Advances in Chromatography: Volume 48

  • 56 107 6
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up

Advances in Chromatography: Volume 48

VOLUME 48 Advances in CHROMATOGRAPHY VOLUME 48 Advances in CHROMATOGRAPHY EDITORS: ELI GRUSHK A Hebrew University

1,489 41 10MB

Pages 424 Page size 441.72 x 698.28 pts Year 2009

Report DMCA / Copyright

DOWNLOAD FILE

Recommend Papers

File loading please wait...
Citation preview

VOLUME 48

Advances in CHROMATOGRAPHY

VOLUME 48

Advances in CHROMATOGRAPHY EDITORS:

ELI GRUSHK A

Hebrew University of Jerusalem Jerusalem, Israel

NELU GRINBERG

Boehringer-Ingelheim Pharmaceutical, Inc. Ridgefield, Connecticut, U.S.A.

Boca Raton London New York

CRC Press is an imprint of the Taylor & Francis Group, an informa business

CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 © 2010 by Taylor and Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, an Informa business 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: 978-1-4200-8453-5 (Hardback) This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint. Except as permitted under U.S. Copyright Law, 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. Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com and the CRC Press Web site at http://www.crcpress.com

Contents Contributors..............................................................................................................vii Chapter 1  Understanding the Retention Mechanism in Reversed-Phase Liquid Chromatography: Insights from Molecular Simulation.............1 Jake L. Rafferty, J. Ilja Siepmann, and Mark R. Schure Chapter 2  Thermodynamic Modeling of Chromatographic Separation.............. 57 Jørgen M. Mollerup, Thomas Budde Hansen, Søren S. Frederiksen, and Arne Staby Chapter 3  Ultra-Performance Liquid Chromatography Technology and Applications..................................................................................99 Uwe D. Neue, Marianna Kele, Bernard Bunner, Antonios Kromidas, Tad Dourdeville, Jeffrey R. Mazzeo, Eric S. Grumbach, Susan Serpa, Thomas E. Wheat, Paula Hong, and Martin Gilar Chapter 4  Biointeraction Affinity Chromatography: General Principles and Recent Developments........................................................................ 145 John E. Schiel, K. S. Joseph, and David S. Hage Chapter 5  Characterization of Stationary Phases in Supercritical Fluid Chromatography with the Solvation Parameter Model..................... 195 Caroline West and Eric Lesellier Chapter 6  Silica Hydride—Chemistry and Applications................................... 255 Joseph J. Pesek and Maria T. Matyska Chapter 7  Multidimensional Gas Chromatography........................................... 289 Peter Quinto Tranchida, Danilo Sciarrone, and Luigi Mondello Chapter 8  Sample Preparation for Chromatographic Analysis of Environmental Samples.................................................................... 329 Tuulia Hyötyläinen v

vi

Contents

Chapter 9  Sample Preparation for Gas Chromatography Using Solid-Phase Microextraction with Derivatization................................................. 373 Nicholas H. Snow Index....................................................................................................................... 389

Contributors Bernard Bunner Core Technology Waters Corporation Milford, Massachusetts Tad Dourdeville Core Technology Waters Corporation Milford, Massachusetts Søren S. Frederiksen Diabetes API, Modelling & Optimisation Novo Nordisk A/S Bagsværd, Denmark Martin Gilar Biopharmaceutical Operations Waters Corporation Milford, Massachusetts Eric S. Grumbach Chemical Operations Waters Corporation Milford, Massachusetts David S. Hage Department of Chemistry University of Nebraska–Lincoln Lincoln, Nebraska Thomas Budde Hansen Protein Separation & Virology Novo Nordisk A/S Gentofte, Denmark Paula Hong Chemical Operations Waters Corporation Milford, Massachusetts

Tuulia Hyötyläinen Maj and Tor Nessling Foundation Helsinki, Finland K. S. Joseph Department of Chemistry University of Nebraska–Lincoln Lincoln, Nebraska Marianna Kele Chemical Operations Waters Corporation Milford, Massachusetts Antonios Kromidas Chemical Operations Waters Corporation Milford, Massachusetts Eric Lesellier Institute of Organic and Analytical Chemistry University of Orleans Orléans, France Maria T. Matyska Department of Chemistry San Jose State University San Jose, California Jeffrey R. Mazzeo Biopharmaceutical Operations Waters Corporation Milford, Massachusetts Jørgen M. Mollerup Prepchrom Klampenborg, Denmark vii

viii

Contributors

Luigi Mondello Department of Pharmacochemistry University of Messina Messina, Italy

Susan Serpa Chemical Operations Waters Corporation Milford, Massachusetts

Uwe D. Neue Chemical Operations Waters Corporation Milford, Massachusetts

J. Ilja Siepmann Department of Chemistry and Department of Chemical Engineering and Materials Science University of Minnesota Minneapolis, Minnesota

Joseph J. Pesek Department of Chemistry San Jose State University San Jose, California Jake L. Rafferty Department of Chemistry University of Minnesota Minneapolis, Minnesota John E. Schiel Department of Chemistry University of Nebraska–Lincoln Lincoln, Nebraska Mark R. Schure Theoretical Separation Science Laboratory The Dow Chemical Company Spring House, Pennsylvania Danilo Sciarrone Department of Pharmacochemisty University of Messina Messina, Italy

Nicholas H. Snow Department of Chemistry and Biochemistry Seton Hall University South Orange, New Jersey Arne Staby CMC Project Planning & Management Novo Nordisk A/S Gentofte, Denmark Peter Quinto Tranchida Department of Pharmacochemistry University of Messina Messina, Italy Caroline West Institute of Organic and Analytical Chemistry University of Orleans Orléans, France Thomas E. Wheat Chemical Operations Waters Corporation Milford, Massachusetts

the 1 Understanding Retention Mechanism in Reversed-Phase Liquid Chromatography: Insights from Molecular Simulation Jake L. Rafferty, J. Ilja Siepmann, and Mark R. Schure Contents 1.1 Introduction.......................................................................................................2 1.2 Thermodynamic-Based Models of Reversed-Phase Liquid chromatography (RPLC)...................................................................................3 1.2.1 Solvophobic theory...............................................................................4 1.2.2 Lattice and self-consistent field theories...........................................8 1.2.3 Group contribution methods.................................................................8 1.2.4 Lipophilic view based on comparison to n-hexadecane transfer..........................................................................9 1.3 Outstanding Problems in Understanding the Reversed-Phase Liquid Chromatography (RPLC) Retention Mechanism............................................ 11 1.3.1 The simulation approach.................................................................... 12 1.3.2 The driving forces for Reversed-Phase Liquid Chromatography (RPLC): Solvophobic or Lipophilic?................................................... 12 1.3.3 Do chains lie extended away or cover the surface?........................ 15 1.3.4 Where is the solvent?..........................................................................20 1.3.5 Partition or adsorption?...................................................................... 21 1.3.6 Effects of embedded polar groups.....................................................25 1.3.7 Determination of the phase volumes..................................................26 1.3.8 Pressure and pore curvature effects................................................... 27 1.3.9 General observations for the bonded-phase–solvent–solute environment........................................................................................ 27

1

2

Advances in Chromatography: Volume 48

1.4 Simulation Methodology.................................................................................28 1.4.1 Simulation and the Theory of Liquids.................................................28 1.4.2 Previous molecular simulations of Reversed-Phase Liquid Chromatography (RPLC) systems...................................................... 29 1.4.2.1 Transferable potentials for phase equilibria force field............................................................................ 31 1.4.2.2 Monte Carlo Methods for Molecular Simulation..................34 1.4.2.3 Gibbs Ensemble Method....................................................... 36 1.4.2.4 Configurational-Bias Monte Carlo........................................ 38 1.4.2.5 Application of the CBMC method in the Gibbs ensemble to simulating Reversed-Phase Liquid Chromatography (RPLC)..................................................... 42 1.4.3 Analysis and Presentation of Data....................................................... 45 1.4.3.1 Gauche Defect Statistics....................................................... 45 1.4.3.2 Order Parameter.................................................................... 45 1.4.3.3 Heterogeneity in System Composition..................................46 1.4.3.4 Solute Distribution Coefficients and Transfer Free Energies........................................................................46 1.5 Reflections....................................................................................................... 47 Acknowledgments..................................................................................................... 48 References................................................................................................................. 48

1.1  Introduction The search for the retention mechanism of reversed-phase liquid chromatography (RPLC) has had a long and interesting history. The mechanism appears to be elusive [1–4] because retention measurements of model compounds on retentive phases have mostly been utilized to infer the mechanism. For example, if one injects a series of compounds, the elution order generally reflects the extent of molecular interactions with the stationary phase: more favorable interactions with the retentive phase take place for molecules with longer retention times. In this review, we refer to the stationary phase interchangeably as a retentive phase and to the mobile phase as a solvent phase; it makes no difference to the thermodynamics of retention whether a phase is moving or not. As one may expect, but is not necessarily true in all cases, larger molecules usually have longer retention times and it has been rationalized that the magnitude of the solute’s interactions with the retentive phase is simply a function of its size. As we will discuss shortly, this guideline is riddled with problems as it is does not take into account specific interactions of the solute with both the retentive phase and the solvent components. The term reversed-phase appears to be first used by Howard and Martin [5] where these pioneers described the surface derivatization of a siliceous material with dichlorodimethyl silane to chemically bind a hydrophobic moiety to the support surface. Further refinement of the bonding chemistry has been a continual process and has been documented recently by Kirkland [6] and the history of the terminology has been given by Melander and Horváth [7]. We loosely use the term RPLC here as liquid chromatography with a waxy hydrophobic phase, typically C18, and polar solvents that are

Retention Mechanism in Reversed-Phase Liquid Chromatography

3

generally a binary mixture of water and an organic modifier. Many different solvent systems have been utilized for RPLC but most common are aqueous mixtures with methanol, acetonitrile, or tetrahydrofuran. With this description it is easy to see that normal phase chromatography utilizes polar moieties on/in the retentive medium, such as bare silanols, bonded amino groups or cyano groups, etc., and the mobile phase consists of less polar solvents of varying degrees, such as n-hexane, although methanol– water or acetonitrile–water mixtures can also be utilized. The majority of separations conducted with high performance liquid chromatography (HPLC) are carried out using RPLC columns [8–10]. This point was made over 25 years ago where authors stated [11] “It has become quite trivial to write that reversed-phase liquid chromatography (RPLC) is certainly the most popular chromatographic technique.” Refinements in this technique drove researchers to make models of the retention process and to refine these models. The early work in this area was thermodynamic in nature with models utilizing lumped interactions. As time passed these interactions became more explicitly detailed. A number of these approaches have been reviewed in detail [12]. The most popular of these, the solvophobic theory of Horváth and coworkers, [3,7,13–16] which will be discussed in detail below, took the center stage in the development of a theory of RPLC. It would take nearly 30 years after the initial presentation of this theory for the evolution of particle-based simulation methodology to allow the a priori determination of the RPLC mechanism [17] without invoking a predetermined mechanism; a necessity and limitation of any thermodynamic theory. In this chapter we review some of the salient features of past theoretical attempts at describing the RPLC process. We then highlight the results of our RPLC investigations that utilize advanced molecular simulation methods and provide a molecular-level detail of the retention mechanism that is currently not attainable through experiment or theory. We present our simulation approach, ­configurational-bias Monte Carlo in the Gibbs ensemble using transferable force fields, which is the method of choice for these investigations. We also discuss the future aspects of molecular simulation in RPLC and liquid chromatography to solve outstanding questions in retention mechanisms. In this regard, this chapter is a continuation of ideas, concepts, and techniques illustrated in a chapter published in Advances in Chromatography by one of us (MRS) ten years ago, entitled “Particle Simulation Methods in Separation Science” [18]. However, our main emphasis here is on the RPLC mechanism and how simulation allows a detailed understanding of this extraordinary separation technique. We intend this chapter to be useful as a teaching tool for those interested in this mechanism and as a brief historical account as to how the research of this mechanism has evolved. This is not intended to be a comprehensive review of the older theoretical work, but should serve as an indication as to where this field is moving.

1.2 Thermodynamic-based models of Reversedphase liquid chromatography (RPLC) The intention of RPLC model development from the late 1970s through to the early 1990s was two-fold: one was to elucidate the retention mechanism and thereby aid in the development of novel RPLC systems and the other was to predict retention order and

4

Advances in Chromatography: Volume 48

thereby provide the ability of solute identification through retention time measurement. Today this latter goal is largely irrelevant because of the power of mass spectrometry and the ease with which LC and mass spectrometry are integrated, enabling the identification of small and medium-size molecules with very few assumptions. One of the common ways to model RPLC was one where empirical values were fitted to equations with predetermined forms. This style of lumped constant ­thermodynamic model can be effective within a certain homologous series of solutes and with very well defined constraints. These models have been reviewed by Kaliszan [12] for work done up to 1987. However, these models can never be rigorous because they neglect a number of physical and chemical facts:





1. The interactions between solvent and bonded-phase chains are represented at a highly empirical and structureless level that neglects specific interactions that may result in an enrichment of one of the solvent components at the bonded-phase–solvent interface, near the silica substrate, or within the bonded-phase region. 2. These models are not atomistic; any change in the solute structure cannot reflect the complex interactions with the solvent and retentive phase (such as specific steric hindrance) but rather some parameter is expected to change within the model. These parameters may have been previously parameterized from experimental data. This is especially critical with molecules that possess complex shapes, with chain branching and multiple functional groups. Orientation effects due to the anisotropic nature of the retentive phase cannot be taken into account. 3. Secondary interactions of the solute with the support material, typically silica, are mimicked on an empirical level with no ability to properly account for the competition between the solvent and the solute for these sites.

Nonetheless, a great deal of experimental data was generated and is still being generated that has been useful in recognizing how chromatographic parameters and analyte structure affect retention. We will now discuss a few of the most successful and recognized theories from the thermodynamic model era.

1.2.1  Solvophobic theory Of all of the theories proposed for RPLC, the solvophobic theory is the most well known. Rather than review this theory [3,7,13–16], the highlights will be given here with emphasis on the resulting limitations of this and other thermodynamic theories of retention in RPLC. However, it needs to be stressed that in the absence of a molecular-detailed view of the RPLC process, these simpler thermodynamic theories offered some interesting insights and served for many years as a valuable guide for RPLC development. It must be realized that thermodynamic models are bookkeeping systems. This is why thermodynamics is versatile and can be used to describe anything from traffic flow [19] to flowing liquids [20]. But there is almost no way to describe molecular systems that have complex interactions in a detailed thermodynamic model. In the

Retention Mechanism in Reversed-Phase Liquid Chromatography

5

Gas phase −∆Gstationary

∆Gmobile Mobile phase

∆Gretention

Stationary phase

Figure 1.1  The thermodynamic cycle which identifies the process of solute retention, solvation, and the ideal vapor reference state in RPLC.

bookkeeping domain, the RPLC models have attempted to isolate each effect by phase and by accounting for solvent–solute, retentive phase–solvent, and retentive phase–solute interactions. This is nearly impossible to do quantitatively for complex systems in the liquid domain and this is why most fundamental studies of liquids have utilized molecular simulation [21–23] as a method to study many-body systems. The solvophobic theory in essence predicts that the most favorable free energy contribution driving the solute retention process is the decrease in the solvent–solute interaction when the solute transfers into the retentive phase. The descriptor “most favorable” here refers to the most negative free energy change within the process [24,25], and a negative free energy change refers to one that increases the ­retention time. The process is described by a thermodynamic cycle, shown in Figure 1.1, which is essential in any thermodynamic theory for accounting purposes. The solvophobic theory of Horváth and coworkers is based on Sinanog˘lu’s theory [26,27]. This theory was adapted by Horváth and coworkers for use in liquid chromatography applications although its potential utilization is much wider than this; its use [26,27] is oriented toward liquid–liquid partitioning processes particularly in biologically relevant applications. The thermodynamic calculation of Δ G °retention is accomplished by accounting for the other free energies in Figure 1.1 such that:

   ∆Gretention = ∆Gmobile − ∆Gstationary

(1.1)

where the superscript ° denotes a standard state. Here is should be noted that the use of a standard state is only a choice of a particular reference system, but that chromatographic separation usually takes place at concentrations far away from the usual standard state of unit molar concentration. Hence, the standard state notation is not used in many subsequent equations. The RPLC solvophobic theory further invokes a thermodynamic model where the two free energy terms on the right hand side (rhs) of Equation 1.1 are expanded such that:  ∆Gmobile = ( ∆Gcav,AL − ∆Gcav,A − ∆Gcav,L )



+ ( ∆Gint ,AL − ∆Gint ,A − ∆Gint,L ) + ∆∆Gmix + ∆Gred − RT ln

RT VE

(1.2)

6

Advances in Chromatography: Volume 48

where ΔGcav,AL is the free energy change of cavity formation for the solute–­retentive phase complex, ΔGcav,A is the free energy of cavity formation for the solute, and ΔGcav,L is the free energy of cavity formation for the bonded retentive phase. The term ΔGint,AL is the free energy for the solute–retentive phase complex with the solvent, ΔGint,A is the free energy for the solute–solvent complex, and ΔGint,L is the free energy of the solvent–retentive phase interaction. The term ΔΔGmix is the free energy of mixing of solute and solvent molecules of different sizes, ΔGred is the reduction of ΔG°vapor due to the presence of the solute, R is the gas constant, T is absolute temperature, and VE is the molar volume of the solvent [15]. The last term in the above equation represents the volume change for the process. The solvation process for each solute consists of two steps; the cavity formation of a solute molecule in the solvent system and the interaction of the solute with the surrounding solvent. This model assumes that these free energies are independent; i.e., these terms are separable. As stated by Horváth et al. [15], the energy of cavity formation is dependent on the size of the solute molecule and the cohesive energy of the solute–solvent interaction. As is the usual case with thermodynamic theories, there is a lack of an explicit description of the molecular geometry and structure. Furthermore, the authors of this theory assume that the change in free volumes for the solute-chain interaction and the unbound chain cancel each other out [15]. Combining Equations 1.1 and 1.2 yields:  ∆Gretention = ( ∆Gcav,AL − ∆Gcav,A − ∆Gcav,L )



+ ( ∆Gint ,AL − ∆Gint ,A − ∆Gint ,L ) + ∆∆Gmix + ∆Gred − RT ln



(1.3)

RT  + ∆Gvapor VE

The solvophobic theory of RPLC continues to define these free energy terms individually using simple equations that attempt to tie these individual terms with experimentally measurable parameters. For example, ΔGcav,A, the free energy of ­cavity formation of the solute, is equated to:

∆Gcav,A = κ Eg γ E ∆ AA

(1.4)

where κ Eg is the curvature correction to the surface tension of the solute, γE , and ΔA A is the molecular surface area of the solute. These types of equations have no easily measurable analog. For example, κ Eg and ΔA A are used throughout the solvophobic theory of RPLC but there is no easy way to make these measurements. The free energy expansions of other terms are given in the original papers. One advantage of this theory is that these equations can imply the retention order of compounds in homologous series. For example, as ΔA A increases, the free energy of retention decreases causing retention to increase, as implied in Equations 1.3

Retention Mechanism in Reversed-Phase Liquid Chromatography

7

and 1.4 and explicitly stated in the well-known relations used in chromatography [28,29]:

 − ∆Gretention  K = exp   RT 

(1.5)

k ′ = Kφ

(1.6)

and

where k′ is the commonly used retention measure that gives the number of column volumes after the void volume of a retained solute, K is the distribution coefficient in units of molar concentrations (or number densities), and φ is the phase ratio of the column, i.e., the ratio of the mobile (solvent) phase volume to the retentive phase volume. We will show below that the retentive phase volume is ill-defined in this problem and may be very difficult to measure. Molecular surface areas can be calculated with molecular modeling tools [23]. However, there exists a similar equation for the free energy of cavity formation for the solute–retentive phase complex, and this implies that there is an average value for this in spite of the fact that there may be a host of different locations and geometries for the solute complex in the retained state due to the anisotropic and microheterogeneous nature of the retentive phase. The average value of this may suffice but it is this lack of detail that makes these types of theories semi-empirical at best. Another example where the theory shows useful trends is that the surface tension shows up as a multiplicative term in the cavity expressions and in general the higher the bulk surface tension between the solute and the solvent, the more retained will be the solute. A number of examples for simple solutes and solvent systems show this trend. For example, an increase in the methanol content of the solvent will decrease the retention factor because the surface tension between a solute and the solvent decreases with increasing methanol content. However, it should be noted here that these surface tensions can only be measured for planar surfaces of bulk phases and that preferential solvation in solvent mixtures [30–33] and curvature induced microphase segregation in solvent mixtures [34] can lead to significant differences between the molecular-scale and its macroscopic analog. This greatly hampers the usefulness of bulk surface tensions for thermodynamic models of the retention process. The solvophobic theory, however, has a number of deficiencies. The most important deficiency is that there is a distinct lack of detail about the influence of the retentive phase on retention. Experimentally, it has been known for years that the solvent composition can alter properties of the retentive phase [35–39], such as the amount of sorbed solvent. The theory cannot accommodate this observation nor predict this a priori. The solvophobic theory of RPLC is focused on the mobile (solvent) phase in determining the retentive properties and disregards the retentive phase. Since it is well known that embedded polar groups (EPGs) [40–43] significantly modify the retention properties of solutes, this seems to invalidate the solvophobic theory of RPLC. It is also well known that the length of the bonded-phase alkyl chains influences retention. It has often been said that it would be possible to include the

8

Advances in Chromatography: Volume 48

retentive phase in free energy-like equations. However, this was perhaps the best attempt at explaining chromatographic retention at the time of its development. As we will show shortly, simulation is much better suited for the purpose of determining the mechanism of RPLC. The parameters required to apply a theory, which may not be measurable experimentally, need not be explicitly defined in a simulation.

1.2.2 Lattice and self-consistent field theories Lattice theories are used to simplify certain statistical mechanics problems by ­forcing a molecular architecture into structural constraints which occupy vertices on a lattice. Configurational statistics are then utilized to determine chain conformation and solute retention. This approach was taken for RPLC by Martire and Boehm [44]. Other applications of this approach include studying liquid crystal problems. This theory has the simplifying assumption that the retentive phase has liquid crystalline, rather than interfacial, organization [2]. Solvent properties are defined through semiempirical energy interactions. This theory leads typically to broad-brush learnings regarding chain ordering although for the short chains used in chromatography the theory has been criticized because the model tends to overestimate the ordering of chains [2]. Ben-Naim [45] has recently stated: “Today, lattice theories have almost disappeared from the scene of the study of liquids and liquid mixtures.” In another form of theory, but closer to a simulation, Böhmer, Koopal, and Tijssen [46], and Tijssen and coworkers [47], used a lattice theory adopted from the selfconsistent field theory for adsorption (SCFA) developed by Scheutjens and Fleer [48]. Both aliphatic and amphiphilic solute molecules are predicted to be distributed non­uniformly in the grafted layer and are accumulated in the interfacial region between the chains and solvent, respectively. It is quite interesting that such detail can be obtained in a lattice theory of this kind. However, it is not clear what parameters should be chosen for such important variables as the Flory–Huggins χ parameters. It is common to build polymer theories using such coarse-grain parameters and then express the results over ranges of these parameters. To our knowledge, no further work has been done using the SCFA approach to study the RPLC mechanism. Specific molecular structure is difficult to graft into a theoretical or simulation approach which is inherently coarsegrained. However, it is well known in polymer physics and chemistry that this theoretical approach is quite useful for studying polymer adsorption where atomic-level detail is not essential to describe trends driven by changes in χ or the chain length.

1.2.3  Group contribution methods Many of the past attempts at the theory of RPLC have grouped terms such that homologous series of a certain molecular structure can be calculated once the retention ­factor of the primary molecule is determined. A number of methods have been explored but the most interesting application [49–51] of this uses the UNIFAC model. UNIFAC is an abbreviation of UNIversal Functional Activity Coefficient, which is a semiempirical system, used for estimating activity coefficients of mixtures. UNIFAC uses the functional groups present on the solute to attempt to predict the equilibrium properties of a mixture. For such a simple approach, the accurate prediction of retention

9

Retention Mechanism in Reversed-Phase Liquid Chromatography

is a tall order but this approach should give the correct retention order when solutes contain different group contributions. In fact, when plots are made of the logarithm of the retention factor k′, as a function of the volume percent of the organic modifier acetonitrile, the agreement for certain solutes is adequate [51]. However, there are no mechanistic insights that can be gleaned from this form of study; the numbers involved are entirely thermodynamic and offer little detail.

1.2.4 Lipophilic view based on comparison to n-hexadecane transfer Carr and coworkers [52,53] conducted some key experimental work that shed light on the thermodynamic driving force of RPLC retention. These studies used a bulk ­liquid n-hexadecane phase as an analog to the typical RPLC stationary phase because this avoids the difficult measurement of the phase ratio, φ. We will discuss the pitfalls of measuring the volume of the retentive chains or the phase ratio below. Their first paper [52] measured the distribution coefficient, K, of a homologous series of alkylbenzenes, from benzene through n-butylbenzene, between n-hexadecane and various solvent phases, including methanol–water, acetonitrile–water, isopropanol– water, and tetrahydrofuran–water systems. Head-space gas chromatography was used to obtain partition and distribution coefficients. Carr and coworkers realized that they must use a thermodynamic cycle, shown in Figure 1.1, to interpret their results. Furthermore, they focused on the incremental  , to negate any dependence on the free energy of transfer of a methylene group, ∆GCH 2 phase ratio in these n-hexadecane–solvent distribution coefficient measurements via:

 ∆GCH = − RT ln( K n+1 / K n ) 2

(1.7)

where Kn + 1 is the distribution coefficient of an alkylbenzene having one more ­methylene group than the alkylbenzene yielding Kn. Using what these authors called an energy level diagram, they were able to construct the free energy of transfer of a methylene group between a methanol–water solvent of varying composition and the n-hexadecane phase (see Figure 1.2). This figure shows that the partitioning from a gas phase into liquid n-hexadecane is favorable (i.e., possesses a negative free energy of transfer) and is much larger in magnitude than the incremental free energy of partitioning into the solvent for all solvent compositions. Only for water-rich solvents (≤ 20% methanol by volume) is the transfer of the methylene increment from the gas phase to the solvent phase unfavorable, but the magnitude of this solvophobic effect is quite small. For solvent compositions with volume fraction of methanol f MeOH ≥ 30% commonly used in RPLC, the transfer of a methylene group is favorable from gas to solvent, i.e., not solvophobic, but the free energy difference between the n-hexadecane and the gas phase is much more favorable than the transfer from gas to solvent. Even for neat water solvent, the unfavorable transfer of a methylene group from gas to solvent is associated with only a relatively small free energy change of ≈ + 0.6 kJ/mol. This suggests that the driving force for the partitioning of nonpolar molecules such as the alkylbenzenes is the lipophilic interaction with n-hexadecane and not the usually solvophilic interaction with the solvent phase (for f MeOH ≥ 30%). This observation

10

alkylbenzenes (25°C. kCal/mol)

Free energy of transfer of a –CH2 group of

Advances in Chromatography: Volume 48 0.2

Water 20

0.0

40

30

10

Gas

50

–0.2 70 –0.4 100

90

60

80

–0.6

C16 % MeOH

Figure 1.2  Carr and coworkers measured incremental transfer free energy diagram for a ­methylene group between bulk methanol–water solvent, n-hexadecane, and gas phase. (From Carr, P. W., Li, J., Dallas, A. J., Eikens, D. I., and Tan, L. C., J. Chromatogr. A, 656, 113–133, 1993. With permission.)

is contrary to the solvophobic theory and has been referred to as the revisionist view. Furthermore, Carr and coworkers compared this data to certain RPLC experiments and showed that the magnitude of the transfer free energies is very similar. They were also cautionary about suggesting that polar molecules would have the same driving forces. In the case of benzene, the vapor to water free energy of transfer is favorable, but the transfer to hexadecane from the vapor phase is ≈ 4 times more favorable. This study shed new light on the driving force for nonpolar molecule solutes and was in stark disagreement with the solvophobic theory which deemphasized the role of the retentive phase. In fact, other studies [54,55] (the second reference is from Carr’s own laboratory) suggested that the solvent controlled the energetics of retention. The good agreement found for methylene group partitioning into a bulk n-hexadecane and for the RPLC transfer process has been used to argue that, in a thermodynamic sense, the RPLC retention mechanism appears to be well described by a partition process. In a following study from Carr’s laboratory [53], the free energies of transfer were broken down into the enthalpy and entropy components that make up the free energy through the well-known relationship: ΔG° = ΔH°−TΔS°. The authors again used n-­hexadecane as a model for a retentive phase material noting that n-­octadecane is a solid at room temperature whereas n-hexadecane is a liquid. Their results indicated that favorable enthalpic contributions were much larger for the transfer of ­solutes from gas to n-hexadecane than were the unfavorable entropic part of the free energy. The gas to solvent transfer was enthalpically favorable as the entropic part was minimally unfavorable. The transfer from solvent to the hexadecane retentive phase was also enthalpically favorable for nonpolar solutes with a small unfavorable component from the entropic part. Hence, the entropic component appears to reduce the favorability of retention but is rather small in magnitude. These authors also suggested based on these data that lipophilic forces are the driving force for retention and that the retention mechanism is

Retention Mechanism in Reversed-Phase Liquid Chromatography

11

more like a partitioning effect than an adsorption effect. This is most difficult to prove, however, because these thermodynamic free energies cannot be used to infer a mechanism (but only to disprove a mechanism), let alone that most of these measurements were made on a system that is a liquid and may differ from the anisotropic environment of chemically bonded chains [56]. Comparison with RPLC measurements was made and the numbers are of a similar trend. However, it was suggested that the lack of good phase ratio measurements have hindered making good direct comparisons. Perhaps the most important point made in the studies from Carr’s laboratory [52,53] is that the use of a vapor-phase standard state for these measurements is of tantamount importance. Others [57–60] have emphasized this point; in one study [54] Martire and coworkers used the pure solute liquid as the reference state. As Carr and coworkers have stated [53]: “The use of such excess thermodynamic quantities inevitably ‘bury’ the interactions that are common to the initial and final states.” Although Carr and ­coworkers solved the driving force mystery for liquid n-hexadecane retentive phases, the mechanism of RPLC was only established by inference and only for nonpolar solutes.

1.3 Outstanding problems in understanding the Reversed-Phase liquid chromatography (RPLC) retention mechanism None of the thermodynamic-based theoretical approaches discussed above can answer the most basic question of whether a molecule adsorbs or partitions into the retentive phase. Here it should be noted that adsorption and partition contributions can be separated by investigating the dependence of the retention factor on the interfacial or surface area or on the thickness of the retentive phase. Although this is possible for gas–liquid chromatography [61–63], it is obviously not possible for RPLC, where the thickness of the retentive phase cannot be varied without changing the nature of the bonded chains. Thus, partition and adsorption can only be determined from knowledge of the solute location. To be clear, adsorption occurs when the solute lies at the interface between bonded chains and solvent (albeit one may argue that a secondary adsorption site would be a location near the silica substrate). Partition in RPLC refers to the solute occupying a space within the bonded phase (neither close to the substrate nor the interface with the solvent). It is well known that n-octadecane itself is virtually insoluble in water with its solubility estimated to be approximately 7 ppb by volume [64] and 4.3 × 10−10 by molfraction [65], but the solubility of water in a liquid alkane is much higher and does not depend as strongly on the chain length of the solvent [66,67]. Nevertheless, it is reasonable to suspect that water will not be found in significant concentration within the bonded-phase layer. We will see shortly, through the use of simulation in which one can pinpoint the location and the spatial concentration, that the solvent penetration into a bonded octadecyl silane (ODS) phase differs markedly from a bulk n-octadecane phase. Furthermore, the thermodynamic theories are incapable of pinpointing the ­driving forces of retention. This is most unfortunate because this has been an outstanding problem for many years. Fortunately, simulation can also solve this problem without an a priori biasing of the problem, i.e., without first constructing a set of equations that presuppose a mechanism.

12

Advances in Chromatography: Volume 48

In addition, a number of workers have speculated on whether the bonded chains lie flat on the surface (i.e., phase collapse) or whether the chains stand up. Again, simulation can provide precise information on the overall bonded-phase distribution and the conformation of individual grafted silanol chains. We have used modern simulation methods to study the problems described above and hope that the reader will agree that simulation has resolved many of the mysteries of the RPLC retention mechanism. The following subsections review some of our most important findings. Following these insights we will present a detailed description of the simulation methodology, thereby allowing the reader to ­understand why certain methodologies are particularly suited to unraveling the RPLC retention mechanism.

1.3.1 The simulation approach We have recently examined the chain conformation, solvent penetration, and retention thermodynamics [17,68–73] of various RPLC systems via particle-based simulations using efficient sampling Monte Carlo algorithms and accurate force fields. These studies investigated the effects of mobile (solvent) phase composition for methanol–water [17,68,69] and acetonitrile–water mixtures [73], of the grafting density of dimethyl octadecylsilane bonded phases [71,72], of the inclusion of embedded ether and amide functionalities in the bonded-phase chains [70], of the chain lengths for bonded alkyl chains [74,75], and of pore curvature [74]. Most of these studies utilized bonded-phase chains with dimethyl side chains and a main chain with a backbone consisting of 18 carbon, nitrogen, or oxygen atoms grafted onto a planar slit pore with exposed (1 1 1) surfaces of a slab of β-cristobalite [76] and at a column pressure of 1 atm. Further details on the system setup are given in Section 1.4.2.5 and our works referenced in this paragraph. We will now present selected results of these studies in a form that provides the answers to the outstanding questions in RPLC mechanism research described in the previous section. Specifically, results will be shown for a 2.9 μmol/m2 ODS phase in contact with pure water, 33% mol fraction methanol, 67% molfraction methanol, and 33% molfraction acetonitrile (denoted hereafter as systems ODS-2.9/WAT, ODS2.9/33M, ODS-2.9/67M, and ODS-2.9/67A, respectively), 1.6 and 4.2 μmol/m2 ODS phases in contact with 50% methanol (systems ODS-1.6/50M and ODS-4.2/50M), and 2.9 μmol/m2 EPG phases in contact with 33% methanol (systems Amide-2.9/33M and Ether-2.9/33M). Some of the quantities presented in the discussion of these systems may be somewhat foreign to the reader. For this reason we include, toward the end of this chapter, Section 1.4.3 dealing with the definitions of these quantities and how they are computed from the simulation data.

1.3.2 The driving forces for Reversed-Phase Liquid Chromatography (RPLC): Solvophobic or Lipophilic? The thermodynamic cycles (or free energy diagrams) for RPLC systems consisting of an ODS bonded phase with a grafting density of 2.9 μmol/m2 in contact with both methanol–water and water–acetonitrile solvent mixtures are compared in Figure 1.3

13

Retention Mechanism in Reversed-Phase Liquid Chromatography WAT

–1

33A 67M

–2 ODS

C16

ODS

C16

ODS

C16

C16

ODS

C16

ODS 33A

ODS WAT

C16 ODS

ODS 33M

–2 –3 0

C16

–10

–30

C16

MET

Vapor

0

–20

0

33M

–1

–3

∆GOH (kJ/mol)

1 Vapor

0

2

∆GCH (kJ/mol)

1

67M

MET

–10 –20 –30

Figure 1.3  Computed incremental transfer free energy diagrams for a methylene group (top) and a hydroxyl group (bottom) between methanol–water (solid black lines) and acetonitrile–water (solid gray lines) solvent mixtures, solvent saturated bulk n-hexadecane (C16) and bonded ODS phases, and gas phase. For comparison of the simulated data to experiment, the data of Barman is shown as dashed lines. (From Barman, B. N., A thermodynamic investigation of retention and selectivity in reversed-phase liquid chromatographic systems, PhD thesis, Georgetown University, 1985.)

to those involving partitioning between the same solvent phases and a bulk liquid n-hexadecane phase.* We prefer to provide the solvent concentrations in mole fraction units because this measure is independent of temperature and pressure and lends itself better to an analysis of the local concentration in subregions of the RPLC system. The meaning of these free energy diagrams is as follows. First, we choose a helium gas (i.e., an ideal gas) phase as our reference state and assign the Gibbs free energy of a solute to zero in this state, although this choice of a reference state energy is entirely arbitrary. This reference state is used for both the alkane incremental free energy and the alcohol incremental free energy. The need for this reference state was first recognized by Carr and coworkers [52], as discussed above, and later used by Vailaya and Horváth [15]. Incremental transfer free energies are used here because they can be determined with very high precision and allow direct comparison to experimental data without knowledge of the phase ratio. A homologous series of n-alkane solutes is used to determine the incremental transfer free energy of a methylene group, and comparison of the n-alkane and primary alcohol solutes with the same number of carbon atoms allows one to determine the incremental transfer free energy of the hydroxyl group. The use of the incremental free energy of ­retention

* A description of how these free energies are computed is given in Section 1.4.3.4.

14

Advances in Chromatography: Volume 48

further takes the solute chain length out of the problem but also implies that the Martin equation is valid for the methylene increment:

∆G (n-alkane) = n∆GCH2 + C0

(1.8)

where n is the number of methylene segments in an n-alkane solute, ∆GCH2 is the incremental transfer free energy per methylene group, and C0 is a constant. This relationship states that there is a linear relationship between the free energy of retention and solute carbon length. Indeed, for the solutes we have utilized so far this relationship holds true. Similarly it is assumed that the hydroxyl increment is independent of chain length and given by

∆G (1-alkanol) = ∆GOH + ∆G (n-alkane)

(1.9)

where both solutes posses the same number of carbon atoms. In the case of the alkane solutes, shown in the top of Figure 1.3, we note that only the incremental transfer free energy of a methylene group from the reference vapor phase into pure water is positive. A positive free energy is often referred to as ­unfavorable because it describes a transfer process that yields an equilibrium concentration higher in the “from” phase than the “to” phase [24,25]. In the special case of a transfer from a vapor phase to a solvent phase, this type of unfavorable process is also referred to as solvophobic (i.e., expulsion from the solvent to a vapor phase is a favorable process). All other transfer processes observed for the methylene increment in the RPLC thermodynamic cycle (see Figure 1.3) possess negative free energies of transfer for the solute going from the vapor phase to the solvent, to the ODS bonded phase, and to the n-hexadecane phase. We say these interactions are favorable due to the negative free energies of transfer. Note that the magnitude of the transfer free energy of the methylene group to the retentive phases (both the ODS bonded phase and the liquid n-hexadecane phases) is always larger than those to the solvent phase. This denotes that the lipophilic interaction is driving retention, i.e., the interaction of the solute with the retentive phase is more favorable (or alternatively the free energy of transfer is more negative) for retention than is interaction with the solvent phase. This lipophilic interaction is the driving force for nonpolar molecules in RPLC. Interestingly, the transfer thermodynamics are very similar for acetonitrile–water solvents and methanol–water solvents on a mole fraction basis, which is used for comparison here. Also note that the incremental free energy of transfer to the bonded ODS phase and to the liquid n-hexadecane phase are quite similar and appears to be insensitive to the solvent composition. We will return to this point shortly as this similarity could be used to infer that the solute-ODS interaction may be similar to the bulk partitioning to a liquid n-hexadecane phase (at least for small alkane solutes). We will show shortly that this is not the case. But this highlights the problem with inferring mechanism from thermodynamic values; we will show that this mode of inference is incorrect. Now we look at the lower panel in Figure 1.3 for the thermodynamics of the transfer of a hydroxyl group. The free energy map for the hydroxyl group differs from that for the methylene group in many important ways. First, the transfer free energy is favorable for a hydroxyl group going from the vapor reference state to any solvent

Retention Mechanism in Reversed-Phase Liquid Chromatography

15

phase and the process is most favorable for the transfer to neat water solvent. Second, for all solvent compositions, this free energy of transfer to the solvent is more favorable than transferring to the ODS bonded phase and far more favorable than to the liquid n-hexadecane phase. In other words, the free energy diagram indicates that the hydroxyl group would much rather be in the solvent than in either chain system, i.e., there is a solvophilic contribution of the hydroxyl group to retention. On a molecular level what would drive the hydroxyl group to favor the solvent over the chain? An analysis of the hydrogen bonding of alcohol solutes in the various phases demonstrates that the formation of hydrogen bonds govern the transfer thermodynamics [17]. It should also be noted that an increase of the methanol content makes the hydroxyl group’s interaction with the solvent-saturated ODS bonded phase less favorable, whereas the opposite is true for the solvent-saturated n-hexadecane phase [17]. This is also explained by the number of hydrogen bonds per alcohol solute that decreases with increasing methanol concentration in the ODS bonded phase while it increases in the n-hexadecane phase due to clustering of methanol for higher methanol molfractions [77]. The number of hydrogen bonds in the ODS bonded phase exceeds the number of hydrogen bonds in the n-hexadecane phase by factors of 36 to 1.5 for phases in contact with neat water and neat methanol, respectively [17]. It should be noted that these differences only hold for an ODS bonded phase without endcapping and at intermediate grafting density. The striking differences between the methylene group and hydroxyl group transfer make it clear that RPLC retention cannot be reduced to a single mechanism that holds for all analytes. Furthermore, the striking differences between an ODS bonded phase and a liquid n-hexadecane phase for the hydroxyl group make it clear that the anisotropic chain grafting and the presence of substrateODS and ODS-solvent interfaces cannot be neglected in a description of the RPLC thermodynamics. In closing this section, we also note that the computed incremental free energies for the transfer from the mobile phase to the ODS phase are in excellent agreement with experimental retention data of Barman [78]. The numerical values of these incremental free energies have been previously reported by us [17]. The agreement provides validation of the accuracy of the force field used in our Monte Carlo simulations (see Section 1.4.2.1) and gives confidence for the microscopic-level analysis.

