic selectivity without the need of changing the type of column, pH, or organic modifier. The retention behavior of basic compounds containing primary, secondary, tertiary, and quaternary amines can be enhanced as a function of the concentration of chaotropic

Chaotropic Effects in RP-HPLC

305

NH2+ Cl−

H3C

S O

S

SO2NH2

O

Abs.

H3C

1 mM PF6−, N = 7,674, Tf = 1.3 5 mM PF6−, N = 11,718, Tf = 1.2 10 mM PF6−, N = 11,988, Tf = 1.1 25 mM PF6−, N = 12,056, Tf = 1.1 0

5

10

15

20

25 30 Time (min)

35

40

45

50

FIGURE 7.10 Effect of hexafluorophosphate concentration on analyte retention, peak efficiency, N(h/2), and tailing factor. Chromatographic conditions: Column: Zorbax Eclipse XDBC8, mobile phase: 90%, 0.1% (v/v) phosphoric acid + xPF6 [1 to 25 mM]: 10% acetonitrile, flow rate: 1.0 ml/min, temperature: 25˚C, analyte load: 1 μg, wavelength: 254 nm. Reprinted with permission from [12].

mobile phase additives (ClO4, PF6, BF4, CF3CO2) at a low pH or at the pH below the analyte pKa [8, 15]. However, different inorganic counteranions at equimolar concentrations lead to a concomitant increase in retention as well as peak symmetry and increased loading capacity. This result was first observed when the chaotropic approach was implemented for the analysis of substituted pyridines, aromatic amines, and ophthalmic pharmaceutical compounds [43]. Later, Roberts also observed similar effects [14] during the analysis of primary, secondary, and tertiary benzyl amines and antidepressants. The analysis of Dorzolamide HCl at pH = 2 with phosphoric acid shows early elution. The addition of hexafluorophosphate to the mobile phase leads to an enhancement of the retention. Figure 7.10 is an overlay of Dorzolamide HCl chromatograms at four increasing PF6 concentrations. As the concentration increased, peak tailing decreased and peak efficiency and analyte retention increased. Figure 7.11 shows the effect of different counteranions on basic analyte retention. Depending upon the desired selectivity between a neutral component and a charged basic analyte, a particular chaotropic counteranion could be employed. Moreover, if a method is to be developed with a chaotropic additive that does not have a buffering capacity, a buffer such as phosphate can be employed and the increase in retention can be modified by the addition of the salt of the chaotropic additive, as shown in Figure 7.2. This approach is particularly useful, especially if other ionogenic species are present in the pharmaceutical mixture. The retention of only the protonated basic compounds can be selectively altered by judicious choice of type and concentration of chaotropic mobile phase additive without any further mobile phase pH adjustment. Chaotropic approach could be used for the separation

Advances in Chromatography

N (h/2)

306

14000 12000 10000 8000 6000 4000 2000 0

OH O H2PO4− CIO4− 0

20 40 Counteranion (mM)

NH2

HO

BF4− PF6−

HN

60

Abs.

0.1 v/v% H3PO4−, N = 4,568, Tf = 1.7 52 mM H2PO4−, N = 6,705, Tf = 1.2 25 mM BF4−, N = 8,280, Tf = 1.2 27.2 mM ClO4−, N = 8,865, Tf = 1.1 27 mM PF6−, N = 11,642, Tf = 1.0

0

5

10

15

20 25 30 Time (min)

35

40

45

50

FIGURE 7.11 Effect of counteranion type and concentration on analyte retention, peak efficiency, N(h/2), and tailing factor, Tf. Chromatographic conditions: Column: Zorbax Eclipse XDB-C8, mobile phase: 75% aqueous: 25% acetonitrile, flow rate: 1.0 ml/min, temperature: 25˚C, wavelength: 225 nm. Reprinted with permission from [12].

of very polar basic compounds as a fast screening method for the resolution of closely eluting basic species without resorting to changing pH, mobile phase composition, or type of column (Figure 7.12). These methods are especially useful in reaction monitoring where only a few species are present and reaction conversion needs to be determined. The separation selectivity of basic compounds with varying pKa can be adjusted using concomitant variations in the pH and concentration of chaotropic mobile phase additive (Figure 7.13). The retention of basic compounds that are not fully protonated decreases as pH is decreased; however, for basic compounds that are fully protonated (pH < 2 units from analyte pKa in hydroorganic mixture), their retention would increase and reversals in elution order could be obtained. The separation selectivity of a mixture of acidic, basic, and neutral compounds can be altered with the addition of chaotropic mobile phase additives (Figure 7.14). The retention of the basic compounds can be increased by the addition of chaotropic counterions in the mobile phase while the retention of neutral and acidic compounds is generally unaffected. This effect is particularly useful during the development of impurity profile methods in the pharmaceutical industry, where the retention of a polar protonated basic impurity can be adjusted such that adequate separation selectivity is obtained when nonionic impurities are present in the drug substance. In Figure 7.14, the retention of protonated basic compounds, metoprolol and labetalol, increases while the retention of phenol (in its neutral state) remains constant.

Chaotropic Effects in RP-HPLC

Compounds A: Theophylline

307

pKa >9

B

6.7 9.3 9.8

C

70 H (adj 2 O (pH . HC = lO ) 3.0) : 30 4

D

A+B

+50 mM NaClO4

MeC N

B: 2, 4 Lutidine C: Benzylamine D: Phenylethylamine

A

C

A+C +10 mM NaClO4

D

B

A+B

D

+5 mM NaClO4

D C

no salt added 0

1

2

3

4

FIGURE 7.12 Chromatographic conditions: Column: Luna C18(2), mobile phase: 70% aqueous: 30% acetonitrile. pH adjusted to pH = 3 with perchloric acid + x mM ClO4 adjusted with NaClO4. Flow rate: 1.0 ml/min, temperature: 25˚C.

A Phenylethylamine

pKa = 9.83

B o-Chloroaniline

pKa = 2.64

pH = 1.84 pH = 1.58 2

B A

(10 mM ClO4−) (25 mM ClO4−) 4

6 Retention volume

8

10

FIGURE 7.13 Effect of perchloric acid on the retention of basic analytes of varying pKa. Reprinted with permission from [44].

The effect of different chaotropic mobile phase additives can also assist the chromatographer in achieving adequate retention and resolution of critical pairs in complex mixtures. As shown in Figure 7.15, at an equimolar concentration of chaotropic mobile phase additives, the greatest increases in retention and resolution between critical pairs of components was achieved employing a 30 mM PF6 mobile phase additive. An increase in retention and an increase in the peak efficiency were

308

Advances in Chromatography

5

phenol

labetolol (base) 4

p-toluenesulfonic acid benzene sulfonic acid

3 k

phenol (neutral) labetolol

2

metoprolol

metoprolol (base)

1

p-toluene sulfonic acid (acid)

Chromatographic conditions benzene sulfonic acid (acid) Column: 15 cm ⫻ 0.46 cm Zorbax Eclipse XDB-C18 30 40 50 60 70 Eluent: 70% Aqueous/30% MeCN ClO4− (mM) Aqueous: Water + xHClO4 + yNaClO4 pH = 3.0 Flow rate: 1 ml/min

0

−1

0

10

20

FIGURE 7.14 Retention factor of acidic, neutral, and basic analytes vs. perchlorate concentration. Reprinted with permission from [8].