1.3.3 Do chains lie extended away or cover the surface? Spectroscopic techniques have long been used to probe the bonded chain conformation and the solvent environment [79–82]. However, spectroscopic techniques ­usually involve averaging over time and space and hence offer only limited insight into the conformation of individual grafted chains, the penetration of solvent into various regions of the retentive phase, and the spatial and orientational preferences of solute molecules within and near the retentive phase. We have examined the structure of the bonded chains in a number of publications [17,68–73] and will present some of the salient results of these studies here. One of the primary advantages of molecular simulation is that knowledge of the explicit positions of all atomic (or interaction) sites allows one to produce highly

16

Advances in Chromatography: Volume 48

detailed snapshots of a complex chemical system. However, for non-crystalline systems, such as in RP liquid chromatography, one should always keep in mind that the properties of the system are described by ensemble averages over millions of configurations and that snapshots are a visualization aid. In principle, every configuration visited during the production period of a Monte Carlo or molecular dynamics simulation contributes equally to the ensemble average. Since system sizes used typically in molecular simulations are quite small (containing a few thousand molecules), fluctuations between different snapshots can be quite large and a given snapshot will not be able to represent the ensemble averages of all important properties. Snapshots of eight RPLC systems spanning a range of mobile phase compositions and grafting densities for ODS bonded phases and for two bonded phases with EPGs are shown in Figure 1.4. These snapshots were taken toward the end of the production simulations. In this section, we will focus on the six systems with bonded ODS chains and the results for the EPG chains will be discussed in Section 1.3.6. As can be seen, the ODS chains are in a disordered conformational state and, with the exception of the system with the lowest grafting density, the chain backbone shows a preferential orientation closer to the substrate normal than parallel to the substrate. The conformational disorder allows for some back folding of individual chains and a rough interface between the ODS chains and the solvent region. For the lowest grafting density (1.6 μmol/m2) or for high methanol concentration (67% MeOH), significant solvent penetration is evident. Please also note in Figure 1.4 the coordinate z, which will used in subsequent discussion, is given below the snapshots. The value z = 0 Å is the position of the silica surface (defined as the outermost silicon substrate atom) and z increases as one moves away from this silica surface. One can look at many different statistical quantities with this type of simulation and a representative set is discussed below. One of the most revealing statistics to characterize the retentive phase structure is density profiles, as shown in Figure 1.5. These profiles show the density of stationary phase CHx segments and solvent molecules as a function of z (see Section 1.4.3.3). When moving from z = 0 Å to larger z, for the intermediate ODS grafting density, the interior part of the bonded-phase region shows very little change with solvent composition, but the density in the tail region softens somewhat with increasing methanol concentration or for replacing methanol with acetonitrile at the same concentration. As subtle as the chain density profile appears to respond to changes in the solvent composition, changes in the grafting density lead to significant differences. The lower grafting density results in a decrease of both the thickness and the carbon density, whereas the higher grafting density mostly leads to an increase of the bonded-phase thickness caused by steric effects. Complementary information can be gleaned from the average z location of the terminal methyl group, as shown in Table 1.1. Overall, it is clear that the ODS chains do not lie flat when in contact with neat water nor approach an all-trans position when the organic modifier content is increased. Another important conformational measure is the fraction of gauche defects (defined in Section 1.4.3.1). As can be seen in Figure 1.6, the fraction of gauche defects along the backbone is quite insensitive to changes in the solvent composition; a result that is not unexpected given that the conformation of an isolated ­n-octadecane chain also shows little dependence on solvent (including solvation in

17

Retention Mechanism in Reversed-Phase Liquid Chromatography

0

12

z (Å)

24

0

12

z (Å)

24

36

Figure 1.4  (See color insert following page 248.) Simulation snapshots of stationary phase configurations taken from simulations of various RPLC systems. The left column, from top to bottom, shows snapshots for systems ODS-2.9/WAT, ODS-2.9/33M, ODS-2.9/67M, and ODS-2.9/33A. The right column, from top to bottom, shows snapshots for systems ODS1.6/50M, ODS-4.2/50M, Amide-2.9/33M, and Ether-2.9/33M. The silica substrate and grafted alkyl chains are shown as tubes with oxygen in orange, silica in yellow, and CH x groups in gray. Methanol, acetonitrile, and water are shown in the ball and stick representation with oxygen in red, hydrogen in white, nitrogen in green, and methyl groups in blue. Solutes are shown as large spheres with CHx groups in cyan, oxygen in red, and hydrogen in white.

18

Advances in Chromatography: Volume 48 1.2

ODS-2.9/WAT

ODS-1.6/50M

ODS-2.9/33M

ODS-4.2/50M

ODS-2.9/67M

Amide-2.9/33M

ODS-2.9/33A

Ether-2.9/33M

0.8 0.4 0 0.8

ρ (z) (g/mL)

0.4 0 0.8 0.4 0 0.8 0.4 0

0

12

z (Å)

24

0

12

z (Å)

24

36

Figure 1.5  (See color insert following page 248.) The density profiles for the grafted chains (black), water (red), methanol (blue), and acetonitrile (green) in the retentive phase. The eight panels depict ensemble averages for the same eight systems shown in Figure 1.4. The total system density, computed as the sum of bonded phase and solvent densities, is shown in purple. The interfacial region (all panels), defined by the range where the total solvent density falls between 10 and 90% of its bulk value, is shaded in gray while the Gibbs dividing surface fitted to the total solvent density is shown by the dashed orange vertical line. The location of z = 0 Å corresponds to the silica surface.

a bulk n-octadecane phase) [30,31]. Changes in the ODS grafting density lead to different conformational distributions for the first few dihedral angles closest to the silanol linker [71]. The fractions of gauche defects averaged over all dihedral angles are compared in Table 1.1 and, again, it is clear that this quantity is not significantly altered for the range of chromatographic parameters considered here.

19

Retention Mechanism in Reversed-Phase Liquid Chromatography

Table 1.1 Summary of Some Average Stationary Phase Structural and Interfacial Propertiesa System

f  bgauche

Snc

z dGDS

δ eint

ODS-2.9/WAT ODS-2.9/33M ODS-2.9/67M ODS-2.9/33A ODS-1.6/50M ODS-4.2/50M Amide-2.9/33M Ether-2.9/33M

0.251 0.271 0.281 0.271 0.262 0.274 0.291 0.311

−0.142 −0.101 −0.082 −0.022 −0.101 0.142 0.171 0.071

14.51 14.82 14.74 14.93 12.41 18.81 18.06 16.76

3.74 5.34 101 6.33 7.68 5.93 7.04 7.64

a

b c d e

Subscripts indicate the standard error of the mean in the final digit as computed from four independent simulations. Fraction of gauche defects averaged over all 15 backbone torsions. Order parameter averaged over all sixteen 1–3 backbone vectors. Location of the Gibbs dividing surface relative to the silica surface (Å). Width of the interfacial region (Å).

1.00

ODS-2.9/WAT ODS-2.9/33M ODS-2.9/67M ODS-2.9/33A

fgauche

0.75 0.50

ODS-1.6/50M ODS-4.2/50M Amide-2.9/33M Ether-2.9/33M

0.25 0.00

S

0.50 0.00 –0.50 0

4

8

12 16 4 8 Dihedral angle or 1−3 vector index

12

16

Figure 1.6  The ensemble average fraction of gauche defects (top two panels) and the orientational order parameter (bottom two panels) as a function of the location along the chain backbone.

20

Advances in Chromatography: Volume 48

Another measure of the chain conformation is the orientational order ­parameter, S, which is defined in Section 1.4.3.2. At the intermediate grafting density, the lower part of the ODS chain prefers backbone orientations perpendicular to the substrate S > 0, whereas the upper part prefers orientation parallel to the substrate S  106 M-1. Symbols and abbreviations: kd, dissociation rate constant; kd,app, apparent dissociation rate constant with possible contributions from both mass transfer and chemical interactions; ka, association rate constant; ka,app, apparent association rate constant with possible contributions from both mass transfer and chemical interactions; ka,conc, concentration-dependent apparent association rate constant; k−1, reverse mass transfer rate constant for movement of a solute from the stagnant mobile phase to the flowing mobile phase in a column; k1, forward mass transfer rate constant for movement of a solute from the flowing mobile phase to the stagnant mobile phase; k1,conc, ­concentration-dependent apparent forward mass transfer rate constant. A correction for the plate height contribution to stagnant mobile phase mass transfer (Hsm) correction can be made by means of theoretical calculations or experimental estimates. The procedures that are listed above are those that have been utilized in the literature. The results in this method can be extrapolated to infinite dilution to correct for any concentration dependence in the results.

Non-linear zonal elution Non-linear frontal analysis Non-linear frontal analysisd Non-linear zonal elution

Zonal elution peak fitting Frontal analysis curve fitting Frontal analysis moment analysis Peak decay method

a

Linear zonal elution

Linear zonal elution

Application Conditions

Plate height measurements Peak profiling

Method

Table 4.3 Comparison of Methods for Kinetic Measurements by Biointeraction Affinity Chromatography

Biointeraction Affinity Chromatography 167

168

Advances in Chromatography: Volume 48

4.3.2 Linear Elution Methods 4.3.2.1 Plate Height Measurements The measurement of plate heights and chromatographic band broadening was the earliest approach developed for using affinity chromatography in the study of reaction rates for biological systems. The concept of a theoretical plate in chromatographic theory was first developed by Martin and Synge (Martin and Synge 1941) and has since been the subject of many theoretical approaches for describing band broadening (Chen 1988; Felinger 2008; Giddings 1965; Grushka et al. 1975; Horvath 1978; Horvath and Lin 1976). This technique views a chromatographic column as being divided into a number of equally sized regions (i.e., theoretical plates) that each represent a single interaction between the analyte and stationary phase. The height of a theoretical plate (i.e., the plate height, H) is the corresponding average distance that a given analyte travels through the column between each of these interactions. The number of theoretical plates (N) and total plate height (H or Htotal) that are observed for an analyte in a chromatographic system can, in turn, be related to the measured variance (σ 2R) and retention time (tR) of the analyte, as shown in Equation 4.19, N=



t R2 L    H = 2 σR N

(4.19)

where L is the length of the column. A number of different processes contribute to the broadening of a chromatographic profile as an analyte travels through a column, each of which can be described by its own plate height term. For instance, the plate height contribution due to mobile phase mass transfer and eddy diffusion (Hm) describes the broadening that occurs due to differential migration paths and interparticle flow profiles. The plate height contribution due to the longitudinal diffusion (HL) describes broadening due to axial diffusion of solutes. Two other important contributions to the measured band broadening are the plate height contributions due to stagnant mobile phase mass transfer (Hsm) and stationary phase mass transfer (Hk), which are related to the mass transfer and interaction processes, respectively, that were given earlier in Equation  4.18. The result of each of these processes is that one subset of the analyte population moves faster than another, resulting in broadening of the peak profile of the analyte (Giddings 1991). Equation 4.20 shows how the total observed plate height (Htotal) for an analyte on a column is equal to the sum of the individual contributions of these processes to the broadening of the profile for the analyte.



H total =

L ⋅ σ 2R = H m + H L + Hsm + H k t R2

(4.20)

As its name implies, the plate height method (or band broadening method) makes use of plate height and band broadening measurements to obtain information on the

169

Biointeraction Affinity Chromatography

rate of an analyte–ligand interaction. In this method, it is typically assumed that the contribution of HL to the total plate height is negligible and that Hm is constant, as is often true at the flow rates that are commonly used in affinity chromatography. The following equations are then used to provide estimates of Hsm and Hk or to measure these terms from independent band broadening studies with retained and non-retained solutes, respectively (Walters 1987).





Hsm

 V  2 ⋅ u ⋅ VP  1 + M ⋅ k   VP  = k−1 ⋅ VM (1 + k )2

Hk =

2⋅u ⋅ k kd ⋅ (1 + k )2

2



(4.21)

(4.22)

In these equations, u is the linear velocity of the flowing mobile phase, VP is the pore volume of the support, VM is the column void volume, k is the retention factor of the analyte, and all other terms are as described previously. In the first step of the plate height method, Equations 4.20 to 4.22 combine to give the following expression for a non-retained solute (k = 0); this makes it possible to estimate k−1 and the plate height terms Hm and Hsm (as represented by the terms on the right) for the analyte.

H Total = H m +

2 ⋅ u ⋅ VP k−1VM

(4.23)

To use this relationship, the analyte of interest is injected onto an inert control column or a non-retained species is injected onto a column containing the desired ligand. These injections are made at various flow rates to obtain plate height measurements at k = 0. A plot of Htotal versus u is then made and examined by linear regression to give k−1 from the best-fit slope and Hm from the best-fit intercept. In the second step of the plate height method, injections of the analyte of interest are made on the column containing immobilized ligand over the same flow rate range. The k−1 value determined from Equation 4.23 is then used along with the analyte retention data to obtain an estimate of the value for Hsm at each flow rate by using Equation 4.21. These resulting values of Hsm and previously determined estimate of Hm are then subtracted in Equation 4.20 from the total measured plate height for the analyte to give the plate height contribution due to stationary phase mass transfer (Hk) at each flow rate. A plot of Hk versus the term [u k/(1 + k)2] is then prepared according to Equation 4.22 (see Figure 4.9). The slope of this plot should have an intercept of zero, and the inverse of the slope can be used to provide the dissociation rate constant for the analyte/ligand system. The association rate constant (ka) can then be calculated using an independent measure of K A (e.g., from the frontal analysis or zonal elution equilibrium methods described earlier). In this approach

170

Advances in Chromatography: Volume 48

Htot (cm)

(a)

0.06

0.04

0.02

0 (b)

0

4

8 12 u (cm/min)

16

20

0.04

Hs (cm)

0.03

0.02

0.01

0

0

2

6 4 u k´/(1 + k´)2 (cm/min)

8

Figure 4.9  Typical results obtained in the plate height method for (a) plots of the total plate height (Htot) versus linear velocity (u) using the injection of D-tryptophan onto an immobilized HSA column, and (b) the corresponding plot of the plate height contribution due to stationary phase mass transfer (Hk ) versus u k/(1 + k)2 after correcting the data for other band broadening contributions. (From Yang, J. and Hage, D. S., J. Chromatogr. A, 766, 15–25, 1997. With permission.)

the values of kd and ka that are obtained are true chemical interaction rate constants because the mass transfer contribution has been eliminated in the correction for Hsm. The conditions that are required to allow for accurate measurements of rate constants by this approach have been described previously (Walters 1987). An early application of the plate height method involved its use to examine the rates at which various sugars bind to concanavalin A (Anderson and Walters 1986). More recently, this technique has been used to measure rate constants for drug and solute interactions with the protein HSA (Loun and Hage 1996; Yang and Hage 1997). The information obtained from this latter type of study has been shown to be important in optimizing chiral separations based on HSA, including the effects of

Biointeraction Affinity Chromatography

171

varying the temperature, pH, and solvent polarity on drug–protein interactions (Loun and Hage 1996; Yang and Hage 1997). This kinetic data has also been shown to be useful in describing the pharmacokinetics of drugs that bind to HSA (Berezhovskiy 2006; Smith et al. 2000) and in developing new assays for measurement of free drug or hormone fractions in serum (Clarke et al. 2005; Ohnmacht et al. 2006). 4.3.2.2 Peak Profiling An approach that is closely related to the plate height method is the technique of peak profiling. The theory for this approach in affinity chromatography was first reported in 1975 by Denizot and Delaage, who based their work on the molecular dynamic theory of chromatography that was previously developed by Giddings and Eyring (Denizot and Delaage 1975; Giddings and Eyring 1955). In this method, the retention time of an analyte from a column and the distribution of this analyte (i.e., as given by its peak variance and second statistical moment) are seen as being dependent on the number of interactions that occur as the analyte passes through the column and on the corresponding rates of analyte–ligand association and dissociation. Both measurements of the retention time for an analyte (tR) and the elution time of a non-retained solute (tM) are made on the same system by this approach. These elution times are then used with variances observed for the peaks of the analyte (σ 2R) and non-retained species (σ 2M) in Equation 4.24 to calculate the dissociation rate constant (kd,app) for this interaction (Denizot and Delaage 1975),

kd ,app =

2 ⋅ t M2 ⋅ (t R − t M ) σ 2R ⋅ t M2 − σ 2M ⋅ t R2

(4.24)

where kd,app values calculated using Equation 4.24 are for data measured at a single linear velocity and will be referred to as such in this review. An early application of the peak profiling method involved its use in studying the kinetics of bovine neurophysin II (BNP II) self association and the binding of this agent with the neuropeptide Arg8-vasopressin (AVP). In these experiments, BNP II was immobilized on either non-porous or porous glass beads. However, it was found that rate constants determined by this initial approach were underestimated by orders of magnitude compared to solution phase studies, especially when using porous supports (Swaisgood and Chaiken 1987, 1986). One likely reason for these deviations include the presence of differences in the stagnant mobile phase mass transfer contributions for the retained and non-retained species, and the use of relatively low flow rates and fraction collection for these studies. Later, this same method was utilized as part of an HPLC system to examine the dissociation kinetics of sugars injected onto an immobilized concanavalin A column (Muller and Carr 1984). Further analysis of this system indicated these results were also underestimated due, in part, to working under non-linear conditions (Muller and Carr 1984; Wade et al. 1987). Later work examined the binding of L-tryptophan with immobilized HSA by this approach in a system that used smaller columns, a monodisperse HPLC support material, and high injection flow rates, giving good agreement with literature values (Talbert et al. 2002). However, this method still assumed that the value of Hsm was either negligible or similar for the retained and non-retained species.

172

Advances in Chromatography: Volume 48

A hybrid version of this approach that combines features of both peak profiling and band broadening measurements was recently developed to better understand the chromatographic variables that might affect rate constant measurements using Equation 4.24 (see Figure 4.10). In this method, Equation 4.24 was rearranged into a form that is expressed in terms of the plate heights for the retained and non-retained species (HR and HM), the retention factor for the retained species, and the linear velocity of the mobile phase (Schiel, Ohnmacht et al. 2009). HR − HM =



2⋅u ⋅ k = Hk kd ,app ⋅ (1 + k )2

(4.25)

Inject analyte and void marker at multiple flow rates

Absorbance

Sample injector

Pump

Absorbance

Equation 4.25 predicts that a plot of HR −HM versus [u k/(1 + k)2] should be linear with a best-fit slope from which the dissociation rate constant for the analyteligand interaction can be estimated. The results of this approach were compared to those obtained by the traditional peak profiling measurements at a single linear velocity by using the L-tryptophan/HSA system as a model. It was found that the single point method resulted in a k d,app value that varied with linear velocity, while the use of Equation 4.25 and data obtained at multiple linear velocities allowed for more precise and accurate estimates of dissociation rate constants (Schiel, Ohnmacht et al. 2009).

0 100 200 300 400 500 600 Time (s)

0

10 20 30 40 50 60 Time (s)

tR (or M)

Detector σ2R(orM)

Affinity column

M R

H -H

Measured slope

Combine multicolumn data 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Calculate HR, HM, k, and u from peak moments

Repeat with columns containing supports with different particle diameters

0

0.2

0.4

0.6 0.8 1 2 6 d 2 (cm × 10 )

1.2

1.4

1.6

0.06 0.05 0.04 0.03 0.02 0.01 0

0

0.01

P

k = d

2 intercept

Figure 4.10  Scheme for a multi-column peak profiling method.

0.02 0.03 0.04 2 u k/(1 + k) slope =

2 k

d,l

0.05

0.06

173

Biointeraction Affinity Chromatography

It is possible to expand Equation 4.25 to differentiate between binding of an analyte with an immobilized ligand or non-specific interactions with the chromatographic support, as is shown in Equation 4.26 for a two-site binding model (Schiel, Papastavros et al. 2009).



HR − HM =

u⋅k (1 + k )2

 2 ⋅ α L 2 ⋅ αn  = H k ,L + H k ,n ⋅ +  kdL ,app kdn ,app 

(4.26)

In this equation, the subscript L refers to the immobilized ligand, the subscript n refers to non-specific interaction with the support, and α is the fraction of total retention due to interaction at site i. This expanded equation has been successfully used to examine the kinetics of binding between immobilized HSA and the drugs impramine and propranolol, which both had significant non-specific binding to the chromatographic support employed in these experiments (Schiel, Papastavros et al. 2009). Equation 4.25 can also be expanded and modified to correct for the contribution of stagnant mobile phase mass transfer in peak profiling studies, as illustrated in Equation 4.27 (Schiel, Ohnmacht et al. 2009).



HR − HM =

u⋅k (1 + k )2

 2 d 2 (2 + 3 ⋅ k )  ⋅ + p 60 ⋅ γ ⋅ D   kd

(4.27)

In this relationship, dp is the particle diameter of the support packing material, γ is the tortuosity factor, and D is the diffusion coefficient of the analyte. This expanded form of Equation 4.25 has been compared with Equation 4.25 in the rate constant measurements for the L-tryptophan/HSA system by using a series of columns that contained supports with different particle diameters. The overall process used in this work is illustrated by the example in Figure 4.10. In this method, a graph was made by using the slopes for plots of HR − HM versus [u k/(1 + k)2] and the known values of dp for each column. This graph made it possible to correct for stagnant mobile phase mass transfer (as represented by the term in parenthesis to the far right of Equation 4.27) and provided an intercept that gave the true dissociation rate constant for the analyte–ligand interaction (Schiel, Ohnmacht et al. 2009). 4.3.2.3 Practical Considerations Many of the practical considerations that were described for zonal elution in equilibrium and thermodynamic measurements also apply to the plate height and peak profiling methods. The conditions needed to obtain linear elution behavior can be determined in these methods by first injecting increasing sample concentrations and finding the range that provides consistent analyte retention and peak shape. The appropriate column size can then be selected to give retention times that are in a usable range for such studies, with higher affinity analytes requiring smaller columns to provide reasonable elution times. In addition, the use

174

Advances in Chromatography: Volume 48

of small diameter packing materials can be used to reduce any contributions to rate ­constant estimates due to stagnant mobile phase mass transfer. Corrections for errors that are due to stagnant mobile phase mass transfer are most accurately made by using the multi-column approach in peak profiling; however, this approach does increase the analysis time because experiments must be performed on multiple columns. It is important in both the plate height and peak profiling methods to use the true retentions and variances of the experimental peaks. This is typically achieved by measuring the first statistical moment (giving the retention time) and second statistical moment (giving the variance) for each peak. It is also necessary to correct for any extra-column elution time or band broadening that may be present, which can be measured by injecting the analyte onto the chromatographic system when all ­components except the column are present (Hage and Chen 2006). The plate height and the traditional peak profiling methods both require that the analyte–ligand interaction (as represented by the plate height term Hk) is the dominant factor contributing to the overall measured plate height and variance for the analyte. Theoretical studies have shown that the relative contribution of Hk to the total plate height is highest when k = 1 for an analyte (Walters 1987). This condition can be reached by varying the amount of ligand in the column or by adding a competing agent to adjust retention for the analyte (Hage and Chen 2006). The multi­column approach in peak profiling is capable of correcting for these Hsm-related errors by allowing kinetic values to be measured in cases even where the contribution of ­stationary phase mass transfer is relatively small. The peak profiling method can increase the throughput of kinetic measurements compared to the plate height method. In addition, there is no need to individually estimate each plate height contribution in the peak profiling methods, thus simplifying the data analysis and minimizing the possibility of propagated errors. One possible limitation of the peak profiling method versus the plate height method is that higher linear velocities are often used in the former method. These higher linear velocities mean that relatively sharp peaks must be produced in the technique, which requires a fast sampling rate to minimize the effect of any electronic dampening on the measured peak variance.

4.3.3 Non-Linear Elution Methods 4.3.3.1 Non-Linear Peak Fitting The plate height and peak profiling methods described in the last section both use dilute amounts of analyte and linear elution conditions to examine the kinetics of analyte–ligand interactions. Similar methods have also been developed that allow the use of much higher analyte concentrations in such work (i.e., non-linear ­elution conditions), thus affording easier analyte detection. The first group of these non-linear methods uses theoretical descriptions of eluting peak profiles and fitted peak parameters to obtain information from experimental peaks on analyte/ligand kinetics.

Biointeraction Affinity Chromatography

175

Equation 4.28 shows one common equation that is used to fit peaks obtained under non-linear conditions in zonal elution studies (Thomas 1944; Wade et al. 1987).



 2 a1x  − x − a / a   a1  x ⋅ I1  a  e 1 2  a0 2  y = [1 − e(− a3 / a2 ) ] ⋅  a3  1 − T  a1 , x  1 − e − a3 / a2  ]  a a  [   2 2

(4.28)

In this equation, y is the intensity of the measured signal, x is the reduced retention time, T is a switching function, and I1 is a modified Bessel function. The terms a 0 through a3 are the best-fit parameters used in fitting this equation to an experimental peak. These parameters are then used to estimate the value of the rate constants and equilibrium constant for the analyte–ligand interaction. For instance, the values of k, kd,app, and K A are determined directly from the fit parameters by using the relationships k = a1, kd,app = 1/a2tM, and K A = a3/Co, where tM is the column void time and Co is the concentration of injected solute multiplied by the sample width (Moaddel and Wainer 2007). The apparent value of the association rate constant can then also be determined by using the expression ka,app = kd,appK A. Equation 4.28 was developed by extending the theory of non-linear frontal application to an infinitely narrow zonal injection (i.e., impulse input) (Thomas 1944; Wade et al. 1987). The model used to obtain this equation assumes that the axial dispersion contribution to band broadening (i.e., HL ) is negligible and that stagnant mobile phase mass transfer (i.e., Hsm) can be included as part of apparent values for k a,app and k d,app. This method was initially used by Wade et al. to study the binding of pNp-mannoside to an immobilized concanavalin A column (Wade et al. 1987). More recently this approach has been used by Wainer and co-workers to characterize the binding of various inhibitors to immobilized nicotinic acetylcholine receptor (nAChR) membrane affinity HPLC columns (Jozwiak et al. 2002; Jozwiak et al. 2003; Jozwiak et al. 2007; Jozwiak et al. 2004; Moaddel et al. 2005; Moaddel and Wainer 2007). An example of this type of experiment is given in Figure 4.11 for two enantiomeric inhibitors, levomethorphan and dextromethorphan. The relative retention and broadening of these two peaks is a function of the affinity and kinetic parameters that characterize the α3β4 nAChR-inhibitor interaction. The best-fit parameters for these peaks yielded a slower dissociation rate constant for dextromethorphan than levomethorphan, which was in agreement with other methods and highlights the influence of kinetics on this inhibitory effect. Similar experiments on other inhibitors were used to develop quantitative-structure activity relationships (QSARs) for these inhibitor/nAChR interactions. This approach has also been used to perform studies on immobilized heat shock protein 90, a molecular chaperone protein that has been noted to have increased activity in some types of cancer (Marszall et al. 2008). A relationship similar to Equation 4.28 was derived by Lee and Chuang to describe peak shapes for the non-specific elution of an otherwise irreversibly retained solute

176

Advances in Chromatography: Volume 48 1200

LM

Signal intensity (TIC)

1000 800

DM

600 400 200 0

0

20

40 60 Retention time (min)

80

100

Figure 4.11  Representative chromatograms obtained under non-linear conditions and analyzed by peak fitting for levomethorphan (LM) and dextromethorphan (DM) on an immobilized α3β4 nAChR membrane affinity column. (From Jozwiak, K., Hernandex, S. C., Kellar, K. J., and Wainer, I., J. Chromatogr. B, 797, 373–379, 2003. With permission.)

(Lee and Chuang 1996). The derivation of this expression was similar to that used in obtaining Equation 4.28 (Wade et al. 1987); however, it was now assumed in the initial boundary conditions that the amount of injected sample was small compared to the column binding capacity and it was assumed that analyte movement from the column began only after the mobile phase was changed to an elution buffer. Data obtained for the use of a pH step to elute human IgG from immobilized protein A on non-porous silica were fit to the equation. A good fit to the peaks for IgG were obtained, giving an estimated dissociation rate constant of 1.5 s−1 for IgG from protein in the presence of pH 3.0, 0.1 M phosphate buffer (Lee and Chuang 1996). The same technique and equation have been used to examine the elution of lysozyme from a Cibacron Blue 3GA affinity column in the presence of various concentrations of sodium chloride in the mobile phase (Lee and Chen 2001). 4.3.3.2 Frontal Analysis The kinetics of solute–ligand interactions have also been measured by fitting profiles that have been generated when using frontal analysis and affinity chromatography. These experiments involve the continuous application of a mobile phase containing the analyte of interest, as opposed to a finite injection band as is used in zonal elution (see Figures 4.1 and 4.2). Many of the models and expressions that are used for this purpose are based on the initial work of Thomas (1944). This model gives an apparent rate constant for analyte binding to the column (ki,app), in which it is assumed that mass transfer is infinitely fast and analyte adsorption is described by second-order Langmuir kinetics based on interaction at a single type of homogeneous ligand binding site (Golshan-Shirazi and Guichon 1992; Mao et al. 1991; Thomas 1944). This general model has been used to examine the adsorption of lysozyme to a Cibacron Blue F3GA column containing non-porous support particles (NPPAM) (Mao et al. 1991). Although a Langmuir adsorption mechanism was assumed to be

Biointeraction Affinity Chromatography

177

present, no assumption was made as to whether mass transfer or adsorption was the rate limiting step in analyte retention. Instead, mass transfer was treated as a film resistance mechanism. Simulations with this model were used to help understand the effects of particle diameter, concentration, flow rate, and the association rate constant on frontal analysis profiles. Such simulations were then used to examine the binding of lysozyme to Cibacron Blue 3GA. As is commonly seen for simple Langmuir models, the resulting fit was good at the beginning of the breakthrough curve, but ligand heterogeneity was thought to result in deviations from the fitted model at the end of the breakthrough curve (Mao et al. 1991). The binding of HSA to affinity columns containing immobilized antibodies was examined by using frontal analysis along with a Langmuir adsorption model and a rectangular isotherm, in which dissociation of the analyte was assumed to be negligible on the timescale of the experiment (i.e., kd was essentially zero) (Renard et al. 1995). A flow rate dependence of the measured apparent association rate constant was observed due to the contribution of mass transfer. Columns of various loading capacity were used along with the following equation to correct for such contributions,

1 ka ,app

=

1 qxVM + ka Fnmt

(4.29)

in which nmt is an overall mass transfer coefficient dependent on the packing and column dimensions, F is the flow rate, VM is the void volume, and qx is the loading capacity of the column. A plot of 1/ka,app versus qx was then prepared, and the intercept used to calculate the true association rate constant ka for the HSA-antibody interaction, giving results in good agreement with those reported in the literature (Renard et al. 1995). A similar approach for correcting for mass transfer contributions has been utilized to study the interaction of fibrinogen with immobilized peptides (de Lucena et al. 1999). Langmuir models based on two independent binding sites have been employed in some studies to examine the rates of analyte–ligand interactions by frontal analysis. This bi-Langmuir model was found to be useful in comparing the adsorption of HSA to various chromatographic stationary phases (Jaulmes et al. 2001) and in describing the adsorption of ß-lactoglobulin to immobilized polyclonal antibodies (Puerta et al. 2002). However, repeated injections indicated that the interaction of ß-lactoglobulin with the polyclonal antibodies was not irreversible over the timescale of the experiment, making it impossible to determine equilibrium binding constants under such conditions. This lead to the development of a sequential frontal analysis system which made use of two sequential frontal applications of the analyte that were separated by a rinsing step of a predefined duration (Puerta et al. 2006). During this rinsing step some, but not all, of the adsorbed analyte was able to desorb as a function of the dissociation rate constant. This effect alters the results of the second frontal application due to incomplete washing of the binding sites, and allowed the use of the simultaneous fitting of both frontal curves by a bi-Langmuir adsorption model to provide the apparent association and dissociation rate constants for this system (Puerta et al. 2006).

178

Advances in Chromatography: Volume 48

Another interesting treatment of non-linear frontal analysis data has involved the use of linear chromatographic theory to determine concentration-dependent rate constants, and then to extrapolate these values to infinite dilution. According to chromatographic theory, the following equation is true under linear elution conditions (Hethcote and Delisi 1982a, 1982b; Munro et al. 1993; Winzor et al. 1991; Winzor 2006).

kd =

2 (VA − VA* ) dσ 2A /dF

(4.30)

In this equation, VA is the breakthrough volume of the analyte, and VA* is the breakthrough volume of a non-retained solute, F is the flow rate, and σ 2A is the variance of the breakthrough curve. A plot of σ 2A versus F is prepared for data obtained by frontal analysis at a range of flow rates and gives a slope which yields (dσ 2A/ dF)app. The effects of slow mass transfer can be corrected for empirically in this approach by making analyte injections on a column with no immobilized ligand (Munro et al. 1993) or by using an analytical solution (Munro et al. 1994) to obtain the corrected value for dσ 2A /dF. This process is done for a range of concentrations, each of which will result in a different value for dσ 2A /dF. Each dσ 2A /dF result is then used to calculate a kd,conc, which is corrected for mass transfer contributions but is still concentration-dependent due to the concentration range that is used in frontal analysis. The resulting values of kd,conc at various concentrations are then extrapolated to infinite dilution to give the true value of kd. This general approach has been used to examine the binding of sugars with immobilized concanavalin A (Munro et al. 1994, 1993). 4.3.3.3 Split-Peak Method The split-peak method is based on the idea that there is a given probability during any separation process that a small fraction of analyte will elute non-retained from the column without interacting with the stationary phase. This phenomenon, which is illustrated in Figure 4.12a, is known as the split-peak effect. The following equation derived under linear elution conditions shows how the presence of either slow mass transfer (as represented by 1/k1 Ve) or slow adsorption (represented by 1/ka mL ) will affect the fraction of solute that elutes non-retained from the column (f) when the split-peak effect occurs.

−1 1   1 = F +  k1Ve ka mL  ln f

(4.31)

In this equation, F is the flow rate, mL is the active moles of immobilized ligand, and Ve is the excluded volume, k1 is the forward mass transfer rate constant and ka is the association rate constant for analyte–ligand binding (Hage et al. 1986). The advantage of using such an equation for rate constant determinations is that it is simple to perform and only requires area measurements. The main limitation of this approach is that it is typically limited to analytes with slow dissociation kinetics (generally

179

Biointeraction Affinity Chromatography

Absorbance, 280 nm

(a)

Diol-bonded silica column, 2.0 ml/min Protein G column, 2.0 ml/min 1.5 ml/min 1.0 ml/min 0.5 ml/min

0 (b)

6

12 18 Time after injection (sec)

24

0.6

–1/lnf

0.4

0.2

0.0

0.0

1.0 2.0 Flow-rate (ml/min)

3.0

Figure 4.12  (a) Representative split-peak data for the injection of immunoglobulin G (IgG) onto a diol-bonded control column or a protein G column and (b) analysis of such data by using a plot of −1/ln f versus F, as prepared according to Equation 4.31. (From Rollag, J. G. and Hage, D. S., J. Chromatogr. A., 795, 185–198, 1998. With permission.)

indicating the presence of high affinity interactions), as is assumed in the derivation of Equation 4.31. Although Equation 4.31 was originally derived for work under linear elution conditions, it has been used under non-linear elution conditions by extrapolating its results to infinite dilution. This extrapolation is done by measuring the peak areas of the retained and non-retained analyte fractions over a range of flow rates. The ratio of the peak areas at each flow rate is then analyzed according to Equation 4.31 by making a plot of −1/ln f versus F for a series of analyte concentrations (see Figure 4.12b). The measured slope at each analyte concentration is then extrapolated to an infinitely dilute sample by using linear regression (Hage et al. 1986; Walters 1987). As can be seen from Equation 4.31, the slope of such a plot is expected to be a function of both mass transfer and adsorption rate constants. The process which is making the dominant contribution to this slope can be determined by comparing the measured slope of this plot with independent estimates of the mass transfer

180

Advances in Chromatography: Volume 48

contribution made by injecting the same analyte on an inert control column or on a column with rapid association kinetics (Hage et al. 1986; Walters 1987). The split-peak method was used by Hage et al. to study the binding kinetics of rabbit IgG on various affinity columns containing immobilized protein A. The results suggested that the rate limiting process for analyte retention (i.e., mass transfer or adsorption) was dependent on the support characteristics and immobilization method (Hage et al. 1986). This information made it possible to determine the apparent rate constant for IgG binding to protein A and to optimize the performance of protein A affinity supports for the analysis of clinical samples (Hage et al. 1986; Walters 1987). The presence of non-linear elution conditions is known to increase the size of the split-peak effect compared to the response that is predicted by Equation 4.31 under linear conditions (Hage et al. 1986). Computer simulations have been used to determine the extent of these deviations for a homogeneous ligand under both mass transfer- and adsorption-limited conditions (Hage and Walters 1988). Equation 4.31 and the simulations for an adsorption-limited case have also been expanded to consider the effect of a heterogeneous ligand on the split-peak method (Rollag and Hage 1998). The binding of IgG on columns that contained recombinant single domain protein A and/or protein G were used as experimental models in these studies, as illustrated in Figure 4.12. These studies have confirmed the use of linear extrapolation to infinitely dilute samples as a means that can accurately determine association rate constants for either homogeneous (Hage and Walters 1988) or heterogeneous ligand systems (Rollag and Hage 1998) displaying adsorption limitedkinetics. For example, independent estimates of apparent association rate constants measured on columns containing only protein A or protein G were used to calculate the values that were expected for a column containing a known mixture of these two ligands. This approach gave good agreement with the extrapolated value of 2.5 (± 0.1) x 106 M−1s−1 that was found when using the results in Figure 4.12 (Rollag and Hage 1998). It is possible in some cases to obtain an exact solution for the split-peak effect under non-linear elution conditions. The most common example is given in Equation 4.32, which has been derived in various forms to describe the split-peak effect for a homogeneous ligand with an adsorption-limited rate for analyte retention and essentially irreversible adsorption of the analyte under the sample application conditions (Hage et al. 1993; Jaulmes and Vidal-Madjar 1991; Renard et al. 1995; Renard and Vidal-Madjar 1994; Rollag and Hage 1998; Vidal-Madjar et al. 1997).

f=

So ln [1 + ( e Load A / So − 1) e −1/ So ] Load A

(4.32)

In this expression, Load A is the relative moles of analyte applied versus the total moles of active ligand in the column, and So is a combination of system parameters (referred to as the split-peak constant), where So = F/(ka,app mL ). This equation eliminates the need to extrapolate to infinite dilution for homogeneous ligands by incorporating the degree of sample overload directly into the mathematical description of the split-peak effect. This method has been employed to study the binding kinetics

Biointeraction Affinity Chromatography

181

of HSA to various antibody systems (Renard et al. 1995; Renard and Vidal-Madjar 1994; Vidal-Madjar et al. 1997) and to describe chromatographic-based competitive binding immunoassays that are based on the split-peak effect (Hage et al. 1993). 4.3.3.4 Peak Decay Method The peak decay method is a technique for rate constant measurements where all or most of the ligand sites in a column are first saturated with analyte. In one form of this method, a high concentration of an agent that competes with the analyte for ligand binding sites is later applied to the column. This creates a situation in which once an analyte dissociates from the ligand it is unlikely to rebind. High flow rates are also typically utilized to ensure that once a dissociated analyte molecule enters the flowing mobile phase it will tend to elute from the column rather than reenter the stagnant mobile phase layer. Under these conditions, the observed rate of elution for the analyte should follow a curve that is related to its rate of dissociation from  the immobilized ligand if mass transfer is fast versus this dissociation rate (Walters 1987). The theoretical description of the peak decay method was first reported in 1987 (Moore and Walters 1987). This was done with a model in which band broadening due to longitudinal diffusion and mobile phase mass transfer was considered to be negligible compared to stagnant mobile phase mass transfer and stationary phase mass transfer (i.e., Hsm and Hk are large compared to HL and Hm). The dissociation of the analyte from the ligand and diffusion of the analyte out of the stagnant phase were also treated as irreversible processes on the time scale of the experiment. If dissociation is the slower of these latter two processes, it is possible to use Equation 4.33 to find the apparent dissociation rate constant for the analyte during its release from the column by using slope of a plot of the natural logarithm of the peak response versus time (Moore and Walters 1987).



 dm Ee  ln  = ln ( m Eo kd ) − kd t  dt 

(4.33)

In this equation, Ee represents the analyte in the flowing mobile phase that is eluting from the column and Eo represents the initial quantity of analyte that was bound to the column. A similar expression can be derived for the case in which the mass ­transfer of the analyte from the stagnant mobile phase to flow mobile phase is the rate limiting case in analyte release, in which the term kd is replaced with k−1 in Equation 4.33. Computer simulations have been used to find the chromatographic conditions that are needed to obtain accurate estimates of kd by this approach. This method has also been tested by using it to examine the dissociation of sugars from immobilized concanavalin A. It was found in this work that the peak decay method did allow for accurate and precise dissociation rate constants to be obtained for this system as long as sufficiently high flow rates were used for the measurements (Moore and Walters 1987). This early work with concanavalin A represented a system with ­moderate-to-high affinity that required the use of a competing agent for analyte elution. In this

182

Advances in Chromatography: Volume 48

approach it is necessary for the competing agent to not produce any signal that might interfere with the detection of the decay response for the analyte (Walters 1987). It has recently been shown that the peak decay method can also be used with weak to moderate affinity systems without requiring any competing agent (Chen 2003). This was accomplished by using short affinity columns (e.g., 2.5 mm in length) and high flow rates to reduce the probability of analyte re-association to the point where no competing agent was necessary. This non-competitive peak decay method was used to measure the dissociation rates of R- and S-warfarin from immobilized HSA. Figure 4.13 shows representative chromatograms for this system and logarithmic peak decay profiles that were analyzed according to Equation 4.33. This method was found to be relatively fast to perform and gave dissociation rate constants for R- and S-warfarin with HSA (0.56 and 0.66 (± 0.01) s −1, respectively) that were in good agreement with previous measurements for this system (Chen 2003; Chen et al. 2009).

Response

(a)

0

5

10

0

5

10

15

20

25

15

20

25

ln (response)

(b)

Time (s)

Time (s)

Figure 4.13  Results for a peak decay analysis as represented by (a) the original chromatograms and (b) logarithmic response for 100 µL injections of racemic warfarin onto an inert control column (dashed lines) or immobilized HSA column (solid lines) at 4 mL/min and 25°C at pH 7.0. (Adapted from Chen, J., Chromatographic studies of drug–protein interactions, PhD dissertation, University of Nebraska, 2003.)

Biointeraction Affinity Chromatography

183

4.3.3.5  Combined Assay Methods A few studies have combined several methods for kinetics measurements to study biological interactions by affinity chromatography. One example is work that has been conducted using frontal analysis, the split-peak method, and peak decay analysis to examine analyte retention and elution on immobilized antibody columns (Hage et al. 2006; Nelson 2003). The method was first used to measure the apparent association rate constants for the binding of 2,4‑dichlorophenoxyacetic acid (2,4-D) and related herbicides to immobilized monoclonal anti−2,4-D antibodies. A fit of the split-peak expression in Equation 4.32 made it possible to estimate the association rate constants for this system under the same application conditions, while analysis of the breakthrough point of the frontal analysis curve according to Equation 4.11 provided an estimate of the association equilibrium measurements for the same interaction. The mobile phase was then changed to a pH 2.5 buffer for elution of the retained analytes. It was found that this elution could be modeled as a first order decay process, thus allowing the peak decay method to be used to find the apparent dissociation rate constant for the analyte under these solution conditions (Hage et al. 2006; Nelson 2003). A similar approach has recently been utilized to examine the binding and elution of an anti-thyroxine aptamer to an immobilized thyroxine ­column to study the effects of mobile phase ionic strengths on the association kinetics of the aptamer-thyroxine complex (Moser 2005). 4.3.3.6 Practical Considerations Non-linear methods for kinetic studies in affinity chromatography are well suited to analytes that are highly soluble and/or produce only a limited signal when passed through a detector. In addition, these non-linear methods can be particularly useful when only a small amount of ligand is present in an affinity column, thus making column overloading likely to occur. As was noted for the linear elution methods, it is desirable to use small particle diameter packing materials (or even non-porous materials) in these studies if the goal is to obtain accurate estimates of analyte–ligand dissociation rates. The use of such materials will help to minimize contributions due to stagnant mobile phase mass transfer and make it easier to examine the kinetics of the analyte–ligand interaction. The approach that should be used for a given system will depend on such factors as the affinity of this system and the degree of accuracy that is desired. Non-linear peak fitting is a relatively fast method and can be used to quickly measure rate constants for a large number of compounds. However, this method is based on an iterative fit to a limited data set and does include stagnant mobile phase contributions as part of the apparent rate constants that it measures. Frontal analysis methods are also easy to perform but are limited by analyte availability and cost. The split-peak method requires only peak area measurements and works well for high affinity analytes, but this approach requires conditions where a portion of the analyte elutes non-retained (e.g., the use of small columns and high injection flow rates). The peak decay method can be used with either weak or moderate affinity analytes but relies on the assumption that a molecule is not able to re-adsorb during elution, as is achieved by working at high flow rates and with small columns.

184

Advances in Chromatography: Volume 48

Regardless of the method that is chosen for such studies, care must be taken in experimental design to ensure that all assumptions made in the particular theory are valid under the experimental conditions used. Care must also be taken to ensure that the reported values reflect the desired rate constant for the system, whether it is an apparent rate constant or one that specifically describes the rate of mass transfer, dissociation, or adsorption. Apparent rate constants can be useful for comparing different columns or in optimizing separations, but appropriate corrections must be made if the goal is to instead obtain a rate constant that reflects the true interaction rate for the analyte–ligand pair.

4.4  Conclusions It has been shown in this review how affinity chromatography can be a valuable tool in the study of biological interactions. Reactions that can be examined by this method vary from simple one-site binding to competitive interactions, multisite binding, and allosteric processes. Some important advantages of using affinity chromatography to study these processes include the ability of this method to reuse a biological ligand for a large number of experiments, the good precision and accuracy of this approach, and the capability of modern affinity columns to be used in HPLC systems for high-throughput measurements. All of these features have lead to a growing interest in affinity chromatography as a tool in characterizing biological interactions. Many techniques have been described for the investigation of biological interactions by affinity chromatography. For instance, equilibrium and thermodynamic studies can be conducted by using either zonal elution or frontal analysis methods. These techniques have been used to obtain information on the binding and competition of solutes for ligands; to examine the effect of changing pH, temperature, or solution conditions on these interactions; and to identify and characterize binding sites on ligands. In addition, frontal analysis methods have been shown to be useful tools in combination with mass spectrometry for screening mixtures of compounds (e.g., drug candidates) in their ability to bind to a given immobilized ligand. A variety of approaches for studying the kinetics of biological interactions have also been reported for use in affinity chromatography. These kinetic tools include plate height measurements, peak profiling, peak fitting in zonal elution or frontal analysis, the split-peak method, and the peak decay method. These methods make it possible to perform kinetic studies for systems that range from weak-to-moderate affinities to those with high affinities. Some of these methods are performed under linear elution conditions (e.g., plate height measurements), while others can be used under non-linear conditions (e.g., frontal analysis with peak fitting or peak decay analysis). A variety of applications for biointeraction affinity chromatography were ­discussed in this review. Some examples included the use of this method to examine the interactions of drugs with serum proteins, sugars with lectins, receptors with their inhibitors, antibodies with antigens, and drug candidates with target proteins. The range of uses for biointeraction affinity chromatography and the continued development of

Biointeraction Affinity Chromatography

185

new techniques in this field indicate that this area will continue to be a valuable tool in the characterization of biological interactions.

Acknowledgments This work was supported, in part, by the National Institutes of Health under grants R01 GM044931 and R01 DK069629.

Abbreviations Analyte Analyte in the flowing mobile phase Analyte-protein complex (or analyte complex with a binding agent L) Non-linear peak fitting parameters Fraction of ligand-bound analyte Amount of immobilized active ligand (frontal affinity application) Diffusion coefficient Support particle diameter Represents analyte in the flowing mobile phase that is eluting Represents the initial quantity of analyte that was bound Fraction of solute eluting non-retained (split-peak fraction); free-­fraction of analyte F Flow rate I1 Modified Bessel function H or Htotal Total plate height Hk Plate height contribution due to stationary phase mass transfer HL Plate height contribution due to longitudinal diffusion Hm Plate height contribution due to mobile phase mass transfer and eddy diffusion HM Total plate height for a non-retained species HR Total plate height for a retained species Hsm Plate height contribution due to stagnant mobile phase mass transfer k Retention factor k1 Rate constant describing movement of a solute from the flowing mobile phase to the stagnant mobile phase k−1 Rate constant describing movement of a solute from the stagnant mobile phase to the flowing mobile phase in a column K A Association equilibrium constant ka Association rate constant ka,app Apparent association rate constant which lumps mass transfer and chemical interaction rate information K d Dissociation equilibrium constant k d Dissociation rate constant kd,app Apparent dissociation rate constant which lumps mass transfer and chemical interaction rate information A AE A-L a 0…a3 b B t D d p Ee Eo f

186

Advances in Chromatography: Volume 48

kd,conc Apparent dissociation rate constant due to concentration dependence of non-linear analyte application and use of linear theory kdL,app Apparent dissociation rate constant for interaction with the immobilized ligand (in situations where the analyte displays non-specific ­binding with the support) kdn,app Apparent dissociation rate constant for interaction with the support (in situations where the analyte displays non-specific binding with the support) kmax Retention factor at maximum kmin Retention factor at minimum L Column length (or ligand in the case of an analyte–ligand interaction) Load A Relative moles of solute applied to the column versus the total moles of active ligand m Slope of a linear plot to examine peak profiling data mL Mole of active immobilized ligands n Number of theoretical plates n1…nn Fraction of each type of site in the column nmt Global mass transfer coefficient qx Column loading capacity R Ideal gas law constant So Split-peak constant, where So = F/ka,app mL for a homogeneous ligand T Absolute temperature t M Column void time (corrected for extra column time) tR Retention time for an analyte on a column (corrected for extra column time) u Linear velocity of the mobile phase VA Analyte breakthrough volume (frontal application) VA* Non-retained species breakthrough volume (frontal application) VM Column void volume V0 Breakthrough volume of void marker (frontal affinity application) V P Pore volume VR Retention volume of analyte (zonal application) Ve Excluded volume x Reduced retention time y Signal intensity κ Kinetic factor for peak profiling γ Tortuosity factor α Ratio of binding capacities for two binding sites (mL1/mL2) αL Fraction of k due to interaction with an immobilized ligand αn Fraction of k due to non-specific interactions with the support β Ratio of association equilibrium constants for two binding sites (Ka2/Ka1) ∆G Change in free energy ∆H Change in enthalpy ∆S Change in entropy σ 2A Variance of analyte breakthrough curve (frontal application)

Biointeraction Affinity Chromatography

187

σ 2M Peak variance for a non-retained species (corrected for extra-column variance) σ 2R Peak variance for a retained species (corrected for extra-column variance)

References Allenmark, S. 1991. Chromatographic enantioseparation: Methods and applications. 2nd ed. New York: Ellis Horwood. Anderson, D. J., and R. R. Walters. 1986. Equilibrium and rate constants of immobilized concanavalin A determined by high-performance affinity chromatography. J. Chromatogr. 376:69–85. Arata, Y., N. Ishii, M. Tamura, T. Nonaka, and K.-I. Kasai. 2007. Identification of the amino acid residue in the nematode galectin lec-1 responsible for its unique sugar binding property: Analysis by combination of site-directed mutagenesis and frontal affinity chromatography. Biol. Pharm. Bull. 30 (11):2012–2017. Arnold, F. H., S. A. Schofield, and H. W. Blanch. 1986. Analytical affinity chromatography, I: Local equilibrium theory and the measurement of association and inhibition constants. J. Chromatogr. 355:1–12. Asnin, L., K. Kaczmarski, and G. Guiochon. 2008. Features of the adsorption of naproxen enantiomers on weak chiral anion-exchangers in nonlinear chromatography. J. Chromatogr. A. 1192:62–73. Berezhovskiy, L. M. 2006. The influence of drug kinetics in blood on the calculation of oral bioavailability in linear pharmacokinetics: The traditional equation may considerably overestimate the true value. J. Pharm. Sci. 95:834–848. Chaiken, I. M., ed. 1987. Analytical affinity chromatography. Boca Raton: CRC Press. Chan, N., D. Lewis, M. Kelly, E. S. M. Ng, and D. C. Schriemer. 2007. Frontal affinity ­chromatography-mass spectrometry for ligand discovery and characterization. In Mass spectrometry in medicinal chemistry, edited by K. T. Wanner and G. Hofner. Weinhein: WILEY-VCH Verlag, pp.217–246. Chattopadhyay, A., T. Tian, L. Kortum, and D. S. Hage. 1998. Development of tryptophan­modified human serum albumin columns for site-specific studies of drug-­protein ­interactions by high-performance affinity chromatography. J. Chromatogr. B 715: 183–190. Chen, J. 2003. Chromatographic studies of drug-protein interactions, PhD dissertation, University of Nebraska, Lincoln. Chen, J., and D. S. Hage. 2004. Quantitative analysis of allosteric drug-protein binding by biointeraction chromatography. Nat. Biotechnol. 22:1445–1448. Chen, J., J. E. Schiel and D. S. Hage. 2009. Non-competitive peak decay analysis of drug– protein dissociation by high-performance affinity chromatography and peak profiling. J. Sep. Sci. 32: 1632–1641. Chen, Y. 1988. A mathematical theory of the random walk model in chromatography. J. Radioanal. Nuclear Chem. 123:667–681. Clarke, W., J. E. Schiel, A. Moser, and D. S. Hage. 2005. Analysis of free hormone fractions by an ultrafast immunoextraction/displacement immunoassay: Studies using free thyroxine as a model system. Anal. Chem. 77:1859–1866. Dalgaard, L., J. J. Hansen, and J. L. Pedersen. 1989. Resolution and binding site determination of DL-thyronine by high-performance liquid chromatography using immobilized albumin as chiral stationary phase: Determination of the optical purity of thyroxine in tablets. J. Pharm. Biomed. Anal. 7:361–368. De Lucena, S. L., R. G. Carbonell, and C. C. Santana. 1999. Peptide affinity chromatography process for adsorption fibrinogen. Powder Technology 101:173–177.

188

Advances in Chromatography: Volume 48

Denizot, F. C., and M. A. Delaage. 1975. Statistical theory of chromatography: New outlooks for affinity chromatography. Proc. Natl. Acad. Sci. U.S.A. 72:4840–4843. Domenici, E., C. Bertucci, P. Salvadori, et al. 1990. Synthesis and chromatographic properties of an hplc chiral stationary phase based upon human serum albumin. Chromatographia 29:170–176. Domenici, E., C. Bertucci, P. Salvadori, S. Motellier, and I. W. Wainer. 1990. Immobilized serum albumin: Rapid HPLC probe of stereoselective protein-binding interactions. Chirality 1:263–268. Domenici, E., C. Bertucci, P. Salvadori, and I. W. Wainer. 1991. Use of a human serum albumin based high-performance liquid chromatography chiral stationary phase for the investigation of protein binding: Detection of the allosteric interaction between warfarin and benzodiazepine binding sites. J. Pharm. Sci. 80:164–166. Dunn, B. M., and I. M. Chaiken. 1974. Quantitative affinity chromatography: Determination of binding constants by elution with competitive inhibitors. Proc. Natl. Acad. Sci. U.S.A. 71:2382–2385. Ettre, L. S. 1993. Nomenclature for chromatography. Pure Appl. Chem. 65:819–872. Felinger, A. 2008. Molecular dynamic theories in chromatography. J. Chromatogr. A. 1184:20–41. Fitos, I., J. Visy, M. Simonyi, and J. Hermansson. 1992. Chiral high-performance liquid ­chromatographic separations of vinca alkaloid analogs on alpha1-acid glycoprotein and human serum albumin columns. J. Chromatogr. 609:163–171. Giddings, J. C. 1965. Dynamics of chromatography. New York: Marcel Dekker. Giddings, J. C. 1991. Unified separation science. New York: John Wiley. Giddings, J. C., and H. Eyring. 1955. A molecular dynamic theory of chromatography. J. Phys. Chem. 59:416–421. Golshan-Shirazi, S., and G. Guichon. 1992. Comparison of the various kinetic models of nonlinear chromatography. J. Chromatogr. 603:1–11. Grushka, E., L. R. Snyder, and J. H. Knox. 1975. Advances in band spreading theories. J. Chromatogr. Sci. 13:25–37. Gustavsson, P.-E., and P.-O. Larsson. 2006. Support materials for affinity chromatography. In Handbook of affinity chromatography, edited by D. S. Hage. New York: Taylor & Francis, pp.15–33. Hage, D. S. 1998. Affinity chromatography. In Handbook of HPLC, edited by E. Katz, R. Eksteen, and N. Miller. New York: Marcel Dekker, pp.483–498. Hage, D. S. 2002. High-performance affinity chromatography: A powerful tool for studying serum protein binding. J. Chromatogr. B 768:3–30. Hage, D. S., ed. 2006. Handbook of affinity chromatography. New York: Taylor & Francis. Hage, D. S., and J. Chen. 2006. Quantitative affinity chromatography: Practical aspects. In Handbook of affinity chromatography, edited by D. S. Hage. New York: Taylor & Francis. Hage, D. S., and P. F. Ruhn. 2006. An introduction of affinity chromatography. In Handbook of affinity chromatography, edited by D. S. Hage. New York: Taylor & Francis, pp.3–13. Hage, D. S., and A. Sengupta. 1998. Studies of protein binding to nonpolar solutes by using zonal elution and high-performance affinity chromatography: Interactions of cisand trans-clomiphene with human serum albumin in the presence of β-cyclodextrin. Anal. Chem. 70:4602–4609. Hage, D. S., D. H. Thomas, and M. S. Beck. 1993. Theory of a sequential addition competitive binding immunoassay based on high performance immunoaffinity chromatography. Anal. Chem. 65:1622–1630. Hage, D. S., and S. A. Tweed. 1997. Recent advances in chromatographic and electrophoretic  methods for the study of drug-protein interactions. J. Chromatogr. B 699:499–525.

Biointeraction Affinity Chromatography

189

Hage, D. S., and R. R. Walters. 1988. Non-linear elution effects in split-peak chromatography I. Computer simulations for the cases of irreversible diffusion- and adsorption-limited kinetics. J. Chromatogr. 436:111–135. Hage, D. S., R. R. Walters, and H. W. Hethcote. 1986. Split-peak affinity chromatographic studies of the immobilization-dependent adsorption kinetics of protein A. Anal. Chem. 58:274–279. Hage, D. S., H. Xuan, and M. A. Nelson. 2006. Application and elution in affinity chromatography. In Handbook of affinity chromatography, edited by D. S. Hage. New York: Taylor and Francis. Hethcote, H., and C. Delisi. 1982a. Determination of equilibrium and rate constants by affinity chromatography. J. Chromatogr. 248:183–202. Hethcote, H., and C. Delisi. 1982b. Non-equilibrium model of liquid column chromatography I. Exact expressions for elution profile moments and relation to plate height theory. J. Chromatogr. 240:269–281. Hethcote, H., and C. Delisi. 1987. Chromatographic theory and application to quantitative affinity chromatography. In Analytical affinity chromatography, edited by I. M. Chaiken. Boca Raton: CRC Press. Horvath, C. 1978. Band spreading in liquid chromatography. General plate height equation and a method for the evaluation of the individual plate height contributions. J. Chromatogr. 149:43–70. Horvath, C., and H.-J. Lin. 1976. Movement and band spreading of unsorbed solutes in liquid chromatography. J. Chromatogr. 126:401–420. Itakura, Y., S. Nakamura-Tsuruta, J. Kominami, et al. 2007. Systematic comparison of oligosaccharide specificity of ricinus communis agglutinin I and erythrina lectins: A search by frontal affinity chromatography. J. Biochem. 142:459–469. Iwaki, J., T. Minamisawa, H. Tateno, et al. 2008. Desulfated galactosaminoglycans are potential ligands for galectins: Evidence from frontal affinity chromatography. Biochem. Biophys. Res. Comm. 373:206–212. Jaulmes, A., and C. Vidal-Madjar. 1991. Split-peak phenomenon in nonlinear chromatography. 1. A theoretical model for irreversible adsorption. Anal. Chem. 63:1165–1174. Jaulmes, A., C. Vidal-Madjar, and A. Pantazaki. 2001. Adsorption kinetics of human serum albumin on various kinds of supports. Chromatographia 53:S417–S423. Jewell, R. C., K. L. R. Brouwer, and P. J. Mcnamara. 1989. α1 Acid glycoprotein high ­performance liquid chromatography column (Enantio-Pac) as a screening tool for ­protein binding. J. Chromatogr. 487:257–264. Joseph, K. S., A. C. Moser, S. Basiaga, J. E. Schiel, and D. S. Hage. 2009. Evaluation of alternatives to warfarin as probes for Sudlow site I of human serum albumin: Characterization by high-performance affinity chromatography. J. Chromatogr. A 1216: 3492–3500. Jozwiak, K., J. Haginaka, R. Moaddel, and I. Wainer. 2002. Displacement and nonlinear ­chromatographic techniques in the investigation of interaction of noncompetitive inhibitors with an immobilized α3β4 nicotinic acetylcholine receptor liquid chromatographic stationary phase. Anal. Chem. 74:4618–4624. Jozwiak, K., S. C. Hernandex, K. J. Kellar, and I. Wainer. 2003. Enantioselective interactions of dextromethorphan and levomethorphan with the α3β4-nicotinic acetylcholine receptor: Comparison of chromatographic and functional data. J. Chromatogr. B 797:373–379. Jozwiak, K., S. Ravichandran, J. R. Collins, R. Moaddel, and I. Wainer. 2007. Interaction of noncompetitive inhibitors with the α3β2 nicotinic acetylcholine receptor investigated by affinity chromatography and molecular docking. J. Med. Chem. 50:6279–6283. Jozwiak, K., S. Ravichandran, J. R. Collins, and I. Wainer. 2004. Interaction of noncompetitive inhibitors with an immobilized α3β4 nicotinic acetylcholine receptor investigated by affinity chromatography. Quantitative-structure activity relationship analysis and molecular docking. J. Med. Chem. 47:4008–4021.

190

Advances in Chromatography: Volume 48

Kasai, K., and S. Ishii. 1975. Affinity chromatography of trypsin and related enzymes, I: Preparation and characteristics of an affinity adsorbent containing tryptic peptides from protamine as ligands. J. Biochem. 78:653–662. Kim, H. S., and D. S. Hage. 2005. Chromatographic analysis of carbamazepine binding to human serum albumin. J. Chromatogr. B 816:57–66. Kim, H. S., and D. S. Hage. 2006. Immobilization methods for affinity chromatography. In Handbook of affinity chromatography, edited by D. S. Hage. New York: Taylor & Francis, pp.35–78. Kimura, T., K. Nakanishi, T. Nakagawa, A. Shibukawa, and K. Matsuzaki. 2005a. Highperformance frontal analysis of the binding of thyroxine enantiomers to human serum albumin. Pharm. Res. 22:667–675. Kimura, T., K. Nakanishi, T. Nakagawa, A. Shibukawa, and K. Matsuzaki. 2005b. Simultaneous determination of unbound thyroid hormones in human plasma using high performance frontal analysis with electrochemical detection. J. Pharm. Biomed. Anal. 38:204–209. Kimura, T., A. Shibukawa, and K. Matsuzaki. 2006. Biantennary glycans as well as genetic variants of α1-acid glycoprotein control the enantioselectivity and binding affinity of oxybutynin. Pharm. Res. 23:1038–1042. Kovarik, P., R. J. Hodgson, T. Covey, M. A. Brook, and J. D. Brennan. 2005. Capillary-scale frontal affinity chromatography MALDI tandem mass spectrometry using protein-doped monolithic silica columns. Anal. Chem. 77:3340–3350. Lee, W.-C., and C.-H. Chen. 2001. Predicting the elution behavior of proteins in affinity chromatography on non-porous particles. J. Biochem. Biophys. Methods 49:63–82. Lee, W.-C., and C.-Y. Chuang. 1996. Performance of pH elution in high-performance affinity chromatography of proteins using non-porous silica. J. Chromatogr. A. 721:31–39. Lei, G., R. Yang, X. Zeng, Y. Shen, X. Zheng, and Y. Wei. 2007. Use of frontal chromatography to measure the binding interaction of berberine chloride with bovine serum albumin. Chromatographia 66:847–852. Loun, B., and D. S. Hage. 1992. Characterization of thyroxine-albumin binding using ­high-performance affinity chromatography I: Interactions at the warfarin and indole sites of albumin. J. Chromatogr. 579:225–235. Loun, B., and D. S. Hage. 1994. Chiral separation mechanisms in protein-based HPLC ­columns. 1. Thermodynamic studies of (R)- and (S)-warfarin binding to immobilized human serum albumin. Anal. Chem. 66:3814–3822. Loun, B., and D. S. Hage. 1996. Chiral separation mechanisms in protein-based HPLC ­columns. 2. Kinetic studies of (R)- and (S)-warfarin binding to immobilized human serum albumin. Anal. Chem. 68:1218–1225. Mallik, R., and D. S. Hage. 2006. Affinity monolith chromatography. J. Sep. Sci. 29:1686–1704. Mallik, R., H. Xuan, G. Guiochon, and D. S. Hage. 2008. Immobilization of α1-acid glycoprotein for chromatographic studies of drug–protein binding II. Correction for errors in association constant measurements. Anal. Biochem. 376:154–156. Mao, Q. M., A. Johnston, J. G. Prince, and T. W. Hearn. 1991. High-performance liquid chromatography of amino acids peptides and proteins. CXII. Predicting the performance of non-porous particles in affinity chromatography of proteins. J. Chromatogr. 548:147–163. Marszall, M. P., R. Moaddel, K. Jozwiak, M. Bernier, and I. Wainer. 2008. Initial synthesis and characterization of an immobilized heat shock protein 90 column for online determination of binding affinities. Anal. Biochem. 373:313–321. Martin, J. P., and R. L. M. Synge. 1941. A new form of chromatogram employing two liquid phases. Biochem. J. 35:1358–1368. Miller, J. H. M., and G. A. Smail. 1977. Interaction of the enantiomers of warfarin with human serum albumin, peptides and amino acids. J. Pharm. Pharmacol. 29:33–33.

Biointeraction Affinity Chromatography

191

Moaddel, R., K. Jozwiak, R. Yamaguchi, and I. W. Wainer. 2005. Direct chromatographic ­determination of dissociation rate constants of ligand-receptor complexes: Assessment of  the interaction of noncompetitive inhibitors with an immobilized nicotinic acetylcholine receptor-based liquid chromatography stationary phase. Anal. Chem. 77:5421–5426. Moaddel, R., and I. Wainer. 2007. Conformational mobility of immobilized proteins. J. Pharm. Biomed. Anal. 43:399–406. Moore, R. M., and R. R. Walters. 1987. Peak-decay method for the measurement of dissociation rate constants by high-performance affinity chromatography. J. Chromatogr. 384:91–103. Moser, A. 2005. Development and use of new affinity ligands for pharmaceutical analysis, PhD dissertation, University of Nebraska, Lincoln. Muller, A. J., and P. W. Carr. 1984. Chromatographic study of the thermodynamic and kinetic characteristics of silica-bound concanavalin A. J. Chromatogr. 284:33–51. Munro, P. D., D. J. Winzor, and J. R. Cann. 1993. Experimental and theoretical studies of rate constant evaluation by affinity chromatography. Determination of rate constants for the interaction of saccharides with concanavalin A. J. Chromatogr. 646:3–15. Munro, P. D., D. J. Winzor, and J. R. Cann. 1994. Allowance for kinetics of solute partitioning in the determination of rate constants by affinity chromatography. J. Chromatogr. A. 659:267–273. Nakamura-Tsuruta, S., N. Uchiyama, and J. Hirabayashi. 2006. High-throughput analysis of lectin-oligosaccharide interactions by automated frontal affinity chromatography. Methods Enzymol. 415:311–325. Nakano, N. I., Y. Shimamori, and S. Yamaguchi. 1982. Binding capacities of human serum albumin monomer and dimer by continuous frontal affinity chromatography. J. Chromatogr. 237:225–232. Nelson, M. A. 2003. Studies of portable immunochromatographic methods for analysis of pesticide residues, PhD dissertation, University of Nebraska, Lincoln. Ng, E. S. M., F. Yang, A. Kameyama, et al. 2005. High-throughput screening for enzyme inhibitors using frontal affinity chromatography with liquid chromatography and mass spectrometry. Anal. Chem. 77:6125–6133. Noctor, T. A. G., M. J. Diaz-Perez, and I. W. Wainer. 1993. Use of a human serum albuminbased stationary phase for high-performance liquid chromatography as a tool for the rapid determination of drug-plasma protein binding. J. Pharm. Sci. 82:675–676. Noctor, T. A. G., C. D. Pham, R. Kaliszan, and I. W. Wainer. 1992. Stereochemical aspects of benzodiazepine binding to human serum albumin, I: Enantioselective high performance liquid affinity chromatographic examination of chiral and achiral binding interactions between 1,4-benzodiazepine and human serum albumin. Mol. Pharmacol. 42:506–511. Noctor, T. A. G., and I. W. Wainer. 1992. The in situ acetylation of an immobilized human serum albumin chiral stationary phase for high-performance liquid chromatography in the examination of drug-protein binding phenomena. Pharm. Res. 9:480–484. Noctor, T. A. G., I. W. Wainer, and D. S. Hage. 1992. Allosteric and competitive displacement of drugs from human serum albumin by octanoic acid, as revealed by high ­performance liquid chromatography, on a human serum albumin-based stationary phase. J. Chromatogr. 577:305–315. Ohnmacht, C. M., J. E. Schiel, and D. S. Hage. 2006. Analysis of free drug fractions using near-infrared fluorescent labels and an ultrafast immunoextraction/displacement assay. Anal. Chem. 75:7547–7556. Parikh, I., and P. Cuatrecasas. 1985. Affinity chromatography. Chem. Eng. News 63:17–29. Puerta, A., A. Jaulmes, M. De Frutos, J.-C. Diez-Masa, and C. Vidal-Madiar. 2002. Adsorption kinetics of β-lactoglobulin on a polyclonal immunochromatographic support. J. Chromatogr. A 953:17–30.

192

Advances in Chromatography: Volume 48

Puerta, A., C. Vidar-Madjar, A. Jaulmes, J.-C. Diez-Masa, and M. De Frutos. 2006. Frontal analysis for characterizing the adsorption-desorption behavior of β-lactoglobulin on immunoadsorbents. J. Chromatogr. A 1119:34–42. Renard, J., C. Vidal-Madiar, B. Sebille, and C. Labresle. 1995. Chromatographic kinetic measurements of human serum albumin adsorption on monoclonal antibodies. J. Mol. Rec. 8:85–89. Renard, J., and C. Vidal-Madjar. 1994. Kinetic study of the adsorption of human serum ­albumin on immobilized antibody using the split-peak effect in immunochromatography. J. Chromatogr. A. 661:35–42. Rollag, J. G., and D. S. Hage. 1998. Non-linear elution effects in split-peak chromatography. II. Role of ligand heterogeneity in solute binding to columns with adsorption-limited kinetics. J. Chromatogr. A. 795:185–198. Ruta, J., C. Ravelet, J. Désiré, J.-L. Décout, and E. Peyrin. 2008. Covalently bonded DNA aptamer chiral stationary phase for the chromatographic resolution of adenosine. Anal. Bioanal. Chem. 390:1051–1057. Schiel, J. E., R. Mallik, S. Soman, K. S. Joseph, and D. S. Hage. 2006. Applications of silica supports in affinity chromatography. J. Sep. Sci. 29:719–737. Schiel, J. E., C. M. Ohnmacht, and D. S. Hage. 2009. Measurement of drug-protein dissociation rates by high-performance affinity chromatography and peak profiling. Anal. Chem. 81: 4320–4333. Schiel, J. E., E. Papastavros, and D. S. Hage. 2009. Rapid determination of drug-protein ­dissociation rates by high performance affinity chromatography: Analysis of multisite systems with non-specific interactions. Submitted. Schill, G., I. W. Wainer, and S. A. Barkan. 1986. Chiral separations of cationic and anionic drugs on an α1-acid glycoprotein-bonded stationary phase (Enantio-Pac). II: Influence of mobile phase additives and pH on chiral resolution and retention. J. Chromatogr. 365:73–88. Schriemer, D. C. 2004. Biosensor alternative: Frontal affinity chromatography. Anal. Chem. 76 (23):440A–448A. Schriemer, D. C., D. R. Bundle, L. Li, and O. Hindsgaul. 1998. Micro-scale frontal affinity chromatography with mass spectrometric detection: A new method for the screening of compound libraries. Angewandte Chemie Int. Ed. 37:3383–3387. Scouten, W. H. 1981. Affinity chromatography: Bioselective adsorption on inert matrices. New York: Wiley. Sebille, B., and N. Thuaud. 1978. Study of binding of low-molecule weight ligand to biological macromolecules by high-performance liquid chromatography. J. Chromatogr. 167:159–170. Sengupta, A., and D. S. Hage. 1999. Characterization of minor site probes for human serum albumin by high performance affinity chromatography. Anal. Chem. 71:3824–3827. Slon-Usakiewicz, J. J., J.-R. Dai, W. Ng, et al. 2005. Global kinase screening. Applications of frontal affinity chromatography coupled to mass spectrometry in drug discovery. Anal. Chem. 77:1268–1274. Smith, Q. R., C. Fisher, and D. D. Allen. 2000. The role of plasma protein binding in drug delivery to brain. In Blood-brain barrier: Drug delivery and brain pathology, edited by D. Kobiler, S. Lustig, and S. Shapira. New York: Kluwer Academic/Plenum Publishers. Swaisgood, H. E., and I. M. Chaiken. 1986. Analytical high-performance affinity chromatography: Evaluation by studies of neurophysin self-association and neurophysin-peptide hormone interaction using glass matrices. Biochemistry 25:4148–4155. Swaisgood, H. E., and I. M. Chaiken. 1987. Analytical affinity chromatography and characterization of biomolecular interactions. In Analytical affinity chromatography, edited by I. M. Chaiken. Boca Raton: CRC Press.

Biointeraction Affinity Chromatography

193

Tachibana, K., S. Nakamura, H. Wang, et al. 2006. Elucidation of binding specificity of jacalin toward o-glycosylated peptides: Quantitative analysis by frontal affinity chromatography. Glycobiology 16:46–53. Talbert, A. M., G. E. Tranter, E. Holmes, and P. L. Francis. 2002. Determination of drug-­protein binding kinetics and equilibria by chromatographic peak profiling: Exemplification of the method using L-tryptophan and albumin. Anal. Chem. 74:446–452. Thomas, H. C. 1944. Heterogeneous ion exchange in a flowing system. J. Am. Chem. Soc. 66:1664–1665. Turkova, J. 1978. Affinity chromatography. Amsterdam: Elsevier. Tweed, S. A., B. Loun, and D. S. Hage. 1997. Effects of ligand heterogeneity in the characterization of affinity columns by frontal analysis. Anal. Chem. 69:4790–4798. Vidal-Madjar, C., A. Jaulmes, M. Racine, and B. Sebille. 1988. Determination of binding equilibrium constants by numerical simulation in zonal high-performance affinity chromatography. J. Chromatogr. 458:13–25. Vidal-Madjar, C., A. Jaulmes, J. Renard, D. Peter, and P. Lafaye. 1997. Chromatographic study of the adsorption kinetics of albumin on monoclonal and polyclonal immunoadsorbents. Chromatographia 45:18–24. Wa, C., R. L. Cerny, and D. S. Hage. 2006. Identification and quantitative studies of protein immobilization sites by stable isotope labeling and mass spectrometry. Anal. Chem. 78:7967–7977. Wade, J. L., A. F. Bergold, and P. W. Carr. 1987. Theoretical description of nonlinear chromatography, with applications to physicochemical measurements in affinity chromatography and implications for preparative-scale separations. Anal. Chem. 59:1286–1295. Walters, R. R. 1985. Affinity chromatography. Anal. Chem. 57:1099A–1114A. Walters, R. R. 1987. Practical approaches for the measurement of rate constants by affinity chromatography In Analytical affinity chromatography, edited by I. M. Chaiken. Boca Raton: CRC Press. Winzor, D. J. 2004. Determination of binding constants by affinity chromatography. J. Chromatogr. A. 1037:351–367. Winzor, D. J. 2006. Quantitative affinity chromatography: Recent theoretical developments. In Handbook of affinity chromatography, edited by D. S. Hage. New York: Taylor & Francis. Winzor, D. J., P. D. Munro, and J. R. Cann. 1991. Experimental and theoretical studies of rate constant evaluation for the solute-matrix interaction in affinity chromatography. Anal. Biochem. 194:54–63. Xuan, H., and D. S. Hage. 2005. Immobilization of α1-acid glycoprotein for chromatographic studies of drug-protein binding. Anal. Biochem. 346:300–310. Yang, J., and D. S. Hage. 1993. Characterization of the binding and chiral separation of Dand L- tryptophan on a high-performance immobilized human serum albumin column. J. Chromatogr. 645:241–250. Yang, J., and D. S. Hage. 1996. Role of binding capacity versus binding strength in the separation of chiral compounds on protein-based high-performance liquid chromatography ­columns: Interactions of D- and L-tryptophan with human serum albumin. J. Chromatogr. A. 725:273–285. Yang, J., and D. S. Hage. 1997. Effect of mobile phase composition on the binding ­kinetics of chiral solutes on a protein-based high-performance liquid chromatography ­column: Interactions of D- and L-tryptophan with immobilized human serum albumin. J. Chromatogr. A. 766:15–25.

of 5 Characterization Stationary Phases in Supercritical Fluid Chromatography with the Solvation Parameter Model Caroline West and Eric Lesellier Contents 5.1 Introduction................................................................................................... 195 5.2 Experimental Conditions............................................................................... 201 5.2.1 Chromatographic System.................................................................. 201 5.2.2 Choice of Chromatographic Conditions............................................202 5.2.3 Selection of Columns......................................................................... 203 5.2.4. Selection of a Set of Test Compounds...............................................206 5.2.5 Data Analysis..................................................................................... 218 5.3 Choice of Solvation Descriptors for Supercritical Fluids.............................. 225 5.4 A General Database for Packed Column SFC............................................... 228 5.4.1 Variation of the System Constants among the Stationary Phases..... 228 5.4.2 A Visual Representation of the Database.......................................... 232 5.4.3 ODS Phases.......................................................................................240 5.4.4 Method Development with the Solvation Parameter Model..............244 5.4.5 Predictive Capability of the Models.................................................. 247 5.5 Conclusion.....................................................................................................248 References............................................................................................................... 249

5.1  Introduction Supercritical fluid chromatography (SFC) has long held an uncertain place among other separation techniques. However, recently, packed column SFC (pSFC) has become more attractive again to drug discovery [1–4] because it offers aqueous-free purification capabilities with a reputed green solvent (carbon dioxide), together with 195

196

Advances in Chromatography: Volume 48

high speed. As a matter of fact, the lower eluent viscosities and higher solute diffusivities in SFC often translate into increased efficiency and shorter ­separation times when compared with high-performance liquid chromatography (HPLC). Shorter analysis time is also the result of the use of larger flow rates (typically 3 to 5  ml min−1) in comparison to HPLC, as the low viscosity of the fluid induces only small pressure drops along the chromatographic column. Nearly all current research in pSFC is performed with column packings prepared for HPLC, as those work equally in SFC. Only a few manufacturers produce columns dedicated to pSFC. Thus, just as for HPLC, most stationary phases are silicabased, chemically bonded or encapsulated, or polymeric; and all available in a wide range of chemistries. The most widely used are bare silica and silica-based sorbents of the monomeric type, bonded with 3-aminopropyl-, 3-cyanopropyl-, or a spacer bonded propanediol-siloxane, thus mostly normal phase HPLC (NP-HPLC) stationary phases. The 2-ethylpyridine phase introduced by Princeton Chromatography is one of the only original packing designed specifically for SFC, which has faced a certain success. Besides, numerous applications (see for example references [5–13]) have shown in the past that nonpolar phases as octadecylsiloxane-bonded silica (ODS) phases can also be very useful in SFC as they provide improved separations compared to reversed-phase HPLC (RP-HPLC). All stationary phase chemistries are indeed useful in developing an SFC method. As a matter of fact, due to the variety of possibilities, the initial choice of a chromatographic system (mobile phase and stationary phase) in SFC is a complex problem. Actually, all stationary phases available for HPLC, and any solvent that is miscible to carbon dioxide (and compatible with the stationary phase) can be combined. Unfortunately, this wide diversity is generally associated to a global lack of knowledge of the interactions established between the analytes and the chromatographic system. Knowledge of the behavior in liquid phase is often of no help as the absence of water in the mobile phase causes drastic differences between reversedphase HPLC (RPLC) and SFC behaviors. Thus only SFC studies are helpful when developing an SFC method. As there are few clear guidelines for the choice of a stationary phase for a particular analyte, often more than one phase needs to be examined in order to obtain a suitable resolution. Luckily, rapid column equilibration makes it easy to use fast gradient analysis or to change the chromatographic parameters as well as the stationary phase for rapid method development. The approach of changing separation selectivity with the stationary phase is especially useful and made easy by modern systems of column selection provided by most SFC systems. With such systems, columns with different stationary phases are rapidly scouted using automatic column switching with a single or different mobile phase. However, in too many cases, column selection is based on which column worked best previously. Besides, changing the column for another one that would provide the same selectivity is of no use. Therefore, a tool was needed to determine which stationary phases are most orthogonal and would thus provide complementary selectivities in order to obtain the required separations. Despite some studies describing the relationship between the chromatographic behaviors and the compound chemical structure, or the eluotropic strength of varied CO2-modifier mobile phases in regards to the one measured in HPLC, performed

Characterization of Stationary Phases

197

either with ODS [14–18] or polar stationary phases [19–21], there has been surprisingly little work done to understand and compare the interactions involved in SFC separations when the stationary phase is varied. Indeed, the primary goal of most pSFC studies was to provide an understanding of the role of solvent modifiers and additives on retention properties. Whenever one separation was investigated on different stationary phases, they were only compared under a limited set of conditions. While this information is useful for that particular type of separation, it does not provide information as to which stationary phase is most appropriate for a different separation problem. Furthermore, a direct comparison of literature data from different publications cannot be made easily because most published studies are application studies intended for different solutes, so different operating conditions were used to optimize the separations. As chromatographic effects are compound-dependent, a column superior in one application can appear worse in another. The retention and selectivity properties of different phases were compared by Schoenmakers et al. [22]. However, this work was carried out with the intent to evaluate stationary phases for use with pure CO2 as a mobile phase, to retain compatibility with flame ionization detection. As a matter of fact, the partial conclusions on the elution and on the peak deformation are only valid for these operating conditions, as some stationary phases that were found inappropriate can be very useful if a polar modifier is used in the mobile phase. For instance, the amino phase did not perform well in pure CO2, while it is generally found appropriate when used with a mixed mobile phase. However, the use of a modifier is common practice in pSFC today, thus studies carried out in pure CO2 have no reason for living anymore. Heaton et al. [23] proposed a retention model for pSFC based on solubility parameters dissociated into components such as dispersive, dipole–dipole, dipole-induced dipole, and hydrogen bonding. This approach was promising but the authors did not furnish any interpretation of the results, their intention was simply to evaluate the capability of the method for retention prediction. Judging from this confused situation, a standardized method for the characterization and comparison of pSFC systems was required. Consequently, we investigated a number of columns containing various bonding chemistries and obtained from a variety of manufacturers. Our objective was to acquire semi-quantitative knowledge of the physico-chemical interactions governing retention and separations in a given chromatographic system. We wished to be able to compare different stationary phases, but also different mobile phases. Indeed, as the stationary phase is solvated and, as this solvation can be selective when the mobile phase is binary or ternary, the three-dimensional structure and the solvation state of the stationary phase depend on the composition of the mobile phase. As a matter of fact, the chromatographic behavior of a stationary phase depends on the mobile phase used. Thus the chosen testing procedure had to allow knowing the chromatographic properties of the stationary phase equilibrated with a given mobile phase. Besides, the interest in comparing mobile phase effects with one stationary phase was also related to our will to compare HPLC and SFC on a reasonable basis. Among the possible criteria used to evaluate the stationary phase properties, the physical parameters of the column (carbon load, particle size, specific surface area…) generally show no simple and direct correlation with their performances,

198

Advances in Chromatography: Volume 48

and thus are not very informative on the final separation quality. The spectrometric analyses (NMR, infrared, or x-ray fluorescence spectroscopic techniques) are employed to study the surface of the stationary phase but the relationship with chromatographic performance is variable. Only the observation of the chromatographic behavior of varied compounds can inform on the selectivity of a chromatographic system. However, a method was required to extract microscopic information (the nature of the molecular interactions established between the solutes and stationary and mobile phases) based on macroscopic data (chromatographic retention factors). No standard test has been accepted and prevails for the characterization of chromatographic systems but some of them are more popular than others. In order to compare SFC observations with HPLC, we chose to focus on existing testing ­procedures, which relevance had already been established, rather than trying to ­conceive a new test for the chromatographic systems we wished to characterize. Besides, an acceptable testing procedure had to be relatively rapid, reproducible, and cheap. A variety of such tests established for HPLC or GC can be found in the literature. The key-solute method has been widely used, particularly for the characterization of ODS phases in HPLC. It is based on the use of a small set of solutes, which behavior (retention factor or separation factors between two solutes) is supposed to be representative of the retention of any possible solute, or characterizes a particular property of the stationary phase as bonding density, or the accessibility to residual silanol groups. However, depending on the chosen solutes, the results are not always consistent between two tests. The choice of compounds itself can be dubious. Methods exist that can help selecting the best solute set, as principal component analysis (PCA), but chromatographic common sense is often the most efficient tool for this selection. PCA is a powerful tool for the comparison of a great amount of data. However the interpretation of the results is generally not simple because each principal component may contain information from several different parameters, thus its significance often remains abstract. The reduction of the number of parameters can help in simplifying the problem and improving the understanding, but this operation can be a delicate one. Quantitative structure-retention relationships (QSRRs) are another alternative to the above methods. They consist of extracting precise data for the characterization of chromatographic systems from a great number of key-solutes. Judging from the abundant literature on QSRRs, it is clear that they allow comparisons of chromatographic systems in an efficient manner. Two types of data are necessary to construct a QSRR: chromatographic retention data for a sufficiently large group of compounds, and a group of data supposedly reflecting the physico-chemical properties of the said solutes. This last point, the representation of molecular structures, is the very heart of QSRRs. Indeed, it is necessary to convert the molecular structure into mathematical data; that is to say to calculate solute descriptors. Among QSRRs, the solvation parameter model using Abraham descriptors has gained acceptance as a general tool to explore the factors affecting retention in chromatographic systems [24–26]. The retention of selected probes can be related

Characterization of Stationary Phases

199

through this relationship, also known as linear solvation energy relationship (LSER), to specific interactions by the following equation, with the notation that is now commonly adopted:

log k = c + eE + sS + aA + bB + vV

(5.1)

In this equation, capital letters represent Abraham solute descriptors, related to particular interaction properties, while lower case letters represent the system constants, related to the complementary effect of the phases on these interactions. c is the model intercept term, which when the retention factor is used as the dependent variable is dominated by the phase ratio, i.e., the ratio of stationary and mobile phase volumes. The following terms are represented on Figure 5.1. E is the excess molar refraction (calculated from the refractive index of the molecule) and models polarizability contributions from n and π electrons; S is the solute dipolarity / polarizability; A and B are the solute overall hydrogen-bond acidity and basicity; V is McGowan characteristic volume in units of cm3 mol–1/100. The system constants (e, s, a, b, v), obtained through a multilinear regression of the retention data for a certain number of solutes with known descriptors, reflect the magnitude of difference for that particular property between the mobile and stationary phases. Thus, if a particular coefficient is numerically large, then any solute having the complementary property will interact very strongly with either the mobile phase (if the coefficient is negative) or the stationary phase (if the coefficient is positive). Moreover, Equation 5.2 can be deduced from Equation 5.1:

log α = eΔE + sΔS + aΔA + bΔB + vΔV

(5.2)

Where α is the separation factor between two solutes and ΔX represents the difference in the X descriptor between these two solutes. Consequently, the coefficients (e, s, a, b, v) do not reflect only the retention properties of the chromatographic system but also its selectivity toward any particular molecular interaction. The possible use of the above equations naturally depends on the availability of Abraham descriptors in the scientific literature. Those are now accessible for a wide range of solutes (about 4000) but for the model to have practical utility it will always be necessary to determine them for new solutes. This can be achieved through experiments (for all descriptors) [24], simple calculation (for E and V) [27–28], or with a fragment method of calculation (for E, S, A, B, and L) [29–30], which is the base of a software program (Absolv, Pharma Algorithms) [31]. Furthermore, the solvation parameter model uses parameters issued from chromatographic retention data obtained in GC and HPLC, as well as solvent–solvent equilibrium constants. As a matter of fact, it is principally based on data issued from partition processes. However, no assumption is made on the processes involved in the investigated system and it has been successfully used to describe phenomena ­relevant to adsorption processes. Although the statistical quality of models for adsorption processes are not as good as for partition systems, the results are chemically sound [32]. A consequence of this is the wide applicability of the model to

O

O

O O

S Dipole–dipole dipole-induced dipole

O

H O X O O

H O

A B Hydrogen-bonding with Hydrogen-bonding with a H-donor solute a H-acceptor solute

O

B:

Figure 5.1  Principle of the solvation parameter model: interactions related to each solute descriptor.