A

400

A

D

C

B

E

mAU

A C B D C A+ B D

100

F

F

D

300

200

30 mM PF6− E

C

B

30 mM BF4−

A-Atenolol B-Nedolol C-Acebutolol D-Metoprolol E-Labetalol F-Propanlol

F E

30 mM CF2COO−

F

E 30 mM H2PO4−

0 0

5

10

15

20

25

30

min

FIGURE 7.15 Chromatograms of a mixture of beta blockers with different inorganic anions in the mobile phase. Chromatographic conditions: Column: Zorbax Eclipse XDB-C18 (150 × 4.6 mm), mobile phase: Aqueous (pH 3.0)-acetonitrile (70:30), flow rate: 1 ml/min. Detection: UV at 225 nm. Reprinted with permission from [15].

obtained, leading to an increase in the resolution of critical pairs of components (A/B and C/D). The retention of four tetracylcines and flumequine was obtained using an isocractic separation on a Zorbax SB-C18 column [13]. The enhancement of retention was obtained using potassium perchlorate in the mobile phase (Figure 7.16). Greater

Chaotropic Effects in RP-HPLC

H3C CH3 H

HO

309

CH3

H 3C

N

OH H

OH

CH3 N

CH3 H

HO

H OH

NH2

NH2

OH OH

O

OH

O

OH

O OH

O

Oxytetracycline (OTC)

O

OH

O

Tetracycline (TC)

CH3 H3C N 2O

N

CH3

CH3 N

OH H

H OH

O F

NH2 O

OH

OH OH

Flumequine (E)

O

O

OH

O

Doxycyline

kFL 2

3.5

1.5

k

1

3

0.5

0

2.5 0

10

20

30

KClO4 conc. (mM)

FIGURE 7.16 Dependence of tetracyclines and flumequine retention factor vs. perchlorate concentration maintaining constant oxalic acid concentration (0.0925 M). Column Zorbax SB-C18, 55% aqueous (0.0925 M + x mM ClO4): 30% (v/v) MeCN + 15% (v/v) MeOH, (diamond) OTC, (square) TC, (triangle) CTC, (circle) DC, and (asterisk) FL. Reprinted with permission from [13].

retention was obtained for all compounds with perchlorate additive compared to other mobile phases containing HCl and TFA, which are weaker chaotropic reagents. The retention of basic compounds can be altered using different chaotropic additives: for example, if adequate separation selectivity is obtained between 2methylbenzylamine and 4-methylbenzylamine with perchlorate as mobile phase additive but the desired retention is not achieved, hexafluorophosphate can be used

310

Advances in Chromatography

50 45

2-MeBA (PF6−) 4MeBA (PF6−) 2MeBA (ClO4−) 4MeBA (ClO4−)

40 35

k′

30 25 20 15 10 5 0 0

10

20

30 − − mM (ClO4 or PF6 )

40

50

60

FIGURE 7.17 Chromatographic conditions: Column: Zorbax Eclipse XDB-C8, mobile phase: 95% aqueous: 5% acetonitrile. pH adjusted with perchloric acid for closed circles and closed triangles and effective counteranion concentration adjusted with NaClO4. pH adjusted with phosphoric acid for open circles and open triangles and effective counteranion concentration adjusted with KPF6. Flow rate: 1.0 ml/min, temperature: 25˚C, analyte load: 0. 0.5 μg, wavelength: 225 nm. Peak #

Peak Name

% MeCN

1 2 3 4 5 6 7 8

Dorzolamide HCI Compound X 2-methyl benzyl amine 3-methyl benzyl amine 4-methyl benzyl amine Labetalol Propranolol Alprenolol

10 10 5 5 5 25 25 25

(Figure 7.17). With hexafluorophosphate the same separation selectivity can be obtained at a lower salt concentration with further enhancement of the analyte retention. The separation selectivity between two protonated basic compounds is independent of the type of chaotropic mobile phase additive employed (BF4, ClO4, and PF6; Tables 7.1, 7.2, and 7.3, respectively). During the development of an HPLC method, the effect of pH on the separation always needs to be determined prior to any other changes in the mobile phase parameters. If a method was originally developed employing phosphoric acid and if a new impurity arises during the optimization of the drug synthesis or during a long-term or accelerated stability study, and this impurity is present above a level where it needs to be identified, a mass-spectroscopy (MS) compatible method needs

Chaotropic Effects in RP-HPLC

311

TABLE 7.1 Effect of BF4-Concentration on Selectivity Concentration BF4– (mM) α

1

5

10

25

1,2 3,4 4,5 6,7 7,8

1.73 1.32 1.00 1.62 1.11

1.74 1.32 1.00 1.62 1.11

1.75 1.33 0.99 1.62 1.11

1.76 1.34 0.99 1.63 1.11

TABLE 7.2 Effect of ClO4-Concentration on Selectivity Concentration ClO4– (mM) α

2.7

5.4

13.6

27.2

40.8

54.3

1,2 3,4 4,5 6,7 7,8

1.33 1.01 1.62 1.11

1.25 1.07 1.62 1.11

1.76 1.33 0.99 1.64 1.12

1.76 1.34 0.99 1.65 1.12

1.76 1.35 0.99 1.62 1.13

1.78 1.32 0.98 1.65 1.11

TABLE 7.3 Effect of PF6-Concentration on Selectivity Concentration PF6– (mM)

1,2 3,4 4,5 6,7 7,8

1

5

10

25

1.81 1.27 1.02 1.62 1.10

1.82 1.34 1.00 1.68 1.10

1.83 1.35 0.99 1.65 1.13

1.83 1.35 0.99 1.70 1.14

to be developed. In a low pH region (less than 2), the most common volatile mobile phase additive employed is TFA; however, TFA can also act as a chaotropic anion as it interacts with basic compounds in their protonated form and increases in retention are observed. Tracking of the retention of unknown protonated basic impurities using MS-compatible mobile phases could be difficult since the elution

312

Advances in Chromatography

25

20

15

A

B

? ?

10 B

A 5 0.2% HClO4 0.1% TFA 0.2% H3PO4 0 0 1

2

3

4

5

6

7

8

9

10

FIGURE 7.18 Column: Phenomenex Luna C18(2) 150 × 4.6 mm, eluent composition: 70% aqueous: 30% MeCN. Bottom chromatogram with 0.2 v/v% phosphoric acid. Middle chromatogram with 0.1 v/v% TFA (MS compatible method). Top chromatogram with 0.2 v/v% perchloric acid.

order of these new impurities compared to existing separation obtained with phosphoric acid could have changed. The following example in Figure 7.18 shows a method originally developed with phosphoric acid (pH = 1.9) for an active (neutral compound) containing two new impurities and the identity of these impurities needed to be performed using MS analysis. Once TFA replaced phosphoric acid, the chromatographic profile for the active and two impurities was changed compared to analysis with phosphoric acid. It was not known was if only the retention of impurity A increased and reversed elution order with impurity B, or if the retention of both impurities A and B increased (note that the retention of the first impurity peak using TFA mobile phase was similar to that of impurity B using the phosphoric acid mobile phase). However, another off-line method was developed with perchloric acid to confirm if the ionogenic nature of impurity B is also basic in nature. The retention of both impurities A and B further increase suggesting that both impurities are basic compounds. Moreover, the selectivity between compounds A and B remained essentially constant when both TFA and perchloric acid mobile phases were used, aiding the chromatographer in the elucidation of tracking the new impurity peaks.

7.8 CONCLUDING REMARKS The effect of small inorganic ions of lipophilic nature (chaotropic agents) on the retention of basic analytes in reversed-phase HPLC is similar to some extent to the effect of amphiphilic mobile phase additives (ion-pairing agents). Chaotropic counteranions essentially introduce reversible secondary equilibria in chromatographic systems without irreversible modification of the surface and significant alteration of the retention of neutral analytes. These counteranions facilitate mass-transfer by

Chaotropic Effects in RP-HPLC

313

disruption of the solvation of charged analytes and introduce weak electrostatic components in the retention process, allowing for flexible alteration of the separation selectivity and enhancement of apparent efficiency. The use of chaotropic counterions for a chromatographic separation is beneficial as a method development strategy. These modifiers could replace the need for changing column type or addition of hydrophobic “ion-pairing” reagents. Further studies are needed to fully elucidate the detailed mechanism of chaotropic mobile phase additives.