E π–π interactions dipole-induced dipole dispersive

O

O

O V Dispersive interactions cavity effects (in condensed phases)

O

200 Advances in Chromatography: Volume 48

Characterization of Stationary Phases

201

diverse and complex processes, as evidenced by the abundant literature. This was perfectly suited to our needs, as we expected to be able to compare very different stationary phase chemistries, ranging from bare silica to ODS phases. At the time we started our studies, there were only a few reported SFC applications of the use of the solvation parameter model: retention has been studied on ODS phases [33], cyanopropylsiloxane-bonded silica [34–35], divinylbenzene-ODS [36–37], polydimethylsiloxane [38–39], and porous graphitic carbon (PGC) [40–42] stationary phases. These models were all established with different mobile phases, temperatures, and pressure conditions, again rendering all comparisons quite difficult. Indeed, since the system constants represent the difference in sorption interactions for the solute in the mobile and stationary phase, any meaningful comparison of stationary phases must be made for the same mobile phase composition. Besides, these studies were mostly concerned with understanding the role of solvent modifiers and/or additives on retention properties, not much in understanding the stationary phase contribution. Moreover, different model types (as will be discussed below) further complicate the comparisons. In the following, we will discuss some particular aspects of the use of the solvation parameter model in pSFC. First of all, we will focus on the experimental conditions selected for this study then on the choice of solvation descriptors. Then we will present a database of stationary phases characterized with the solvation parameter model, all under the same supercritical conditions, and the way to exploit it for particular applications.

5.2 Experimental conditions 5.2.1 Chromatographic System Chromatographic separations were carried out using equipment manufactured by Jasco (Tokyo, Japan, supplied by Prolabo, Fontenay-sous-Bois, France). Two model 980-PU pumps were used, one for carbon dioxide (N45 quality, contained in gaseous tank) and a second for the modifier (methanol). The modifier pump performed control of the mobile phase composition. The pump head used for pumping carbon dioxide was cooled to −5°C by a cryostat (Julabo F10c, Seelbach, Germany, supplied by Touzart et Matignon, les Ulis, France). When the two mobile phase solvents (methanol and CO2) were mixed, the fluid was introduced into a dynamic mixing chamber PU 4046 (Pye Unicam, Cambridge, UK) connected to a pulsation damper (Sedere, supplied by Touzart et Matignon). The injector valve was supplied with a 20  μL loop (model 7125 Rheodyne, Cotati, CA, USA). Injection volumes ranged from 1 to 5 µL. The columns were thermostated by an oven (Jetstream 2 Plus, Hewlett-Packard, Palo Alto, USA), regulated by a cryostat (Haake D8 GH, Karlsruhe, Germany). The detector was a UV-visible HP 1050 (Hewlett-Packard), with a high-pressure resistant cell. The detection wavelength was 254 nm. After the detector, a Jasco 880-81 pressure regulator (supplied by Prolabo, Fontenay-sous-Bois, France) controlled the outlet column pressure. The outlet regulator tube (internal diameter 0.25 mm) was heated to 60°C to avoid ice formation during the CO2 depressurization.

202

Advances in Chromatography: Volume 48

Chromatograms were recorded using Azur software version 4.6 (Datalys, France). Methanol (MeOH) was HPLC grade and provided by Carlo Erba (Milan, Italy).

5.2.2 Choice of chromatographic conditions As our aim was to investigate the differences in stationary phase properties, it was important to choose some operating conditions that would be suitable to the wide variety of stationary phase chemistries that needed to be investigated. The operating conditions also needed to be consistent with common practice of pSFC today. On another hand, subcritical conditions were preferred to supercritical conditions for a variety of reasons. A supercritical fluid is defined as a fluid above its critical pressure and temperature. The name of SFC is thus very unfortunate as it lets those who are not familiar with the technique think that the fluid must be in its supercritical state. Actually, the fulfilment of both conditions (pressure and temperature) is not absolutely necessary as one can also work in the subcritical state, when only one of the two conditions is respected. Indeed, particularly in enantioselective pSFC, column temperatures of 25–30°C are frequently selected as providing the optimum resolution. Under these conditions, a carbon dioxide-based mobile phase will be below its critical temperature and will thus be subcritical. Besides, the addition of a modifier to the carbon dioxide mobile phase causes an increase in the critical parameters. For instance, if the critical temperature for pure CO2 is 31°C, for a MeOH-CO2 18:82 (v/v) mobile phase the critical temperature is increased to 75°C [43]. Thus studies carried out with a composition gradient varying the percentage of modifier from 5 to 40% while maintaining the pressure and temperature constant may well start with a supercritical fluid and end with a subcritical fluid. Fortunately, there are no significant changes in properties on going from super- to subcritical temperatures at moderate or high pressure [44–45], and there would be no reason for a change, as the so-called transition does not exist: there is a continuum of the state of matter when going from subcritical to supercritical region. Consequently, in this paper, the term SFC will be applied to chromatography carried out with both subcritical and supercritical fluids. Another good reason for working in subcritical conditions is that temperatures are very mild and, as a result, thermolabile solutes can be analyzed and the columns are very stable and have prolonged lifetimes. Consequently, we set the temperature at 25°C. The outlet pressure was maintained constant (15 MPa). This means that the internal pressure is not strictly constant for all columns, depending on column length, particle size, or porosity. However, modified subcritical fluids are less compressible than supercritical fluids, thus less susceptible to pressure variations, and most of our studies were performed with columns having the same geometrical dimensions, filled with 5 μm particles. The choice of mobile phase is highly significant because, as explained above, the solvation parameter model characterizes a whole chromatographic system, comprising the mobile phase and the stationary phase equilibrated with the said mobile phase. It was important to choose a mobile phase with a composition comparable to the mobile phases currently in use by pSFC chromatographers. According to

Characterization of Stationary Phases

203

common practice observed, and to the solubility of studied compounds, methanol was chosen as reference modifier. Studies carried out in SFC showed that the few percents of modifier (typically from 0 to 5%) induce significant retention decrease [16,23,33,36,39,46]. This retention decrease is less and less significant when the percentage of modifier is further increased. Different causes have been attributed to these observations. First of all, the first percents of modifier strongly adsorb onto the stationary phase, thereby reducing the interactions between the solutes and the stationary phase. Secondly, the addition of modifier is responsible for increased eluting strength of the mobile phase, due to two different processes: the great increase in the possible interactions between the solute and the mobile phase components and the slight variation in the density of the mobile phase. All in all, these factors together contribute to decreased retention. It is clear that the modification induced by the addition of co-solvent is therefore largely dependent on the nature of this solvent and on its ability to interact with the stationary phase or with the analyzed solutes. Besides, in the aim of characterizing stationary phases, it is not advisable to use large proportions of modifier in the mobile phase because, in this case, the influence of the stationary phase itself in the separation process is lessened, while the influence of the mobile phase is increased. Thus stationary phases become more alike at high modifier percentage. In order to be able to see significant differences in the stationary phase behaviors, it was found reasonable to use no more than 10% modifier in the mobile phase. Another important point is that the chosen mobile phase should allow measuring appropriate retention factors for all columns: the elution strength must be sufficient, so that the analysis time remains reasonable, but not too important otherwise the precision on the measurement of retention factors is poor. The 10% methanol content of the mobile phase rendered it possible. For the same reason, total flow rate was kept constant at 3 ml min−1. In addition to the modifier, it is common practice in pSFC to use small proportions of acids or bases, then called additives. Although they generally appear to provide improved peak shapes, the mechanisms of their action is still not well understood and seems to depend on the stationary phase and solutes used. We were concerned by the fact that different additives would adsorb onto the stationary phase to different extents, further complicating the understanding of already complex phenomena. Consequently, we chose not to use any additive in the mobile phase.

5.2.3  Selection of Columns All stationary phases presented in this work are commercially available. They are presented in Table 5.1. We were interested in scanning a large diversity of stationary phases, bonded or polymeric coated, based on silica, carbon, or polymers, and with different types of bonding chemistries, as can be seen on Figure 5.2. Sometimes, more than one column of each type is used. For example, three columns from different manufacturers were used to evaluate phenyl-hexyl-siloxane bonded silica phases.

C4 C8 C12 C18 C18-C RPH MIX PE1 PE2 PE3 CHL FD SI PEG PVA AMD DIOL NH2

Abbreviation

Butylsiloxane-bonded silica Octylsiloxane-bonded silica Dodecylsiloxane-bonded silica Octadecylsiloxane-bonded silica Octadecylsiloxane-bonded type-C silica Octadecyl- and phenylsiloxane-bonded silica Octadecyl- and phenylpropyl-bonded silica Amide-embedded hexadecylsiloxane-bonded silica Ether-sulfonamide-embedded hexadecylsiloxane-bonded silica Carbamate-embedded hexadecylsiloxane-bonded hybrid silica Cholesteryl-bonded silica Fluorodecylsiloxane-bonded silica Silica gel Poly(ethylene glycol) bonded on silica Poly(vinyl alcohol) bonded on silica Polyamide gel bonded on silica Propanediol-bonded silica Aminopropyl-bonded silica

Nature of the Stationary Phase

Table 5.1 Varied Stationary Phases Characterized in this Study Trade Name Uptisphere C4 Uptisphere C8 Synergi Max RP Kromasil C18 100 Cogent Bidentate C18 Uptisphere RPH Nucleodur Sphinx RP Supelcosil ABZ + Plus Acclaim Polar Advantage XBridge Shield Cosmosil Cholester Chromegabond Fluorodecyl Kromasil SIL 100 Discovery HS PEG YMC-Pack PVA-Sil TSK-gel Amide Diol Amino

Manufacturer Interchim Interchim Phenomenex Eka Nobel MicroSolv Technologies Interchim Macherey-Nagel Supelco Dionex Waters Nacalai Tesque ES Industry Kromasil Supelco YMC Tosoh Princeton Chromatography Princeton Chromatography

204 Advances in Chromatography: Volume 48

CN EP PGC PS OPHE DP DP-X C3P C6P-L C6P-G C6P-Z NAP PYE DNAP PNP PFP PBB

Cyanopropyl-bonded silica 2-Ethylpyridine-bonded silica Porous graphitic carbon Polystyrene-divinylbenzene Phenyl-oxypropyl-bonded silica Diphenyl-propyl-bonded silica Diphenyl-propyl-bonded silica Phenyl-propyl-bonded silica Phenyl-hexyl-bonded silica Phenyl-hexyl-bonded hybrid silica Phenyl-hexyl-bonded silica Naphtyl-ethyl-bonded silica 2-Pyrenyl-ethyl-bonded silica Dinitroanilido-propyl-bonded silica p-Nitrophenyl-bonded silica Pentafluorophenyl-propyl-bonded silica Pentabromobenzyl-oxypropyl-bonded silica

Cyano Ethylpyridine Hypercarb PLRP-S Synergi Polar RP Pursuit Diphenyl Pursuit XRs Diphenyl Uptisphere PH Luna Phenylhexyl Gemini Phenylhexyl Zorbax Eclipse + Phenylhexyl Cosmosil π-NAP Cosmosil 5PYE Uptisphere DNAP Nucleosil NO2 Discovery HS F5 Cosmosil 5PBB

Princeton Chromatography Princeton Chromatography Thermo-Hypersil Keystone Polymer Lab Phenomenex Varian Varian Interchim Phenomenex Phenomenex Agilent Nacalai Tesque Nacalai Tesque Interchim Macherey-Nagel Supelco Nacalai Tesque

Characterization of Stationary Phases 205

206

Advances in Chromatography: Volume 48 H2N C18

C12

C8

H

C4

Si

O n

PS

nH

AMD R

Si

PEG

R R Si R

SI F

F

Si

F F

F

F

F

F

F

F

F

O

F

CHL

O

NH

Si

PE1

O

Si

S

Si

PE2

Si

F F

F

PGC

F

Si

C18-C NH

Si

Si

Si

PNP

PE3

RPH

Si

O Si

MIX

NO2

NO2

Si

DNAP Br Br

F

O Si

Si

HN

C3P

2

NH2

NO2

O

O

NH

Si

DP DP-X

C6P-L C6P-G C6P-Z

OH Si

CN

Si

FD

OH

CN

Si

EP

F

H n

DIOL

N

F

H O O O Si Si Si O O O O

N

O

F

F F

O

PVA

O Si OH O O Si O O Si OH

Si

S

O n

F

F Br

F

F

Br Br O

Si

Si

Si

Si

Si

OPHE

NAP

PYE

PFP

PBB

Figure 5.2  Chemical structures of the stationary phases compared in this study. (See Table 5.1 for identification of the columns.)

All columns were 250*4.6 mm, apart from PGC (100*4.6 mm) and PS (150*4.6 mm). All columns were 5 μm, apart from C12, OPHE (4 µm), DIOL, NH2, CN and EP (6 µm). We put a special emphasis on ODS-bonded stationary phases, as they are the most widely used type in HPLC and a great diversity of bonding chemistries exist. The ODS phases selected, together with the known properties, can be found in Table 5.2. The columns were chosen for their representativeness of the possible treatments and bonding modes present in modern ODS phases. We were particularly interested in phases possessing a polar function in the endcapping group, or embedded in the alkyl chain. Unfortunately, not all manufacturers are willing to divulge the ­functionality, bonding technology and composition of their commercially available stationary phase column chemistries. As a matter of fact, some of the ODS phases selected have proprietary structure. By comparison of their system constants with those in the database, we hoped to shed some light on their chemistry.

5.2.4  Selection of a Set of Test Compounds The number of solutes used in our testing protocol has increased with time. Indeed, introducing new columns, with different retention behaviors, rendered it necessary

Stationary Phase Name

Zorbax StableBond C18 Zorbax Rx C18 Zorbax Extend Zorbax Eclipse XDB Kromasil C18

Gammabond C18 Uptisphere ODB Uptisphere HDO

Uptisphere TF Uptisphere NEC Nucleosil 50 C18 Nucleosil 100 C18 Nucleosil 5 C18 AB

Chromolith C18 XTerra MS C18 Atlantis dC18 Capcell Pak C18 Uptisphere RPH



1 2 3 4 5

6 7 8

9 10 11 12 13

14 15 16 17 P1

Merck Waters Waters Shiseido Macherey-Nagel

Interchim Interchim Macherey-Nagel Macherey-Nagel Macherey-Nagel

ES Industries Interchim Interchim

Agilent Agilent Agilent Agilent EKA Nobel

Manufacturer

Table 5.2 ODS Phases Characterized in this Study

300 175 330 300 330

310 320 450 350 350

na 320 320

180 180 185 180 350

Specific Surface Area (m2 g−1)

17 15.5 12 15 15

14 16 14 14 25

na 17 18

10 12 12.1 10.3 21.4

Carbon Content (%)

130 125 100 120 120

na 120 50 100 100

na 120 120

80 80 80 80 100

Pore Diameter (Å)

hybrid silica, trifunctional silane difunctional silane coated polymer ODS and phenylsiloxane

polymeric layer monofunctional silane monomeric layer monomeric layer crosslinked polymeric layer

coated polymer monofunctional silane monofunctional silane

monofunctional DibuC18 monofunctional silane propylene bidentate monofunctional silane monomeric layer

Bonding Type

(Continued)

Y Y Y   Y

N N N Y

Y Y

N N Y Y Y

Endcapping

Characterization of Stationary Phases 207

Synergi Hydro RP

HyPurity Aquastar

Hypersil Gold AQ

Hydrosphere

YMC Pack ODS AQ Zorbax Bonus RP

Prevail Amide C18 Alltima HP C18 Amide Acclaim Polar Advantage Acclaim Polar Advantage II

P8

P10

P11

P12 P13

P14 P15 P16 P17

Aquasil C18 Prevail C18 Alltima HP C18 AQ

P5 P6 P7

P9

Nucleodur Sphinx RP

Cogent C18 Bidentate Platinum C18 EPS

P2

P3 P4

Stationary Phase Name



Table 5.2  (Continued)

Grace Vydac Grace Vydac Dionex Dionex

YMC Dupont

YMC

Thermo-Electron

Thermo-Electron

Phenomenex

Thermo-Electron Grace Vydac Grace Vydac

MicroSolv Technology Grace Vydac

Macherey-Nagel

Manufacturer

350 200 300 300

330 180

340

220

200

474

310 350 450

350 100

340

Specific Surface Area (m2 g−1)

na 12 17 17

14 9.5

12

na

10

19

12 15 20

16 5

15

Carbon Content (%)

110 190 120 120

120 80

120

175

190

80

100 110 100

100 300

110

Pore Diameter (Å)

amide-embedded amide-embedded ether-sulfonamide C16 amide-embedded

polar endcapped amide-C14, sterically protected

surface-enhanced polar selectivity

polar endcapped

polar endcapped

polar endcapped

polar endcapped monomeric layer polar endcapped

bidentate CH2Si2 low bonding density

mixed ODS and propylphenyl

Bonding Type

Y Y Y

Y

Y Y

N

Y

Endcapping

208 Advances in Chromatography: Volume 48

Stability BS-C23 ne

Uptisphere PLP Nucleosil Nautilus C18 Synergi Fusion RP Discovery RP Amide Ascentis RP Amide Suplex pKb Supelcosil LC-ABZ Supelcosil ABZ + -Plus

Polaris C18 Ether Polaris C18 A

Polaris C18 B

Polaris C18 Amide Symmetry Shield RP 18 XBridge Shield

P18

P19 P20 P21 P22 P23 P24 P25 P26

P27 P28

P29

P30 P31 P32

Varian Waters Waters

Varian

Varian Varian

Interchim Macherey-Nagel Phenomenex Supelco Supelco Supelco Supelco Supelco

Cluzeau

250

180 340 185

na

na 180

320 350 475 200 450 170 170 170

na

na 17.6 17

na

na na

na 16 na 11 19.5 12.5 12 12

100

180 100 135

na

na 180

120 100 80 180 100 120 120 120

quaternary ammonium-C16

amide embedded carbamate-embedded hybrid silica, carbamate-embedded

unknown, possibly polar embedded

ether-embedded polar embedded

amide-embedded unknown, possibly polar embedded mixed classical and polar embedded amide-C16 amide-embedded amide-embedded amide-embedded amide-embedded

Y Y

Y

Y Y Y Y N Y Y

Characterization of Stationary Phases 209

210

Advances in Chromatography: Volume 48

to introduce new solutes. For instance, our initial set had been established for characterization of PGC [40], which is a highly retentive stationary phase. Then, when less retentive phases were characterized, it was necessary to inject different solutes that would be sufficiently retained to allow for precise measurement of their retention factors. Such compounds are different for different stationary phases: for instance, when a nonpolar phase needs to be characterized, nonpolar compounds are more retained, whereas polar compounds are more retained on polar phases. Thus, for most stationary phases in Table 5.1, only a subset of the solutes in Table 5.3 was used. In the end, we believe to have achieved a set of test solutes that is sufficiently wide and diverse for the characterization of phases of all available polarities. The solutes in the final set are presented in Table 5.3, along with their Abraham descriptors. The latter were extracted from an in-house database, based on a great variety of published works. Besides, there are some essential rules to follow in order to obtain meaningful results from multiple linear regression analysis. One is that the set of probe solutes must be sufficiently large to ensure the statistical significance of the calculated system constants. A rule of thumb indicates that a minimum of four solutes per variable should be used, although it is clearly better to over-determine the system by using more input retention factors. In our case, we have chosen to work with a much larger solute set. The system constants, particularly in small data sets, are strongly influenced by statistical outliers. This is another reason for increasing the initial data set so experimental errors have less weight on the final equation. Actually, in SFC, the time required to generate data is favorable for the collection of larger data sets with minimal additional effort. Size of the solute set is not the only requirement: an equilibrated set of solutes should have a wide variety of chemical functions, so much so that the introduction of additional solutes would not significantly modify the results. This means that the chosen solutes must differ in physico-chemical properties and have different threedimensional structures. While building our solute set, we have been careful to introduce a great variety of functional groups, sizes, and shapes. Figure 5.3, showing the repartition of the solutes of Table 5.3 in each descriptor space, evidences this point. It is clear from this figure that the solutes are distributed in such a manner that each descriptor covers a wide range. Clustering should be avoided as much as possible. The only exception to this rule is the A descriptor, because, due to the very definition of this parameter, a large proportion of solutes have A values equal to zero. Besides, an essential rule of QSRRs is that the variables employed in the regression be independent, that is to say the descriptors used in one equation should be as orthogonal as possible. Cross-correlation must be avoided because it results in difficulties in the interpretation of the coefficients, as the multiple linear regression analysis is unable to distinguish between correlated descriptors. Thus it is necessary that the probe solutes be chosen so as to minimize correlation between the variables. This point is demonstrated in Table 5.4, representing the correlation matrix for the solutes in Table 5.3. The largest correlation coefficient observed is 0.727, calculated between the E and S descriptors. It is well known to the chromatographers using the solvation parameter model that a certain correlation exists between the E and S descriptors, particularly when only aromatic solutes are used, which is the case here

211

Characterization of Stationary Phases

Table 5.3 Chromatographic Solutes and LSER Descriptors E, S, A, B, V, and L No.

Compound

E

S

A

B

V

L

1

Benzene

0.610

0.52

0.00

0.14

0.7164

2.786

2

Toluene

0.601

0.52

0.00

0.14

0.8573

3.325

3

Ethylbenzene

0.613

0.51

0.00

0.15

0.9982

3.778

4

Propylbenzene

0.604

0.50

0.00

0.15

1.1391

4.230

5

Butylbenzene*

0.600

0.51

0.00

0.15

1.2800

4.730

6

Pentylbenzene

0.594

0.51

0.00

0.15

1.4209

5.230

7

Hexylbenzene

0.591

0.50

0.00

0.15

1.5620

5.720

8

Heptylbenzene

0.577

0.48

0.00

0.15

1.7029

6.219

9

Octylbenzene

0.579

0.48

0.00

0.15

1.8438

6.714

10

Nonylbenzene

0.578

0.48

0.00

0.15

1.9847

7.212

11

Decylbenzene

0.579

0.47

0.00

0.15

2.1256

7.708

12

Undecylbenzene*

0.579

0.47

0.00

0.15

2.2665

8.159

13

Dodecylbenzene

0.571

0.47

0.00

0.15

2.4074

8.600

14

Tridecylbenzene

0.570

0.47

0.00

0.15

2.5483

9.132

15

Tetradecylbenzene

0.570

0.47

0.00

0.15

2.6892

9.619

16

Allylbenzene

0.717

0.60

0.00

0.22

1.0961

4.136

17

Cumene

0.602

0.49

0.00

0.16

1.1391

4.084

18

t-Butylbenzene

0.614

0.49

0.00

0.16

1.2800

4.730

19

o-Xylene*

0.663

0.56

0.00

0.16

0.9980

3.939

20

m-Xylene

0.623

0.52

0.00

0.16

0.9980

3.839

21

p-Xylene

0.613

0.52

0.00

0.16

0.9980

3.839

22

Naphthalene

1.340

0.92

0.00

0.20

1.0854

5.161

23

1-Methylnaphthalene

1.344

0.90

0.00

0.20

1.2260

5.802

24

2-Methylnaphthalene*

1.304

0.92

0.00

0.20

1.2260

5.617

25

1-Ethylnaphthalene

1.371

0.87

0.00

0.20

1.3670

6.226

26

2-Ethylnaphthalene

1.331

0.87

0.00

0.20

1.3670

6.203

27

Aniline

0.955

0.96

0.26

0.50

0.8162

3.934

28

N, N-dimethylaniline*

0.957

0.84

0.00

0.47

1.0980

4.701

29

Phenylurea*

1.110

1.40

0.77

0.77

1.0730

30

Pyridine*

0.631

0.84

0.00

0.52

0.6753

3.022

31

2-ethylpyridine

0.613

0.70

0.00

0.49

0.9570

3.844

32

Indazole

1.180

1.25

0.54

0.34

0.9050

33

Carbazole

1.787

1.42

0.47

0.26

1.3150

34

Acridine*

2.356

1.32

0.00

0.58

1.4130

35

Nicotinamide

1.010

1.09

0.63

1.00

0.9317

36 37

Caffeine o-Toluidine

1.500 0.966

1.60 0.92

0.00 0.23

1.35 0.45

1.3630 0.9570

7.638

7.352 4.442 (Continued)

212

Advances in Chromatography: Volume 48

Table 5.3  (Continued) No.

Compound

E

S

A

B

V

L

38

m-Toluidine*

0.946

0.95

0.23

0.55

0.9570

4.463

39

p-Toluidine

0.923

0.95

0.23

0.52

0.9570

4.452

40

1-Naphtylamine

1.670

1.26

0.20

0.57

1.1850

6.490

41

Benzoic acid

0.730

0.90

0.59

0.40

0.9317

4.395

42

Isophthalic acid

0.940

1.46

1.14

0.77

1.1470

6.108

43

Trimesic acid

1.140

1.84

1.71

1.10

1.3623

44

1-Naphtoic acid*

1.200

1.27

0.52

0.48

1.3007

45

1-Naphtylacetic acid

1.300

1.35

0.54

0.40

1.4416

46

Anisole

0.708

0.75

0.00

0.29

0.9160

3.890

47

Benzaldehyde

0.820

1.00

0.00

0.39

0.8730

4.008

48

Naphthylaldehyde

1.470

1.19

0.00

0.47

1.2420

49

Acetophenone*

0.818

1.01

0.00

0.48

1.0139

4.501

50

Propiophenone

0.804

0.85

0.00

0.51

1.1548

4.971

51

Valerophenone

0.795

0.95

0.00

0.50

1.4366

5.902

52

o-Methylacetophenone

0.780

1.00

0.00

0.51

1.1550

53

m-Methylacetophenone

0.806

1.00

0.00

0.51

1.1550

5.167

54

p-Methylacetophenone*

0.842

1.00

0.00

0.52

1.1550

5.081

55

Methylbenzoate

0.733

0.85

0.00

0.48

1.0726

4.704

56

Ethylbenzoate

0.689

0.85

0.00

0.46

1.2140

5.075

57

Propylbenzoate

0.675

0.80

0.00

0.46

1.3540

5.718

58

Butylbenzoate*

0.668

0.80

0.00

0.46

1.4953

6.210

59

Dimethylphthalate

0.780

1.41

0.00

0.88

1.4290

6.051

60

Diethylphthalate

0.729

1.40

0.00

0.88

1.7110

61

Dipropylphthalate*

0.713

1.40

0.00

0.86

1.9924

62

Dibutylphthalate

0.700

1.40

0.00

0.86

2.2700

63

Naphthylacetate

1.130

1.25

0.00

0.62

1.4416

64

Coumarine

1.060

1.79

0.00

0.46

1.0620

6.023

65

Benzonitrile

0.742

1.11

0.00

0.33

0.8711

4.039

66

Cyanonaphthalene*

1.190

1.25

0.00

0.41

1.2401

67

Naphtylacetonitrile

1.430

1.44

0.00

0.53

1.3810

68

Nitrobenzene

0.871

1.11

0.00

0.28

0.8906

4.557

69

Nitronaphthalene*

1.600

1.51

0.00

0.29

1.2596

6.816

70

o-Nitrotoluene

0.866

1.11

0.00

0.28

1.0320

4.878

71

m-Nitrotoluene

0.874

1.10

0.00

0.25

1.0320

5.097

72

p-Nitrotoluene*

0.870

1.11

0.00

0.28

1.0320

5.154

73

o-Nitrobenzylalcohol

1.059

1.11

0.45

0.65

1.0900

74

m-Nitrobenzylalcohol

1.064

1.35

0.44

0.64

1.0900

75

p-Nitrobenzylalcohol*

1.064

1.39

0.44

0.62

1.0900

6.794

213

Characterization of Stationary Phases

Table 5.3  (Continued) No.

Compound

E

S

A

B

V

L

76

o-Nitrophenol

1.045

1.05

0.05

0.37

0.9490

4.760

77

m-Nitrophenol

1.050

1.57

0.79

0.23

0.9490

4.692

78

p-Nitrophenol

1.070

1.72

0.82

0.26

0.9490

5.876

79

Chlorobenzene

0.718

0.65

0.00

0.07

0.8288

3.657

80

Bromobenzene

0.882

0.73

0.00

0.09

0.8910

4.041

81

Iodobenzene*

1.188

0.82

0.00

0.12

0.9750

4.502

82

1-Fluoronaphthalene

1.320

0.82

0.00

0.18

1.1030

83

1-Chloronaphthalene

1.540

0.92

0.00

0.15

1.2078

6.154

84

1-Bromonaphthalene

1.670

0.97

0.00

0.17

1.2604

6.567

85

1-Iodonaphthalène

1.840

1.04

0.00

0.20

1.3436

86

1-Phenylethanol

0.784

0.83

0.30

0.66

1.0570

4.431

87

Benzyl alcohol*

0.803

0.87

0.39

0.56

0.9160

4.221

88

Naphthalene methanol

1.640

1.19

0.27

0.64

1.2850

89

Naphthalene ethanol*

1.670

1.21

0.23

0.72

1.4259

7.046

90

Phenol

0.805

0.89

0.60

0.30

0.7751

3.766

91

Eugenol

0.946

0.99

0.22

0.51

1.3540

92

Vanillin*

1.040

1.33

0.32

0.67

1.1313

5.730

93

Pyrocatechol

0.970

1.10

0.88

0.47

0.8338

4.450

94

Resorcinol

0.980

1.00

1.10

0.58

0.8340

4.618

95

Hydroquinone*

1.000

1.00

1.16

0.60

0.8337

4.827

96

Phloroglucinol

1.355

1.12

1.40

0.82

0.8925

97

α-Naphtol

1.520

1.05

0.61

0.37

1.1441

6.140

98

β-Naphtol*

1.520

1.08

0.61

0.40

1.1440

6.124

99

o-Chlorophenol

0.853

0.88

0.32

0.31

0.8975

4.178

100

m-Chlorophenol

0.909

1.06

0.69

0.15

0.8975

4.773

101

p-Chlorophenol*

0.915

1.08

0.67

0.20

0.8975

4.775

102

o-Cresol

0.840

0.86

0.52

0.30

0.9160

4.218

103

m-Cresol

0.822

0.88

0.57

0.34

0.9160

4.310

104

p-Cresol*

0.820

0.87

0.57

0.31

0.9160

4.312

105

2,3-Dimethylphenol

0.850

0.85

0.52

0.36

1.0569

4.866

106

2,4-Dimethylphenol*

0.840

0.80

0.53

0.39

1.0570

4.770

107

2,5-Dimethylphenol

0.840

0.79

0.54

0.37

1.0570

4.774

108

2,6-Dimethylphenol

0.860

0.79

0.39

0.39

1.0570

4.680

109

3,4-Dimethylphenol

0.830

0.86

0.56

0.39

1.0570

4.980

110

3,5-Dimethylphenol

0.820

0.84

0.57

0.36

1.0570

4.856

111

o-Isopropylphenol

0.822

0.79

0.52

0.44

1.1978

112

m-Isopropylphenol

0.811

0.92

0.55

0.46

1.1978 (Continued)

214

Advances in Chromatography: Volume 48

Table 5.3  (Continued) No.

Compound

E

S

A

B

V

L

113

p-Isopropylphenol

0.791

0.89

0.55

0.38

1.1978

114

Biphenyl

1.360

0.99

0.00

0.26

1.3242

115

1-Phenylnaphthalene

1.910

1.08

0.00

0.30

1.6932

116

Diphenylméthane

1.220

1.04

0.00

0.33

1.4651

6.313

117

Benzophenone

1.447

1.50

0.00

0.50

1.4810

6.852

118

Acenaphthene

1.604

1.05

0.00

0.22

1.1726

6.469

119

Acenaphtylene

1.750

1.14

0.00

0.26

1.2156

6.415

120

Fluorene

1.588

1.03

0.00

0.20

1.3570

6.928

121

Phenanthrene*

2.055

1.29

0.00

0.26

1.4540

7.632

122

Anthracene

2.290

1.34

0.00

0.26

1.4540

7.568

123

9-Methylanthracene

2.290

1.30

0.00

0.26

1.5950

124

Fluoranthene

2.377

1.53

0.00

0.20

1.5850

8.702

125

Pyrene

2.808

1.71

0.00

0.29

1.5850

8.833

126

Chrysene*

3.027

1.73

0.00

0.36

1.8230

127

Benz[a]anthracene

2.992

1.70

0.00

0.33

1.8230

128

Tetracene

2.847

1.70

0.00

0.32

1.8230

129

Benzo[a]pyrene*

3.625

1.98

0.00

0.44

1.9536

130

Perylene

3.256

1.76

0.00

0.42

1.9536

131

Binaphthyl

2.820

1.81

0.00

0.31

2.0622

132

Triphenylene

3.000

1.71

0.00

0.42

1.8234

133

o-Terphenyl

2.194

1.61

0.00

0.38

1.9320

134

p-Terphenyl

2.194

1.61

0.00

0.38

1.9320

6.014

11.736

Note: E: Excess molar refraction; S: Dipolarity/polarizability; A: Hydrogen bond acidity; B: Hydrogen bond basicity; V: McGowan’s characteristic volume; L: logarithm of the gas-hexadecane partition coefficient. Compounds marked with an asterisk are the test solutes used to assess the predictive capability of the models.

to allow for UV detection [26]. For this reason, a number of positional aromatic isomers were introduced as they allow limiting the correlation between E and S. However, it should be pointed out that covariance estimated through the correlation coefficient is somewhat overestimated because this coefficient can be strongly influenced by a few points acting as levers, while the rest of the points would be scattered. Indeed, the correlation between the E and S coefficients among the group of solutes we have selected can be observed on Figure 5.4. It is clear from this figure that the largest PAH solutes (solutes 121 to 134 in Table 5.3) largely contribute to the correlation observed. When they are removed, the correlation coefficient between E and S falls down to 0.561. Thus the correlation observed is not really as important as Table 5.4 would suggest. A better correlation coefficient, expressing a total covariance that would be more representative of the whole distribution and

0

5

10

15

20

25

30

35

40

45

50

0

0.2 0.4 0.6 0.8 B value

0.4 0.7 1 1.3 1.6 1.9 2.2 2.5 2.8 3.1 E value

Number of solutes

Number of solutes

1

0

5

10

15

20

25

30

1.2

0.4 0.6 0.8

0

5

10

15

20

25

30

35

Number of solutes 0

10

20

30

40

50

60

70

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 A value

0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5 V value

1 1.2 1.4 1.6 1.8 S value

Figure 5.3  Distribution of descriptor values among the test set in Table 5.3.

0

5

10

15

20

25

30

35

Number of solutes

40

Number of solutes

45

Characterization of Stationary Phases 215

216

Advances in Chromatography: Volume 48 2.5

S

2.0

1.5

1.0

0.5

0.0 0.0

E 0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Figure 5.4  Plot of the S descriptor vs. the E descriptor for the solutes in Table 5.3. Open diamonds are solutes 1 to 120; black diamonds are solutes 121 to 134.

Table 5.4 Covariance Matrix for the Solute Set in Table 5.3 R E

E  

S

0.727

A

−0.131

S

A

0.727

−0.131

0.027

0.352

0.611

0.168

0.473

0.169

0.367

0.374

−0.363

−0.178

  0.168

 

B

V

L

B

0.027

0.473

0.374

 

−0.048

0.007

V L

0.352 0.611

0.169 0.367

−0.363 −0.178

−0.048 0.007

  0.842

0.842  

less influenced by levers would be useful but we have not found anything satisfying so far. For the ODS phases in Table 5.2, only the 29 solutes in Table 5.5 were analyzed. In this case, we chose to work with a smaller set of compounds, knowing that the precision of the results is lesser than when larger sets of solutes are used. However, the compounds were chosen so as to retain diversity and absence of cross-­correlation as described above. Besides, comparison of the models obtained with the larger (Table 5.3) and smaller (Table 5.5) sets of solutes on several ODS phases showed no significant difference, at the 95% confidence level. Therefore, we consider this small set as perfectly valid and representative of the possible interactions occurring between the solutes and the chromatographic systems.

217

Characterization of Stationary Phases

Table 5.5 Reduced Test Set for ODS Phases E, S, A, B and V No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Compound Benzene Toluene Ethylbenzene Propylbenzene Butylbenzene Pentylbenzene Allylbenzene Anisole Methyl benzoate Benzaldehyde Acetophenone Benzonitrile Nitrobenzene Chlorobenzene Bromobenzene Naphtalene Biphenyl 1-Phenylethanol Benzyl alcohol o-Cresol m-Cresol p-Cresol Phenol Resorcinol Phloroglucinol Benzoic acid Isophthalic acid Aniline N,N-Dimethylaniline

E

S

A

B

V

0.610 0.601 0.613 0.604 0.600 0.594 0.717 0.708 0.733 0.820 0.818 0.742 0.871 0.718 0.882 1.340 1.360 0.784 0.803 0.840 0.822 0.820 0.805 0.980 1.355 0.730 0.940 0.955 0.957

0.52 0.52 0.51 0.50 0.51 0.51 0.60 0.75 0.85 1.00 1.01 1.11 1.11 0.65 0.73 0.92 0.99 0.83 0.87 0.86 0.88 0.87 0.89 1.00 1.12 0.90 1.46 0.94 0.84

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.30 0.39 0.52 0.57 0.57 0.60 1.10 1.40 0.59 1.14 0.26 0.00

0.14 0.14 0.15 0.15 0.15 0.15 0.22 0.29 0.48 0.39 0.48 0.33 0.28 0.07 0.09 0.20 0.26 0.66 0.56 0.46 0.34 0.31 0.30 0.58 0.82 0.40 0.77 0.50 0.47

0.7164 0.8573 0.9982 1.1391 1.2800 1.4209 1.0961 0.9160 1.0726 0.8730 1.0139 0.8711 0.8906 0.8288 0.8910 1.0854 1.3242 1.0570 0.9160 0.9160 0.9160 0.9160 0.7751 0.8340 0.8925 0.9317 1.1470 0.8162 1.0980

Note: E: Excess molar refraction; S: Dipolarity/polarizability; A: Hydrogen bond acidity; B: Hydrogen bond basicity; V: McGowan’s characteristic volume.

To possibly reduce the testing procedure, we have devised an extremely reduced set of nine test solutes [47]. This set allows estimating the solvation parameter coefficient, when rapid information is required. However, the precision of the results is naturally less satisfying than with larger sets. The solutes were obtained from a range of suppliers. Solutions were prepared in methanol, or methanol-tetrahydrofurane for the least soluble polynuclear aromatic hydrocarbons and long-chain alkylbenzenes. Solutions were prepared in such

218

Advances in Chromatography: Volume 48

a concentration that no variation of retention factor is observed on the PGC phase when further diluted solutions are injected. As PGC has the smallest specific surface area of all phases investigated, no saturation occurred on the other stationary phases. The sunscreen molecules were kindly provided by L’Oreal (Chevilly Larue, France).