REFERENCES 1. Bartha, Á., Vigh, G., and Ståhlberg, J. Extension of the electrostatic retention model of reversed-phase ion-pair chromatography to include the simultaneous effect of the organic modifier and the pairing ion. J. Chromatogr. A, 1990, 506, 85–96. 2. Ståhlberg, J. Retention models for ions in chromatography. J. Chromatogr. A, 1999, 855, 3–55. 3. Bartha, Á. and Ståhlberg, J. Electrostatic retention model of reversed-phase ion-pair chromatography. J. Chromatogr. A, 1994, 668, 255–284. 4. Chen, J.-G. and Weber, S.G. Electrical double-layer models of ion-modified (ionpair) reversed-phase liquid chromatography. J. Chromatogr. A, 1993, 656, 549–576. 5. Bartha, Á., Vigh, G., Billiet, A.H., and de Galan, L. Studies in reversed-phase ionpair chromatography. IV. The role of the chain length of the pairing ion. J. Chromatogr. A, 1984, 303, 29–38. 6. Bartha, Á. and Vigh, G. Studies in reversed-phase ion-pair chromatography. V. Simultaneous effects of the eluent concentration of the inorganic counter ion and the surface concentration of the pairing ion. J. Chromatogr. A, 1987, 396, 503–509. 7. Melander, W.R. and Horvath, Cs. Reversed-phase chromatography. High Performance Liquid Chromatography, Advances and Perspectives, Vol. 2, 1980, Academic Press, NewYork, 114–303. 8. LoBrutto, R., Jones, A., Kazakevich, Y.V., and McNair, H.M. Effect of the eluent pH and acidic modifiers on the HPLC retention of basic analytes. J. Chromatogr., 2001, 913, 175–189. 9. LoBrutto, R., Jones, A., and Kazakevich, Y.V. Effect of counteranion concentration on HPLC retention of protonated basic analytes. J. Chromatogr., 2001, 913, 191–198. 10. Collins, K.D. and Washabaugh, M.W. The Hofmeister effect and the behaviour of water at interfaces. Quar. Rev. Biophys., 1985, 8, 323–422. 11. Cacace, M.G., Landay, E.M., and Ramsden, J.J. The Hofmeistrer series: Salt and solvent effects on interfacial phenomena. Quar. Rev. Biophys., 1997, 30, 241–277. 12. Pan, L., LoBrutto, R., Kazakevich, Y.V., and Thompson, R. Influence of inorganic mobile phase additives on the retention, efficiency and peak symmetry of protonated basic compounds in reversed-phase liquid chromatography J. Chromatogr. A, Vol. 1049, 17(1-2), 2004, 63–67. 13. Pilorz, K. and Choma, I. Isocratic reversed-phase high-performance liquid chromatographic separation of tetracyclines and flumequine controlled by a chaotropic effect. J. Chromatogr. A, Vol. 1031, 2004, 303–305. 14. Roberts, J.M., Diaz, A.R., Fortin, D.T., Friedle, J.M. and Piper, S.D. Influence of the Hofmeister series on the retention of amines in reversed-phase liquid chromatography. Anal. Chem., 2002, 74, 4927–4932.

314

Advances in Chromatography

15. Jones, A., LoBrutto, R., and Kazakevich, Y. Effect of the counter-anion type and concentration on the liquid chromatography retention of β-blockers. J. Chromatogr. A, 2002, 964(1-2), 26, 179–187. 16. Gritti, F. and Guiochon, G. Role of the buffer in retention and adsorption mechanism of ionic species in reversed-phase liquid chromatography: I. Analytical and overloaded band profiles on Kromasil-C18. J. Chromatogr. A, 2004, 1038, 53–66. 17. Gritti, F. and Guiochon, G. Effect of the ionic strength of salts on retention and overloading behavior of ionizable compounds in reversed-phase liquid chromatography: I. XTerra-C18. J. Chromatogr. A, 2004, 1033, 43–55. 18. Gritti, F. and Guiochon, G. Effect of the ionic strength of salts on retention and overloading behavior of ionizable compounds in reversed-phase liquid chromatography: II. Symmetry-C18. J. Chromatogr. A, 2004, 1033, 57–69. 19. Horvath, Cs., Melander, W., Molnar, I., and Molnar, P. Enhancement of retention by ion-pair formation in liquid chromatography with nonpolar stationary phases. Anal. Chem., 1977, 49, 2295. 20. Sokolowski, A. Zone formation in Ion-Pair HPLC. I. Effects of adsorption of organic ions on established column equilibria. Chromatographia, 1986, 22, 168. 21. Sokolowski, A. Zone formation in Ion-Pair HPLC. II. System peak retention and effects of desorption of organic ions on established column equilibria. Chromatographia, 1986, 22, 177. 22. Hagglund, I. and Stahlberg, J. Ideal model of chromatography applied to charged solutes in reversed-phase liquid chromatography. J. Chromatogr. A, 1997, 761, 3–11. 23. Kazakevich, Y.V., LoBrutto, R., and Vivilecchia, R. Reversed-phase HPLC behavior of chaotropic counteranions, J. Chromatogr. A, 2005, 1064, 9. 24. Kazakevich, Y.V, LoBrutto, R., Chan, F., and Patel, T. Interpretation of the excess adsorption isotherms of organic eluent components on the surface of reversed-phase adsorbents: Effect on the analyte retention. J. Chromatogr. A, 2001, 913, 75–87. 25. Buckenmaier, S.M.C., McCalley, D.V., and Euerby, M.R. Overloading study of bases using polymeric RP-HPLC columns as an aid to rationalization of overloading on silica-ODS phases. Anal. Chem., 2002, 74, 4672. 26. Wirth, M.J., Swinton, D.J., and Ludes, M.D. Adsorption and diffusion of single molecules at chromatographic interfaces. J. Phys. Chem. B, 2003, 107, 6258. 27. Gritti, F., Gotmar, G., Stanley, B.J., and Guiochon, G. Determination of single component isotherms and affinity energy distribution by chromatography. J. Chromatogr. A, 2003, 988, 185. 28. Fornstedt, T., Zhong, G., and Guiochon, G. Peak tailing and slow mass transfer kinetics in nonlinear chromatography. J. Chromatogr. A, 1996, 742, 55. 29. Fornstedt, T., Zhong, G., and Guiochon, G. Peak Tailing and mass transfer kinetics in linear chromatography. J. Chromatogr. A, 1996, 741, 1. 30. Gotmar, G., Fornstedt, T., and Guiochon,G. Peak Tailing and mass transfer kinetics in linear chromatography dependence on the column length and the linear velocity of the mobile phase. J. Chromatogr. A, 1999, 831, 17. 31. Giddings, J.C. Unified Separation Science. Wiley Interscience, New York, 1991. 32. Dolan, J.W. and Snyder, L.R. Troubleshooting LC Systems. Humana Press, Clifton, N.J., 1989. 33. McCalley, D.V. Influence of sample mass on the performance of reversed phase columns in the analysis of strongly basic compounds by high performance liquid chromatography. J. Chromatogr. A, 1998, 793, 31.

Chaotropic Effects in RP-HPLC

315

34. McCalley, D.V. Selection of suitable stationary phases and optimum conditions for their application in the separation of basic compounds by reversed-phase HPLC. J. Sep. Sci., 2003, 26, 187. 35. Riedo, F. and Kovats, E.S. Adsorption from liquid mixtures and liquid chromatography. J. Chromatogr. A, 1982, 239, 1. 36. Huber, J.F.K. and Gerritse, R.G. Evaluation of dynamic gas chromatographic methods for the determination of adsorption and solution isotherms. J. Chromatogr. A, 1971, 58, 137. 37. Wang, H. L., Duda, J.L., and Radke, C.J. Solution adsorption from liquid chromatography. J. Colloid. Interf. Sci., 1978, 66, 152. 38. Koster F.J. and Findenegg, G.H. Adsorption from binary solvent mixtures onto silica gel by HPLC frontal analysis. Chromatographia, 1982, 15, 743. 39. DeWitte R.S., McBrien, M., and Kolovanov, E. Intelligent Optimization of HPLC Separations Based on Chemical Structures Benefiting from a Unified Knowledge Base. Presentation by Advanced Chemistry Development, Toronto, Canada, March 5-8, 2001, New Orleans, Pittcon 2001. 40. Ascah, T.L. and Feibush, B. Novel, highly deactivated reversed-phase for basic compounds. J. Chromatogr. A, 1990, 506, 357. 41. Cox, G.B. and Stout, R.W. Study of the retention mechanism for basic compounds on silica under “pseudo-reversed-phase” conditions. J. Chromatogr. A, 1987, 384, 315. 42. Nawrocki, J. The silano l group and its role in liquid chromatography. J. Chromatogr. A, 1997,779, 29. 43. LoBrutto, R. and Kazakevich, Y.V. Effect of chaotropic anions on analyte retention. In Proc. 22nd International Symposium on High Performance Liquid Phase Separations and Related Techniques, St. Louis, MO, May 2-8, 1998. 44. LoBrutto, R. and Kazakevich Y. Practical Problem Solving in HPLC, Ch. 5, WileyVCH, 2000.

8

Chromatography of Difficult and WaterInsoluble Proteins with Organic Solvents Andrew J. Alpert PolyLC, Inc., Columbia, Maryland

CONTENTS 8.1 Introduction ..................................................................................................317 8.2 Modest Levels of Organic Solvents with Water-Soluble Proteins..............318 8.3 High Levels of Organic Solvents with Water-Soluble Proteins..................318 8.4 High Levels of Organic Solvents with Water-Insoluble Proteins ...............323 8.5 Discussion ....................................................................................................327 References..............................................................................................................327

8.1 INTRODUCTION If a protein is not soluble in aqueous mobile phases, chromatography is usually performed with added solubilizing agents such as detergents or chaotropes (e.g., urea, guanidinium hydrochloride). These additives have a number of disadvantages: 1. They are incompatible with many modern instrumental methods such as mass spectroscopy and thus must be removed prior to further analysis. 2. The quality of the chromatography could be suboptimal, reflecting such properties as the tendency of detergents to form different structures in solution (e.g., micelles vs. lamellar bilayers) depending on the concentration of detergent and salt. Detergents can also minimize the differences in perceived composition of proteins, part of the objective in SDS-PAGE electrophoresis but a drawback in chromatography. 3. The additives can also cause a significant increase in the viscosity and backpressure of the mobile phase. 4. The additives interfere with detection.