5.2.5 Data Analysis Retention factors (k) were calculated based on the retention time tr, determined using the peak maximum (even when tailing did occur, on some columns, for some of the most acidic and basic solutes) and on the hold-up time t0 measured on the first negative peak due to the unretained dilution solvent. The system constants for each chromatographic system, presented in Table 5.6, were obtained by multiple linear regression analysis on the logarithm of the measured retention factors (log k). Compared to our previously published works, some additional compounds were injected to improve the predictive capability (as explained above) thus the results may be slightly different. However, none of our previous conclusions regarding chromatographic behaviors is called into question by the new models established. Multiple linear regression analysis and statistical tests were performed using the XL Stat software (Addinsoft, New York, NY, USA). The quality of the fits was estimated using the overall correlation coefficient (R), adjusted determination coefficient (R 2adj), standard error in the estimate (SE) and Fischer F statistic. The statistical significance of individual coefficients was evaluated using the t-ratio, which is defined as the ratio of the regression coefficient to its standard error. In each case, a few outliers were eliminated from the set, as their residuals were too high. These solutes were different for each stationary phase and most of the time showed no common property, apart for some basic solutes, as will be detailed later. On another hand, solutes that were not retained enough were also eliminated, and these are different depending on the nature of the stationary phase. For instance, the large alkylbenzenes are not retained enough on the polar phases while the very polar solutes such as phloroglucinol are not retained enough on nonpolar phases. Independent variables E, S, A, B, and V, which were not statistically significant, with a confidence interval of 99.9%, were eliminated from the model. The fits were all of reasonable quality, R ranging from 0.930 to 0.995, SE varying from 0.029 to 0.229. These results are reasonably good and confirm that the solvation parameter model adequately describes retention even when applied to a wide variety of columns. The somewhat poor fit obtained on PGC will be explained below. Deviations from the experimental values for the predicted values are generally within the uncertainty indicated by the model fits but, for some solutes, there is a systematic trend in the data. Indeed the residuals of the fit are consistent from column to column: most often, those compounds that are actually less retained than predicted by the model, or those that are more retained than predicted by the model, or those that are well predicted, are the same on all stationary phases. This fact can

0.220

−1.193

0.013 −0.998

0.019 −0.771

0.020 −0.779

0.019 −0.829

0.028 −0.809

0.039 −0.950

0.013 −1.190

0.026 −0.822

0.022 −1.171

C4

  C8

  C12

  C18

  C18-C

  RPH

  MIX

  PE1

  PE2

  PE3

0.015 0.577

0.017 0.527

0.008 0.661

0.022 0.309

0.016 0.470

0.014 0.658

0.012 0.587

0.012 0.379

0.008 0.322

e

c

Stationary Phase

0.028 −0.214

0.032 −0.091

0.016 −0.317

0.032 −0.106

0.028 −0.261

0.025 −0.447

0.023 −0.459

0.025 −0.242

0.017 −0.268

−0.170

s

0.017 0.795

0.022 0.499

0.010 1.263

0.024 −0.156

0.022 −0.407

0.020 0.199

0.018 −0.457

0.018 −0.372

−0.278



a

0.030 −0.462

0.033 −0.428

0.017 −0.427

0.040 −0.254

−0.411

0.028 −

0.029 −0.479

0.027 −0.444

0.020 −0.219

−0.111

b

0.014 0.308

0.016 0.305

0.008 0.345

0.039 0.328

0.018 0.373

0.012 0.276

0.012 0.439

0.012 0.313

0.008 0.272

0.230

v

108

116

113

118

76

115

110

115

121

114

n

Table 5.6 System Constants and Statistics for the 35 Columns in Table 5.1

0.995

0.989

0.991

0.994

0.992

0.982

0.995

0.990

0.986

0.985

R

0.991

0.987

0.981

0.988

0.983

0.963

0.989

0.980

0.971

0.968

R2adj

0.037

0.053

0.058

0.029

0.054

0.066

0.043

0.044

0.045

0.030

SE

2244

1017

1151

1860

863

748

1968

1120

801

860

F

(Continued)

1.15

0.90

1.56

0.55

0.87

0.87

1.09

0.80

0.61

0.38

u

Characterization of Stationary Phases 219

0.053 −1.354

0.064 −1.113

0.029 −1.080

0.038 −1.118

0.068 −0.715

0.072 −1.033

  SI

  PEG

  PVA

  AMD

  DIOL

  NH2

0.078

0.030 −1.069

  FD

 

0.017 −1.135

c

  CHL

Stationary Phase

0.044

0.033 0.475

0.040 0.460

0.015 0.499

0.026 0.535

0.036 0.341

0.024 0.352

0.020 0.282

0.010 0.595

e

Table 5.6  (Continued)

0.083

0.301

0.074 −

0.290

0.054 −

0.081 0.256

0.336

0.031 −

0.021 −0.172

s

0.063

0.057 1.349

0.056 0.928

0.038 1.368

0.036 1.189

0.055 1.236

0.039 1.196

0.023 0.457

0.013 0.328

a

0.106

0.081 0.901

0.094 0.945

0.044 1.184

0.056 0.878

0.075 −0.223

0.056 1.011

0.035 0.961

0.021 −0.625

b

0.074

0.077 −0.678

0.059 −0.561

0.030 −0.618

−0.477

0.060 −

0.060 −0.571

0.018 −0.501

0.010 0.469

v

109

111

125

110

84

100

98

116

n

0.962

0.933

0.974

0.985

0.981

0.967

0.930

0.982

R

0.922

0.865

0.946

0.969

0.960

0.931

0.859

0.962

R2adj

0.160

0.156

0.154

0.088

0.083

0.117

0.100

0.064

SE

255

177

435

839

496

269

149

590

F

1.85

1.51

2.00

1.64

1.33

1.74

1.21

1.05

u

220 Advances in Chromatography: Volume 48

0.086 −2.175

0.140 −0.669

0.052 −1.092

0.028 −1.489

0.028 −1.067

0.019 −1.217

0.020 −1.068

0.013 −1.120

0.016 −1.142

  PGC

  PS

  OPHE

  DP

  DP-X

  C3P

  C6P-L

  C6P-G

  C6P-Z

0.013

0.052 −1.057

  EP

 

−0.993

CN

0.006

0.006 0.338

0.006 0.319

0.008 0.326

0.007 0.280

0.010 0.311

0.016 0.297

0.024 0.346

0.106 0.762

0.050 1.552

0.028 0.588

0.339











0.032 −

0.099

0.121 −

0.094 0.391

0.053 0.564

0.372

0.013 −

0.084

0.017 −

0.014 0.218

0.022 0.218

0.022 0.311

0.033 0.185

0.091 0.132

0.069 0.593

0.039 1.053

0.696

0.015

0.017 −0.163

0.014 −0.249

0.021 −0.125

0.018 0.121

0.030 0.076

0.038 0.213

0.047 0.101

0.151 −0.279

0.120 −0.307

0.064 0.790

0.449

0.009

0.011 0.306

0.009 0.312

0.014 0.285

0.015 0.279

0.024 0.261

0.017 0.209

0.055 0.192

0.090 0.268

0.071 1.145

0.040 −0.692

−0.346

125

125

119

135

113

119

132

68

91

120

118

0.989

0.988

0.990

0.975

0.987

0.967

0.972

0.987

0.936

0.964

0.965

0.979

0.975

0.980

0.949

0.974

0.933

0.943

0.973

0.868

0.926

0.929

0.038

0.040

0.036

0.053

0.040

0.064

0.065

0.077

0.229

0.191

0.110

1884

1232

1926

629

1042

411

435

595

119

297

305

(Continued)

0.48

0.52

0.45

0.47

0.47

0.52

0.46

0.86

2.08

1.70

1.03

Characterization of Stationary Phases 221

0.333

−1.243

0.028 −1.362

0.083 −1.577

0.060 −1.357

0.058 −1.101

0.021 −1.245

0.050

NAP

  PYE

  DNAP

  PNP

  PFP

  PBB

 

0.038 −

0.060 0.482

0.060 0.267

0.069 0.510

0.027 0.398

0.132

s

0.037

0.023 0.214

0.045 −0.054

0.043 1.010

0.048 0.750

0.021 0.145

0.126

a



0.073 −

0.071 0.622

0.077 0.501

0.277



b

0.045

0.045 0.479

0.045 −0.313

0.033 −0.226

0.083 −0.086

0.017 0.364

0.307

v

82

115

114

107

73

124

n

0.985

0.956

0.973

0.983

0.978

0.977

R

n is the number of solutes considered in the regression R is the multiple correlation coefficient R2adj is the adjusted correlation coefficient SE in the standard error in the estimate, F is Fischer’s statistic u is the solvation vector length according to Equation 5.7 and the numbers in second lines represent 99.9% confidence limits Dashes indicate coefficients that were not included in the model due to statistical insignificance

0.020

0.020 0.728

0.031 0.169

0.031 0.470

0.047 0.737

0.015 0.333

e

c

Stationary Phase

Table 5.6  (Continued)

0.970

0.912

0.943

0.965

0.953

0.953

R2adj SE

0.102

0.076

0.121

0.117

0.086

0.066

F

872

296

377

588

291

621

u

0.90

0.60

1.32

1.27

0.71

0.49

222 Advances in Chromatography: Volume 48

223

Characterization of Stationary Phases

Normalized residual

2.0 1.5 1.0 0.5 0.0 –0.5 –1.0 –1.5 –2.0

Anthracene

Fluoranthene

Phenanthrene

p-nitrobenzylalcohol

m-nitrobenzylalcohol

Anisole

o-nitrobenzylalcohol

p-xylene o-toluidine

o-xylene

m-xylene

Pentylbenzene

Propylbenzene Butylbenzene

Toluene

Ethylbenzene

PFP OPHE C4 PE2 AMD

Figure 5.5  Plot of the normalized residuals calculated for PFP, OPHE, C4, PE2, and AMD for 16 representative solutes.

be observed on Figure 5.5, where the normalized residuals for 16 compounds have been plotted for 5 columns displaying very different chemistries and polarities. More columns and compounds could have been used, which would have shown exactly the same tendencies but we have chosen to simplify the figures for the purpose of clarity. Thus the observed deviations are obviously not related to experimental errors but might originate from limited adequacy and/or precision of the used descriptors or from some molecular interactions that are not accounted for by the solvation parameter models. As far as limited adequacy of the descriptor is concerned, we have to point out that descriptors available in the literature are not always very accurate, which seems normal judging by the variety of partition and chromatographic processes used to calculate them. Concerning the molecular interactions that are not accounted for by the solvation parameter model, electrostatic interactions may sometimes occur. Indeed, in most cases, the compounds needing to be excluded were of varied nature and no systematic trend was observed. However, for some columns (as PFP, C18-C or C4), some basic solutes (among solutes 27 to 40) had to be removed, as they were extreme outliers and were largely more retained than what would be expected, based on the model calculations. These are the only N-containing bases in the solute set. Since oxygen-containing compounds of similar capacity for H-bond and dipole-type interactions (as 1-phenylethanol and benzyl alcohol, for instance) are not influenced to the same extent, we presume that this additional retention results from a contribution to retention that is not considered by the model, such as electrostatic interactions with residual silanol groups (in the non-endcapped phases) or other possibly ionized groups. Indeed, the solvation parameter model—in the form employed here—uses descriptors characteristic of the neutral form of the molecule. It is not expected to provide accurate predictions of chromatographic properties of solutes in a fully or partially ionized form. Different authors have suggested additional terms for ionizable solutes [48–53] but these descriptors require knowledge of the pH and pKa of

224

Advances in Chromatography: Volume 48

the species, while the pH of the carbon dioxide-methanol mobile phase is unknown. However, some studies carried out with pH indicators tend suggest that it could be acidic [54–56], possibly below 5. Even if the pKa of the silanol groups might be larger in the subcritical mobile phase than in aqueous environment, it is reasonable to think that it should remain lower than 5. Therefore, silanol groups are expected to be partly dissociated in the CO2-MeOH subcritical mobile phase. On the other hand, a number of N-containing basic solutes would be present in their cationic form, thus would establish electrostatic interactions with the anionic silanol groups. As no more precise information is available, we have to admit that electrostatic interactions occur but that we can so far not evaluate them. In the same manner, some acidic solutes (as benzoic acids) may be present in their anionic form, which would explain why they may be badly predicted on some stationary phases, where attractive or repulsive electrostatic interactions would occur. Besides, the adsorption phenomenon on the surface could also partly explain the worse correlations obtained for SI and PGC. Selective adsorption and steric requirements for sorption at sites of different sorption energy may not be perfectly modeled. In those cases, when it is reasonable to think that adsorption is the principal retention mechanism, the molecular volume may not perfectly model the contact surface area for non-planar molecules [57]. For these solutes, the trends in the residual graph is generally different from the one observed on other stationary phases. This can be seen on Figure 5.6, where the normalized residuals for 17 solutes are plotted for PGC, SI, and C18. For those solutes which show generally small normalized residual on C18, thus cannot be suspected for having wrong descriptors, the deviation on PGC and SI is larger. Thus, when working with adsorbents, replacing the molecular volume by a contact surface area may improve the fit and reduce the residuals for angular molecules. We already had the opportunity to discuss this point for PGC [40]. However, as we wish to compare the results between all stationary phases, we chose to use the molecular volume in all cases.

Normalized residual

1.5 1.0 0.5 0.0 –0.5 –1.0 –1.5 –2.0 –2.5 –3.0 3,5-Dimethylphenol

2,4-Dimethylphenol 2,5-Dimethylphenol 2,6-Dimethylphenol 3,4-Dimethylphenol

Bromobenzene Iodobenzene 2,3-Dimethylphenol

Chlorobenzene

Cumene Diethylphthalate Dipropylphthalate Dibutylphthalate

Toluene Ethylbenzene Propylbenzene

Benzene

SI PGC C18

Figure 5.6  Plot of the normalized residuals calculated for SI, PGC, and C18 for 17 ­representative solutes.

225

Characterization of Stationary Phases

Nevertheless, the results are reasonably good and, it is worth noting that the sign and magnitude of each regression coefficient obtained are always in accordance with the chemical nature of the stationary–mobile phases system.

5.3 Choice of solvation descriptors for supercritical fluids The solvation parameter model can also be used with a second equation, represented below:

log k = c + eE + sS + aA + bB + lL

(5.3)

Equation 5.1 is applied to processes involving condensed phases while Equation 5.3 is applied to processes in gaseous phases. As a consequence, the solvation parameter model can be used either with Equation 5.1 or with 5.3, depending on the density of the mobile phase used. The case of supercritical fluids is critical regarding the choice of the most appropriate equation because the density of these fluids varies with a number of operating parameters. The choice of one or the other of the equations is thus not trivial. Some supercritical fluid studies were carried out with the L descriptor [33–39,58], others with the V descriptor [59–61]. Some authors [60] suggest that this choice is of no importance because the two descriptors would be supposedly correlated. However, this is not the case, as evidenced by Figure 5.7, where the L descriptor is plotted against the V descriptor for 600 solutes. The covariance essentially exists when a single family of solutes is considered, as homologous series, for instance, but not when a variety of solutes are considered, as must be the case for the calculation of QSRRs. Consequently, V and L are not interchangeable. Since the properties of supercritical fluids are intermediate between gases and liquids, both V and L could be appropriate descriptors. To choose between the two, it is important to understand the nature of each of these descriptors. 12

L

R2 = 0.7485

10 8 6 4 2 0 –2

V 0.0

0.5

1.0

1.5

2.0

2.5

3.0

Figure 5.7  Plot of the gas-hexadecane partition coefficient (L) vs. McGowan’s volume (V) for 600 solutes.

226

Advances in Chromatography: Volume 48

V is McGowan’s molecular volume. Expressed in cm3 mol–1/100, it represents the volume of one mole of the solute, when the molecules are immobile. It can be calculated very simply, using the equation below [62]:

V=

ΣVatoms − ΣVbonds 100

(5.4)

where ΣVatoms represents the sum of the volumes of all atoms in the molecule and ΣVbonds the sum of the volumes of all bonds. As the volume of one bond is considered to be always the same, whatever the atoms it links and whatever the multiplicity of the bond (single, double, or triple), we only need to know the number of bonds (B). The latter can be determined with the following equation:

B = N – 1 + R

(5.5)

where N is the number of atoms and R the number of rings. L represents the logarithm of partition coefficient between gas and hexadecane at 25°C. It is thus reasonable to think that L is related to the boiling point of the solute. This is evidenced by Figure 5.8a, where L values for 300 solutes have been plotted versus their boiling points. In this case, the correlation coefficient is large (0.921). When V values for the same 300 solutes are plotted against the same boiling point values the correlation is significantly lower (Figure 5.8b). However, in the subcritical conditions we have chosen, retention is not related to volatility, contrary to what occurs in the supercritical state at elevated temperatures [63]. This is a first argument for preferring the V descriptor. Besides, one problem encountered when working with the L descriptor is that L has a different dimension than the other Abraham descriptors. The other descriptors roughly range from 0 to 5, while L ranges from 0 to 30 [27]. As a consequence, the interpretation of the coefficients issued from the multiple linear regression analysis is not as immediate because the largest coefficient is not necessarily associated to the strongest interactions. Replacing V by L generally induces an increase in the associated system constant, with a concomitant decrease in all other system constants. This can be partly explained by the lack of independence of L with the other Abraham descriptors. Indeed, we have shown that L is strongly correlated to the boiling point. However, volatility of a solute is not only associated to its molecular weight but also to the interactions existing between the solutes, depending on their physico-chemical properties. As a result, it can be seen in the covariance matrix in Table 5.4 that the correlation coefficients relating L to E and S are larger than those relating V to E and S. Consequently, the information contained in L is partly redundant with the information contained in E and S. In other words, L would be less pure than V. Thus when V is replaced by L, the associated system constant is increased, not because the dispersion interactions really are stronger than estimated with V, but because the l coefficient is associated to a blend of interactions: dispersive, charge-transfer, dipole–dipole. The use of L instead of V might induce erroneous and/or confuse interpretation of the chromatographic system.

227

Characterization of Stationary Phases (a) 10

L

8

R2 = 0.9209

6 4 2 0 –200 (b) 3.0 V

bp(°C) –100

0

100

200

300

2.5

400

500

R2 = 0.5745

2.0 1.5 1.0 0.5 0.0 –200

bp(°C) –100

0

100

200

300

400

500

Figure 5.8  Plots of (a) the gas-hexadecane partition coefficient (L); (b) McGowan’s volume (V) vs. boiling point (bp) for 300 solutes.

In short, there are, in our case, different important reasons to choose V rather than L: • L is correlated to volatility, a factor that is not relevant to explain retention in subcritical conditions. • L is not on the same scale as the other Abraham descriptors, thus the amplitudes of the system constants are not directly comparable to estimate the relative importance of each type of interaction toward the retention behavior. • L is itself a composite descriptor, associating information related to molecular volume, polar, and polarizable characters, further complicating the interpretation of the results. For all these reasons, we chose to work with V instead of L. Besides, Lagalante and Bruno [59] have found it important to introduce an additional descriptor to achieve an exact description of processes occurring in supercritical

228

Advances in Chromatography: Volume 48

fluids. The authors base their reasoning on the fact that, if solvents remain at a constant temperature and pressure when used in the liquid state, it is not the case when they are in the supercritical state because the eluting strength is modulated through changes in pressure and temperature. When the density of the fluid changes, variations in the polarity parameter π* of the mobile phase are observed: the polar or polarizable character of the fluid varies with its density. The other descriptors supposedly remain constant. The authors considered that this should be taken into account in the solvation parameter model when applied to supercritical fluids. Indeed, the polar character of supercritical fluids varies linearly with density, and faces a sudden change of slope when the density is changed from gas-like to liquid-like. But the fact that other characters would remain constant with density changes is questionable. This was verified for pure carbon dioxide fluids but not for carbon dioxide-modifier mixtures, which are commonly used in pSFC today. If density varies, be it through a change in temperature, pressure, flow rate, or mobile phase composition, and if the modifier presents an acidic or basic character (as is the case of most modifiers commonly used), it seems logical that the acidic and basic character of the mobile phase should change with density, and not only polarity. In this case, other additional descriptors would be required to account for these phenomena. In our case, the fluid density variations are strongly limited by the subcritical conditions and by the fact that the chromatographic mobile phase and operating conditions are kept constant to compare the stationary phases. Thus we chose to use the classical Abraham equation, without any additional descriptor.

5.4 A general database for packed column SFC The original database described in our previous works [64–65] has since been expanded and upgraded. In the current database, 35 columns of varying chemistries and 49 ODS phases are included.

5.4.1 Variation of the system constants among the stationary phases Since the descriptors represent the solute effect on various solute-phase interactions, the coefficients obtained from the multiple linear regression analysis correspond to  the complementary effect of the stationary and mobile phases on these interactions. The regression coefficients are thus very important, because they will encode chromatographic system properties. As the chromatographic conditions and mobile phase are kept constant, it is reasonable to think that the regression coefficients encode stationary phase properties. The coefficients can then be regarded as constants characterizing the stationary phase. The intercept, c, is not assigned any chemical significance. It represents a part of the retention factors that is not accounted for by the solvation parameters. Therefore, the c coefficients are not easily compared or interpreted, and they will always be omitted in this section. The various stationary phases may be compared relative to their regression coefficients to establish a relative order of selectivities toward specific types of molecular interactions. Indeed, as discussed in the introduction, the system constants are

229

Characterization of Stationary Phases

also related to the system’s selectivity through Equation 5.2. Therefore, to enhance the separation of compounds differing in their X property, one should choose the chromatographic system where the x coefficient is the largest. For instance, to separate compounds differing primarily in their molecular volume, as is the case in homologous series, it is advisable to choose a chromatographic system with a large v coefficient. Indeed, along a series of homologues, the E, S, A, and B descriptors will be almost the same, and the only descriptor to vary will be V. Then the v coefficient will be the only system constant of importance. Thus, a first possibility is to observe the system constants one by one through histogram plots (Figure 5.9). We have shown in previous works how small differences in the bonding chemistry of the stationary phase can result in large differences in individual system constants [66–68]. At the same time, it is interesting to evaluate the part taken by each type of interaction in the dispersion of the studied chromatographic systems. Indeed, each coefficient represents a certain proportion of the initial information which can be estimated through the calculation of the percentage of variance explained. The five types of interactions (e, s, a, b, and v) can be associated to five axes, in the same manner as five principal components in a PCA. On each axis, the percentage of variance explained can be calculated very simply with a formula similar to the one used in PCA. First of all, the barycenter G of all points must be determined (using equal weights for all chromatographic systems), then the percentage of variance is expressed by:



∑ (M ′ − G) %var = ∑ (M − G) i i

i i

2



(5.6)

2

where Mi-G is the distance between a chromatographic system Mi and G and Mi′-G is the distance between the projection of the Mi system on the axis considered and G. The e coefficient shows the tendency of the phase to interact with solutes through π and n electron pairs. The e axis carries only about 6% of the variance observed among the 35 columns. One reason for the small variance is the fact that all columns display positive e coefficients (Figure 5.9a). This indicates that the respective interactions are always stronger in the stationary than in the mobile phase. No general tendency can be found to explain the variations of the e coefficient. This, and the fact that the e coefficient is always positive, is possibly related to its composite nature. Indeed the E descriptor represents the capability of the solute to interact through π and n electrons thus can be associated to dispersive, dipoleinduced dipole, and π–π interactions. Another reason for the poor variance could be the fact that the solute set is only constituted of aromatic molecules, thus providing little opportunity to observe a variance in the e interactions: each type of phase (polar, nonpolar, or aromatic) can present high or low e coefficient. PGC is the only stationary phase to exhibit significantly different e-type interactions from the other stationary phases with an e value being twice as large as the largest e value calculated for all other phases. The extended planar aromatic surface

230

Advances in Chromatography: Volume 48

of PGC, covered with π electrons, is responsible for this, as we discussed in previous works [68]. The s coefficient gives the tendency of the phase to interact with dipolar and/ or polarizable solutes. The s coefficient is the second least important coefficient in terms of variance between the columns. Indeed, the s axis carries about 14% of the variance. As can be seen on the histogram plot (Figure 5.9b), the s coefficient varies from small negative values (for alkyl-type phases) to moderate positive values (for polar and aromatic phases). Negative values indicate that dipole–dipole interactions (a) e 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.0

(b) s

PFP C4 C3P FD DP MIX DP-X C6P-G C8 C6P-L PYE NAP C6P-Z CN PEG OPHE SI C12 DIOL RPH PNP NH2 AMD PE2 PVA PE3 C18 EP CHL C18-C PE1 PBB DNAP PS PGC

0.2

Column

0.8 0.6 0.4 0.2 0.0

–0.2

–0.6

(c) a

C18 C18-C PE1 C8 RPH C12 PE3 CHL C4 MIX PE2 FD PVA DIOL PS DP DP-X C3P C6P-L C6P-G C6P-Z PBB OPHE NAP PEG PNP AMD NH2 SI CN PGC PYE PFP DNAP EP

–0.4

Column

1.5 1.0 0.5 0.0

–1.0

C18 RPH C12 C8 MIX PFP C4 C6P-L C6P-Z C6P-G NAP PS PYE OPHE C18-C PBB DP-X C3P DP CHL FD PE2 PGC CN DNAP PE3 DIOL PNP EP PVA SI PEG PE1 NH2 AMD

–0.5

Column

Figure 5.9  Histogram plots of the system constants in Table 5.6, in order of increasing value, of a) the e coefficient; b) the s coefficient; c) the a coefficient; d) the b coefficient; e) the v coefficient.

231

Characterization of Stationary Phases (d) b 1.5 1.0 0.5 0.0

–1.0

CHL C18 PE3 C12 PE2 PE1 RPH PGC PS MIX C6P-G PEG C8 C6P-Z C6P-L C4 C18-C PFP PBB NAP DP-X OPHE C3P DP PYE CN DNAP PNP EP PVA NH2 DIOL FD SI AMD

–0.5

(e) v

Column

1.5 1.0 0.5 0.0

–1.0

EP NH2 AMD SI DIOL FD PVA CN PFP PNP DNAP PEG OPHE DP C4 DP-X PS C8 C18-C C3P C6P-L PE2 C6P-Z NAP PE3 C6P-G C12 MIX PE1 PYE RPH C18 CHL PBB PGC

–0.5

Column

Figure 5.9  (Continued)

are more favorable in the mobile phase when the stationary phase is nonpolar, which is in accordance with chemical sense. The stationary phases are equally parted between negative, positive, and zero values of the s coefficient. A negative coefficient does not mean that no s-type interactions (dipole–dipole and dipole-induced dipole interactions) occur, but that they are equally strong between the solute and stationary phase and between the solute and mobile phase. The limited variations of the s coefficient can possibly be explained by the fact that the E and S descriptors are partly correlated, thus some information contained in the s coefficient is already described in the e coefficient. The a, b, and v axes carry 26, 29, and 25% of the variance, respectively, thus they are nearly equivalent. The a coefficient denotes the hydrogen-bond basicity of the phase, because acidic solutes will interact with a basic phase. Only a few columns display a negative a coefficient (Figure 5.9c). Most of them are alkyl-bonded silica phases possessing no polar group to provide a hydrogen-bonding acceptor site. The largest a values are generally found among polar phases but it is worth noting that polar-embedded ODS phases as PE1 (amide-embedded) and PE3 (carbamate) can also display a strong basic character, whereas PE2 (ether-sulfonamide) displays a smaller a value. The b coefficient is a measure of the hydrogen-bond acidity of the phase, because basic solutes will interact with an acidic phase. The situation for the b coefficient is a little different (Figure 5.9d) as about a half of the phases investigated possess

232

Advances in Chromatography: Volume 48

a negative b coefficient (mostly alkyl-bonded phases and some aromatic phases as phenyl-hexyl bonded phases), with the other half possessing a positive b coefficient (mostly polar phases). Variance among the positive values is twice as large as among the negative values. Comparing the three polar embedded ODS columns, one can remark they display close values, whatever the polar embedded group nature. In our column set, there are three stationary phases grafted with the same bonded group (phenyl-hexyl) but three different silica gels. In particular, C6P-G is bonded on hybrid organic/inorganic silica in which silane bridges are replaced by ethane bridges. Consequently, there should be less residual silanol groups on the silica surface. Indeed C6P-G displays lower acidity than the two other phenyl-hexyl phases (C6P-L and C6P-Z), probably due to this special silica. The v coefficient is a combination of exoergic dispersion forces that make a positive contribution and an endoergic cavity term that make a negative contribution. Cavity effects in the mobile phase should be negligible in our case because the cohesiveness of the supercritical mixture used here is low. Therefore, dispersive interactions and cavity effects mainly occur with the stationary phase. Only a third of the columns tested display a negative v coefficient and all of them are polar phases (Figure 5.9e). Indeed, it generally requires more energy to create a cavity in a polar stationary phase than in a nonpolar stationary phase as the latter is less cohesive. The other stationary phases, moderately polar (aromatic) and nonpolar (alkyl) phases have a positive v coefficient. In this case, the dispersion interactions dominate. Thus the v coefficient seems to be a useful measure of the hydrophobicity of a stationary phase. This is also clearly visible on Figure 5.10, which represents the separation of some alkylbenzenes on five stationary phases with increasing v values. On the PEG phase, where the v coefficient is zero, the five homologues coelute. On the four other phases, which are all alkylsiloxane-bonded silica phases with increasing alkyl chain length, the v values are increasing with the chain length and it is clear that the separation of the homologues is improving when the v value is increasing. Again, by far the largest value is the one calculated for PGC, with its highly dispersive surface. While all these observations match intuition, the solvation parameter model provides a quantitative estimate of the relative importance of these effects.

5.4.2 A visual representation of the database As described above, using the solvation parameter model, the different chromatographic systems can be compared based on raw values of the regression coefficients. However, to compare the 35 columns based on numerical values or using five histogram plots is a complicated task. Thus we have devised a way to plot the results so as to represent the five-dimensional repartition of the chromatographic systems in the five-dimensional space of selectivities defined by the solvation descriptors [69]. We have also looked for numerical tools, which would help measuring similarities between stationary phases in an objective manner. The spider diagram (Figure 5.11) is very simply plotted: each chromatographic system is placed at the extremity of the normalized solvation vector positioned from the origin defined as the center of the five-branched-star. This means that the

Minutes

2

1

Minutes

C4

2

1 Minutes

C8

2 1

2 Minutes

C12

3

1

C18

2 Minutes

3

Figure 5.10  Chromatograms of alkylbenzene compounds, with carbon number in the alkyl chain ranging from 11 to 15 on five selected columns with increasing v value. Mobile phase: CO2-MeOH 90:10 (v/v). Temperature: 25°C. Outlet pressure: 15 MPa. Flow rate: 3 ml min−1

1

PEG

Characterization of Stationary Phases 233

234

Advances in Chromatography: Volume 48 RPHC12 C18 C8

C6P-L C6P-G PGC PBB NAP

C4

v

e MIX C6P-Z

C18-C

PS

PE2

OPHE DP-X C3P

PFP

CHL

s

PE3

PYE

PE1 DP

DNAP PEG

PNP

EP CN

PVA DIOL FD

AMD

SI NH2

a

b

Figure 5.11  Spider diagram for a five-dimensional representation of the solvation parameter models in Table 5.6. The stationary phases are identified according to the abbreviations in Table 5.1. Bubble size is related to the vector length (u), calculated with Equation 5.7. Ellipses circle the columns, which were found similar through the calculation of J (Equations 5.9 to 5.11) Chromatographic conditions as in Figure 5.10.

chromatographic systems are plotted according to the normalized values of their system constants. Indeed, plots of data of very different absolute magnitudes can exhibit skewed data classification. Consequently, a range-scaling transformation of some sort is required. One way to do this is to divide e, s, a, and b by the v coefficient, as has been suggested by other authors [25,70]. This would allow assessment of the relative part of interactions—other than dispersive—to retention. However, in our case, contrary to what occurs in RPLC, the v coefficient is not always positive but it can be negative or zero. Thus, this operation does not really make sense. On another hand, the contribution of one type of interaction relative to all interactions established in a system can be evaluated if each coefficient is divided by u, defined by the following equation:

ui = ei2 + si2 + ai2 + bi2 + vi2

(5.7)

u is the length of the solvation vector associated to the chromatographic system. We have shown in previous works [69] that the vector length is a valuable tool to compare the amplitude of the interactions developed in a chromatographic system. It was calculated for each column tested and the values appear in the last column in Table 5.6. It must be noted that u is not correlated to the total retentive power of a chromatographic system as it takes both positive and negative coefficients into account in the same manner: interactions established with the stationary or mobile

Characterization of Stationary Phases

235

phase are both considered. Furthermore, it does not take into account the value of the c system constant, thus the phase ratio is not included in the comparison, only the interaction terms. Thus each x coefficient is divided by the vector length (u) to normalize the data before plotting the spider diagram. This way, the experimental behaviors are consistent with the observations based on the figure: when two columns are close on the spider diagram, they provide nearly identical elution order of the analytes. The size of the bubbles is related to u, thus the intensity of the interactions is also apparent on the figure. Thus, when two columns are close but have different bubble sizes, it indicates that the analytes are eluted in the same order on the two columns, but with larger retention factors are larger separation factors on the column with the larger u value. The way of reading this figure is not obvious at first sight as anyone familiar with PCA plots would be tempted to interpret the proximity of a point and an axis as an indication of the dominant factor contributing to retention on this stationary phase. However, it is absolutely not the case. The centered star composed of the five solvation axes is only represented to indicate the origin of the referential space and the directions that allowed placing the points in the figure, but only the distances between the points are significant, not their positions regarding the centre or the axes. Moreover, the position of the points is not indicative of the amplitude of each type of interaction and cannot distinguish the difference in the sign of the coefficients, as would be the case with a radar plot. This plot is more advantageous than the plots produced by a PCA, as only one figure represents all the information, while PCA generally results in more than two principal components, therefore requires two or more figures to exhibit all the information. Besides, any new chromatographic system can be added in a very simple manner, without the need of calculating the position of every other point, as is the case with PCA. Moreover, this plot can be used to compare numerous phases together, without the need for a reference phase. However, it must be mentioned that the five axes are not independent on the twodimensional paper and that the view of the five-dimensional space offered by this two-dimensional figure can be somewhat distorted. In any case, the figure cannot stand for itself and it is always preferable to use it in conjunction with Table 5.6, before drawing any conclusions. Nevertheless, we must point out that, so far, this figure has never contradicted the observations based on the values of the solvation coefficients. Furthermore, the angle between two solvation vectors (ω) associated to two chromatographic systems can be calculated according to the following equation, based on the solvation parameter model coefficients of the two systems noted i and j:

  ω *ω cos θij =  i  j = ωi * ω j

eie j + si s j + ai a j + bi b j + vi v j ei2 + si2 + ai2 + bi2 + vi2 e 2j + s 2j + a 2j + b 2j + v 2j

(5.8)

This method was first introduced by Ishihama and Asakawa to evaluate the similarity between liquid–liquid partition systems [71].

236

Advances in Chromatography: Volume 48

The angle between two columns provides a mean to measure the informational equivalence of different chromatographic systems: the wider the θ angle, the more different the systems are. Conversely, when the angle is close to zero, it indicates that the vectors have the same direction (although not necessarily the same length) and this would mean that the interactions established in these two systems are ­proportionally identical; therefore the two chromatographic systems would provide identical elution order of the solutes. We have shown in previous works that the proximities and distances between the columns on the spider diagram are perfectly consistent with calculated values of the θ angles [68,70,72]. Moreover, the similarity between two chromatographic systems is evaluated through the calculation of the J similarity factor, determined through Equations 5.9–5.11:

J = cos θij – cos (θdi + θdj)

(5.9)



 DD D2  D2   cos(θdi + θdj ) =  1 −  i 2   1 −  j 2  −  i  j ω ωi   i ωj ωj  

(5.10)



D = TINV(1−0.99, N) · SEav

(5.11)

where TINV is the inverse of the Student’s t-distribution for the specified degrees of freedom N, and SEav is the average of the standard errors of the solvation parameter model coefficients. In Equation 5.9, when J is positive, the systems compared are found to be similar; in the opposite case, they are considered to be different. It is worth noting that the angle alone is not sufficient to judge the similarity between two chromatographic systems, as it does not take into account the confidence limits associated to the system constants. There is no absolute limit angle that would provide a decision for similarity, because the calculation of J takes into account a sphere of uncertainty depending on the standard error associated to each coefficient and the degrees of freedom related to the number of solutes that were used to calculate the solvation coefficients (Equation 5.11). Thus, there is a further interest in injecting more solutes to achieve a higher precision in the comparisons: when the number of solutes injected is increased, the degrees of freedom increase, thus the D factor decreases. As a matter of fact, the uncertainty spheres are smaller, and thus the limit angle for dissimilarity is smaller. Under-determining the system could lead to considering two columns similar when they are not, just because their positions are not determined with enough precision. For instance, we will show later in this paper that the discrimination obtained on ODS phases is not as good as the one obtained here, because the solvation parameter models for ODS phases were only determined with 29 solutes. The θij angles existing between the solvation vectors associated to all the stationary phases characterized above through the use of the solvation parameter model and the J similarity factors between them were calculated. At the 99% confidence limit, some columns were found to be similar. Thus, the groups of stationary phases that were judged to be similar are circled in Figure 5.11.

237

Characterization of Stationary Phases

θ=19° θ=7°

OPHE

DP-X C3P

C3P

Figure 5.12  Two-dimensional representation of the solvation vectors associated to C3P, DP-X and OPHE. The circles indicate the uncertainty spheres (D) calculated with Equation 5.11. The θ angle values were calculated with Equation 5.8.

As an example of how the spider diagram representation can be related to practical solute retention, let us compare three aromatic columns: C3P, DP-X, and OPHE. The three of them are placed in close positions in the diagram, but only C3P and DP-X are circled, indicating that they were found to be similar through the calculation of J. Indeed, as appears on Figure 5.12, the vectors representing C3P and DP-X point in directions separated by a 7° angle and the uncertainty spheres cover a common angle, revealing that the two columns cannot be discriminated. Looking at Figure 5.12, it is clear that the two columns provide highly correlated retention factors (R2 = 0.985). Moreover, because the vector lengths are identical for the two columns (u = 0.467), indicating identical interaction strengths, the slope of the correlation curve is close to 1. All in all, C3P and DP-X are interchangeable. On another hand, Figure 5.12 shows that the vector for OPHE points toward a slightly different direction, establishing a 19° angle with the C3P solvation vector. Moreover, the uncertainty spheres do not cover any common angle thus the phases were found to be dissimilar through the calculation of J. This is in accordance with Figure 5.12, as some more dispersion is found among the retention factors between C3P and OPHE (R2 = 0.900). However, as the vector length for OPHE (u = 0.459) is close to that of C3P (u = 0.467), the slope of the regression line is still close to 1. As a result, C3P and DP-X cannot be exchanged for OPHE as the latter would provide slight differences in the separation of complex mixtures. First of all, the spider diagram allows us to evaluate the extent of the separation space that can be employed for separations in pSFC and to observe what part of the selectivity space is occupied by existing phases. A first observation is that the supercritical chromatographic systems are scattered in a wide selectivity space, mostly along a diagonal line. The regions of empty space on the plot provide a clear indication that the characterized columns do not occupy the selectivity space uniformly: they only fill a fraction of the selectivity space and allow targets for new stationary phases with complementary separation properties to be identified. For example, the PFP phase is the only one to occupy the top right of the figure, and no stationary phase is present at the bottom left. To fill this latter space, a stationary phase providing a high b value together with a low a value would be required. This means producing a stationary phase, which would be highly acidic but not basic. A closest examination of Figure 5.11 allows synthesizing the behaviors of the numerous tested columns. The nonpolar alkyl bonded phases of varying chain length (from C8 to C18) are located in the same area, indicating that they develop

238

Advances in Chromatography: Volume 48

identical interactions. As expected, the alkyl chain length does not modify the phase ­properties, but induces changes in the intensity of these interactions (see the differences in bubbles size or in u values in Table 5.6). The most polar phases are located in the opposite area of the spider diagram, underlying that these phases develop totally different interactions from the ones developed by the alkyl bonded phases. Between these two groups, there are numerous phases with different characters. Most of them are aromatic phases, but the polar embedded ODS phases (PE1, PE2, PE3) are also in the middle part of the figure. Secondly, the spider diagram can serve as a tool for stationary phase selection during method development. Indeed, the observation of individual system constants with histogram plots is helpful when simple separation problems are encountered: as discussed above, homologous series are best separated using a chromatographic system providing high v value; compounds differing in the aromatic character or number of double bonds are best separated using a chromatographic system providing a high e value, etc. However, most of the time, more than one chemical function is varied among the analytes, thus several terms must be taken into account and the combination of interactions is most important. On the spider diagram, when two columns are close, it means that their five normalized system constants (e/u, s/u, a/u, b/u, v/u) are close. Indeed, as the distances between the different columns in Figure 5.11 are consistent with the calculated values of the θ angle, two close columns have their solvation vectors pointing in the same direction thus proportionally identical system constants. When normalized system constants are identical, the two columns are judged similar by the calculation of the J similarity factor indicating that there is no difference in the effective selectivity of the two systems: the analytes will be eluted in highly similar order. However, in the course of method development, it is advisable to screen chromatographic systems providing different effective selectivities, thus being distant on the spider diagram. After identifying the most suitable stationary phase for a separation, near neighbors can be investigated for optimization. For instance, all polar phases are very close on the diagram and display quite small angles between them, because they globally show an identical pattern of interactions (an increase in polarity causes an increase in retention while an increase in the molecular volume causes a decrease in retention) with a slightly different blend of polar interactions, but they are not all found to be similar, thus small differences in the elution order can occur. More precisely, when largely different compounds need to be separated, the differences between the polar phases will not be obvious but, for the separation of closely spaced compounds, EP will not always retain the same order as those observed on SI and one column cannot be substituted for the other when the separation of complex mixtures are of interest. Thus, in the course of method development, initial screening of all polar phases is of no use: it is more valuable to screen stationary phases that are scattered in the whole spider diagram then, if a polar phase seems to be the most appropriate, screening of phases providing slightly different effective selectivities can be useful. Besides, it can be an advantage to possibly exchange a column for another one that would provide identical elution order with higher selectivity, as is the case when CN is exchanged for EP. As a matter of fact, the high selectivity of a column can be an advantage for the separation: high selectivity

239

Characterization of Stationary Phases CN o

EP

m

SI

o

m

m

o p

p p

1

2 Minutes

3

1

2

3

Minutes

4

5

1

Minutes

2

Figure 5.13  Chromatograms of ortho-, meta, and para-nitrophenols on three polar columns. Chromatographic conditions as in Figure 5.10.

is important in that more selective columns do not need as many plates to achieve a specified resolution and thus shorter columns or higher flow rates can be used. A detailed example is provided (Figure 5.13) with the separation of nitrophenols. This choice is significant because, apart from the molecular volume, all other descriptors vary among the three isomers (see descriptor values in Table 5.3). The solvation vectors of CN and EP are only separated by a 6° angle, indicating that the two columns provide proportionally identical interactions thus will elute the analytes in the same order. However, the length of the solvation vectors is not identical (uEP = 1.696 while uCN = 1.029) indicating that EP establishes stronger interactions than CN. Judging from Equations 5.1 and 5.2, EP should provide both larger retention factors and larger separation factors (α) than CN. Indeed, the retention factors for o−, m−, and p-nitrophenols are, respectively, −0.279, 0.404, and 0.507 on EP; −0.485, 0.090, 0.170 on CN and the logarithms of separation factors between o/m and m/p are, respectively, 0.683 and 0.103 on EP; 0.575 and 0.080 on CN. Thus changing CN for EP is a means to retain identical elution order while increasing separation factors, where need be SI, on the contrary, is not judged to be similar to the other two columns, although it displays a small angle with each of them (15°). This is indicative of a slightly different blend of interactions, which can result in different separations. Figure 5.13 indeed shows that the separation of m− and p-nitrophenols is not complete with this column, while they were baseline resolved with EP and CN. Thus when a separation is not perfect with CN or EP, it can be worth testing SI. Regarding the screening of orthogonal columns, we have shown how this diagram can be used to build a panel of complementary stationary phases to achieve

240

Advances in Chromatography: Volume 48

separations in a screening method [72]. It is recognized today that SFC is especially well suited for these screening methods, as elution gradients (and subsequent equilibration) can be performed in a few minutes, due to the high fluid diffusivity. In practice, any approach to method development in pSFC would have to include mobile phase optimization. Therefore, some care is needed in comparisons of stationary phases at a single reference mobile phase composition. The few studies available [73] and our own experience indicate that selectivity changes with mobile phase composition are dependent on the identity of the stationary phase. The phase constants generally (but not always) decline with increasing percentage of modifier in the mobile phase, often in a non-linear manner and with different slopes. For this reason, it cannot be assumed that the selectivity comparison of stationary phases in the chromatographic conditions used here would be exactly the same in other chromatographic conditions. This point is currently under investigation.

5.4.3 ODS phases In this part, we will put a special emphasis on ODS phases, as the large diversity of chemistries existing are the object of many characterization tests [74]. Indeed, the structures of the ODS phases are very varied and can lead to very diverse selectivities: • Types of silica base: A, B (high purity), or C (surface covered with Si-H groups), organic/inorganic hybrid silica, silica covered with a polymethylsilicone polymer layer pore diameter (from 60 to 300 Å´). • Surface area (from 180 to 450 m2/g). • Functionality of the bonding (mono- or polymeric phases). • Bonding density (from 1.5 to 3.6 μmol/m2). • Endcapping treatment: nature of the endcapping reactant, hydrophilic endcapping, bonded chains with steric protection and bidentate bonding. • Horizontal polymerization of the bonded chains. • Embedded polar groups (amide, urea, carbamate, quaternary ammonium, ether, or sulfonamide). The majority of these processes are intended to produce base-deactivated packings, that is to say to reduce the accessibility to residual silanol groups to basic compounds, and to favor the stability of the silica at the high pH often required to avoid the ionic form of these basic compounds. The columns and their potentially very different selectivity require a classification in order to facilitate the selection of appropriate stationary phases for a given application. We have shown in previous works [65,75] how the solvation parameter model can help in this task, in the same manner as for the other types of phases described above. The database presented here has since been updated. The ODS phases in Table 5.2 were characterized with the solvation parameter model, using the 29 solutes in Table 5.5. In the same manner as for the varied phases above, angles and similarities were calculated between the solvation vectors, using Equations 5.8 to 5.11. The results are presented on the spider diagram (Figure 5.14).

241

Characterization of Stationary Phases

1

P11

P8

11 6 12 10

P21

P3 P29

P7

P28 P16

P9

P6 P27

v

e

P2 P1 P10 P12 17

2 to 9 13 to 16

P5

P32 P31 P4

P30 P15

P24 P20

P26 P19

b

P23 P17

s

P22 P13 P25 P14 P18

a

Figure 5.14  Spider diagram for a five-dimensional representation of the solvation parameter models of the ODS columns. The stationary phases are identified according to the numbers in Table 5.2. White bubbles are the classical ODS phases (1 to 17), black bubbles are the polar-type ODS phases (P1 to P32). Bubble size is related to the vector length (u), calculated with Equation 5.7. Ellipses circle the columns, which were found similar through the calculation of J (Equations 5.9 to 5.11). Chromatographic conditions as in Figure 5.10.