317

318

Advances in Chromatography

Organic solvents can frequently be used as alternative additives in the mobile phase with the elimination of some or all of the problems listed above. The quality of the resulting chromatography can be remarkably good, even for difficult proteins. This review will address the benefits of organic solvents in such applications.

8.2 MODEST LEVELS OF ORGANIC SOLVENTS WITH WATER-SOLUBLE PROTEINS A common convention in biochemistry is that organic solvents denature proteins and thus should be avoided. For example, exposing phosphofructokinase to 70% acetonitrile (ACN) will lead to irreversible loss of its tertiary structure and enzymatic activity. However, many proteins will regain their biological properties when the organic solvent is removed, particularly true of smaller proteins (less than 30 KDa). Furthermore, the chromatography of water-soluble proteins can often be improved by inclusion of modest levels of organic solvent in the mobile phases. Figure 8.1 shows the separation of glycosylation variants of recombinant α-bungarotoxin (8 KDa) by cation-exchange chromatography. In the absence of organic solvent, the variants elute in a continuum. With 40% ACN in both mobile phases, the variants are well-resolved [1]. This concentration window — 30 to 40% organic solvent — seems to be optimal for ion-exchange of a number of growth factors and lymphokines, including heregulin [2] and the sialylation variants of interleukin 2 [3]. However, the optimum concentration of organic solvent must be determined on a case-by-case basis. For example, basic fibroblast growth factor variants are well-separated by cationexchange HPLC [4] but addition of even low concentrations of organic solvent leads to the irreversible loss of biological activity. Presumably the organic solvent affects the chromatography via modest changes in the secondary or tertiary structure of the protein. In the process, portions of the protein structure become available to the stationary phase that formerly were not. This increase in selectivity can be accompanied by an increase, decrease, or no change in the overall retention time of the protein, depending on whether charged groups are made more or less available to the stationary phase by the structural shift. For examples of these trends with peptides, see Figure 6 in Alpert and Andrews [5].

8.3 HIGH LEVELS OF ORGANIC SOLVENTS WITH WATER-SOLUBLE PROTEINS When the mobile phases contain more than 60% organic solvent, proteins are retained in proportion to their content of polar residues, a phenomenon called hydrophilic interaction chromatography (HILIC) [6]. If the stationary phase is polar but uncharged, chromatography occurs purely through hydrophilic interactions, functioning as the inverse of reversed-phase chromatography (RPC); decreasing organic gradients are used and additives that promote retention in one mode have the opposite effect in the other. The mechanism involves partitioning between the predominantly nonaqueous mobile phase and a stagnant layer of water adsorbed by

Chromatography of Difficult and Water-Insoluble Proteins

319

0.030

Absorbance (280 nm)

0.025

0% ACN

0.020 0.015 0.010 0.005 0.000

Absorbance (280 nm)

0.03

Glycosylated forms

40% ACN

Nonglycosylated forms (native) (truncated)

0.02

0.01

0.00

0

10

20

30 Time (min)

40

50

60

FIGURE 8.1 Effect of organic solvent on cation-exchange of recombinant α-bungarotoxin. (top) No acetonitrile used. (bottom) 40% acetonitrile in both mobile phases. Conditions — Column: PolyCAT A, 200 × 4.6-mm, 5-μm, 300-Å. Gradient: 50–300 mM ammonium acetate, pH = 6.0, in 60 minutes. Flow rate: 1 ml/min. (Courtesy of Robert Rogowski, Brown University, Providence, RI.)

the polar stationary phase in the form of water of hydration. The more polar the stationary phase, the less organic solvent is needed to obtain a given degree of retention. This mode of chromatography has been used for proteins [7], peptides [6, 8–10], sugars and oligosaccharides [6, 11] and polar small molecules in general [12–15]. While the phrase “hydrophilic interaction chromatography” has been in use since 1990 [6], this mode has been used for HPLC of sugars since 1975 [16, 17] and a paper from 1967 correctly described the mechanism of chromatography in the case of Sephadex eluted with a predominantly organic mobile phase [18]. If the stationary phase is charged, the hydrophilic interaction is superimposed on the electrostatic effects as an independent force. These conditions can be quite useful for some protein separations; histones are an instructive example [19]. Histones tend to be both basic and somewhat hydrophobic. Proteins and polypeptides with these characteristics tend to aggregate in aqueous media but can often be

320

Advances in Chromatography

dissociated by organic solvents. This effect is evident in Figure 8.2, showing the separation of chicken erythrocyte histone H1 protein variants with a cation-exchange column with 0, 40 and 70% ACN in the mobile phases. With 70% ACN, the column resolves peaks corresponding to the six nonallelic H1 variants. The additional peaks reflect allelic polymorphism in this pooled sample from a number of different animals. Figure 8.3 plots retention time as a function of percent of ACN. Initially, retention decreases with an increase in the fraction of ACN, passing through a minimum at 40% and then again increasing. Figure 8.4 shows the circular dichroism spectra of H1 protein at varying levels of ACN and NaClO4 (a chaotropic salt). A continuous loss of protein tertiary structure occurs with increasing concentration of either additive, probably accounting for the decrease in retention up to 40% ACN (the ACN causes conformational changes that rotate some basic residues away from contact with the stationary phase). While those conformational changes continue at concentrations of ACN greater than 40%, the higher concentrations of ACN also promote hydrophilic interaction that causes increasing retention of the protein despite the possible decrease in electrostatic attraction. The result is the U-shaped curve of Figure 8.3. U-shaped curves of this sort are rather frequently encountered (and infrequently accounted for) in the literature. Examples of such curves can even be found for retention of proteins on nonendcapped reversed-phase columns, although it is necessary to use more than 90% organic solvent before the hydrophilic interactions become significant. With such high levels of organic solvents in a mobile phase, restrictions exist on which salts can be used for gradients with ion-exchange columns. An effective salt for the purpose is NaClO4. It is readily soluble in 70% ACN, is transparent at low wavelengths and, being a chaotrope, promotes the solubilization of proteins. However, it must not be combined with a potassium salt (e.g., KH2PO4) as potassium perchlorate is much less soluble than the sodium salt. Proteins and peptides are retained in HILIC in proportion to their content of polar residues. Basic amino acids are the most polar of all (or at least promote retention the most in this mode), followed by phosphorylated amino acids [6]. Retention due to the other amino acids decreases in order from asparagine and serine to phenylalanine and tryptophan. Figure 8.5 shows the separation of phosphorylation variants of histone H1 by cation-exchange HPLC. This protein is quite basic, so even the most heavily phosphorylated variant is well-retained. In the absence of organic solvent, the most highly phosphorylated variants elute first (top) since the phosphate groups are negatively charged and repel the stationary phase electrostatically. When 70% ACN is present in both mobile phases, the order of elution is inverted (bottom). Electrostatic repulsion still exists between the phosphate residues and the stationary phase under these conditions but hydrophilic interaction due to the polar phosphate residues is stronger than the electrostatic repulsion, leading to a net increase in retention when a phosphate residue is present. With 40% ACN, both forces are in balance and all variants coelute (middle) [20] .

Chromatography of Difficult and Water-Insoluble Proteins

321

(a)

60

200

Absorbance 214 nm

(b)

(c)

100

200 (d)

0

100 Time (min)

200

FIGURE 8.2 Effect of organic solvent on cation-exchange (CEX) of chicken erythrocyte histone H1 variants. (a) Reversed-phase HPLC of the pooled histone H1 variants. (b) CEX with 0% acetonitrile. (c) CEX with 40% acetonitrile in both mobile phases. (d) CEX with 70% acetonitrile in both mobile phases. Conditions (CEX) — Column: PolyCAT A, 200 × 4.6-mm, 5-μm, 1000-Å. Gradient: 4-hour linear gradient, 380–590 mM NaClO4. Buffers: (b and c) 10 mM Na2HPO4, pH = 6.5; (d) 10 mM propionic acid, pH = 6.5. The insert in (c) shows a gradient with 230–440 mM NaClO4. Flow rate: 0.8 ml/min. From [19].

322

Advances in Chromatography

140

Retention time (min)

120 100 80

IEX

60

IEX + HILIC

40 20 0

0

20 40 60 Buffer acetonitrile content (% v/v)

80

FIGURE 8.3 Retention of chicken erythrocyte histone H1 in CEX vs. percent acetonitrile. The data points are for the first variant to elute. Retention times to the left of the dashed line reflect electrostatic effects almost exclusively. Retention times to the right of the dashed line reflect a combination of electrostatic effects and hydrophilic interaction. Conditions — 10 mM Na2HPO4, pH = 6.5, was used as the buffer for mobile phases containing 0–50% acetonitrile, while 10 mM propionic acid, pH = 6.5, was used for mobile phases containing 55–70% acetonitrile. All other conditions were as in Figure 8.2. From [19].

10 mM Na-PO4 0.6 M NaClO4 70% CH3CN 70% CH3CN + 0.6 M NaClO4

10−3 ⫻ [θ] degrees.cm2 .dmol−1

5.0

−0

−5.0

−10.0

200

220 Wavelength (nm)

240

FIGURE 8.4 Effect of denaturants on the circular dichroism spectra of chicken erythrocyte histone H1. Pooled histone protein was dissolved in 10 mM Na2HPO4, pH = 6.5 containing the indicated additives. The spectra reflect increasing α-helix content with either additive and additive effects with both. From [19].

Chromatography of Difficult and Water-Insoluble Proteins

p2

323

p1 + p0 IEX 0% ACN

Absorbance

p3

Mixed mode 40% ACN

Protamine sulfate

p0 p1 p2

HILIC 70% ACN p3

0

10

20 Time (min)

30

40

FIGURE 8.5 CEX of phosphorylation variants of H1b Histone from Raji cells. (top) 0% acetonitrile. (middle) 40% acetonitrile in both mobile phases. (bottom) 70% acetonitrile in both mobile phases. Conditions — Column: PolyCAT A, 200 × 4.6-mm, 5-μm, 300-Å. Gradient: 0–40% B in 5 minutes, then 40–70% B in 30 minutes. Mobile Phase A: 10 mM NaH2PO4, pH = 3.0, plus acetonitrile as indicated. Mobile Phase B: A + 1 M NaClO4. In the case with 40% acetonitrile, the buffer was 10 mM triethylamine methylphosphonate (pH = 3.0) instead of NaH2PO4, and the gradient was 0–15% B in 5 min., then 15–45% B in 30 minutes. Flow rate: 1.0 ml/min. (Courtesy of Herbert Lindner, Univ. of Innsbruck, Innsbruck, Austria.)

8.4 HIGH LEVELS OF ORGANIC SOLVENTS WITH WATER-INSOLUBLE PROTEINS The in vivo milieu of water-insoluble proteins is frequently nonaqueous or depleted in water, as in membranes or an air-water interface. Such proteins frequently respond well to the running conditions of HILIC. Figure 8.6 shows the analysis of a sample of lung surfactant protein (LSP) in a 500:1 emulsion with lipid (this preparation is sprayed into the lungs of premature infants too young to make their own LSP in order to prevent hyaline membrane disease). The chromatography was performed at pH = 3 to insure that the protein would have a net positive charge and thus be retained on the strong cation-exchange (SCX) column. A stationary phase with pore diameter of 1000 Å was used to insure free diffusion of the proteins into and out of the pores (as was true with the histone runs in Figure 8.2).

324

Advances in Chromatography

DAD1 B, Sig=215,4 Ref=off (K:\HPCHEM\4\DATA\20516A5\003–0401, D) 140

Lecithins, steroids, & other lipids

120

100

mAU

80 Surfactant protein variants 60

40

20

0

−20 0

10

20

30

40 min

50

60

70

FIGURE 8.6 Lung surfactant protein in a lipid emulsion. The sample was bovine lung surfactant protein B in an emulsion with 500 parts lipid. Conditions — Column: PolySULFOETHYL A, 200 × 4.6-mm, 5-μm, 1000-Å. Mobile Phase A: 0.1% methylphosphonic acid + 5 mM NaClO4, pH = 3.0, with 70% acetonitrile. Mobile Phase B: Same, but 1 M NaClO4. Gradient: 5 minutes at 0% B, then 0–100% B in 30 minutes. Flow rate: 1 ml/min. Detection: 215 nm. (Courtesy of Richard Hartwick, PharmAssist Analytical Laboratory, South New Berlin, NY.)

Lipids and detergents are poorly-retained in the HILIC mode, here eluting in a large peak in the void volume. The salt gradient was then started and the LSP eluted in several variant peaks. These appear to involve various deletions at either terminus (data not shown).

Chromatography of Difficult and Water-Insoluble Proteins

325

SDS, Coomassie blue, and salts

Column: PolyHYDROXYETHYL Aspartamide 200 ⫻ 4.6 mm (5 μm; 200-Å) Gradient: 70 0% n-propanol in 50 mM formic acid

100

50

% n-propanol

A280

45-kD mitochondrial membrane protein

0 0

10

20

30

Time (min)

FIGURE 8.7 Hydrophilic interaction chromatography of an integral membrane protein. Conditions — Column: PolyHYDROXYETHYL A, 200 × 4.6-mm, 5-μm, 200-Å. Gradient: 70–0% n-propanol in 50 mM formic acid in 10 minutes after an initial hold for 15 minutes. Flow rate: 0.5 ml/min. (Courtesy of Paul Jenö, Univ. of Basel, Basel, Switzerland.)

Figure 8.7 shows the purification of an integral membrane protein that had been electroeluted. The objective was to eliminate the SDS and salts prior to further analysis. As with Figure 8.6, the SDS and salts eluted in or near the void volume. In this case, a neutral column was used for pure HILIC and so the protein was eluted with a decreasing gradient of organic solvent. Additives have the opposite effect in HILIC that they do in RPC. Chaotropes promote retention in RPC but promote elution in HILIC. Thus, use of trifluoroacetic acid (TFA) often leads to poor retention of small solutes in HILIC (see Yoshida [21] for a comparison with formic and acetic acid as additives). In such cases, the use of a buffering salt as the additive is advisable. If a volatile mobile phase is desired for compatibility with mass spectroscopy or evaporative light scattering detection, one can use ammonium acetate or formate. Concentrations of 5 to 20 mM have been used for this purpose. If the application requires a buffering salt that is transparent at 215 to 225 nm, convenient choices are triethylamine phosphate or sodium

326

Advances in Chromatography

1.0

Absorbance 280 nm

0.8

0.6 PrPsc 0.4

0.2

0.0 0

10

20

30

Time (minutes)

FIGURE 8.8 Hydrophilic interaction chromatography of scrapie prion protein. A brain extract from a sheep with scrapie was treated with Proteinase K, then centrifuged. The pellet was solubilized with 0.01 M Tris.Cl, pH = 8.0, containing 2 mM EDTA, 5% SDS, and 10% hexafluoro-2-propanol (HFIP). It was diluted with 10 vol. of Mobile Phase A and applied to the column. Conditions — Column: PolyHYDROXYETHYL A, 200 × 4.6-mm, 5-μm, 300Å. Gradient: 0% B for 8 minutes, 0–100% B over 15 minutes and 100% B for an additional 10 minutes. Mobile Phase A: 100% acetonitrile containing 50 mM HFIP and 0.1% TFA. Mobile Phase B: Same but with water in place of acetonitrile. Flow rate: 0.5 ml/min. (Courtesy of Mary Jo Schmerr, NADC/ARS/USDA, Ames, IA.)

methylphosphonate. Now, even the most nonpolar proteins generally have numerous polar residues. A HILIC column interacts with such residues and tends to ignore the hydrophobic residues. Consequently, retention of proteins in pure HILIC is almost always adequate; the more usual problem is to get them to elute. Therefore, it is common to include a chaotrope in the mobile phases used for such applications. The 50-mM formic acid used in Figure 8.7 frequently yields good results. If a protein is extremely insoluble in water, it can be helpful to include 50 mM hexafluoro-2propanol as well in the mobile phases; its effects with formic acid are additive (viz., the results in Figure 8.4 with ACN and NaClO4). In extreme cases (Figure 8.8), some TFA can be added as well. The resulting mobile phases are volatile and thus compatible with modern methods of instrumental analysis. Pathogenic prion proteins have a solubility in water that has been compared to that of nylon. Following a prolonged extraction process, they are generally solubilized with detergent. However, the detergent precludes the antigen-antibody interaction needed for immunoassay, a serious limitation in the detection of these proteins. Figure 8.8 shows the purification by pure HILIC of such a preparation from an extract of the brain of a sheep with scrapie. As in Figure 8.6 and Figure 8.7, the detergent and lipid contaminants eluted in the void volume while the scrapie prion