Again, the phases that were judged to be similar are circled. As mentioned above, the discrimination obtained in this case is inferior due to the smaller number of solutes analyzed resulting in larger uncertainty spheres. However, the solvation parameter model allows a fine evaluation of dipole–­dipole and hydrogen-bonding interactions occurring on these phases, either due to the residual silanol groups on non-endcapped phases, or to the polar embedded (amide, carbamate, sulfonamide, ether…) and hydrophilic endcapping groups. As a matter of fact, the non-endcapped phases (10 to 12 in Table 5.2) are well discriminated from the endcapped or “protected” phases. The case of phases possessing a hydrophilic endcapping group (P5 to P12 in Table 5.2) is more confuse as the phases identified as such are scattered. The variety of possible treatments and the absence of clear indications from the manufacturers do not help in understanding the differences observed. The polar-embedded phases (P13 to P33 in Table 5.2), on the contrary, are generally clearly distinguished from all others. In some cases, different polar groups can be discriminated. For instance, the ether-sulfonamide group from P16 and the ether group from P27 are clearly distinguished from the amide-embedded group, which is most commonly found. However, the carbamate groups are not discriminated from the amide groups. It is also worth noting that P28 and P29 (Polaris A and Polaris B), which both have undisclosed structures, are also clearly discriminated from the amide-embedded phases. Thus it can be concluded that they must not possess any amide group. As an example of the different selectivities, which can be expected when varying the nature of the ODS bonding chemistry, we analyzed three non-steroidal

242

Advances in Chromatography: Volume 48

a­ nti-inflammatory drugs (NSAID) (aspirin, phenylbutazone, and sulindac) having very different structures (Figure 5.15). Figure 5.16 shows how the elution order changes when a well-protected column (Uptisphere ODB No.7, Atlantis dC18 No.16) is changed for a non-endcapped phase (Uptisphere NEC No.10), for a phase with a hydrophilic endcapping group (Aquasil C18 No.P5, Alltima HP C18 AQ No.P7), then for a polar-embedded phase. The effect of the nature of the embedded group is seen with the reversal of elution order between a sulfonamide-embedded phase (Acclaim Polar Advantage No.P16) and amide-embedded phases (Alltima HP C18 Amide No. P15, Acclaim Polar Advantage II No.P17, Uptisphere PLP No.P19, Discovery RP Amide No.P22, Ascentis RP Amide No.P23, and Supelcosil ABZ + Plus No.P26). Moreover, the chromatograms provided in Figure 5.17 show how the retention and selectivities vary among the different ODS phases. This all shows that even when comparing structurally related stationary phases, differences in their intermolecular interaction abilities can easily be detected, providing a better semi-quantitative understanding of their behavior. [26] F O

O

OH

HO

N

O

O

N O

S

O

Aspirin (A)

O

Phenylbutazone (P)

Sulindac (S)

Figure 5.15  Structures of the non-steroidal anti-inflammatory drugs analyzed as an example of different selectivities of ODS phases. Aspirin Sulindac Phenylbutazone

4

Elution order

3

2

1

7

16

10

P5

P7

P16 P15 P17 Column no.

P19

P22

P23

P26

Figure 5.16  Elution orders of the NSAID in Figure 5.15 on twelve selected ODS columns with varying selectivities. The stationary phases are identified according to the abbreviations in Table 5.2. Chromatographic conditions as in Figure 5.10.

2 Minutes

A P

1.5 Minutes

A

S

P

3

S

1

16

1

P23

1.5 Minutes

A

S

2

A

A

S P

S

2.5

1 2 Minutes

1.5 2 Minutes

P

P

8 A

1.5

P

P22

1

10

S

2.25 3 Minutes

A

1.5 2 Minutes

P

S

P

P19

1

P5 A

1

2

P

8 12 Minutes

A

3 4 Minutes

S

5

16

S

Figure 5.17  Chromatograms of three non-steroidal anti-inflammatory drugs on nine selected ODS columns with varied bonding chemistries (see Figure 5.15 for identification of the compounds). The stationary phases are identified according to the abbreviations in Table 5.2. Chromatographic conditions as in Figure 5.10.

1

P16

1

7

Characterization of Stationary Phases 243

244

Advances in Chromatography: Volume 48

However, it must be pointed out that only little discrimination is possible between the classical, endcapped or protected ODS phases (1 to 9, 13 to 17 in Table 5.2) with the solvation parameter model, although these phases are known to have different chromatographic behaviors, and particularly different steric selectivities. It is a wellknown defect of the solvation parameter model that steric impedance to insertion into the bonded phases is totally absent from Abraham descriptors: there is no term related to molecular shape as, in the calculation of the molecular volume (V), no account is taken for three-dimensional shape, only the number of atoms and bonds is taken into account. Regarding ODS phases, we have solved this problem by combining the solvation parameter model with another characterization test, which can provide an evaluation of steric selectivity, based on the analysis of carotenoid pigments in SFC [76]. The two tests are complementary as the carotenoid test has no ability to distinguish between different sources of polar interactions [77], while the solvation parameter model can. For other types of phases, different solutions are needed. A new descriptor in the solvation parameter model, which could account for shape selectivity without being redundant with the already existing descriptors, could be introduced. This is a difficult task as, among the varied descriptors having been introduced to estimate shape, no one offers a universal description. In general, the parameters describing molecular shape cannot be shortened to a unique term, unless the solutes are limited to a single structural family. We will address this point in future works.

5.4.4 Method development with the solvation parameter model At this time, the use of the solvation parameter model for systematic selectivity optimization is only poorly developed. With the following examples, we wish to demonstrate how a clever selection of columns based on utilization of the above results can help in achieving a separation of a real complex mixture. A set of seven sunscreen molecules was selected as test mixture (Figure 5.18). These compounds are classically encountered in cosmetic products in which they are combined at different concentrations, following maximum concentration authorized by regulatory authorities around the world. Five of the chosen molecules (OCR, BDM, BZ3, ES, and EMC) have related chemical structures, while the last two (ET and EMT) also have common features. This choice of molecules is an interesting one because it allows estimating the orthogonality or similarity of the stationary phases toward similar structures, a situation that is representative for impurity profiling of drugs. Besides, we were interested in the role of multifunctional and high molecular mass compounds as ET and EMT, because the great majority of real sample solutes have higher molecular mass and more functional groups than the test compounds that served to establish the solvation coefficients. It must be noted that ET and EMT are possibly ionized basic compounds, thus the phases which were suspected to establish ionic interactions were not tested on this sample. First of all, the elution orders of the seven sunscreen molecules on eight stationary phases covering a wide range of selectivities are presented in Figure 5.19. The varied elution orders obtained are a clear indication of the variety of separations that can be achieved when stationary phases are chosen in distant places of the spider

245

Characterization of Stationary Phases OH

BZ3

O

O

O CN

MeO

O

OCR

O

O O

O OMe

MeO

O

BDM

EMC

ES

OH

O

O O

O N N

OH

N

HN

OH

N

EMT

H N

N

O

N NH

ET

O

OMe

O

O

Figure 5.18  Structures of the sunscreen molecules analyzed as an example of complex mixture for method development. 8 7

Elution order

6 5 4 3 2 1 0

MIX

C6P-L

BDM

PE1 BZ3

C4 C3P Column EMC

EMT

OPHE

EP

ES

ET

NH2 OCR

Figure 5.19  Elution orders of the sunscreen molecules (see Figure 5.18 for identification of the compounds) on eight selected columns with varying selectivities. The stationary phases are identified according to the abbreviations in Table 5.1. Chromatographic conditions as in Figure 5.10.

diagram. Besides, the close elution orders obtained on columns issued from close regions of the diagram (as EP and NH2; MIX and C6P-L; C3P and OPHE) are also consistent with all above comments. On the other hand, one can remark that retention of BZ3 strongly depends on the phase polarity, due to its hydroxyl function, in

246

Advances in Chromatography: Volume 48 C4

OCR

OCR+BDM

EMC BZ3

ET + EMT

OCR

MIX

ET

BDM

BDM EMC + BZ3

EMC ES + BZ3

EMT ES

0.75

EMC + BZ3

1.5 Minutes

2.25

2

4

6 Minutes

8

2

4

6 8 Minutes

10

OCR + ES + EMC

OCR

C3P

OPHE BDM

BDM

ET

EMT

ET

PE1

BZ3

BZ3 EMC C3

ET–EMT

ES

OCR + BDM

ES

C6P-L

ET EMT

EMT 2

4

6

8 10 Minutes

12

14

2

4

6

EMT-BZ3 + OCR

8 10 Minutes

12

BZ3 + OCR

EP

ET

EMC

14

2

BDM

ET

ES

BDM 2

4

8

NH2

EMT

EMC

ES

4 6 Minutes

6 8 Minutes

10

12

1

2

3 Minutes

4

5

Figure 5.20  Chromatograms of the seven sunscreen molecules (identified in Figure 5.18) on eight columns with varying selectivities. The stationary phases are identified according to the abbreviations in Table 5.1. Chromatographic conditions as in Figure 5.10.

accordance to the stationary phase chemical structure. However, chromatographic behaviors are difficult to estimate, because EMT which has two hydroxyl groups is the less retained compound on PE1, whereas BZ3 was one of the most retained on this phase. One can suppose that the great size of EMT hinders it to deeply penetrate into the stationary phase, avoiding the interactions between the hydroxyl groups and the polar embedded group of the bonded chain, which is located close to the silica surface. Secondly, the observation of the chromatograms obtained on these eight columns (Figure 5.20) show the varied qualities of peak shapes that can be observed. In the course of a method development, the stationary phase providing the best start for

Characterization of Stationary Phases

247

optimization, based on resolution and peak shapes, would be selected. In the case presented here, several columns appear to be possible starters: MIX, OPHE, or NH2 would all be potential candidates. It is worth pointing out that these phases are situated in three totally different regions of the spider diagram. This is a clear indication that all stationary phase polarities can be helpful in developing a separation method in pSFC.

5.4.5 Predictive capability of the Models A tool to predict the behavior of compounds on column would facilitate the selection of the stationary phase before compounds are analyzed. However, the solvation parameter model is not expected to provide predictions of retention to an accuracy that would be chromatographically useful for method development (± 0.03 in ln k) [26]. Indeed, prediction of retention requires that a suitable model be generated. Then the retention for any solute with known solute descriptors, or solutes for which reasonable estimates of the descriptors are possible, could be estimated by simple arithmetic. The accuracy of the prediction depends both on the uncertainty in the solute descriptors and the model system constants. The average uncertainty of the model system constants can be estimated through the standard error in the estimate of the multiple linear regression analysis. In our pSFC database, they range from 0.029 to 0.229. In order to assess the predictive capability of the solvation parameter models established, a second set of models was established, based on a reduced set of analytes. For this purpose, 30 compounds were removed from the initial set (compounds marked with an asterisk in Table 5.3). These compounds were selected so as to cover a wide range of properties (acidic and basic character, size, polarity, and polarizability) and in such a manner that the homogeneous repartition of the descriptor values in the remaining set would not be disturbed. Then, based on this second series of models, the predictive retention factors of the 30 test-compounds were calculated and compared to the measured experimental retention factors. Figure 5.21 is given as an example of these comparisons. The correlation coefficients among all models tested ranged from 0.879 (for the worst PGC phase) to 0.990. The predictive capability is quite good for most columns (see Figure 5.21) and, apart from solutes which elute very close, it would be possible to predict the elution order for most simple compound mixtures. Only the adsorbenttype stationary phases (PGC and SI) show poor correlations, indicating that tentative prediction of elution orders would be more uncertain. One limitation of this demonstration is that the solutes in the validation set are much simpler structurally than the broad, complex array of pharmacologically active compounds that are targets for quantitative prediction of retention and selectivity. Such compounds set two problems regarding retention prediction: the first one is that improved methods for estimating solute descriptors from structure for complex molecules are required; the second is that stereo-induced interactions might occur on certain stationary phases. We will address this question in future works.

248 (a) 1.5 1.0

Advances in Chromatography: Volume 48

R2 = 0.9893

Log ktheo

R2 = 0.9307

Log ktheo

0.5

0.5

0.0

0.0

–0.5

–0.5 –1.0 –1.5 –1.5

(b) 1.0

Log kexpPE3 –1.0

–0.5

0.0

(c) 1.0 0.5

0.5

1.0

–1.0 –1.5 1.5 –1.5

Log kexpSI –1.0

–0.5

0.0

0.5

1.0

R2 = 0.98

Log ktheo

0.0 –0.5 –1.0 –1.5 –2.0 –2.0

Log kexpPVA –1.5

–1.0

–0.5

0.0

0.5

1.0

Figure 5.21  Comparison of the calculated and measured retention factors obtained on (a) PE3 (b) SI and (c) PVA for the compounds marked with an asterisk in Table 5.3, when the models are calculated omitting these solutes.

5.5  Conclusion Modern pSFC mostly relies on a more restricted range of stationary phases than those depicted in this study. In practice polar stationary phases (SI, NH2, DIOL, CN, and EP) virtually monopolize the market. However, we have shown that these phases occupy only a fraction of the selectivity space and provide limited possibilities for selectivity optimization. The point we want to make here is that the nonpolar and moderately polar phases provide an opportunity to extend the selectivity space significantly beyond that which can be explored using polar phases. The solvation parameter model has had a considerable impact on our understanding of the retention mechanism of non-ionic compounds in pSFC. The spider diagram provides a convenient visual classification of packed columns for SFC. It can be used for the rational choice of columns. It is easy to identify those stationary phases with separation properties that are most similar to each other, and phases with less similarity. It is a valuable technical aid to the chromatographer faced with the need to make rapid decisions in SFC. The columns characterized encompass the full range of polarity of commercial packed columns currently available. They span a wide selectivity space, although

Characterization of Stationary Phases

249

there remain empty areas of the selectivity space which would benefit by the introduction of new stationary phase chemistries. Moreover, these studies underline that pSFC with modifier allows us to cover both the normal and the reversed-phase domains. Whereas these two domains are strictly separated when using liquid mobile phases (HPLC), the use of an identical supercritical mobile phase whatever the stationary phases allows the unification of these domains in SFC [64]. It also opens a new field of research for orthogonal separation on line.

References





1. Yaku K., Morishita F. 2000. Separation of drugs by packed-column supercritical fluid chromatography. J. Biochem. Biophys. Methods 43:59–76. 2. White C., Burnett J. 2005. Integration of supercritical fluid chromatography into drug discovery as a routine support tool. J. Chromatogr. A 1074:163–185. 3. Abbott E., Veenstra T.D., Issaq H.J. 2008. Clinical and pharmaceutical applications of packed-column supercritical fluid chromatography. J. Sep. Sci. 31:1223–1230. 4. Brunelli C., Zhao Y.,. Brown M.-H, Sandra P. 2008. Pharmaceutical analysis by supercritical fluid chromatography: optimization of the mobile phase composition on a ­2-ethylpyridine column. J. Sep. Sci. 31:1299–1306. 5. Lesellier E., Gurdale K., Tchapla A. 1999. Separation of cis/trans isomers of β-carotene by supercritical fluid chromatography. J. Chromatogr. A 844:307–320. 6. Lesellier E., Tchapla A. 1999. Retention behaviour of triglycerides in octadecyl packed subcritical fluid chromatography with CO2/modifier mobile phases. Anal. Chem. 71:5372–5378. 7. Lesellier E., Tchapla A. 2000. Separation of vegetable oil triglycerides by subcritical fluid chromatography with octadecyl packed columns and CO2/modifier mobile phases. Chromatographia 51:688–694. 8. Gaudin K., Lesellier E., Chaminade P., Ferrier D., Baillet A., Tchapla A. 2000. Retention behaviour of ceramides in subcritical fluid chromatography in comparison with non-aqueous reversed-phase liquid chromatography. J. Chromatogr. A 883: 211–222. 9. Hoffman B.J., Taylor L.T., Rumbelow S., Goff L., Pinkston J.D. 2004. Separation of derivatized alcohol ethoxylates and propoxylates by low temperature packed-column supercritical fluid chromatography using ultraviolet absorbance detection. J. Chromatogr. A 1034:207–212. 10. Yarita T., Nomura A., Abe K., Takeshita Y. 1994. Supercritical fluid chromatographic determination of tocopherols on an ODS-silica gel column. J. Chromatogr. A 679:329–334. 11. Bamba T., Fukusaki E., Kajiyama S., Ute K., Kitayama T., Kobayashi A. 2001. High resolution analysis of polyprenols by supercritical fluid chromatography. J. Chromatogr. A 911:113–117. 12. Bari V.R., Dhorda U.J., Sundaresan M. 1997. A simultaneous packed-column supercritical fluid chromatographic method for ibuprofen, chlorzoxazone, and acetaminophen in bulk and dosage form. Talanta 45:297–302. 13. Yamada A., Ezaki Y., Matsuo K., Yarita T., Nomura A. 1995. Supercritical fluid chromatography of free resin acids on an ODS-silica gel column. J. Chromatogr. A 709:345–349.

250

Advances in Chromatography: Volume 48

14. Gurdale K., Lesellier E., Tchapla A. 1999. Methylene selectivity and eluotropic strength variations in subcritical fluid chromatography with packed columns and CO2-modifier mobile phases. Anal. Chem. 71:2164–2170. 15. Lesellier E., Bleton J., Tchapla A. 2000. Use of relationships between retention behaviours and chemical structures in subcritical fluid chromatography with CO2/modifier mixtures for the identification of triglycerides. Anal. Chem. 72:2573–2580. 16. Lesellier E., Gurdale K., Tchapla A. 2002. Phase ratio and eluotropic strength changes on retention variations in subcritical fluid chromatography using packed octadecyl columns. Chromatographia 55:555–563. 17. Lesellier E., Tchapla A. 1998. Subcritical fluid chromatography with organic modifiers on octadecyl packed columns: recent developments for the analysis of high molecular organic compounds. In Chromatogr. Sci. 75, Supercritical-fluid chromatography with packed columns, ed. K. Anton and C. Berger, 195–221. Marcel Dekker, Inc, New York, NY. 18. Upnmoor D., Brunner G. 1992. Packed column supercritical fluid chromatography with light-scattering detection. II. Retention behaviour of squalane and glucose with mixed mobile phases. Chromatographia 33:261–266. 19. Jiang C., Ren Q., Wu P. 2003. Study on retention factor and resolution of tocopherols by supercritical fluid chromatography. J. Chromatogr. A 1005:155–164. 20. Berger T.A., Deye J.F. 1991. Effects of column and mobile phase polarity using ­steroids as probes in packed-column supercritical fluid chromatography. J. Chromatogr. Sci. 29:280–286. 21. Combs M.T., Ashraf-Khorassani M., Taylor L.T. 1997. Method development for the separation of sulfonamides by supercritical fluid chromatography. J. Chromatogr. Sci. 35:176–180. 22. Schoenmakers P.J., Uunk L.G.M., Janssen H.-G. 1990. Comparison of stationary phases for packed-column supercritical fluid chromatography. J. Chromatogr. 506:563–578. 23. Heaton D.M., Bartle K.D., Clifford A.A., Klee M.S., Berger T.A. 1994. Retention prediction based on molecular interactions in packed-column supercritical fluid chromatography. Anal. Chem. 66:4253–4257. 24. Abraham M.H., Ibrahim A., Zissimos A.M. 2004. Determination of sets of solute descriptors from chromatographic measurements. J. Chromatogr. A 1037:29–47. 25. Poole C.F., Poole S.K. 2002. Column selectivity from the perspective of the solvation parameter model. J. Chromatogr. A 965:263–299. 26. Vitha M.F., Carr P.W. 2006. The chemical interpretation and practice of linear solvation energy relationships in chromatography. J. Chromatogr. A 1126:143–194. 27. Abraham M.H., Poole C.F., Poole S.K. 1999. Classification of stationary phases and other materials by gas chromatography. J. Chromatogr. A 842:79–114. 28. Zhao Y.H., Abraham M.H., Zissimos A.M. 2003. Fast calculation of van der Waals volume as a sum of atomic and bond contributions and its application to drug compounds. J. Org. Chem. 68:7368–7373. 29. Platts J.A., Butina D., Abraham M.H., Hersey A. 1999. Estimation of molecular linear free energy relation descriptors using a group contribution approach. J. Chem. Inf. Comput. Sci. 39:835–845. 30. Platts J.A., Abraham M.H., Butina D., Hersey A. 2000. Estimation of molecular linear free energy relationship descriptors by a group contribution approach. 2. Prediction of partition coefficients. J. Chem. Inf. Comput. Sci. 40:71–80. 31. http://www.pharma-algorithms.com/absolv.htm 32. Kiridena W., Poole C.F. 1998. Influence of solute size and site-specific surface interactions on the prediction of retention in liquid chromatography using the solvation parameter model. Analyst 123:1265–1270.

Characterization of Stationary Phases

251

33. Pyo D., Li W., Lee M.L., Werkwerth J.D., Carr P.W. 1996. Addition of methanol to the mobile phase in packed capillary column supercritical fluid chromatography Retention mechanisms from linear solvation energy relationships. J. Chromatogr. A 753:291–298. 34. Blackwell J.A., Stringham R.W., Werckwerth J. D. 1997. Effect of mobile phase additives in packed-column subcritical and supercritical fluid chromatography. Anal. Chem. 69:409–415. 35. Cantrell G.O., Stringham R.W., Blackwell J.A., Werckwerth J. D., Carr P.W. 1996. Effect of various modifiers on selectivity in packed-column subcritical and supercritical fluid chromatography. Anal. Chem. 68:3645–3650. 36. Blackwell J.A., Stringham R.W. 1997. Characterization of temperature dependent modifier effects in SFC using linear solvation energy relationships. Chromatographia 46:301–308. 37. Blackwell J.A., Stringham R.W. 1997. Comparison of various bulk fluids and modifiers as near-critical mobile phases on a polymeric column using linear solvation energy relationships. J. High Resolut. Chromatogr. 20:631–637. 38. Weckwerth J., Carr P. 1998. Study of interactions in supercritical fluids and supercritical fluid chromatography by solvatochromic linear solvation energy relationships. Anal. Chem. 70:1404–1411. 39. Pyo D., Kim H., Park J.H. 1998. Modifier effects in open-tubular capillary column supercritical fluid chromatography: retention mechanisms from linear solvation energy relationships. J. Chromatogr. A 796:347–354. 40. West C., Lesellier E., Tchapla A. 2004. Retention characteristics of porous graphitic carbon in subcritical fluid chromatography with carbon dioxide-methanol mobile phases. J. Chromatogr. A 1048:99–109. 41. West C., Lesellier E. 2005. Effects of modifiers in subcritical fluid chromatography on retention with porous graphitic carbon. J. Chromatogr. A 1087:64–76. 42. West C., Lesellier E. 2005. Separation of substituted aromatic isomers with porous ­graphitic carbon in subcritical fluid chromatography. J. Chromatogr. A 1099:175–184. 43. Janssen H.-G., Lou X., 1999. Packed columns in SFC: mobile and stationary phases and further resuirements. In Practical Supercritical Fluid Chromatography and Extraction, ed. M. Caude and D. Thiébaut, 15–52. Harwood Academic Publishers, Amsterdam, NL. 44. Berger T.A. 2005. Density of methanol-carbon dioxide mixtures at three temperatures: comparison with vapour-liquid equilibria measurements and results obtained from chromatography. J. High. Resolut. Chromatogr. 14:312–316. 45. Chester T. 1999. The road to unified chromatography: the importance of phase behaviour knowledge in supercritical fluid chromatography and related techniques, and a look at unification. Microchem. J. 61:12–24. 46. Smith R.M., Sanagi M.M. 1990. Retention and selectivity in supercritical fluid chromatography on octadecylsilyl-silica column. J. Chromatogr. 505:147–159. 47. West C., Lesellier E. 2007. Characterization of stationary phases in supercritical fluid chromatography with the solvation parameter model: V. Elaboration of a reduced set of test solutes for rapid evaluation. J. Chromatogr. A 1169:205–219. 48. Li J. 2002. Prediction of internal standards in reversed-phase liquid chromatography: III. Evaluation of an alternative solvation parameter model to correlate and predict the retention of ionizable compounds. J. Chromatogr. A 982:209–223. 49. Li J. 2004. Prediction of internal standards in reversed-phase liquid chromatography: IV. Correlation and prediction of retention in reversed-phase ion-pair chromatography based on linear solvation energy relationships. Anal. Chim. Acta 522:113–126.

252

Advances in Chromatography: Volume 48

50. Rosés M., Bolliet D., Poole C.F. 1998. Comparison of solute descriptors for predicting retention of ionic compounds (phenols) in reversed-phase liquid chromatography using the solvation parameter model. J. Chromatogr. A 829:29–40. 51. Abraham M.H., Zhao Y.H. 2004. Determination of solvation descriptors for ionic species: hydrogen bond acidity and basicity. J. Org. Chem. 69:4677–4685. 52. Zhao Y.H., Abraham M.H., Zissimos A.M. 2003. Determination of McGowan volumes for ions and correlations with van der Waals volumes. J. Chem. Inf. Comput. Sci. 43:1848–1854. 53. Li J., Sun J., Cui S., He Z. 2006. Quantitative structure-retention relationship studies using immobilized artificial membrane chromatography I. Amended linear solvation energy relationships with the introduction of a molecular electronic factor. J. Chromatogr. A 1132:174–182. 54. Wen D., Olesik S.V. 2000. Characterization of pH in liquid mmixtures of methanol/H2O/ CO2. Anal. Chem. 72:475–480. 55. Toews K.L., Shroll R.M., Wai C.M., Smart N.G. 1995. pH-defining equilibrium between water and supercritical CO2. Influence on SFE of organics and metal chelates. Anal. Chem. 67:4040–4043. 56. Zheng J. 2005. Supercritical fluid chromatography of ionic compounds. PhD diss., Virginia Polytechnic Institute, Blacksburg, VA 57. Reta M., Carr P.W., Sadek P.C., Rutan S.C. 1999. Comparative study of hydrocarbon, fluorocarbon and aromatic bonded RP-HPLC stationary phases by linear solvation energy relationships. Anal. Chem. 71:3484–3496. 58. Wang T., Wang X., Smith R.L. Jr. 2005. Modeling of diffusivities in supercritical carbon dioxide using a linear solvation energy relationship. J. Supercrit. Fluids 35:18–25. 59. Lagalante A.F., Bruno T.J. 1998. Modelling the water-supercritical CO2 partition coefficients of organic solutes using a linear solvation energy relationship. J. Phys. Chem. B 102:907–909. 60. Bush D., Eckert C.A. 1998. Prediction of solid-fluid equilibria in supercritical carbon dioxide using linear solvation energy relationships. Fluid Phase Equilib. 150–151:479–492. 61. Timko M.T., Nicholson B.F., Steinfeld J.I., Smith K.A., Tester J.W. 2004. Partition coefficients of organic solutes between supercritical carbon dioxide and water: experimental measurements and empirical correlations. J. Chem. Eng. Data 49:768–778. 62. Poole S.K., Poole C.F. 2003. Separation methods for estimating octanol-water partition coefficients. J. Chromatogr. B 797:3–19. 63. Smith R.M., Cocks S., Marsin Sanagi M., Briggs D.A., Evans V.G. 1991. Retention in supercritical fuid chromatography on cyano-bonded silica columns, Analyst 116:1281–1285. 64. West C., Lesellier E. 2008. A unified classification of stationary phases for packedcolumn supercritical fluid chromatography. J. Chromatogr. A 1191:21–39. 65. West C., Fougère L., Lesellier E. 2008. Combined supercritical fluid chromatographic tests to improve the classification of numerous stationary phases used in reversed phase liquid chromatography, J. Chromatogr. A 1189:227–244. 66. West C., Lesellier E. 2006. Characterisation of stationary phases in subcritical fluid chromatography by the solvation parameter model: I. Alkylsiloxane-bonded stationary phases. J. Chromatogr. A 1110:181–190. 67. West C., Lesellier E. 2006. Characterisation of stationary phases in subcritical fluid chromatography by the solvation parameter model: III. Polar stationary phases. J. Chromatogr. A 1110:200–213. 68. West C., Lesellier E. 2006. Characterisation of stationary phases in subcritical fluid chromatography by the solvation parameter model: IV. Aromatic stationary phases. J. Chromatogr. A 1115:233–245.

Characterization of Stationary Phases

253

69. West C., Lesellier E. 2006. Characterisation of stationary phases in subcritical fluid chromatography with the solvation parameter model II: Comparison tools. J. Chromatogr. A 1110:191–199. 70. Abraham M.H., Rosés M., Poole C.F., Poole S.K. 1997. Hydrogen bonding. 42. Characterisation of reversed-phase liquid chromatographic C18 stationary phases. J. Phys. Org. Chem. 10:358–368. 71. Ishihama Y., Asakawa N. 1999. Characterization of lipophilicity scales using vectors from solvation energy descriptors. J. Pharm. Sci. 88:1305–1312. 72. West C., Lesellier E. 2008. Orthogonal screening system of columns for supercritical fluid chromatography. J. Chromatogr. A 1203:105–113. 73. Bui H., Masquelin T., Perun T., Castle T., Dage J., Kuo M.-S. 2008. Investigation of retention behaviour of drug molecules in supercritical fluid chromatography using linear solvation energy relationships. J. Chromatogr. A 1206:186–195. 74. Lesellier E., West C. 2007. Description and comparison of chromatographic tests and chemometric methods for packed column classification. J. Chromatogr. A 1158:329–360. 75. Lesellier E., West C. 2007. Combined supercritical fluid chromatographic methods for the characterization of octadecylsiloxane-bonded stationary phases, J. Chromatogr. A 1149:345–357. 76. Lesellier E., Tchapla A. 2005. A simple subcritical chromatographic test for an extended ODS high-performance liquid chromatography column classification. J. Chromatogr. A 1100:45–59. 77. Lesellier E., West C., Tchapla A. 2006. Classification of special octadecyl-bonded stationary phases by the carotenoid test. J. Chromatogr. A 1111:62–70.

Hydride— 6 Silica Chemistry and Applications Joseph J. Pesek and Maria T. Matyska Contents 6.1 Introduction................................................................................................... 255 6.1.1 Background........................................................................................ 255 6.1.2 Synthesis of Silica Hydride................................................................ 256 6.2 Hydride-Based Stationary Phases for HPLC................................................. 257 6.2.1 Synthesis and Characterization......................................................... 257 6.2.2 Stability of Silica Hydride Materials................................................. 258 6.2.3 Chromatographic Properties of Silica Hydride Phases..................... 259 6.3 Applications of Hydride-Based Phases in HPLC..........................................264 6.3.1 Overview............................................................................................264 6.3.2 Reversed-Phase..................................................................................264 6.3.3 Aqueous Normal Phase..................................................................... 267 6.3.4 Dual Mode Retention......................................................................... 272 6.3.5 Organic Normal Phase....................................................................... 275 6.3.6 Microcolumn HPLC.......................................................................... 277 6.4 Hydride-Based Etched Capillaries................................................................ 278 6.4.1 Fabrication......................................................................................... 278 6.4.2 Characterization of Capillary Properties........................................... 279 6.4.3 Dual Separation Modes.....................................................................280 6.4.4 Other Capillary Formats.................................................................... 282 6.4.5 Applications of Etched Chemically Modified Capillaries................. 283 6.5 Conclusions.................................................................................................... 286 Acknowledgments................................................................................................... 287 References............................................................................................................... 287

6.1  Introduction 6.1.1 Background While some facets of silica hydride have been know for many years, its use as a separation medium did not begin until around 1990 [1,2]. However, significant advances in the technology and the development of the material did not begin until after 2000. 255

256

Advances in Chromatography: Volume 48

O O

Si

OH

O

O O

Si

Si

Si

O

O

Si

O

O

O

Si

O

Si

H

Si

H

O

O

Ordinary silica

H

O

O OH

Si O

O OH

O O

O

O

Silica hydride

Figure 6.1  Chemical surface structures of ordinary silica and silica hydride.

The chemical difference between ordinary silica and silica hydride is illustrated in Figure 6.1. The essential feature is that the silanol groups (Si-OH) on ordinary silica have been replaced by Si-H moieties on the hydride silica, which leads to profound differences in the surface properties of the two materials. Silica hydride has unique properties that can be exploited as part of the solid support of stationary phases for HPLC and on the inner walls of capillaries for electrophoretic separation media.

6.1.2  Synthesis of Silica Hydride There are two general approaches for the fabrication of silica hydride surfaces. The first involves the conversion of silanol groups on particulate silica with thionyl chloride to the Si-Cl function and then reducing this with lithium aluminum hydride, as shown in the following sequence [1]:

 ≡ Si-OH + SOCl2 → ≡ Si-Cl  +  SO2 + HCl



 ≡ Si-Cl + LiAlH4 → ≡ Si-H + LiAlH3Cl

This process must be done under relatively inert conditions, since the Si-Cl bond is hydrolytically unstable and easily reverts back to a silanol in the presence of any moisture. Another approach for creating a silica hydride surface utilizes the condensation reaction between silica and triethoxysilane. This method is illustrated by the following reaction [2]: Si OH + (OEt)3Si H



H+

OY Si O Si Η + nΕtΟΗ OY

Y = H or Si depending on the extent of cross-linking This is a single step process that is not sensitive to the presence of water, and in fact a small amount of acid is needed as a catalyst.

257

Silica Hydride—Chemistry and Applications

In both processes there is a number of reaction variables that control the exact nature of the hydride surface produced, i.e., relative number of Si-H groups vs. residual silanols. With the onset of commercialization of silica hydride materials for separation, the exact process used to fabricate each product is a proprietary piece of information as is the precise formulation of ordinary HPLC stationary phases by each individual manufacturer, even though most use some form of organosilanization.

6.2 Hydride-Based Stationary Phases for HPLC 6.2.1  Synthesis and Characterization Silica hydride HPLC stationary phases are based on the use of high purity/low metal content manufacturing technology of commercial silica. As depicted in Figure 6.1, the surface of the hydride material is predominantly populated (> 95% as determined in a study by 29Si CP-MAS NMR [2]) with nonpolar, silicon-hydride (Si-H) groups instead of the polar silanol groups (Si-OH) that dominate the surface of ordinary silica. Further modification of the hydride surface can be made using hydrosilation [3] that produces a bonded stationary phase with specific properties as a separation medium (hydrophobic, hydrophilic, ion-exchange, chiral, etc.): HYDROSILATION

Si H + CH2

CH

R

cat.

Si CH2

CH2

R

cat = catalyst, typically hexachloroplatinic acid or free radical initiator As shown above, one of the advantages to this process is the attachment of the bonded organic moiety to the surface by a stable Si-C bond. This feature leads to the high stability reported in chromatographic experiments [4–6]. While the most common approach for attaching an organic species to the silica hydride involves a terminal olefin, it is also possible to bond molecules with the olefin in a nonterminal position [7], alkynes [8], and other functional groups such as cyano [9]. This versatility in the attachment of organic moieties to the surface hydride leads to the possibility of producing stationary phases not feasible by other bonding methods. One interesting example is the double attachment of the bonded group that has been

O Si O

CH

CH2

R

H

Si

C

Si

C

O

R

O

Si O Structure I

O

  

Structure II

258

Advances in Chromatography: Volume 48

shown to occur when an alkyne is used in the hydrosilation reaction [8]. The two structures postulated on the basis of NMR studies are shown above.

6.2.2  Stability of Silica Hydride Materials The long-term viability of the hydride surface is a crucial aspect of this material. This is a reasonable concern because of the limited stability of small organosilanes in ­aqueous solutions. However, the Si-H moiety on a silica surface is in a completely different chemical environment and does not have the same properties as free silane molecules in solution since it is stabilized by the larger polymer matrix of the silica. A nonchromatographic example of the viability of Si-H has been demonstrated on silicon wafers where silicon hydride groups have been created chemically and then exposed to a variety of aqueous conditions including acid and basic environments [10]. The stability of the Si-H group on a silica surface has been demonstrated both spectroscopically and chromatographically. One test was a series of diffuse reflectance infrared Fourier transform (DRIFT) analyses of several archived silica hydride samples prepared on various commercial silica over a period of more than 10 years. No special precautions were taken since the samples were kept in laboratory drawers in screw-top containers. Therefore, the samples were essentially under atmospheric conditions with no temperature or humidity control. DRIFT spectra of several of these samples that had been prepared between six and eight years previously were obtained and compared to the original spectrum taken right after synthesis. The key feature in each spectrum is the Si-H stretching band at 2250 cm−1 [11,12]. In each case the intensity of this band in the archived sample was identical to the intensity obtained at the time of synthesis, proving that these materials have excellent stability when stored as an unprotected solid. Another test of the hydride stability was obtained by packing this material in a chromatographic column and pumping degassed DI water through it for several hours. The column was then unpacked and the DRIFT spectra before and after the water test were compared. The results of this experiment were that the intensities of the Si-H stretching peaks at 2250 cm−1 before and after the water test were essentially the same, indicating little or no decomposition of the hydride layer under these conditions. Further evidence for the stability of the hydride moiety on the silica surface involved a silica hydride material stored in columns for eight months in 0.05% phosphoric acid. Chromatograms run on one of these columns over an eight-month period showed no noticeable change in retention of the three test solutes, indicating good hydrolytic stability of the hydride under these mobile phase conditions. These results are reinforced by more extensive chromatographic studies on ­hydride-based stationary phases where columns have been used for thousands of ­column ­volumes at low pH (∼2) and high pH (9–10) with little evidence of ­deterioration [5,6]. In one test a bidentate C18 column (fabricated by hydrosilation of silica hydride with 1-octadecyne) was used for more than one thousand column volumes with a variety of samples and several organo/aqueous mobile phases. Subsequently the column was subjected to a 90:10 ammonium formate-ammonia (pH 10)/acetonitrile mobile phase for more than 1000 additional column volumes. The mixture uracil/pyridine/ phenol was periodically injected during the elution of this mobile phase. The k′ of pyridine was determined to be 0.3 + 0.05 while the k′ of phenol was 8.9 + 0.1 over

Silica Hydride—Chemistry and Applications

259

the 1000 column volume test. The DRIFT spectrum of the material from this column after the chromatographic testing was compared to the spectrum obtained immediately after synthesis. Both the carbon-hydrogen stretching bands between 2800 and 3000 cm−1 and the Si-H band at 2250 cm−1 are nearly identical in both spectra, confirming the results of the chromatographic tests that no significant decomposition of the bonded phase and the underlying surface occurred as a result of exposure to the various mobile phases.

6.2.3 Chromatographic Properties of Silica Hydride Phases An example of the fundamental difference between silica hydride and typical bare silica (where silanols are the predominant functional group on the surface) is demonstrated by the measurement of the retention of, pyridine, and phenol under reverse phase conditions. The polarity of ordinary silica results in no retention of uracil and phenol while pyridine is retained because of its basic properties. However, the much less polar silica hydride surface shows only marginal retention of pyridine (slightly separated from uracil) and more significant retention of phenol. The elution order is typical reversed-phase with the most polar compound eluting first and the least polar species being the last eluted. This silica hydride retention behavior can also be compared to a commercial C18 phase that is not endcapped. The presence of a significant number of silanols on the stationary phase results in elution of pyridine after phenol. It has been postulated that the nonpolar nature of the silica hydride surface results in less adsorption of polar solvents (particularly water) on the surface of the separation media. The consequence of this property is that aggressive components in the mobile phase (such as trifluoroacetic acid, phosphate, or bases) are less likely to attack the surface and the bonded moiety providing enhanced stability, and changes in the mobile phase composition can be accomplished efficiently so that the separation system rapidly reaches equilibrium. The latter property is of particular advantage when doing gradient separations because repeated analyses can be done with a minimum of time between runs. For aromatic and polyaromatic hydrocarbons on both C8 and C18 hydride-based columns made from the respective alkynes, it was possible to get reproducible retention for equilibration times after the end of the gradient of less than 5 minutes for all the solutes tested. In HPLC, retention properties are determined by the relative degree of interaction of the solute with the stationary phase and the mobile phase. Adsorption occurs at the silica surface and the bonded phase surface. Partitioning can take place in the solvent layer that forms at the surface of the stationary phase. Depending on the nature of the bonded phase and liquid phase surrounding it, the amount of partitioning will vary. With ordinary silica, water is strongly adsorbed onto the silica surface due to the active silanol groups. This water forms a stable hydration shell which often leads to chromatographic difficulties such as long equilibration times when the mobile phase composition is changed, lack of reproducibility in normal phase, and pH hysteresis. The Si-H groups found on the surface of silica hydride are not prone to such strong water retention as ordinary silica, making them more suitable for organic-normal phase (ONP) separations. The weaker water adsorption also accounts for the negligible or no hysteresis observed when changing pH. This

260

Advances in Chromatography: Volume 48

(a)

(b) 2

1

1.5 1

0.6

2

Norm.

0.8 mAU

1+2

0.4

1 0.5 0

0.2

–0.5

0 2

Min

4

2

Min

4

Figure 6.2  Separation of phenylglycine (1) and phenylalanine (2) on (a) silica hydride and (b) ordinary silica.

effect is in contrast to some bonded phases that often exhibit long-term memory effects. Silica hydride ­materials operate in the normal phase mode with solvent systems ranging from hexane/ethyl acetate all the way to water/acetonitrile. An example of the difference between ordinary silica and silica hydride using a water/acetonitrile mobile phase (aqueous normal phase) is shown in Figure 6.2 for the retention of phenylalanine and phenylglycine. In this example, retention increases on both columns as the amount of acetonitrile in the mobile phase is increased. However, resolution of the  components is only observed on the silica hydride column. Since the only difference between the columns is the hydride surface, the improved normal phase behavior must be the result of the Si-H moieties. Silica hydride-based columns have the unique capability to be used in any of the following three chromatographic modes: aqueous reverse phase (ARP); aqueous normal phase (ANP), defined below; and organic normal phase (ONP). When utilizing mobile phases with an aqueous component, methods having solvent compositions of water/acetonitrile or water/acetone that varies in the concentration of the organic constituent from 0% to between 50 and 70% will result in decreasing retention of hydrophobic analytes as the less polar solvent is increased. With a 100% aqueous mobile phase, retention is  at a maximum for neutral compounds. Under these conditions the hydride columns display typical reversed-phase behavior. What is unique is that when analyzing ionizable or other polar compounds and the acetonitrile or acetone concentration is above 50 to 70%, a second retention maximum occurs at 100% organic. Thus the solute behavior in this region is that the retention time decreases as the more polar solvent (water) increases, indicative of a normal phase mechanism. This section of the solute retention map (tR vs. % organic in the mobile phase) is given the designation aqueous normal phase. An example of such a retention map is shown in Figure 6.3. Therefore, for polar compounds the elution order and/or the retention times can be changed either by varying the pH

261

Retention time (min)

Silica Hydride—Chemistry and Applications 20 18 16 14 12 10 8 6 4 2 0

0

10

20

30 40 50 60 70 % Organic in mobile phase

80

90

100

Figure 6.3  Retention map (tR vs. % organic in the mobile phase) for a compound ­displaying both reversed-phase and aqueous normal phase retention on a silica hydride-based column.