Chromatography of Difficult and Water-Insoluble Proteins

327

protein eluted at the end of the solvent gradient. Experiments with samples spiked with 125I-tagged prion protein indicated that recovery of applied prion protein was 93%. If the objective is to separate sequence variants of the prion protein, a highresolution separation of this sort is appropriate. However, the degree of separation between the prion protein and the interfering detergent and lipids is so great that their separation can be accomplished just as effectively with an inexpensive solidphase extraction cartridge of the stationary phase [22]. This is true for other protein and peptide samples as well. The HILIC mobile phases cause the loss of some of the β-sheet structure responsible for the poor water solubility of the scrapie prion protein; the protein recovered from the HILIC column remains soluble in water and can be detected with an immunoassay. As a rule, water-insoluble proteins elute from an HILIC column in a solvent in which they remain soluble. Optimum levels of organic solvent for gradients must be determined on a case-by-case basis. In the case of core histones, for example, the chromatography fails if the ACN content of the mobile phases deviates as little as 5% from the optimum level — too little causes poor separation of the variants while too much leads to precipitation of the histone on-column.

8.5 DISCUSSION Many protein separations benefit from the inclusion of organic solvents in the mobile phases. Some of the examples presented here cannot be performed any other way. The methods presented here are applicable to complex mixtures of proteins as well as to relatively pure collections of variants. An example would be the fractionation of membrane proteins for proteomics analyses. The interested reader is advised to look up the abstracts of such conventions as that of the Association of Biomolecular Resource Facilities (http://www.abrf.org) or the American Society of Mass Spectroscopy (http://www.asms.org). These conferences have had presentations dealing with this topic in the past several years and will likely have others that do so in the future.

REFERENCES 1. Levandoski, M.M., Caffery, P.M., Rogowski, R.S., Lin, Y., Shi, Q.-L., and Hawrot, E. Recombinant expression of α-bungarotoxin in Pichia pastoris facilitates identification of mutant toxins engineered to recognize neuronal nicotinic acetylcholine receptors. J. Neurochem., 2000, 74(3), 1279–1289. 2. Holmes, W.E., Sliwkowski, M.X., Akita, R.W., Henzel, W.J., Lee, J., Park, J.W., Yansura, D., Abadi, N., Raab, H., Lewis, G.D., Shepard, H.M., Kuang, W.-J., Wood, W.I., Goeddel, D.V., and Vandlen, R.L. Identification of heregulin, a specific activator of p185erbB2. Science, 1992, 256(5060), 1205–1210.

328

Advances in Chromatography

3. Marchese, E., Natalio, V., Maureaud, T., and Ferrara, P. Separation by cation-exchange high-performance liquid chromatography of three forms of Chinese hamster ovary cell-derived recombinant human interleukin-2. J. Chromatogr., 1990, 504(2), 351–358. 4. Shahrokh, Z., Eberlein, G., Buckley, D., Paranandi, M.V., Aswad, D.W., Stratton, P., Mischak, R., and Wang, Y.J. Major degradation products of basic fibroblast growth factor: detection of succinimide and iso-aspartate in place of aspartate15. Pharmaceut. Res., 1994, 11(7), 936–944. 5. Alpert, A.J. and Andrews, P.C. Cation-exchange chromatography of peptides on poly(2-sulfoethyl aspartamide)-silica. J. Chromatogr., 1988, 443, 85–96. 6. Alpert, A.J. Hydrophilic-interaction chromatography for the separation of peptides, nucleic acids and other polar compounds. J. Chromatogr., 1990, 499, 177–196. 7. Jenö, P., Scherer, P.E., Manning-Krieg, U., and Horst, M. Desalting electroeluted proteins with hydrophilic interaction chromatography. Anal. Biochem., 1993, 215(2), 292–298. 8. Lane, T.F., Iruela-Arispe, M.L., Johnson, R.S., and Sage, E.H. SPARC is a source of copper-binding peptides that stimulate angiogenesis. J. Cell Biol., 1994, 125(4), 929–943. 9. Furuya, K., Schegg, K.M., Wang, H., King, D.S., and Schooley, D.A. Isolation and identification of a diuretic hormone from the mealworm Tenebrio molitor. Proc. Natl. Acad. Sci. USA, 1995, 92(26), 12323–12327. 10. Zhang, J. and Wang, D.I.C. Quantitative analysis and process monitoring of sitespecific glycosylation microheterogeneity in recombinant human interferon-α from Chinese hamster ovary cell culture by hydrophilic interaction chromatography. J. Chromatogr. B, 1998, 712(1), 73–82. 11. Alpert, A.J., Shukla, M., Shukla, A.K., Zieske, L.R., Yuen, S.W., Ferguson, M.A.J., Mehlert, A., Pauly, M., and Orlando, R. Hydrophilic-interaction chromatography of complex carbohydrates. J. Chromatogr. A, 1994, 676(1), 191–202. 12. Soltysik, S., Bedore, D.A., and Kensil, C.R. Adjuvant activity of QS-21 isomers. Ann. NY Acad. Sci., 1993, 690, 392–395. 13. Dallet, Ph., Labat, L., Kummer, E., and Dubost, J.P. Determination of urea, allantoin, and lysine pyroglutamate in cosmetic samples by hydrophilic interaction chromatography. J. Chromatogr. B, 2000, 742(2), 447–452. 14. Troyer, J.K., Stephenson, K.K., and Fahey, J.W. Analysis of glucosinolates from broccoli and other cruciferous vegetables by hydrophilic interaction liquid chromatography. J. Chromatogr. A, 2001, 919(2), 299–304. 15. Garbis, S.D., Melse-Boonstra, A., West, C.E., and van Breemen, R.B. Determination of folates in human plasma using hydrophilic interaction chromatography-tandem mass spectrometry. Anal. Chem., 2001, 73(22), 5358–5364. 16. Linden, J.C. and Lawhead, C.L. Liquid chromatography of saccharides. J. Chromatogr., 1975, 105(1), 125–133. 17. Palmer, J.K. A versatile system for sugar analysis via liquid chromatography. Anal. Lett., 1975, 8, 215–224. 18. Wieland, T. and Determann, H. Some recent developments in gel chromatography, with special reference to thin layers. J. Chromatogr., 1967, 28, 2–11. 19. Mizzen, C.A., Alpert, A.J., Lévesque, L., Kruck, T.P.A., and McLachlan, D.R. Resolution of allelic and non-allelic variants of histone H1 by cation-exchange-hydrophilic-interaction chromatography. J. Chromatogr. B, 2000, 744(1), 33–46.

Chromatography of Difficult and Water-Insoluble Proteins

329

20. Lindner, H., Sarg, B., and Helliger, W. Application of hydrophilic-interaction liquid chromatography to the separation of phosphorylated H1 histones. J. Chromatogr. A, 1997, 782(1), 55–62. 21. Yoshida, T. Peptide separation in normal phase liquid chromatography. Anal. Chem., 1997, 69(15), 3038–3043. 22. Schmerr, M.J. and Alpert, A.J. Method and kit for extracting prion protein. US Patent 6,150,172, Nov. 21, 2000.