(removing the charged state for ionizable compounds) or the organic concentration of the mobile phase. With the hydride columns in both the aqueous (using water) and organic (using nonpolar solvents) normal phase modes, the retention time of the analyte decreases as the more polar solvent is increased. The elution order is primarily based on the functionality or ionic state of the solutes. Since the maximum retention of the analytes is at 100% concentration of the least polar solvent, the behavior fits the definition of normal phase. For some basic compounds retention reaches a minimum at both high and low pH. This minimum is shifted to higher percent organic as the pH increases. At lower organic mobile phase compositions the column displays reverse phase behavior, while at higher organic percentages the column functions in the ANP mode. This behavior is in contrast to a typical commercial phase such as C18 or C8 that only displays reversed-phase characteristics and has no ANP retention at all. For acids at high pH, retention is at a minimum for low amounts of organic in the mobile phase, and then increases at about 70% and higher. In a few cases ANP behavior has been observed on hydride-based stationary phases with methanol as the organic solvent if the solute has multiple amine groups. The fundamental reversed-phase properties of silica hydride-based columns are similar to typical commercial stationary phases. The solutes are eluted in the usual reversed-phase order, i.e., from the most polar to the least polar at all pH values. The efficiency (typically around 100,000 plates/m) and peak symmetry (0.98 to 1.15) of these solutes on silica hydride stationary phases is also excellent. Therefore, when necessary, a silica hydride material bonded with an alkyl moiety such as C8 or C18 can be used for reversed-phase applications with separation capabilities similar to those of monomeric stationary phases. Some differences are found since the base material (silica hydride) is not the same as typical commercial phases (ordinary ­silica). These variations in selectivity have been documented in column equivalency tests [13]. The ability of a silica hydride bonded material to function in the normal phase is illustrated by the separation of a group of closely related phenols (Figure 6.4). Two examples of the separation of these four compounds are shown illustrating the

262 (a) 100

Advances in Chromatography: Volume 48 (b)

1.22 1.57 1.87

2.47

100

3.35

%

%

2.27

3.08 0 0.00

1.00

2.00 3.00 Time (min)

4.00

0 0.00

1.00

2.00 3.00 Time (min)

4.00

Figure 6.4  Normal phase separation of phenols on (a) a silica hydride column and (b) a bare silica column. Mobile phase: 10% diethyl ether in hexane. (Adapted from Pesek, J. J., Matyska, M. T., and Sharma, A., J. Liq. Chromatogr. & Rel. Technologies, 31, 134, 2008). (a) 100

(b) 100

1

2

0

3

4

%

3

%

2

1

4

5.00

10.00 15.00 Time (min)

20.00

0

5.00

10.00 15.00 Time (min)

20.00

Figure 6.5  Separation of carbohydrate structural isomer using a 100% aqueous mobile phase containing 0.5% formic acid. (a) first injection and (b) tenth injection.

versatility of the hydride-based stationary phases. In the first example (Figure 6.4a) the separation is accomplished on silica hydride material. The second case (Figure 6.4b) shows the same analysis on an ordinary bare silica column. As demonstrated, the silica hydride-based stationary phase provides adequate separation of the four phenols while the ordinary bare silica column does not. In contrast to what is common practice for many normal phase separations, the mobile phase solvents were not rigorously dried before use and were not placed in an air-tight or purged reservoir system to prevent adsorption of water. It is interesting that both bare silica hydride as well as a C18 modified phase (essentially a reversed-phase column) can provide good normal phase capabilities. A feature of the reversed-phase silica hydride materials is their ability to ­function in a 100% aqueous environment without any detrimental effects such as phase collapse. This problem has been encountered with many C18 phases and is evident by drastically reduced retention in comparison to mobile phases with a small amount of organic component (5–10%) Figure 6.5 shows an example of a separation for some carbohydrate structural isomers on an octadecyl modified silica hydride column using a 100% aqueous mobile phase containing 0.5% formic acid with MS detection. This analysis was repeated with 10 consecutive injections with no change (%RSD  100), an obliged choice related to the extensive time requirements (8 h) for data processing. 7.3.3.2 Spatial Order and Enhanced Sensitivity In recent research, Tranchida et al. exploited orderly structured 2D chromatograms and the high sensitivity of GC × GC for the elucidation of a sample-type of great scientific importance, namely methyl ester-derivatized plasma fatty acids (FAMEs) [61]. The analysis of FAMEs in human plasma is a well-known GC procedure: both apolar and polar stationary phases have been used in the conventional GC separation of plasma FAMEs, with a preference for the more polar capillaries (e.g., polyethylene glycol). In recent years, a series of studies related to human blood or plasma FAMEs

320

3.5

Advances in Chromatography: Volume 48

(×100,000) TIC

3.0 2.5 2.0 1.5 1.0 0.5 26.025 26.050 26.075 26.100 26.125 Spectrum comparison Spectrum1 #Data# data451.QGD R.Time:26.002(Scan#:26403) MassPeaks:82 BasePeak:147.80(1000) RawMode:Averaged 25.999-26.141(26400-26570) BG Mode:Averaged 26.173-26.178(26608-26614) 1000 148 121 82 43 67 91 41 105 133 30 40 50 60 70 80 90 100 110 120 130 140 Spectrum2 #Library# wiley229.lib Entry:26477 Formula:C10 H12 O CAS:140-67-0 MolWeight:148 RetIndex:0 MassPeaks:50 BasePeak:148.00(1000) Compname:Benzene, 1-Methoxy-4-(2-propenyl)-(CAS) p-Allylanisole $$ Anisole, p-allyl-$$ Methyl chavicol $$ 1-Allyl-4-methoxybenzene $$ 4-A 1000 148 39

77

51

65

30 40 50 60 70 Spectrum3 #Calcuration Result# MassPeaks:86 BasePeak:82.00(803) 1000 43 67 41 0 30

40

50

60

70

80

91 90

82

80

100

93

90

117

105

133 110

107

100

110

120

130

121

120

140

136

130

147

140

Figure 7.18  GC-qMS result for a single perfume peak. (From Mondello, L., Casilli, A., Tranchida, P.Q., Dugo, G., and Dugo, P. J. Chromatogr. A, 1067, 235, 2005. With permission.)

321

Multidimensional Gas Chromatography (×100,000) 6.0 TIC 5.5 52 54 53 5.0 4.5 4.0 3.5 3.0 2.5 2.0 55 1.5 1.0 0.5 0.0 26.025 26.050 26.075 26.100 26.125 26.000 Spectrum comparison Spectrum1 #Data# data450.QGD R.Time:26.088(Scan#:26507) MassPeaks:84 BasePeak:147.80(1000) RawMode:Averaged 26.085–26.091(26503–26510) BG Mode:Averaged 26.074-26.080 (26490-26497) 1000 148

41 30

40

77

51 50

65 60

70

80

91 90

105 100

110

121

120

133 130

140

150

Spectrum2 #Library# wiley229.lib Entry:26477 Formula:C10 H12 O CAS:140-67-0 MolWeight:148 RetIndex:0 MassPeaks:50 BasePeak:148.00(1000) CompName:Benzene. 1-methoxy-4-(2-propenyl)-(CAS) p-Allylanisole $$ Anisole, p-allyl-$$ Methl chavicol $$ 1-allyl-4-methoxybenzene $$ 4-A 1000 148

41

51

77

63

30 40 50 60 70 Spectrum3 #Calcuration Result# MassMeaks:79 BasePeak:77.00(124) 1000 40

0 30

40

51

50

65

60

80

77

70

80

91 90

91

90

105 100

110

105

100

110

117 120

121

120

133 130

140

133

130

150

146

140

150

Figure 7.19  GC × GC-qMS result for the peak illustrated in Figure 7.18. Peak identification: 52) estragole; 53) cis-dihydrocarvone; 54) γ-terpineol; 55) neo-dihydrocarveol. (From Mondello, L., Casilli, A., Tranchida, P. Q., Dugo, G., and Dugo, P., J. Chromatogr. A, 1067, 235, 2005. With permission.)

322

Advances in Chromatography: Volume 48

5.4 4.7

Sec

4.1 3.4 2.7 2.1 1.4 0.7 8.3 12.6 16.7 21.2 25.6 29.9 34.2 38.5 42.8 47.1 51.5 55.8 60.1 64.4 68.7 73.0 77.3 81.7 86.0 90.3 Min

Figure 7.20  (See color insert following page 248.) Headspace SPME-GC × GC-qMS result for Arabica coffee volatiles. (From Mondello, L., Tranchida, P. Q., Dugo, P., and Dugo, G., Mass Spectrom. Rev., 27, 101, 2008. With permission.)

have been described, reporting the presence of not more than 30 well known and widely reported FAs [62–66]. In general, the main drawbacks that can be encountered in the GC-FID and/or GC-MS analysis of fatty acid matrices are: (1) the difficulties related to the identification of double-bond positional isomers; for example, the mass spectra of C18:3ω3 and C18:3ω6 are very similar; (2) an additional less-known drawback, related to the separation process, is that the number of fatty acids contained in these sample-types often exceed the separation potential of a 30 m capillary column; and (3) a final problem is related to the limits of detection, because many FAMEs reach the detector at excessively low concentrations to be revealed. In [61] the authors demonstrated that these three disadvantages can be almost completely eliminated by using GC × GC. The 2D chromatogram relative to the GC × GC separation a plasma sample is shown in Figure 7.21. The fatty acids are located within a typical GC × GC band, generated by using an orthogonal column set. The FAMEs are grouped on the basis of their: • Carbon number (the CN14–24 zones are indicated by arrows) • Double-bond number (DB): seven distinct bands, grouping FAMEs in the DB0-6 range. • ω number: FAMEs with the same location of the last double bond are aligned along descending diagonal bands. Of the 65 identified compounds, 36 were identified by using pure standard compounds, while the remaining 29 FAMEs were identified through the highly ordered

323

Multidimensional Gas Chromatography 6.01 5.34 4.68

Sec

4.01 3.34 2.68 2.01 1.35 0.68

ω6 ω3

ω6 ω3

ω6 ω3 ω1 63 64 62 ω6 ω3 59 55 56 53 ω6 ω3 50 57 58 52 51 46 48 49 47 DB6 41 42 44 45 37 38 40 43 DB5 36 31 29 34 35 30 22 DB4 33 21 27 28 20 DB3 18 19 26 14 16 17 DB2 25 13 15 DB1 12 11 9 10 g 67 8 f DB0 IS e 5 d c 34 b 12a

0.02 4.4

C14 C15 7.54

10.67

C16 13.81

C17

C18

16.95

C19

20.08 Min

C20 23.22

C21 26.36

C22

65 60 61 54 39

23

C23 29.49

32

24

C24 32.63

Figure 7.21  (See color insert following page 248.) GC × GC-FID result for plasma FAMEs. Refer to Table 7.2 for peak identification. (From Tranchida, P. Q., Costa, R., Donato, P., Sciarrone, D., Ragonese, C., Dugo, P., Dugo, G., and Mondello, L., J. Sep. Sci., 31, 3347, 2008. With permission.)

structure of the 2D chromatogram (Table 7.2). A series of fatty acids were identified by considering the intersection points of the DB bands with the ω-number diagonals; an example of this simple procedure is shown in the C20 group 2D chromatogram, reported in Figure 7.22: peak 45 corresponds to C20:3ω6, because it is situated within the C20 group 2D space and is located at the intersection point of the DB3 band and the ω6 diagonal. With regards to the benefits of sensitivity enhancement, a series of rather unexpected odd-CN fatty acids (i.e., C11:0, C19:0, C21:0, C19:3, C21:4, C21:5, etc.) were determined at very low relative-% amounts. Furthermore, a (probable) homologous series of unassigned compounds (defined with the letters a, b, c, d, e, f, and g in Figure 7.21) appeared in the nonpolar zone of the space plane. The authors affirmed that these analytes appeared to be n-hydrocarbons with a 2CN difference. Finally, the advantages of the isolation of chemical bleed, especially in trace-amount analysis, were also highlighted: a series of descending streaks, corresponding to modulated 1D stationary-phase release, and separated from the plasma FAMEs, is evident in Figure 7.21. Without a doubt, the most typical and common GC × GC application is that related to fuel samples; apart from the possibility to increase the number of separated analytes, the 2D methodology generates highly structured chromatograms in this type of experiment. As an example, Figure 7.23 illustrates both an orthogonal

324

Advances in Chromatography: Volume 48

Table 7.2 Identification of GC × GC plasma FAMEs. St = Standard compound peak assignment Peak 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

FAME C8:0 C9:0 C10:0 (st) C11:0 (st) C12:0 (st) i-C14:0 C14:0 (st) i-C15:0 (st) a-C15:0 (st) C15:0 (st) i-C16:0 (st) C16:0 (st) i-C17:0 (st) a-C17:0 C17:0 (st) i-C18:0 C18:0 (st) a-C19:0 C19:0 C20:0 (st) C21:0 (st) C22:0 (st) C23:0 (st) C24:0 (st) C14:1ω5 (st) C16:1ω7 (st) C17:1ω7 (st) C18:1ω9 (st) C19:1 C20:1ω9 (st) C22:1ω9 (st) C24:1ω9 (st) C16:2ω6

2-D GC rel.% 0.015 0.015 0.009 0.019 0.097 0.003 1.124 0.019 0.021 0.169 0.035 20.785 0.059 0.085 0.194 0.019 5.326 0.007 0.011 0.036 0.002 0.058 0.021 0.046 0.092 2.511 0.134 18.349 0.036 0.116 0.017 0.058 0.025

Peak

FAME

34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

C17:2 C18:2ω6 (st) C20:2 C20:2ω6 (st) C22:2ω6 (st) C24:2ω6 C18:3ω6 (st) C18:3ω3 (st) C18:3 C19:3 C19:3ω6 C20:3ω6 (st) C20:3ω3 (st) C22:3ω6 C18:4ω3 C20:4ω6 (st) C20:4ω3 (st) C21:4 C22:4ω6 C22:4ω3 C24:4ω6 C20:5ω3 (st) C20:5ω1 C21:5 C22:5ω6 C22:5ω3 (st) C24:5ω3 C24:5 C20:6ω1 C22:6ω3 (st) C23:6 C24:6ω3

2-D GC rel.% 0.013 30.386 0.023 0.267 0.005 0.015 0.451 0.789 0.027 0.013 0.078 1.814 0.033 0.006 0.041 5.091 0.136 0.003 0.178 0.011 0.007 0.428 0.020 0.004 0.089 0.427 0.008 0.004 0.013 1.539 0.021 0.012

Source: Tranchida, P. Q., Costa, R., Donato, P., Sciarrone, D., Ragonese, C., Dugo, P., Dugo, G., and Mondello, L., J. Sep. Sci., 31, 3347, 2008. With permission.

325

Multidimensional Gas Chromatography 6.01 5.34 4.67 4.01

ω6

ω3

ω1

62 55 49

45

Sec

3.34

56 50 36

2.67

46 37 30

20

DB6 DB5 DB4 DB3 DB2 DB1 DB0

2.01 1.34 0.67 0.01 20.2

C20 20.72

21.23

21.75

22.26

22.78 Min

23.29

23.81

24.33

24.84

Figure 7.22  (See color insert following page 248.) Expansion relative to the 2D chromatogram shown in Figure 7.21, illustrating the C20 FAMEs group (see Table 7.2 for peak assignment). (From Tranchida, P. Q., Costa, R., Donato, P., Sciarrone, D., Ragonese, C., Dugo, P., Dugo, G., and Mondello, L., J. Sep. Sci., 31, 3347, 2008. With permission.)

(­apolar–­polar) and inversed (polar-medium polarity) GC × GC-FID analysis on diesel oil [67]. Using the orthogonal combination, the analytes are: (a) subjected to a boiling-point 1D separation with little inter-class distinction (e.g., there is no group separation between monoaromatics and alkanes); (b) subjected to a polaritybased 2D analysis, with satisfactory inter-group separation between diaromatics and monoaromatics, and partial border overlapping between the monoaromatics and the alkanes. The quality of intra-class separation is good for diaromatics and monoaromatics, while, as expected, the secondary polar column showed a low selectivity for the alkanes. Employing the inversed combination, the solutes are: (a) subjected to a polarity-based 1D separation, showing some degree of inter-group separation (e.g., there is partial group isolation between monoaromatics and alkanes); (b) subjected to a medium polarity-based 2D analysis, with good inter-group separation between all chemical classes. However, the quality of intra-class separation cannot be defined as satisfactory for any of the three chemical groups. Considering both applications, it can be concluded that both column combinations generate structured chromatograms; however, the orthogonal approach provided better inter-class separation, while the inversed set was to be preferred for inter-group resolution.

326

Advances in Chromatography: Volume 48

BP20,2tR (s)

7.5

5.0

Di-aromatics

2.5 Mono-aromatics

0.0

Alkanes 13

25

63 38 50 DB-1, 1tR (min)

88

Alkanes

5 BPX-35,2 tR (s)

75

4 3 2

Mono-aromatics

Di-aromatics

1 0

0

10

20

30 40 BP21, 1tR (min)

50

60

70

Figure 7.23  (See color insert following page 248.) GC × GC–FID chromatograms of diesel oil obtained on two different ­column sets, namely, apolar–polar (top) and polar–­medium polarity. (From Adahchour, M., Beens, J., Vreuls, R. J. J., Batenburg, A. M., and Brinkman UATh, J. Chromatogr. A, 1054, 47, 2004. With permission.)

References

1. Golay MJE. Gas chromatography. Academic Press, New York, 1958. 2. James AT, Martin AJP, Biochem. J. 50, 679, 1952. 3. Giddings JC in: Multidimensional Chromatography: Techniques and Applications, H.J. Cortes (Ed.), Marcel Dekker, New York, 1990. 4. Berger TA, Cromatographia 42, 63, 1996. 5. Poole CF, Poole SK, J. Chromatogr. A 1184, 254, 2008. 6. Bertsch W, J. High Resol. Chromatogr. 22, 647, 1999. 7. Schomberg G, J. Chromatogr. A 703, 309, 1995. 8. Consden R, Gordon AH, Martin AJP, Biochem. J. 38, 224, 1944. 9. Liu Z, Phillips JB, J. Chromatogr. Sci. 29, 227, 1991. 10. Boer H, van Arkel P, Chromatographia 4, 300, 1971. 11. Kinghorn RM, Marriott P, Cumbers M, J. High Resol. Chromatogr. 19, 622, 1996. 12. Mondello L, Catalfamo M, Proteggente AR, Bonaccorsi I, Dugo G, J. Agric. Food Chem. 46, 54, 1998. 13. Mondello L, Catalfamo M, Cotroneo A, Dugo G, J. High Resol. Chromatogr. 22, 350, 1999.

Multidimensional Gas Chromatography

327

14. Mosandl A, Bruche G, Askari C, Schmarr H-G, J. High Resol. Chromatogr. 13, 660, 1990. 15. Simmons MC, Synder LR, Anal. Chem. 30, 32, 1958. 16. McEwen DJ, Anal. Chem. 36, 279, 1964. 17. Deans RR, Chromatographia 1, 18, 1968. 18. Jennings W, J. Chromatogr. Sci., 22, 129, 1984. 19. Fenimore DC, Freeman RR, Loy PR, Anal. Chem. 45, 2331, 1973. 20. Sciarrone D, Tranchida PQ, Ragonese C, Schipilliti L, Mondello L. Submitted to J. Sep. Sci. 21. Mondello L, Casilli A, Tranchida PQ, Sciarrone D, Dugo P, Dugo G, LC GC Eur. 21, 130, 2008. 22. Quimby B, McCurry J, Norman W, LC GC Eur. 25, 174, 2007. 23. Luong J, Gras R, Yang G, Cortes H, Mustacich R, J. Sep. Sci. 31, 3385, 2008. 24. Luong J, Gras R, Yang G, Cortes H, Mustacich R, J. Chromatogr. Sci. 44, 253, 2006. 25. Kaiser RE, Leming L, Blomberg L, Reider RI, HRC & CC. 8, 92, 1985. 26. Veriotti T, McGuigan M, Sacks R, Anal. Chem. 73, 279, 2001. 27. Veriotti T, Sacks R, Anal. Chem. 73, 3045, 2001. 28. Veriotti T, Sacks R, Anal. Chem. 75, 4211, 2003. 29. Schoenmakers P, Marriott P, Beens J, LC-GC Eur., 16, 335, 2003. 30. Liu Z, Sirimanne SR, Patterson Jr DG, Needham LL, Phillips JB, Anal. Chem. 66, 3086, 1994. 31. Kinghorn RM, Marriott PJ, J. High Resol. Chromatogr. 22, 235, 1999. 32. Phillips JB, Gaines RB, Blomberg J, van der Wielen FWM, Dimandja J-M, Green V, Granger J, Patterson D, Racovalis L, de Geus H-J, de Boer J, Haglund P, Lipsky J, Sinha V, Ledford Jr EB, J. High Resol. Chromatogr. 22, 3, 1999. 33. Mondello L, Lewis AC, Bartle KD in: Multidimensional chromatography, John Wiley & Sons Limited, West Sussex, 2002. 34. Seeley JV, Kramp F, Hicks CJ, Anal. Chem. 72, 4346, 2000. 35. Bueno Jr. PA, Seeley JV, J. Chromatogr. A 1027, 3, 2004. 36. Seeley JV, Seeley KS, Libby EK, Breitbach ZS, Armstrong DW, Anal. Bioanal. Chem. 390, 323, 2008. 37. Seeley JV, Micyus NJ, McCurry JD, Seeley SK, Am. Lab. 38, 24, 2006. 38. Quimby B, McCurry J, Norman W, LC-GC 25, 174, 2007. 39. Libardoni M, Waite JH, Sacks R, Anal. Chem. 77, 2786, 2005. 40. Sandra P, David F, Klee MS, Blumberg LM, Proceedings of the Dalian International Symposia and Exhibition on Chromatography, June 4–7, 2007, Dalian, China. 41. Marriott P, Dunn M, Shellie R, Morrison P, Anal. Chem. 75, 5532, 2003. 42. Shellie R, Marriott P, Morrison P, Mondello L, J. Sep. Sci. 27, 504, 2004. 43. Beens J, Janssen H-G, Adahchour M, Brinkman UATh, J. Chromatogr. A 1086, 141, 2005. 44. Tranchida PQ, Casilli A, Dugo P, Dugo G, Mondello L, Anal. Chem. 79, 2266, 2007. 45. Oldridge N, Panic O, Górecki L, J. Sep. Sci. 31, 3375, 2008. 46. von Mühlen C, Khummueng W, Alcarez Zini C, Bastos Caramão E, Marriott PJ, J. Sep. Sci. 29, 1909, 2006. 47. Phillips JB, Xu J, J. Chromatogr. A 703, 327, 1995. 48. Frysinger GS, Gaines RB, J. High Resol. Chromatogr. 22, 251, 1999. 49. van Deursen M, Beens J, Janssen H-G, Leclercq PA, Cramers CA, J. Chromatogr. A 878, 205, 2000. 50. van Deursen M, Beens J, Reijenga J, Lipman P, Cramers C, Blomberg J, J. High Resol. Chromatogr. 23, 507, 2000. 51. Mondello L, Tranchida PQ, Dugo P, Dugo G, Mass Spectrom. Rev. 27, 101, 2008.

328

Advances in Chromatography: Volume 48

52. Adahchour M, Beens J, Vreuls RJJ, Brinkman UATh, Trends Anal. Chem. 25, 438, 2006. 53. Adahchour M, Beens J, Vreuls RJJ, Brinkman UATh, Trends Anal. Chem. 25, 540, 2006. 54. Adahchour M, Beens J, Vreuls RJJ, Brinkman UATh, Trends Anal. Chem. 25, 726, 2006. 55. Adahchour M, Beens J, Vreuls RJJ, Brinkman UATh, Trends Anal. Chem. 25, 821, 2006. 56. Dallüge J, van Stee LLP, Xu X, Williams J, Beens J, Vreuls RJJ, Brinkman UATh, J Chromatogr A 974, 169, 2002. 57. Shellie R, Marriott PJ, Anal. Chem. 74, 5426, 2002. 58. Shellie RA, Marriott PJ, Huie CW, J Sep. Sci. 26, 1185, 2003. 59. Mondello L, Casilli A, Tranchida PQ, Dugo G, Dugo P, J. Chromatogr. A 1067, 235, 2005. 60. Ryan D, Shellie R, Tranchida P, Casilli A, Mondello L, Marriott P, J. Chromatogr. A 1054, 57, 2004. 61. Tranchida PQ, Costa R, Donato P, Sciarrone D, Ragonese C, Dugo P, Dugo G, Mondello L, J Sep. Sci. 31, 3347, 2008. 62. Marangoni F, Colombo C, Galli C, Anal. Biochem. 326, 267, 2004. 63. Ohta A, Mayo MC, Kramer N, Lands WEM, Lipids 11, 742, 1990. 64. Rodríguez-Palmero M, López-Sabater MC, Castellote-Bargallo AI, De La Torre-Boronat MC, Rivero-Urgell M, J. Chromatogr. A 778, 435, 1997. 65. Akoto L, Vreuls RJJ, Irth H, Pel R, Stellaard F, J. Chromatogr. A 1186, 365, 2008. 66. Tvrzická E, Vecka M, Stanˇková B, Žák A, Anal. Chim. Acta 465, 337, 2002. 67. Adahchour M, Beens J, Vreuls RJJ, Batenburg AM, Brinkman UATh, J. Chromatogr. A 1054, 47, 2004.

Preparation 8 Sample for Chromatographic Analysis of Environmental Samples Tuulia Hyötyläinen Contents 8.1 Introduction................................................................................................... 329 8.2 Sample Preparation Techniques.................................................................... 331 8.2.1 Drying and Homogenization of Solid Samples................................. 333 8.2.2 Extraction........................................................................................... 333 8.2.2.1 Vapor-Phase Extraction....................................................... 334 8.2.2.2 Liquid Samples................................................................... 335 8.2.2.3 Solid and Semisolid Samples..............................................344 8.3 Cleanup of Extracts....................................................................................... 350 8.3.1 Lipid Removal from Biological Extracts........................................... 350 8.3.2 Sulfur Removal from Sediment Extracts........................................... 351 8.3.3 Fractionation...................................................................................... 351 8.3.4 Derivatization.................................................................................... 353 8.3.5 Online Techniques............................................................................. 355 8.3.6 Selection of Sample Preparation Method Methods........................... 356 8.4 Conclusions and Future Perspectives............................................................ 366 Abbreviations.......................................................................................................... 367 References............................................................................................................... 367

8.1  Introduction Determination of the chemical composition of complex environmental samples is a challenging task, owing to myriad species of compounds, many of them present in only trace amounts. Typically, chromatographic techniques are utilized in the analysis of complex environmental samples. However, most samples cannot be injected directly into the chromatographic system without sample preparation. It is worth stressing that the sample preparation step required before the chromatographic separation largely determines the quality of obtained results. The sample preparation procedure also significantly impacts assay throughput, data quality, and analysis 329

330

Advances in Chromatography: Volume 48

cost. It should also be noted that sampling and sample preparation steps typically account for over 80% of the total analysis time. Therefore, selecting and optimizing an appropriate sample preparation method is essential for successful method development. One of the major problems with environmental samples is that the conventional sample preparation techniques are seldom sufficient in terms of speed, sensitivity, selectivity, and reliability. Several interlaboratory studies have shown that even established laboratories applying well validated methods frequently produce inconsistent data, for example for brominated flame retardants (BFRs), polyfluorinated chemicals (PFCs), and other organic compounds.1 In some cases, interlaboratory results will reveal a relation between the use of nonoptimal techniques or methods and poor results. In many other cases, however, laboratories suffer from multiple difficulties which hinder clear identification of the error source. Issues needing to be addressed are often related to poor sample preparation methods, such as inefficient extraction methods, matrix effects, and the cleanup steps needed to remove them.1 In short, obtaining reliable data depends upon systematic development of techniques and methods, particularly for the sample preparation part. Moreover, sample preparation methods with lower consumption of toxic organic solvents to minimize the generation of hazardous residues and health risks for operators are needed.2,3,4 Sample processing and pretreatment can take a number forms depending on the nature of the sample.5 Typical processes may include sample filtration, centrifugation, distillation, dilution, target amplification, and extraction. Successful execution of these processes is required to ensure that the analyte is present in a form compatible with the analytical system. The most classical sample preparation techniques rely on extraction with solvents, including traditional techniques such as liquid–liquid extraction (LLE) and Soxhlet extraction. The broad polarity range of solvents and general applicability made these techniques popular. However, from an environmental point of view, the use of large amounts of organic and often chlorinated solvents is unfavorable. Furthermore, they often require complex time-consuming multistep procedures which can lead to low accuracy, contamination, and losses of analytes. Various other techniques, such as solid-phase extraction (SPE) and pressurized liquid extraction (PLE), have been developed to replace these traditional sample pretreatment techniques. In addition, solid-phase microextraction (SPME), membrane-based techniques (dialysis, ultrafiltration, supported liquid membrane extraction (SLM)), and other miniaturized extraction techniques have been developed to overcome the problems of traditional methods. At present, the sample pretreatment is often the weakest and most time-consuming part in the analytical procedure. Thus, much effort has been put into exploring the possibilities of miniaturization and automation of the extraction procedures to minimize or eliminate the limitations of the sample preparation. No universal sample preparation technique suitable for all types of sample exists. The sample preparation required is dependent on the nature of analytes, matrix, and final separation method. Naturally, the sample preparation must be tailored to the final analysis. The sample matrix and the type and amount of analytes in the sample are of primary importance. Moreover, a method good for target-compound analysis

Sample Preparation for Chromatographic Analysis of Environmental Samples 331

may not be good for comprehensive chemical profiling of samples. Selectivity of the sample preparation is often a key factor for target-compound analysis while an exhaustive extraction is the better choice for profiling. In the selection of a sample preparation technique, not only the effectiveness needs to be considered but many other factors that affect the analytical scheme.6 The major factors are cost of the equipment, operating costs, complexity of method development, amount of organic solvent required, and level of automation. In addition, the number of samples to be analyzed is also of importance—the question is whether the planned procedure will be unique or whether it will be used in carrying out routine analysis. In the latter case, the techniques facilitating automation and low cost per analysis are preferred. Sample preparation methods are required for large number of different analytes and matrices. Levels of persistent organic pollutants (POPs), radionuclides, and toxic metals are extensively measured in various compartments of the environment, such as water, soil, sediment, air, and biota. POPs include industrial compounds and flame retardants such as polychlorinated biphenyls (PCBs); polychlorinated naphthalenes (PCNs); polybrominated diphenyl ethers (PBDEs); polybrominated biphenyls (PBBs); industrial by-products such as polychlorinated dibenzo-p-­dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs); and organochlorine (OC) pesticides such as dichlorodiphenyltrichloroethane (DDT), hexachlorobenzene (HCB), and hexachlorocyclohexane (HCH). More recently, emphasis has been put on the determination of levels of emerging contaminants, such as surfactants, human and veterinary drugs, fragrances, antiseptics, new brominated/chlorinated flame retardants (beyond PBDEs), sunscreens/UV filters, contaminant dibutylphthalates (DBPs), benzotriazoles, naphthenic acids, perfluorinated surfactants (including perfluorooctanoic sulfates (PFOS) and perfluorooctanoic acid (PFOA)), algal toxins, perchlorate, pesticide degradation products, chiral contaminants, and microorganisms.7,8 Due to the large number of different compounds and matrices, the development of sample preparation schemes is challenging. Recently, much effort has gone into the development of more efficient extractions that could replace the conventional methods, which typically are laborious and time-consuming multistep procedures, requiring much manual handling of the extracts.

8.2 Sample preparation techniques Sample preparation includes several steps, of which the most time- and labor­consuming part is the extraction and further cleanup of the extracts. Different methods are required for different types of samples, as shown in Figure 8.1. Liquid samples are generally easier to handle than solid samples which require exhaustive methods. The first step of sample preparation for solid and semisolid samples is drying and homogenization. For liquid sample, simple filtration is often sufficient. Then, the target analytes are extracted from the sample (solid or liquid), and the extract is usually purified, fractionated, and concentrated before the final analysis, which is typically performed with gas or liquid chromatography. The extraction procedure is dependent on the sample matrix; different methods are used for sediment, soil, plant, tissue, and liquid samples. After extraction, it will usually be necessary to purify and fractionate the extract, because most extraction methods are insufficiently selective and the

332

Advances in Chromatography: Volume 48 Sample Solid

Particle size reduction required?

YES Hard sample

NO

NO

Liquid

Semi-soft, soft sample

Mill, grind, crush Chop, macerate, or pulverize cut or mince

Is the sample homogeneous?

Homogenize sample

Is the whole sample of interest?

NO

Selective extraction by LLE SPE SPME SBSE LPME MASE

YES

YES Take a sample aliquot

Is the whole sample of interest?

NO

NO

YES Sample dissolution by digestion organic solvent aqueus solvent

Column chromatography SPE filtration dialysis

Are analytes volatile?

PLE SLE Soxhlet SAE MAE SFE

Does sample/ extract need purification

YES

YES Headspace SPME SBSE MESI

Is sample/extract suitable for direct analysis?

YES

NO

NO Are analytes thermally/chemically unstable YES

NO

Analyse

Derivatize

Figure 8.1  Typical sample preparation schemes used in environmental analysis.

separation power of the analytical technique not good enough. Extracts ­typically contain several analytes similar to the flame retardants, which may be present in much higher quantities. The fractionation procedures are similar for the different types of extracts.

Sample Preparation for Chromatographic Analysis of Environmental Samples 333

During recent years, many modified, innovated sample preparation ­methods have been developed. This review focuses on the advanced and promising methods and highlights the new trends in developing the methods for sample preparation.

8.2.1 Drying and Homogenization of Solid Samples Drying of solid samples, such as soil, sediment, and sewage sludge, is usually the first step in the analysis. Dry samples are more effectively homogenized, allowing accurate subsampling for parallel analyses for other determinants. In addition, the absence of water in the samples makes the sample matrix more accessible to organic solvents. Because some of the POPs are relatively volatile, both losses and uptake of compounds from air can occur if the drying is done at room temperature or in a heated oven ( 100 ml low low

low

low

Extraction time Solvent

Selectivity Instrumentation cost Level of automation Operator skill

5min–12 h organic, > 50 ml low low

polarnonpolar, semivolatilenonvolatile

Analyte type

solid/ liquid polarnonpolar, volatilenonvolatile

solid

Sample type

1–10 g

10 g

LSE/LLE

Sample size (g)

Soxhlet

moderate

high

5–40 min organic, < 50 ml low high

polarnonpolar, volatilenonvolatile

solid

5–50 g

PLE

Table 8.13 Relevant Figures of Merit of Extraction Methods

moderate

moderate

3–40 min organic, < 50 ml low moderate

solid or liquid polarnonpolar, volatilenonvolatile

1–30 g

SAE

moderate

moderate

5–40 min organic, < 50 ml low moderate

polarnonpolar, volatilenonvolatile

solid

1–10 g

MAE

high

high

high high

relatively polarnonpolar, volatilenonvolatile 20–60 min 0–10 ml

solid

1–10 g

SFE

moderate

low

High low

2–20 min 1–5 ml

solid, semisolid polarnonpolar, volatilenonvolatile

0.5–10 g

MSPD

moderate

high

high low

A few min. 1–5 ml

0.1–1000 ml liquid, gaseous polarnonpolar, volatilenonvolatile

SPE

moderate

high/low

high low

10–120 min 0.1–0–5 ml

0.1–1000 µ l liquid, gaseous polarnonpolar, volatilenonvolatile

LPME/ MASE

low

moderate

high low

30–840 min no solvent

liquid, gaseous nonpolar; volatile to semivolatile

1–1000 ml

SPME/SBSE

360 Advances in Chromatography: Volume 48

1

4

5

23

5

3

4

7

5 7 6

6

5

8

10

10

IS

15

11 14 10

21

23

20

27

Min

25

25

25 29 22 28 3233

30

30

38 40 42

43

15

20

Min

31 36 9 11 1516 24 34 12141718 20 26 30 3537 3940 41 43 19 10 13

9

IS

40

35

40

44 46 48

35

51

46 47 44 45 50 49 51

(i)

(c)

(iii)

C11 hydrocarbons

40.5%

33.2%

13.5%

Other compounds

Sulfur compounds

8.5%

58.8%

26.3%

(ii)

7.5%

Sesquiterpenes

19.4%

60.8%

31.7%

Figure 8.7  Comparison of different extraction methods in the determination of volatile metabolites of the brown alga Dictyopteris membranacea. Total ion chromatogram of the volatile fraction of D. membranacea obtained by (a) HD and SFE, and (b) FMAHD (IS = internal standard). (c) Comparison of the main chemical classes of compounds identified in the different volatile fractions obtained from D. membranacea by: (i) HD, (ii) SFE, and (iii) FMAHD. (From Hattab, M. E., Culioli, G., Piovetti, L., Chitour, S. E., and Valls, R., J. Chromatogr. A, 1143, 1, 2007. With permission.)

0

(b)

0

(a)

Sample Preparation for Chromatographic Analysis of Environmental Samples 361

362

Advances in Chromatography: Volume 48

acids). The HD and the fMAE extracts contained a large amount of compounds with higher volatility. Interestingly, sesquiterpene compounds were found in the extract obtained with fMAE, while these compounds were absent in the volatile fractions from HD and SFE. It is clear that sesquiterpenes, which have a relatively low volatility, were steam distilled under the effect of microwaves energy. This study showed clearly that the choice of the extraction technique could induce dramatically the preferential obtaining of a given chemical class of compounds. In selection of extraction method for liquid samples, the best options are SPE, SPME, SBSE, or LPME techniques. SPE techniques, including miniaturized SPE methods such as MEPS and pipettetip SPE, are adsorptive extraction techniques and they give quantitative recoveries, unlike the abovementioned sorptive extraction techniques, and are therefore well suited to trace analysis. A wide range of SPE materials are available, and not only nonpolar but also polar analytes can be extracted efficiently. These SPE techniques tend to be less selective, and matrix components are often extracted as well. On the other hand, the rather low selectivity is an advantage for profiling type of analysis, i.e., when all or most of the sample components are of interest. Pipette-tip SPE is simple to use but it is performed manually, and is user-dependent. The optimization is easy and flexible, only small volumes of solvents are needed, and as new pipettetip can be used for each extraction, there are no problems with memory effects. MEPS can be employed as an automated online technique. The main advantage of the SPE methods over SPME, SBSE, and LPME is that because the extraction is typically quantitative, the calibration is straightforward. This is in contrast to the other techniques, where various calibration methods are employed, including classical calibration relying on equilibrium extraction or more novel kinetic calibration. For the extraction of volatile and semivolatile compounds, SPME and SBSE are highly useful techniques. The instrumentation of these two techniques is commercially available, and particularly SPME has been widely applied to the analysis of several types of volatile compounds in combination with GC analysis. SPME instrumentation is simpler, as the injection in SBSE requires a special interface for thermal desorption. However, the EE is clearly better in SBSE than in SPME, and SBSE thus more suitable for trace analysis. Both techniques are solvent-free and easy to use, and the EFs are typically high. In addition, SPME and SBSE devices are easily stored and transported, and they can be applied even in on-site sampling. Field sampling is effective because only the fiber or stir-bar with the absorbed analytes needs to be brought back to the laboratory. Transportation of large sample volumes is avoided, and no sampling accessories such as pumps or filters (as needed in on-site SPE) are required. SBSE is the more rugged system for on-site sampling because SPME fibers are quite fragile. In addition, if trace amounts are to be extracted, SBSE gives higher recoveries and thus better sensitivities. Both techniques are very useful for samples where the sample volume is limited, as for example, in the determination of the composition of pore water in marine sediments.132 Although derivatization can improve the extraction of polar analytes, both SPME and SBSE are best suited to the extraction of relatively nonpolar and reasonably volatile analytes.

Sample Preparation for Chromatographic Analysis of Environmental Samples 363

The LPME techniques are not yet as widely applied as SPE, SPME, or SBSE. Of the LPME techniques, membrane assisted systems are slightly more complex systems, but the ruggedness of these systems is clearly better. The membrane-assisted extraction techniques, which apply the same extraction mechanism as LPME, are best suited for semi-volatile to (relatively) nonvolatile analytes. The choice of solvent is less critical than in LPME. The repeatability is typically better than in LPME because there are no problems with drop stability as in LPME, and larger volumes can be extracted. Static membrane-assisted extraction, with either hollow fibers or membrane bags, is inexpensive and simple to use, and since the membranes are typically for single use, there are no problems with cross contamination. The flat-sheet modules are best suited to dynamic extraction of larger sample volumes and to online connection with GC. In the flat-sheet modules the membranes are normally used for several extractions, and careful cleaning between extractions is required to minimize the risk of cross contamination. In a recent study, SMBSE and membrane assisted LLE were compared in the extraction of PAHs, PCBs, PBDEs, PBBs, PEs, NPs, and NPEOs in water samples.133 SBSE was applied with TD-GC–MS and MASE was combined with LVI-PTV-GC–MS. In general, the SBSE method provided better sensitivity, whereas MASE resulted in similar recoveries, but faster extraction. With SBSE, extraction time was 12 hours, and with MASE one hour. Several studies have shown that similar sensitivities are obtained with SPE and SPME or SBSE.134,135 The latter two techniques are particularly suitable for the analysis of relatively nonpolar and volatile analytes which can be analyzed with GC. Since SPME and SBSE are more selective methods than SPE and can be performed in solvent-free mode, the enrichment is clearly higher and well compensates the lower recoveries of the less exhaustive SPME and SBSE. In some cases, higher recoveries can be obtained with SBSE than with SPE, as shown in a method developed for the extraction of PAHs in aquatic samples, where the recovery was improved with use of SBSE.135 In particular, recoveries of the more hydrophobic PAHs (log KOW>5) were noticeably higher with SBSE than with SPE. A further benefit of SBSE is that it is easy to apply as it is nearly solvent-free, and no restriction or cleanup procedures are necessary. In a study where SPME and SBSE were compared for the extraction of PAHs and organochlorine compounds in water and GC-MS was used for analysis, SBSE was found to be more robust and to enable higher recoveries (20.1–97.2%) than SPME (recoveries of 6.3–51.6%).136 Thus, lower detection limits (0.05 and 1.0 ng/L) were obtained for SBSE than for SPME (0.1 - 4.5 ng/L). Also, MAE and PLE can be used for the extraction of liquid samples. In a recent study SPE, PLE, and MAE were compared with LLE in the extraction of organic contaminants (PCBs, OCPs, and PBDEs) from blood matrices.137 Two different MAE techniques, namely, cavity-dispersed MAE and focused microwave-assisted (FME) extractions were applied. Figure 8.8 shows the comparison of the tested methods. Of the tested methods FME method was found to be the most reliable, with ­highest IS recovery and low to moderate variability in the results. Also the precision of the method was generally better than other methods. All methods other than FME presented quantification problems for PBDEs. The highly reproducible concentrated microwave energy from this method is likely the cause of its optimal performance. LLE gave the poorest efficiency, precision, and accuracy of the techniques studied.

364

Advances in Chromatography: Volume 48

Mass fraction (pg/g wet mass)

14000 13000 12000 11000 10000 1200

Certified LLE MAE FME PFE SPE

1000 800 600 400

B 13 15 8 3+ 13 PC 2 PC B 1 7 B 18 0 0+ 19 PC 3 B 18 7 4, 4´ -D D PB E D E 4 PB 7 D E PB 99 D E 10 0

11 8

B

PC

B

PC

PC

PC

B

0

99

200

Figure 8.8  Comparison of extraction efficiencies of five different extraction techniques, in the extraction of PCBs in certified serum (SRM 1589a). (From Keller, J. M., Swarthout, R.  F., Carlson, B. K. R., Yordy, J., Guichard, A., Schantz, M., and Kucklick, J.  R.,  Anal. Bioanal. Chem., 393, 747, 2009. With permission.)