Index A Abraham-type equations for RP-HPLC, 125–127 Accuracy, Process Analytical Technology (PAT), 246 Acetonitrile, 112–113, 126–127, 130–131, 268, 273, 299, 318 Acousto-optical deflection (AOD) scanners, 19 Additives mobile phase, 278–283 organic solvent, 318–327 Affinity chromatography, 178–179 Affinity electrophoresis, 29–30 Anomalies in multidimensional separations, 224–226

B Bioassay microchip designs, 8 Buffers, mobile phase, 114

C Calibration, 99 Capillary electrophoresis (CE), 2, 5–6 Capillary liquid chromatography applications, 65–66 columns and stationary phases, 56–63 temperature programming and instrumentation, 63–65 Capillary sample introduction, 15–17 Casting, 4 Channel filling and sample loading, 8 Chaotropic effect, 292–294 applications in pharmaceutical industry, 304–312 effect of counteranion type and concentration on peak efficiency and asymmetry, 301–304 effect of different counteranions, 297 model, 295–297 retention of counteranions, 298–301 Charge-coupled device (CCD), 22–23, 24 Chiral stationary phases (CSP). See Polysaccharide chiral stationary phases (CSPs)

Chromatographic Hydrophobicity Index (CHI), 106–107, 117, 122–125 Circular dichroism (CD), 263 Colchicines and colchiceinamides, 148 Color maps, 197–199 Columns, liquid chromatography capillary, 56–63 conventional analytical, 49–51 Confocal fluorescence detection with a galvanometric scanner, 19 with a rotary scanner, 19–22 Continuous sample introduction microchip designs, 8 Contour maps, 215 Conventional analytical liquid chromatography, columns and stationary phases, 49–51 Correlation coefficients, 197–199 Counteranions effect of different, 297 retention of, 298–301 type and concentration effect on peak efficiency and asymmetry, 301–304 Countercurrent chromatography (CCC), 156 Critical micellar concentration (CMC), 154

D Designs, multiple-channel microchip bioassay, 8 continuous sample introduction, 8 radial, 6–8 rectilinear, 4–6 Detection, limit of (LOD), 246 Detection methods, 17–18 DNA, 24–29 fluorescence detection with an acousto-optical deflection scanner, 22 with a charge-coupled device (CCD), 22–23, 24 scanning, 19–22 single-point confocal, 18 translation-stage confocal, 18–19 Differential scanning calorimetry (DSC), 264 Dimensionality, sample, 192–193

331

332 Displacements, separation, 188–190 Distribution coefficients, 92–94 DNA (deoxyribonucleic acid) genotyping and, 25–27 multiple-channel electrophoresis and, 30–32 separations involving restriction enzyme digests, 24–25 sequencing, 27–29 Drug lipophilicity, 146–148 Dual-column second-dimension design, 218–224 Dynamic analysis of reactions with multiplechannel electrophoresis, 30–32

E Electronic records and signatures, 250–252 Electrophoresis affinity, 29–30 capillary, 2, 5–6 dynamic analysis of reactions with multiplechannel, 30–32 ELogP method, 100–102 Energy-transfer (ET) tags, 26–27 Enthalpy–entropy compensation (EEC), 46–47 Enzymes, restriction, 24–25 Evaporative light scattering detection (ELSD), 64 Extrapolation method for measuring lipophilicity, 102–105

F Factor analysis (FA), 202–204 Fluorescence detection with an acousto-optical deflection scanner, 22 with a charge-coupled device (CCD), 22–23 Food and Drug Administration (FDA), 238

G Galvanometric scanners, 19 Gas chromatography (GC), 95 Gel permeation chromatography (GPC), 178 Generator columns, 100 Genotyping, 25–27 Glass substrates in microchip microfabrication, 3 Gradient RP-HPLC, 117

H Hereditary hemochromatosis (HHC), 26 High-performance liquid chromatography chaotropic effect, 292–294

Advances in Chromatography applications in pharmaceutical industry, 304–312 effect of counteranion type and concentration, 301–304 effect of different counteranions, 297 model, 295–297 retention of counteranions, 298–301 lipophilicity and, 94–96 correction for ionization, 107–108 eLogP method, 100–102 extrapolation method, 102–105 influence of stationary phase, 108–112 isocratic logk values and, 105 mobile phase composition, 112–117 with octanol-coated columns, 96–100 reversed-phase (RP-HPLC), 102 significance of slope (S) in, 106–107 solvent-generated technique, 100 mobile phase additives, 278–283 reversed-phase (RP-HPLC), 102 Abraham-type equations for, 125–127 Chromatographic Hydrophobicity Index (CHI), 117, 122–125 chromatographic parameters, 117, 118–121 gradient, 117–127 ionic interactions in, 291–292 mobile phase, 112–117 stationary phase, 108–112 studies using RP-HPTLC and, 149–154 Hydrophilic interaction chromatography (HILIC), 318–323 Hydrophilic–lipophilic balance (HLB), 145 drug lipophilicity and, 146–148 and RM values of triazine herbicides, 148 studies using both RP-HPLC and RP-HPTLC, 149–154 Hydrophobicity, 81 Index, Chromatographic, 106–107, 117, 122–125 influence of salts on, 135 organic modifiers and, 112–114 and specific hydrophobic surface area of solute, 134–135

I Immunoassays, affinity electrophoresis for, 29–30 Information theory (IT), 200–202 Injection methods capillary sample introduction, 15–17 optically gated, 13–14 T-type, 9–13 Installation qualification (IQ), 249–250

Index Intercepts. See RM0 value Ionic strength, mobile phase, 114–115 Ionization, 107–108 Ion-pair chromatography, 54–55, 291–292 reversed-phase (RP-IPC), 155 Isocratic logk values, 105

K Kamlet-Taft solvatochromic classification system, 180

L Limit of detection (LOD), 246 Limit of quantitation (LOQ), 246–247 Linearity, Process Analytical Technology (PAT), 247–248 Lipophilicity concept and definition, 81 and determination of hydrophilic–lipophilic balance (HLB), 145–154 drug, 146–148 effect of chemical reactions on partitioning and, 83–87 HPLC measurement of, 94–96 correction for ionization, 107–108 eLogP method, 100–102 extrapolation method, 102–105 generator columns used in, 100 influence of stationary phase on, 108–112 isocratic logk values and, 105 mobile phase composition and, 112–117 with octanol-coated columns, 96–100 reversed-phase, 102 significance of slope (S) in, 106–107 influence of salts on, 135 logD and, 82–83, 92–94 logP and, 90–94 calculations involving, 90–91 shake-flask measurement of, 91–94 studies using both RP-HPLC and RP-HPTLC, 149–154 terms for expressing, 82–83 thin-layer chromatography estimation of, 127–128 influence of mobile phase composition on, 132–133 influence of pH on, 131–132 influence of stationary phase on RM0 value, 133–134 mobile phase, 128–131

333 relationship between intercepts and slope in, 135–143 specific hydrophobic surface area of solute, 134–135 uses of, 81–82 Liquid chromatography capillary applications, 65–66 columns and stationary phases, 56–63 temperature programming and instrumentation, 63–65 conventional analytical applications, 52–56 columns and stationary phases, 49–51 temperature programming and instrumentation, 51–52 countercurrent (CCC), 156 high-performance (HPLC) correction for ionization, 107–108 eLogP method, 100–102 extrapolation method, 102–105 influence of stationary phase on lipophilicity, 108–112 isocratic logk values, 105 lipophilicity measurement by, 94–96 mobile phase composition and lipophilicity, 112–117 with octanol-coated columns, 96–100 reversed-phase, 102 significance of slope (S) in, 106–107 hydrophilic interaction (HILIC), 318–323 micellar (MLC), 154–155 multidimensional separations application of, 208–212, 226–228 assessment of potential of, 185–186 contour maps, 215 designed in accordance with orthogonality, 212–213, 214–215 dual- and quad-column second-dimension designs, 218–224 orthogonality in, 193–207 sample dimensionality and, 192–193 SMO treatment of, 186–188 two-dimensional systems, 191–192, 215–217 types of chromatographic separation displacements in, 188–190 one-dimensional separations, 184–185 retention in, 46–48, 95–96 reversed-phase high-performance (RPHPLC), 102 Abraham-type equations for, 125–127 buffers, 114 effect of temperature on, 116–117 gradient, 117–125

334 ionic strength, 114–115 organic modifiers, 112–114 solute association in organic phase, 116 solute association in water, 116 reversed-phase ion pair (RP-IPC), 155 reversed-phase (RP-LC), 95, 178 selectivity in mobile phase pH, 184 solvent, 179–180 stationary phase, 180–182 temperature, 182–184 temperature effects on retention and selectivity in, 46–48 thin-layer (TLC), 127–128 influence of mobile phase composition on, 132–133 influence of pH on, 131–132 influence of stationary phase on RM0 value, 133–134 relationship between intercepts and slope in, 135–143 relationship between RM and molar fraction X in, 128–131 RM determination and correlation with partition coefficients, 143–145 specific hydrophobic surface area of solute in, 134–135 LogD, 82–83, 92–94, 99 Logk, 102–105 LogP, 90–94 RM determination and correlation with, 143–145

M Masking agents, 115–116 Mass spectrometry (MS), 64 Methanol, 111, 112, 130–131 Methylenetetrahydrofolate reductase (MTHFR), 26 Micellar electrokinetic chromatography (MECK), 81, 155 Micellar liquid chromatography (MLC), 154–155 Microchips, multiple-channel applications affinity electrophoresis, 29–30 DNA separation, 24–29 multiple-channel electrophoresis, 30–32, 34 channel filling and sample loading, 8–9 designs bioassay, 8 continuous sample introduction, 8

Advances in Chromatography radial, 6–8 rectilinear, 4–6 detection methods DNA, 24–29 fluorescence, 22–23, 24 scanning, 19–22 single-point confocal, 18 translation-stage confocal, 18–19 DNA and genotyping, 25–27 separations involving restriction enzyme digests in, 24–25 sequencing, 27–29 early, 2, 4 future perspectives in, 32–35 injection methods capillary sample introduction, 15–17 optically gated, 13–14 T-type, 9–13, 29 microfabrication of glass substrates in, 3 polymer substrates in, 3–4 parallel analysis on, 4–5 Microemulsion electrokinetic chromatography (MEECK), 81 Mobile phase, liquid chromatography. See also Stationary phase, liquid chromatography additives, 278–283 buffers, 114 chaotropic additives, 307–310 composition influence on RM0 value, 132–133 effect of temperature on, 116–117 gradient RP-HPLC, 117 ionic strength, 114–115 masking agents, 115–116 organic modifier, 112–114 pH selectivity, 184 polysaccharide CSPs and, 265–267 relationship between RM and molar fraction X in, 128–131 solute association in organic phase, 116 solute association in water, 116 Molecular modeling of polysaccharide CSPs, 265 Multidimensional separations. See Twodimensional separations

N Nernst’s law of distribution, 82 Normal phase liquid chromatography (NPLC), 178 Normal-phase mode, 268

Index

O Octanol, 87–90 coated columns and high-performance liquid chromatography, 96–100 generator columns and, 100 One-dimensional separations, 184–185 Operational qualification (OQ), 250 Optically gated injection, 13–14 Organic modifiers, 112–114 relationship between RM and molar fraction X of, 128–131 Organic solvent additives with water-insoluble proteins, 323–327 with water-soluble proteins, 318–323 Orthogonality in two-dimensional liquid chromatography, 193–195 combination methods for determining, 204–205, 206–208 correlation coefficients and color maps, 197–199 determination of, 195–197 factor analysis (FA) and, 202–204 information theory (IT) and, 200–202 principal component analysis (PCA) and, 199–200 two-dimensional separations design and, 212–213, 214–215

P Parallel analysis on microchips, 4–5 Partition coefficients, 82, 90–91 retention and, 95–96 RM determination and correlation with, 143–145 slope (S) and, 106–107 Partitioning effect of chemical reactions on, 83–87 solvents, 87–90 Perfluoroctyl stationary phase, 123 Performance qualification (PQ), 250 PH effect on separation, 310 selectivity, mobile phase, 184 and thin-layer chromatography, 131–132 Polar organic mode, 268–274 Polymer substrates in microchip microfabrication, 3–4 Polysaccharide chiral stationary phases (CSPs), 257–259 analyte structure effects, 264–265 effect of side chain substitutions on, 262–263 mechanisms of selectivity, 260–267

335 mobile phase additives, 278–283 mobile phase effects, 265–267 modes of separation normal phase, 267 polar organic, 268–274 supercritical fluid chromatography, 275–278 molecular modeling, 265 reversed phase mode, 274–275 spectroscopic and thermal studies, 263–264 Precision, 245–246 Principal component analysis (PCA), 199–200 Process Analytical Technology (PAT) concepts in, 238–240 continuous improvement and knowledge management tools, 243 electronic records and signatures, 250–252 installation qualification (IQ), 249–250 instrument qualification, 249–250 launch of, 238 method validation, 243–245 accuracy and, 246 limit of detection, 246 limit of quantitation and, 246–247 linearity and, 247–248 precision and, 245–246 range and, 247 specificity, 247 system suitability, 248–249 operational qualification (OQ), 250 performance qualification (PQ), 250 process and endpoint monitoring and control tools, 242–243 tools, 240–242 Programming and instrumentation, temperature capillary liquid chromatography, 63–65 conventional analytical liquid chromatography, 51–52 Proteins water-insoluble, 323–327 water-soluble, 318–323

Q Quad-column second-dimension designs, 218–224 Quantitation, limit of (LOQ), 246–247 Quantitative-structure-retention relationships (QSRR), 113–114

R Radial microchip designs, 6–8

336 Range, Process Analytical Technology (PAT), 247 Rectilinear microchip designs, 4–6 Restriction endonucleases, 24–25 Retention of counteranions, 298–301 extrapolation method and, 102–105 ionic strength and, 114–115 partition coefficients and, 95–96 temperature effects on selectivity and, 46–48 RM0 value of colchicines and colchiceinamides, 148 determination and correlation with partition coefficients, logP, 143–145 determination by RP-TLC, 147–148 influence of mobile phase composition on, 132–133 influence of stationary phase on, 133–134 relationship between slopes (a1) and, 135–143 of triazine herbicides, 148 RNA (ribonucleic acid), multiple-channel electrophoresis and, 32, 34 Rotary scanners, 19–22

S Salts and lipophilicity, 135 Scanning methods of detection, 19–22 Selectivity chaotropic effects and, 306 mobile phase pH, 184 polysaccharide CSPs, 260–267 solvent, 179–180 stationary phase, 180–182 temperature, 182–184 Separation anomalies in multidimensional, 224–226 channels, single, 2 displacement types, 188–190 multidimensional, 185–224 normal phase mode, 268 one-dimensional, 184–185 polar organic mode, 268–274 polysaccharide CSPs modes of, 267–278 reversed-phase mode, 274–275 sample dimensionality and, 192–193 two-dimensional chromatographic systems, 191–224 Sequencing, DNA, 27–29 Serotonin transporter gene (5-HTTLPR), 27 Shake-flask, logP measurement by, 91–94, 143–144 Side chain substitutions, 262–263 Silica-based stationary phases, 108–110 buffers and, 114

Advances in Chromatography Single-point confocal detection, 17–18 Size-exclusion chromatography (SEC), 178 Slope, 106–107 relationship between intercepts and, 135–143 Solvents organic, 318–323 partition, 87–90, 100 Solvophobic interactions, 95 Specific hydrophobic surface area of solutes, 134–135 Specificity, 247 Stationary phase, liquid chromatography, 49–51. See also Mobile phase, liquid chromatography influence on lipophilicity measurements, 108–112 influence on RM0 value, 133–134 polysaccharides and, 257–259 selectivity in, 180–182 silica-based, 108–110, 114 Supercritical fluid chromatography (SFC), 275–278 System suitability, 248–249

T Temperature, 45–46 effect on mobile phase in RP-HPLC, 116–117 effects on capillary columns, 56–63 effects on retention and selectivity, 46–48 programming and instrumentation capillary liquid chromatography, 63–65 conventional analytic liquid chromatography, 51–52 selectivity, 182–184 Tetrahydrofuran (THF), 112–113 Thin-layer chromatography, reverse phase, 127–128 determination of hydrophilic–lipophilic balance (HLB), 145–154 influence of mobile phase composition on RM0 value, 132–133 influence of pH on, 131–132 influence of stationary phase RM0 value, 133–134 relationship between intercepts and slopes in, 135–143 relationship between RM and molar fraction X, 128–131 RM determination and correlation with partition coefficients, 143–145 specific hydrophobic surface area of solute in, 134–135 studies using both RP-HPLC and, 149–154

Index Translation-stage confocal detection, 18–19 Triazine herbicides, 148 Trifluorpropyl siloxane, 112 T-type injections, 2, 29 Two-dimensional separations anomalies in, 224–226 application of, 208–212, 226–228 chromatographic systems, 191–192 contour maps and, 215 designed in accordance with findings of orthogonality studies, 212–213, 214–215 dual- and quad-column second-dimension designs, 218–224 orthogonality in, 193–195 combination methods for determination of, 204–205, 206–208 correlation coefficients and color maps, 197–199 determination of, 195–197 factor analysis (FA) and, 202–204 information theory (IT) and, 200–202 principal component analysis (PCA) and, 199–200 sample dimensionality, 192–193 SMO treatment of, 186–188

337 system design, 215–217 types of chromatographic separation displacements in, 188–190 2,4-dihydroxythiobenzanilides, 148–149 2,2,2,-trifluorethanol (TFE) gradient, 123

U UV detection, 64

W Water acetonitrile systems, 299 insoluble proteins, 323–327 -octanol partitioning, 87–90 soluble proteins, 318–323 solute association in, 116

X Xanthine derivatives, 131–132