For the extraction of solid, environmental samples, the solvent extraction techniques utilizing high temperatures or pressures, microwaves or sonication are typically a better option than SLE or Soxhlet. Of these techniques, PLE and MAE are more exhaustive in nature, and thus, suitable for samples in which the target analytes are tightly bound to the matrix, i.e., sediments and soils. SAE, on the other hand, is well suited for the extraction of less complex samples, such as aerosol particles collected on filters. In further comparison of PLE, MAE, and SAE, the latter in static mode is simple and fast, but it is labor-intensive and requires a skilled operator to obtain reproducible data. DSAE avoids many of the problems of the static mode. MAE and PLE, on the other hand, offer various advantages and disadvantages. While MAE is capable of extracting multiple samples simultaneously in a short time, additional cleanup is required to remove the sample matrix from the analyte-containing solvent, after cooling of the sample vessels. Like DSAE, MAE can also be done in dynamic mode. PLE allows multiple samples to be extracted sequentially in an automated system, but the instrumentation is relatively expensive. Several studies have compared the performance of these new extraction systems for the extraction of solid samples.131,138–141 Results have been slightly different, depending on the study. However, in most cases, under optimized conditions, PLE, MAE, SAE, and SFE give similar or better recoveries than conventional Soxhlet methods. In a recent study, for example, the performances of MAE, SAE, Soxhlet, and PLE were compared for the extraction of PAHs from marine sediment and sewage sludge samples.138 The results are summarized in Table 8.14. In terms of extraction efficiency, extraction time, and amount of solvent required, PLE was superior over the

Sample Preparation for Chromatographic Analysis of Environmental Samples 365

Table 8.14 Comparison of PLE, MAE, and SAE with Soxhlet in the Extraction of PAHs in Sludge Sample Efficiency Time Solvent volume

Soxhlet

PLE

fMAE

SAE**

100 * 180 min 250 ml

80–156% 20 min 20 ml

54–79% 10 + 10 min 30 ml

52–87% 45 + 10 min 60 ml

Source: From Thordarson, E., Jönsson, J. Å., and Emnéus, J., Anal. Chem., 72, 5280, 2000. *Results compared with the value obtained with Soxhlet; ** method not optimized

other techniques. In the comparison it must be pointed out that, in this particular study, sonication was used without any optimization of the experimental conditions. Concentration and cleanup steps were the same for all the techniques. Similar selectivities have been obtained for all the techniques employed, with numerous interfering compounds remaining despite the cleanup step. In another recent study, SAE, MAE, and PLE were compared in the extraction of different BFRs (TBBPA, HBCD congeners, and deca-BDE).140 Almost complete extraction of TBBPA and HBCD was achieved using MAE and PLE and particularly for deca-BDE, the pressurized conditions of PLE gave by far the highest extraction yields. SAE gave the lowest recoveries, particularly for deca-BDE. It was also noticed that mixed polar/nonpolar solvent systems such as isopropanol/n-hexane allow higher extraction rates than polar mixtures such as methanol/isopropanol alone. In another study, two validated European standard methods, based on LSE and LLE, were compared with PLE and SAE in the extraction of pesticides in soil.141 The results showed that with SAE was successful to recover all the selected substances with a good repeatability; however, the extraction efficiency was lower (57.0%) than with PLE (median recovery of 68.3.5%) and the two standard techniques (median recoveries of 72.7% and 65.7%). SAE has been shown to enable efficient extraction of PAHs from biological marine samples. With the optimized ultrasonic extraction procedure, aromatic hydrocarbons from NIST-2977 were extracted with recoveries higher than 80% for most analytes.142 Similarly, for PCBs and organochlorine pesticides in sediments, good recoveries have been obtained with MAE, PLE, and SAE.143 Good recoveries have also been achieved with MAE for simultaneous extraction of PAHs, PCBs, phthalate esters, and nonylphenols in sediments.144 SFE has not been used on a routine basis in the extraction of POPs in laboratory analysis, owing to the high cost of the instruments and the need to optimize a large number of operating parameters for each matrix. However, the greatest advantage of SFE in the analysis of complex biological and environmental samples is the possibility of obtaining highly selective extractions and relatively pure and preconcentrated extracts. SFE has further been compared with Soxhlet extraction for the determination of PCBs and PCDDs in sediment.92 The study showed that concentrations of PCBs obtained by SFE were very similar to those of Soxhlet: agreement was good for 35 congeners out of 38. In another study of SFE where the extraction of PCDD/PCDFs was of interest,

366

Advances in Chromatography: Volume 48

recoveries for lower chlorinated compounds were satisfactory in comparison with Soxhlet while the recoveries of highly chlorinated PCDD/PCDFs were lower.92 Evidently the highly chlorinated congeners are more tightly bound to sediment than the lower ones. A clear advantage of the SFE method was that the SFE method was much faster than the Soxhlet. SFE combined with an extra cleanup on alumina takes only 1.5 h, whereas Soxhlet takes 18 h plus several days of cleanup on different columns. Compared with the other extraction methods for solid samples, MSPD is different in several respects. It is a manual technique, which combines sample disruption and dispersal of the sample onto particles of a solid phase support. Elution is also usually accomplished manually, although instrumental techniques such as PLE can be utilized for elution as well. The MSPD technique is well suited particularly for biological samples and plant-derived samples. The main advantage of the technique is the possibility to use a very small sample sizes and also the solvent consumption is low.

8.7  Conclusions and future perspectives Sample preparation is still the most critical and time-consuming step within the overall analytical procedure needed to obtain accurate determination of POPs in environmental biota samples. Due to the peculiarities of these samples, selection of the appropriate techniques to extract, purify, and preconcentrate the analytes, along with a careful optimization of the corresponding operational parameters, are critical. New analytical tools are continually being developed both for sample preparation and final analysis. Powerful and relatively fast extraction techniques are now available for the diverse sample matrices and analytes. Among the novel extraction techniques for solid samples PLE is gradually replacing conventional Soxhlet extraction for solid and semisolid samples, both because the extraction is much faster and because of the commercial automated systems available. In addition, literature data from the Soxhlet methods can easily be utilized in selecting extraction solvent. However, the comparatively high investment cost of PLE instrument explains why conventional Soxhlet extraction, in combination with adsorption columns and/or GPC for purification and fractionation of extracts, is still widely used, being the sample preparation reference method in numerous applications. (D)SAE and (D)MAE offer efficient extraction of a variety of samples at considerably lower cost. SFE is an excellent method in many respects, but the matrix-dependent extraction mechanism, expensive instrumentation, and rather demanding optimization make the technique unsuitable for large-scale analyses. SFE has nevertheless proven to be an excellent tool for determination of bioavailable fractions of organic pollutants, particularly in sediment. It should also be noted that especially the extracts from solid and semi solid samples require further purification and extract cleanup is usually still done by tedious conventional methods, i.e., using manual column chromatographic approaches. Improvement of this part of the analytical procedure requires much work, as it is becoming the bottleneck of the whole scheme. Among the techniques for sample preparation of liquid samples, LLE methods continue to be widely used, but SPE has fast been gaining ground. The more novel extraction techniques, such as SPME, SBSE, MSPD, and LPME, can be expected

Sample Preparation for Chromatographic Analysis of Environmental Samples 367

to find a place in future, particularly because they can be used for on-site sampling, something that is particularly advantageous in environmental analysis.

ABBREVIATIONS BFR DHS GC HPLC HS LLE LLLME LPME MAE MASE MESI MMLLE MS PAH POP PCB PCN PBDE PBB PCDD PCDF PHWE PLE PME PT OC SAE SBSE SFE SHS SLE SLM SPE SPME

Brominated flame retardant Dynamic headspace Gas chromatography High performance liquid chromatography Headspace Liquid–liquid extraction Liquid–liquid–liquid microextraction Liquid-phase microextraction Microwave assisted extraction Membrane assisted solvent extraction Membrane extraction with sorbent interface Microporous membrane liquid–liquid extraction Mass spectrometry Polyaromatic hydrocarbon Persistent organic pollutant Polychlorinated biphenyl Polychlorinated naphthalene Polybrominated diphenyl ether Polybrominated biphenyl Polychlorinated dibenzo-p-dioxin Polychlorinated dibenzofuran Pressurized hot water extraction Pressurized liquid extraction Polymeric membrane extraction Purge and trap Organochlorine pesticide Sonication assisted extraction Stir-bar sorptive extraction Supercritical fluid extraction Static headspace Solid–liquid extraction Supported liquid membrane Solid-phase extraction Solid-phase microextraction

References

1. S. P. J. van Leeuwen, A. Kärrman, B. van Bavel, J. de Boer, G. Lindström, Environ. Sci. Technol. 40 (2006) 7854. 2. G. Romanik, E. Gilgenast, A. Przyjazny, M. Kaminski, J. Biochem. Biophys. Methods 70 (2007) 253. 3. W. Wardencki, J. Curylo, J. Namiesnik, J. Biochem. Biophys. Methods 70 (2007) 275.

368

Advances in Chromatography: Volume 48

4. T. Hyötyläinen, J. Chromatogr. A 1153 (2007) 14. 5. S. P. J. van Leeuwen, J. de Boer, J. Chromatogr. A 1186 (2008) 161. 6. J. R. Dean, G. Xiong, Trends Anal. Chem. 19 (2000) 553. 7. M. Petrovic, E. Eljarrat, M. J. Lopez de Alda, D. Barcelo, Anal. Bioanal. Chem. 378 (2004) 549. 8. W. Eljarrat, D. Barcelo, Trends Anal. Chem. 22 (2003) 655. 9. Z. Zhang, J. Pawliszyn, Anal. Chem. 65 (1993) 1843. 10. M. E. McComb, R. D. Oleschuk, E. Giller, H. D. Gesser, Talanta 44 (1997) 2137. 11. F. Musshoff, D. W. Lachenmeier, L. Kroener, B. Madea, J. Chromatogr. A 958 (2002) 231. 12. B. Tienpont, F. David, C. Bicchi, P. Sandra, J. Microcol. Sep. 12 (2000) 577. 13. A. L. Theis, A. J. Waldack, S. M. Hansen, M. A. Jeannot, Anal. Chem. 73 (2001) 5651. 14. A. Segal, T. Gorecki, P. Mussche, J. Lips, J. Pawliszyn, J. Chromatogr. A 873 (2000) 13. 15. I. Bruheim, X. Liu, J. Pawliszyn, Anal. Chem. 75 (2003) 1002. 16. P. Tsytsik, J. Czech, R. Carlee, J. Chromatogr. A 1210 (2008) 212. 17. C. Bicchi, C. Cordero, E. Liberto, P. Rubiolo, B. Sgorbini, P. Sandra, J. Chromatogr. A 1148 (2007) 137. 18. Y. Z. Luo, M. Adams, J. Pawliszyn, Analyst 122 (1997) 1461. 19. M. D. Luque de Castro, I. Papaefstathiou, Trends Anal. Chem. 17 (1998) 41. 20. M. D Luque de Castro, J. L Luque-Garcia, E. Mataix, J. AOAC Int. 86 (2003) 394. 21. K. E. Rasmussen, S. Pedersen-Biergaard, Trends Anal. Chem. 23 (2004) 1. 22. J. A. Jönsson, L. Mathiasson, J. Sep. Sci. 24 (2001) 495. 23. J. F. Liu, G. B. Jiang, Y. G. Chi, Y. Q. Cai, Q. X. Zhou, J. T. Hu, Anal. Chem. 75 (2003) 5870. 24. Q. Zhou, H. Bai, G. Xie, G. Xiao, J. Chromatogr. A 1177 (2008) 43. 25. N. Fontanals, R. M. Marcé, F. Borrull., J. Chromatogr. A 1152 (2007) 14. 26. A. Martın-Esteban, Fresenius J. Anal. Chem. 370 (2001) 795. 27. K. Haupt, Anal. Chem. 75 (2003):376A. 28. V. Pichon, J. Chromatogr. A 1152 (2007) 41. 29. J. Chico, S. Meca, R. Companyó, M. D. Prat, M. Granados J. Chromatogr. A 1181 (2008) 1. 30. T. Kumazawa, C. Hasegawa, X.-P. Lee, K. Hara, H. Seno, O. Suzuki, K. Sato, J. Pharm. Biomed. Anal. 44 (2007) 602. 31. T. Kumazawa, C. Hasegawa, X.-P. Lee, A. Marumo, N. Shimmen, A. Ishii, H. Seno, K. Sato, Talanta 70 (2006) 474. 32. C. Hasegawa, T. Kumazawa, X.-P. Lee, M. Fujishiro, A. Kuriki, A. Marumo, H. Seno, K. Sato, Rapid Commun. Mass Spectrom. 20 (2005) 537. 33. W. E. Brewer, U.S. Pat. Appl. Publ. (2002) 8. 34. A. El-Beqqali, A. Kussak, M. Abdel-Rehim, J. Chromatogr. A, 1114 (2006) 234. 35. M. Abdel-Rehim, J. Chromatogr. B, 801 (2004) 317. 36. L. G. Blomberg, Anal. Bioanal. Chem. 393 (2009) 797. 37. F. David, P. Sandra, J. Chromatogr. A (2007) 54. 38. C. Basheer, M. U. Vetrichelvan, S. Valiyaveettil, H. K. Lee, J. Chromatogr. A 1139 (2007) 157. 39. E. Y. Zeng, D. Tsukada, D. W. Diehl, Environ. Sci. Technol. 38 (21) 5737. 40. M. Kawaguchi, R. Ito, K. Saito, H. Nakazawa, J. Pharm. Biomed. Anal. 40 (2006) 500. 41. J. B. Quintana, I. Rodríguez, Anal. Bioanal. Chem. 384 (2006) 1447. 42. G. Ouyang, J. Pawliszyn, J. Chromatogr. A 1168 (2007) 226. 43. Y. Chen, J. O’Reilly, Y. Wang, J. Pawliszyn, Analyst 129 (2004) 702. 44. Y. Chen, J. Pawliszyn, Anal. Chem. 79 (2004) 5897. 45. G. Ouyang, J. Pawliszyn, Anal. Bioanal. Chem. 386 (2006) 1059.

Sample Preparation for Chromatographic Analysis of Environmental Samples 369 46. S.  Risticevic, V.  H.  Niri, D.  Vuckovic, J.  Pawliszyn, Anal. Bioanal. Chem., 393 (2009) 781. 47. F. David, P. Sandra, J. Chromatogr. A 1152 ( 2007) 54. 48. H. Lord, J. Pawliszyn, J. Chromatogr. A 902 (2000) 17. 49. K. E. Miller, R. E. Synovec, Talanta, 51 (2000) 921. 50. E. Psillakis, N. Kalogerakis, Trends Anal. Chem., 21 (2002) 53. 51. L. Xu, C. Bacheer, H. K. Lee, J. Chromatogr. A 1152 (2007) 184. 52. K. E. Rasmussen, S. Pedersen-Bjergaard, Trends Anal. Chem. 23 (2004) 1. 53. E. Psillakis, N. Kalogerakis, Trends Anal. Chem. 22 (2003) 565. 54. D. A. Lambropouloua, T. A. Albanis, J. Biochem. Biophys. Met. 70 (2007) 195. 55. J. Lee, H. K. Lee, K. E. Rasmussen, S. Pedersen-Bjergaard Analytica Chimica Acta (2008) 253. 56. M. Rezaee, Y. Assadi, M.-R. M. Hosseini, E. Aghaee, F. Ahmadi, S. Berijani, J. Chromatogr. A 1116 (2006) 1. 57. S. Berijani, Y. Assadi, M. Anbia, M.-R. M. Hosseini, J. Chromatogr. A 1123 (2006) 1. 58. L. Fariña, E. Boido, F. Carrau, E. Dellacassa, J. Chromatogr. A 1157 (2007) 46. 59. J. Regueiro, M. Llompart, C. Garcia-Jares, J. C. Garcia-Monteagudo, R. Cela, J. Chromatogr. A 1190 (2008) 27. 60. L. Xua, C. Basheera, H. K. Lee, J. Chromatogr. A 1152 (2007) 184. 61. K. Hylton, S. Mitra, J. Chromatogr. A 1152 (2007) 199. 62. S. Anderssen, T. G. Halvorsen, S. Pedersen-Bjergaard, R. E. Rasmussen, L. Tanum, H. Refsum, J. Pharm. Biomed. Anal. 23 (2003) 263. 63. S. Pedersen-Bjergaard, R. E. Rasmussen, Anal. Chem. 71 (1999) 2650. 64. M. Schellin, P. Popp, J. Chromatogr. A 1020 (2003) 153. 65. D. Kou, S. Mitra, Anal. Chem. 75 (2003) 6355. 66. K. Kuosmanen, T. Hyötyläinen, K. Hartonen, M.-L. Riekkola, Analyst, 128 (2003) 434. 67. L. Xua, C. Basheera, H. Lee, J. Chromatogr. A 1152 (2007) 184. 68. M. Gfrer, C. Fernandes, E. Lankmayr, Chromatographia 60 (2004) 681. 69. J. Sanz-Landaluze, L. Bartolome, O. Zuloaga, L. Gonzalez, C. Dietz, C. Camara, Anal. Bioanal. Chem. 384 (2006) 1331. 70. E. Björklund, T. Nilsson, S. Bøwadt, Trends Anal. Chem. 19 (2000) 434. 71. B. E. Richter, B. A. Jones, J. L. Ezzell, N. L. Porter, N. Avdalovic, C. Pohl, Anal. Chem. 68 (1996) 1033. 72. L. Ramos, E. M. Kristenson, U. A. T. Brinkman, J. Chromatogr. A 975 (2002) 3. 73. L. Ramos, J. J. Vreuls, U. A. Th. Brinkman, J. Chromatogr. A 891 (2000) 275. 74. Y. Yang, S. B. Hawthorne, D. J. Miller, Environ. Sci. Technol. 31 (1997) 430. 75. X. Lou, D. J. Miller, S. B. Hawthorne, Anal. Chem. 72 (2000) 481. 76. K. Lüthje, T. Hyötyläinen, M. Rautiainen-Rämä, M.-L. Riekkola, Analyst 130 (2005) 52. 77. Y. Yang, S. Bøward, S. B. Hawthorne, D. J. Miller, Anal. Chem. 67 (1995) 4571. 78. K. Kuosmanen, T. Hyötyläinen, K. Hartonen, M.-L. Riekkola, Analyst 128 (2003) 434. 79. K. Kuosmanen, T. Hyötyläinen, K. Hartonen, J-Å Jönssön, M.-L. Riekkola, Anal. Bioanal. Chem. 375 (2003) 389. 80. T. Hyötyläinen, T. Andersson, K. Hartonen, K. Kuosmanen, M.-L. Riekkola, Anal. Chem. 72 (2000) 3070. 81. K. Kuosmanen, T. Hyötyläinen, K. Hartonen, M.-L. Riekkola, J. Chromatogr. A 943 (2002) 113. 82. Y. Yang, S. B. Hawthorne, D. J. Miller, Environ. Sci. Technol. 31 (1997) 430. 83. M. Numata, T. Yarita, Y. Aoyagi, A. Takatsu, Anal. Chem. 75 (2003) 1450. 84. E. B. Vazquez, P. M. Lopez, L. S. Muniategui, D. R . Prada, F. E. Fernandez, Fresenius J. Anal. Chem. 366 (2000) 283. 85. F. Priego-Capote, M. D. Luque de Castro, Trends Anal. Chem. 23 (2004) 644. 86. C. Domeño, M. Blasco, C. Sánchez, C. Nerín, Analytica Chimica Acta 569 (2006) 103.

370

Advances in Chromatography: Volume 48

87. P. Rodríguez-Sanmartín, A. Moreda-Piñeiro, A. Bermejo-Barrera, P. Bermejo-Barrera, Talanta 66 (2005) 683. 88. N. Ratola, S. Lacorte, A. Alves, D. Barceló, J. Chromatogr. A 1114 (2006) 198. 89. K. E. C. Smith, G. L. Northcott, K. C. Jones, J. Chromatogr. A 1116 (2006) 20. 90. M. Jin, Y. Zhu, J. Chromatogr. A 1118 (2006) 111. 91. US Environmental Protection Agency, Ultrasonic extraction, EPA standard method 3550B (1996a) 92. M. Mannila, J. Koistinen, T. Vartiainen, J. Environ. Mon. 4 (2002) 1047. 93. T. Nilsson, J. Haekkinen Larsson, E. Bjoerklund, Environ. Pollution 140 (2006) 87. 94. R. Kraaij, W. Seinen, J. Tolls, G. Cornelissen, A. Belfroid, Environ. Sci. Technol. 36 (2002) 3525. 95. D. García-Rodríguez, A. María Carro-Díaz, R. Antonia Lorenzo-Ferreira, J. Sep. Sci. 31 (8) 1333. 96. S. A. Baker, J. Biochem. Biophys. Methods 70 (2007) 151. 97. S. Bogialli, B. Milena, R. Curini, A. Di Corcia, A. Lagana, B. Mari, J. Agric. Food. Chem. 53 (2995) 910. 98. J. L. Gomez-Ariza, M. Bujalance, I. Giraldez, A. Velasco, E. Morales, J. Chromatogr. A 946 (2002) 209. 99. S. A. Barker, J. Biochem. Biophys. Methods 70 (2007) 151. 100. S. Bogialli, A. Di Corcia, J. Biochem. Biophys. Methods 70 (2007) 163. 101. A. M. Carro, R. A. Lorenzo, F. Fernandez, R. Rodil, R. Cela, J. Chromatogr. A 1071 (2005) 93. 102. M. R. Criado, D. H. Fernández, I. R. Pereiro, R. R. C. Torrijos, J. Chromatogr. A 1056 (2004) 187. 103. E. M. Kristenson, L. Ramos, U. A. Th. Brinkman, Trends Anal. Chem. 25 (2006) 96. 104. L. Webster, R. J. Fryer, E. J. Dalgarno, C. Megginson, C. F. Moffat, J. Environ. Monit. 3 (2001) 591. 105. H Hagenmeier, J She, T Benz, N Dawidowsky, L Düsterhöft, C Lindig. Chemosphere 25 (1992) 1457. 106. J. F. Focant, C. Pirard, E. De Pauw, Talanta 63 (2004) 1101. 107. F. Destaillats, P.-A. Golay, F. Joffre, M. de Wispelaere, B. Hug, F. Giuffrida, L. Fauconnot, F. Dionisi, J. Chromatogr. A 1145 (2007) 222. 108. W. Brack, T. Kind, H. Hollert, S. Schrader, M. Möder, J. Chromatogr. A 986 (2003) 55. 109. K. V. Thomas, J. Balaam, N. Barnard, R. Dyer, C. Jones, J. Lavender, M. McHugh, Chemosphere 49 (2002) 247. 110. U. Lübcke-von Varel, G. Streck, W. Brack, J. Chromatogr. A 1185 (2008) 31. 111. L. Viglino, K. Aboulfadl, A. Daneshvar Mahvelat, M. Prévost, S. Sauvé, J. Environ. Monit. 10 (2008) 482. 112. S. Rodriguez-Mozaz, M. J. Lopez de Alda, D. Barceló, J. Chromatogr. A 1152 (2007) 97. 113. A. Asperger, J. Efer, T. Koal, W. Engewald, J. Chromatogr. A 960 (2002) 109. 114. K. Stoob, S. P. Singer, C. W. Goetz, M. Ruff, S. R. Mueller, J. Chromatogr. A 1097 (2005) 138. 115. S. Rodriguez-Mozaz, M. J. Lopez de Alda, D. Barceló, Anal. Chem. 76 (2004) 6998. 116. L. Brossa, R. M. Marce, F. Borrull, E. Pocurull, J. Chromatogr. A 963 (2002) 287. 117. J. Slobodnik, S. Ramalho, B. L. M. van Baar, A. J. H. Louter, U. A. T. Brinkman, Chemosphere 41 (2000) 1469. 118. J. Å. Jönsson, L. Mathiasson, Trends Anal. Chem. 18 (1999) 325. 119. E. Thordarson, J. Å. Jönsson, J. Emnéus, Anal. Chem. 72 (2000) 5280. 120. T. Barri, S. Bergstroem, J. Norberg, J. Å. Jönsson, Anal. Chem. 76 (2004) 1928.

Sample Preparation for Chromatographic Analysis of Environmental Samples 371 121. T. Barri, S. BergstrÖm, A. Hussen, J. Norberg, J.-Å. Jönsson, J. Chromatogr. A 1111 (2006) 11. 122. R. Batlle, H. Carlsson, E. Holmgren, A. Colmsjö, C. Crescenzi, J. Chromatogr. A 963 (2001) 73. 123. M. Ericsson, A. Colmsjö, Anal. Chem. 75 (2003) 1713. 124. E. Thordarson, J. Å. Jönsson, J. Emnéus, Anal. Chem. 72 (2000) 5280. 125. B. Li, Y. Yang, Y. Gan, C. D. Eaton, P. He, A. D. Jones, J. Chromatogr. A 873 (2000) 175. 126. M. Shimmo, T. Hyötyläinen, K. Hartonen, M.-L Riekkola, J. Microcol. Sep., 13 (2001) 202. 127. M. Shimmo, H. Adler, T. Hyötyläinen, K. Hartonen, M. Kulmala, M.-L. Riekkola, Atmos. Environ. 36 (2002) 2985. 128. M. Shimmo, K. Saarnio, P. Aalto, K. Hartonen, T. Hyötyläinen, M. Kulmala, M. L. Riekkola, J. Atmos. Chem. 47(3) (2004) 223. 129. T. Hyötyläinen, J. Chromatogr. A, 1186 ( 2008) 39. 130. K. Lüthje, T. Hyötyläinen, M.-L. Riekkola, M.-L., Anal. Bioanal. Chem. 378 (2004) 1991. 131. M. E. Hattab, G. Culioli, L. Piovetti, S. E. Chitour, R. Valls, J. Chromatogr. A 1143 (2007) 1. 132. S. Bondarenko, F. Spurlock, J. Gan, Environ. Toxicol. Chem. 26 (2997) 2587. 133. A. Prieto, O. Telleria, N. Etxebarria, L. A. Fernández, A. Usobiaga, O. Zuloaga, J. Chromatogr. A 1214 (2008) 1. 134. F. Monteil-Rivera, C. Beaulieu, J. Hawari, J. Chromatogr. A 1066 (2005) 177. 135. B. Niehus, Popp, C. Bauer, G. Peklo, H. W. Zwanziger, Intern. J. Environ. Anal. Chem. 82 (2005) 669. 136. P. Popp, C. Bauer, B. Hauser Keil, L. Wennrich, J. Sep. Sci. 26 (2003) 961. 137. J. M. Keller, R. F. Swarthout, B. K. R. Carlson, J. Yordy, A. Guichard,. M. Schantz, J. R. Kucklick, Anal. Bioanal. Chem., 393 (2009) 747. 138. V. Flotron, J. Houessou, A. Bosio, C. Delteil, A. Bermond, V. Camel, J. Chromatogr. A 999 (2003) 175. 139. I. K. Konstantinou, D. G. Hela, D. A. Lambropoulou, V. A. Sakkas, T. A. Albanis, Chromatographia 56 (2002) 745. 140. F. Vilaplana, P. Karlsson, A. Ribes-Greus, P. Ivarsson, S. Karlsson, J. Chromatogr. A 1196–1197 (2008) 139. 141. C. Lesueura, b, M. Gartnera, A. Mentlerc and M. Fuerhacker, Talanta 75 (2008) 284. 142. J. Sanz-Landaluze, L. Bartolome, O. Zuloaga, L. González, C. Dietz, C. Cámara, Anal. Bioanal. Chem. 384 (2006) 1331. 143. M. Numata, T. Yarita, Y. Aoyagi, A. Takatsu, Anal. Sci. 20 (2004) 793. 144. L. Bartolome, E. Cortazar, J. C. Raposo, A. Usobiaga, O. Zuloaga, N. Etxebarria, L. A. Fernandez, J. Chromatogr. A 1068 (2005) 229.

Preparation for 9 Sample Gas Chromatography Using Solid-Phase Microextraction with Derivatization Nicholas H. Snow Contents 9.1 Introduction................................................................................................... 373 9.2 Overview........................................................................................................ 374 9.2.1 Brief History and Timeline of SPME................................................ 374 9.2.2 Derivatization and Gas Chromatography.......................................... 374 9.2.3 Modes of Derivatization with SPME................................................. 376 9.3 Pre-Extraction Derivatization........................................................................ 376 9.4 Simultaneous Extraction and Derivatization................................................. 380 9.5 Post-Extraction Derivatization on the Fiber.................................................. 381 9.6 Post-Extraction Derivatization in the Inlet.................................................... 382 9.7 Summary and Conclusions............................................................................ 386 Acknowledgments................................................................................................... 386 References............................................................................................................... 386

9.1  Introduction Solid-phase microextraction (SPME) has been an important sample preparation technique in gas chromatography for almost 20 years. First developed for the analysis of volatile organic contaminants from water, its application has grown to numerous compound classes, including both volatile and non-volatile analytes. SPME may be combined with classical derivatization reactions to assist in transferring analytes from the sample to the fiber coating, from the fiber coating into and through the GC, or to aid in detection. This chapter summarizes techniques and efforts in the development and application of SPME methods for GC that include derivatization. Pre and post-extraction derivatization and simultaneous extraction and derivatization are possible and have been used for a variety of analytical problems. Using SPME for sample transfer makes derivatization readily automated and much simpler than 373

374

Advances in Chromatography: Volume 48

in classical solution-based methods. The development of derivatization techniques involving SPME necessitates a further examination of classical derivatization reactions applied to GC.

9.2 Overview 9.2.1 Brief History and Timeline of SPME It has been nearly 20 years since the initial development of SPME by Pawliszyn and colleagues.1,2 Over these two decades, the initial concept of extraction on the surface of fused silica fibers has evolved significantly and numerous related analytical techniques have developed. Most importantly, the dramatic rise of interest in SPME has opened a relatively new field of sample preparation: sorptive microextraction, in which many configurations of sorbents and equipment have been used to apply the original basic concept of sample preparation using a solid (or nearly solid) phase sorbent to trap analytes, followed by thermal desorption into a gas chromatograph to eliminate the use of traditional extraction solvents. The fundamental processes and techniques involved in SPME have been reviewed extensively elsewhere; the 1998 text by Pawliszyn is must reading for anyone interested in beginning SPME method development.3–6 SPME was originally conceived for the analysis of volatile organic contaminants from water. Following the initial development of direct immersion methods and ­discussion of the implications of the very small volume fiber phase on the kinetics and ­thermodynamics of liquid–liquid and liquid–solid extraction, additional extraction modes were developed.7 Headspace-SPME was introduced in 1993, followed shortly thereafter in the 1990s by interfacing with HPLC, CE, and various spectroscopic techniques.8–12 Originally, SPME was conducted on a bare fused silica fiber. When it was commercialized in 1993, coated fibers were used, with polydimethylsiloxane (PDMS) coating for nonpolar analytes, and polyacrylate (PA) for polar analytes. Since 1993, development of fiber coatings has been brisk, with several coatings of varying polarity and thickness now available.13 Applications development has also been brisk, with a Sci-finder search using “solid phase microextraction” as ­keywords producing about 3040 references from Chemical Abstracts, and an applications CD provided by the main SPME vendor providing hundreds of references and application notes.14,15 Two books have also extensively reviewed SPME applications.16,17 In examining these references, it is interesting to note the progression of journals over 20 years, from ­chromatography and analytical chemistry journals (although the very first article was in a water research-specific journal) to more applied journals in a wide range of disciplines. This is a strong indication of the acceptance of SPME as a routine and important analytical technique.

9.2.2 Derivatization and Gas Chromatography Gas chromatography is generally applied to the analysis of volatile compounds, with the required level of volatility determined by the mass and chemical ­structure of the stationary phase. Non- or semi-volatile compounds can be analyzed by gas chromatography, but these compounds may require derivatization to form more volatile analogs prior to analysis. In classical packed column gas chromatography, derivatization is quite

375

SPME with Derivatization

common, as the mass of stationary phase in the column is relatively high, promoting strong analyte retention, and the ability to temperature program is limited by the thermal mass of the column itself and by possible bleeding or decomposition of the stationary phase material. Specific derivatization reactions for gas chromatography have been extensively treated in the literature, so except for a few reactions commonly used with SPME, they are not treated here. The classical text by Knapp and the review by Wells provide detailed information on myriad derivatization reactions for chromatography.18,19 In classical liquid–liquid or solid-phase extraction methods, derivatization may be carried out prior to extraction, or following extraction. If carried out following extraction, the derivatizing agent is often added directly to the dried extract. Figure 9.1, adapted from a thorough review of techniques for extending the analytical range of gas chromatography by Kaal and Janssen, illustrates the place of derivatization in extending the range of compounds for gas chromatographic analysis and places it into context with other techniques such as pyrolysis and high temperature GC.20 Derivatization generally allows analysis of more polar compounds by reacting them to form less polar, although often higher molecular weight, analogs. However, Kaal and Janssen also note that the need for derivatization has appeared to decrease over the past 20 years as alternative methods for chromatographic analysis of polar and/or high molecular weight compounds, such as LC-MS have become routine. They further note that the need for high peak capacity or separation ­efficiency for complex biological samples still makes derivatization with GC favorable for many applications, such as metabolomics. The replacement of polar substituent groups with nonpolar substituents can improve chromatographic performance in several ways. First, the substitution of a polar alcohol with a nonpolar silyl ester improves injection performance as there are no longer polar groups present which may hydrogen bond with active glass surfaces in the inlet, causing possible tailing and discrimination during the injection process. Secondly, this substitution of polar groups with nonpolar groups limits reactivity of the compound to active

Pyrolysis MW

Thermochemolysis

HT-GC

LC

GC Derivatization Polarity

Figure 9.1  Diagram showing relationship between analyte molecular weight, polarity, and analysis technique. In gas chromatography, derivatization generally decreases polarity and increases molecular weight of the analyte. (From Kaal, E. and Janssen, H-G., J. Chromatogr. A, 1184, 43–60, 2008. Copyright 2008, Elsevier Science. With permission.)

376

Advances in Chromatography: Volume 48

sites in the column itself, reducing peak tailing. Thirdly, replacement of a polar group with a less polar group such as trimethylsilyl, methyl, or acetyl to the molecule, while it does increase the molecular weight, generally will decrease the energy of intermolecular interactions with the stationary phase, reducing retention on most polar phases while increasing retention slightly on nonpolar phases. Finally, derivatization can improve detection by adding a detector-specific moiety to the analyte or, most commonly in electron impact GC-MS applications, by substituting a nonpolar functional group in place of a reactive proton, making the molecular ion more likely to be observed.

9.2.3 Modes of Derivatization with SPME In 1997, Pan and Pawliszyn thoroughly described the various modes of ­derivatizing when SPME fibers are used for the extracting phase and progress was recently reviewed by Stashenko and Martinez.21,22 They described three possibilities:

1. Derivatization of the analytes prior to extraction. 2. Doping of the fiber with the derivatizing reagent followed by simultaneous extraction and derivatization. 3. Derivatization of the analytes in the fiber following extraction.

Flow charts for these possibilities are illustrated in Figure 9.2, with the additional post-extraction possibility of performing the derivatization directly in the gas chromatographic inlet following desorption from the fiber. The choice of derivatization scheme will depend on a number of factors. Preextraction derivatization is often used when the sample matrix is not complex or not reactive with the derivatization reagents themselves. It may also be used to enhance partitioning of the analytes into the fiber. For situations in which the sample matrix or interferences may react with the derivatizing reagents, post-extraction derivatization in the SPME fiber coating matrix or simultaneous extraction and derivatization within the fiber coating can be used. Often these methods were developed as analogs to classical methods; the choice or pre- or post-extraction derivatization was made based on the original method.

9.3 Pre-extraction Derivatization Pre-extraction derivatization in the sample vial, often termed in situ derivatization involves addition of the derivatizing reagent directly to the sample prior to extraction by the SPME fiber. Shown in Figure 9.2a, the process involves addition of the derivatization reagent to the sample solution, followed by exposure of the fiber to the sample, either by headspace or direct immersion, followed by exposure of the fiber to the GC inlet to desorb the derivatives into the GC. Pre-extraction derivatization is often used for the analysis of small, polar molecules, to make them more amenable to headspace SPME or to the more commonly used nonpolar fibers. In the forensic analysis of amphetamines, while the analytes are generally volatile enough for headspace extraction, derivatives are often desired to ensure effective confirmation of the structures by mass spectrometry.23,24 For the analysis of amphetamines from hair, samples of hair are first digested with 1M sodium hydroxide and then pH adjusted with phosphate to

377

SPME with Derivatization (b)

(a) - Analyte - Derivatizing reagent a.

- Derivative

b.

(c)

1

2

3

Figure 9.2  Flow charts showing SPME derivatization procedures. (a) Derivatization in the sample matrix. 1: Sample in the matrix; 2: Derivatization reagent added and reacts with sample; 3: Derivatives are formed in solution and extracted by SPME either headspace or direct ­immersion. (b) Pre-extraction on-fiber derivatization. 1: Fiber is exposed to headspace of derivatization reagent, reagent is absorbed into fiber matrix; 2: Fiber is exposed to sample; 3: Derivatives are formed in the fiber matrix. (c) Post-extraction on-fiber derivatization: 1: Fiber is exposed to analyte either headspace or direct immersion; 2: Fiber containing analyte is exposed to headspace of derivatization reagent; 3: Derivatives are formed in the fiber matrix. (From Stashenko, E. and Martinez, J., Trends Anal. Chem., 23(8), 553–561, 2006. With permission.)

pH 6.0. This solution is then mixed with a small amount of HFB-Cl (heptafluorobutyryl chloride), ­followed by exposure of a PDMS SPME fiber to the headspace of the solution. The fiber is then desorbed into the gas chromatographic inlet as usual. Figure 9.3 shows a key result from this work: quantitative comparison of derivatization-SPME results with a more classical solid-phase extraction approach. The results were very similar. In situ derivatization followed by SPME has also been used for analysis of haloacetic acids and related compounds in drinking water supplies.25 These compounds are by-products of traditional water purification methods and some are considered by the US EPA as possible or probable human carcinogens. Haloacetic acids are

378

Advances in Chromatography: Volume 48

Concentration (pmol/cm)

45

SPME

30

Extrelut 15

0

0–1

1–2 2–3 3–4 Distance from hair root (cm - cm)

4–5

Figure 9.3  Comparison of amphetamine recovery from hair using SPME with pre-extraction derivatization and solid-phase extraction on EXtrelut. Note the similar responses to both methods for all samples. (From Liu, J., Hara, K., Kashimura, S., Kashiwagi, M., and Kageura, M., J. Chromatogr. B, 758, 95–101, 2001. Copyright 2001, Elsevier Science. With permission.)

derivatized with demethyl or diethyl sulfate in the original aqueous solutions to form methyl esters, which are much more amenable to analysis by GC than the parent compounds. In both examples, the derivatization procedure is quite simple, ­involving just addition of the derivatization reagent to the original sample, immediately ­followed by exposure of the SPME fiber to the headspace in the sample vial. Figure 9.4 shows the TIC and several extracted ion chromatograms showing separations of derivatized haloacetic acids extracted from swimming pool water. These were measured in aqueous solution at approximately 0.2–250 µg/L with detection limits in the approximate 0.01–0.5 µg/L range, depending on the specific analyte. Stashenko and colleagues used pre-extraction derivatization in combination with headspace SPME to determine low molecular weight aldehydes and carboxylic acids from a variety of matrices including air is a shoe factory, car exhaust, foot sweat, breath, and rainwater.26 In this work, they made pentafluorobenzyl derivatives of the carboxylic acids and pentafluorophenyl derivatives of the aldehydes in solution prior to extraction, followed by headspace extraction onto PA fibers. Finally, in another application designed to make the analytes amenable to headspace extraction, Cancho, Ventura, and Galceran derivatized volatile aldehydes in water using PFBHA (O-(2,3,4, 5,6-pentafluorobenzyl)hydroxyamine hydrochloride) prior to headspace SPME extraction.27 They examined C2–C10 aldehydes and thoroughly described method development and validation for both the derivatization reaction step and then the ensuing headspace-SPME extraction using a divinylbenzene-PDMS fiber. They observed separation of E- and Z-isomers of several derivatives and linear ranges of about 0.1–20 µg/L. A standard chromatogram is shown in Figure 9.5. Note the separation of isomeric E- and Z-pairs shown as peaks 2 and 3 for E- and Z-acetaldehyde, 4 and 5 for E- and Z- propanal and the separation from some artifacts of the in-matrix derivatization. Additional details are provided in the reference.

379

SPME with Derivatization 100 TIC

% 0 100 %

m/z 59

1

0 100

2

%

m/z 83

0 100

3

% 0 100

m/z 117

4

% 0 100

m/z 129

5

% 0 100

m/z 173

6

% 0 100

7

% 0 100 % 0 100 % 0 0.0

m/z 163

IS

m/z 165

m/z 251

2.5

m/z 209

8 5.0

7.5 Time (min)

10.0

12.5

15.0

Figure 9.4  Headspace-SPME-GC-MS total-ion chromatogram and single ion chromatograms of methyl haloacetates from swimming-pool water. Compound identifications: 1, monochloroacetic acid; 2, dichloroacetic acid; 3, trichloroacetic acid; 4, bromochloroacetic acid 5, dibromoacetic acid; 6, bromodichloroacetic acid; 7, chlorodibromoacetic acid; and 8, tribromoacetic acid methyl esters; IS, 2,3-dibromopropionic acid methyl ester. Methyl esters were prepared in situ using demethyl or diethyl sulfate. (From Sarrion, M., Santos, F., and Galceran, M., Anal. Chem., 72, 4865–4873, 2000. Copyright, 2000, American Chemical Society. With permission.)

380

Advances in Chromatography: Volume 48 300

1

d d

d

250

10 11 12 7 13 8 15 9 16 14

3

Response (mV)

200

2

150

4 5

d

100

6

50

*

17 18

19 20

23+25

21 22

26 24

0

0

2

4

6

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 Retention time

Figure 9.5  In situ PFBHA–headspace-SPME–GC–ECD chromatogram of a water sample (30 ml) spiked with aldehydes at 5 mg/ l. Extraction was performed by headspace-SPME with DVB–PDMS fiber. Identification of peaks: 2,3: E, Z-acetaldehyde; 4,5: E,Z-propanal; 7,8: E,Z-butanal. *: 1,2-dibromopropane; w: E, Z-2,4,5-trifluoroacetophenone; d: artifacts. (From Cancho, B., Ventura, F., and Galceran, M., J. Chromatogr. A, 943, 1, 2002. Copyright, 2002, Elsevier Science. With permission.)

9.4 Simultaneous Extraction and Derivatization Simultaneous extraction and derivatization may be performed by first doping or loading the SPME fiber with a derivatizing reagent and then exposing the fiber either directly to the analytical sample or to its headspace. Shown in Figure 9.2b, the fiber is first exposed, usually to the headspace vapor of the derivatization reagent or a solution containing the reagent. The fiber containing the reagent is then exposed to the sample, usually in headspace, to avoid extraction of the derivatization reagent into the sample and analytes are extracted and derivatized in a single step. The fiber is then exposed in the inlet of the GC to desorb the derivatives. In their 1995 article, Pan, Adams, and Pawliszyn demonstrated extraction of fatty acids with simultaneous derivatization by reagents pre-doped into the fiber.28 This was the first report anywhere in the literature, of a derivatization reaction being performed within the SPME fiber coating. They studied short chain fatty acids that were too polar to effectively extract into a polar PA fiber as native compounds. The fiber was pre-doped with 1-pyridinyldiazomethane by exposing it directly to a 5 mg/mL solution in hexane for 60 minutes. The fiber, now containing the derivatizing reagent, was exposed to the headspace of a solution containing the fatty acids. The practical and theoretical development of derivatization in a pre-doped fiber, ­presented in this paper, is important reading.

SPME with Derivatization

381

Analysis of formaldehyde is often difficult due to reactivity of the analyte under many extraction and environmental conditions. In one of the classical examples of SPME in the field, Koziel, Noah, and Pawliszyn used a PFBHA doped fiber to extract formaldehyde from indoor air.29 They showed that the pre-doping technique was especially effective because slow kinetics of desorption and saturation of the derivatizing reagent in the fiber allowed complete reaction of PFBHA and formaldehyde to form a more stable oxime. The oxime product is then desorbed into a GC for analysis. Stability of the oxime allowed extraction in the field followed by later analysis in the lab.

9.5 Post-extraction Derivatization on the Fiber The basic procedure for post-extraction on-fiber derivatization is very straightforward and demonstrated in Figure 9.2c. First, SPME is performed on a sample as usual, typically by direct immersion with consideration of the typical extraction conditions, including: choice of fiber phase and film thickness, extraction time and temperature, agitation method and speed, addition of modifiers such as salt and/or pH adjustment to the sample, and sample volume. Following extraction, the fiber, now containing the extracted analyte(s) and matrix components, is exposed to the derivatizing reagent. Since typical gas chromatographic derivatizing reagents are commonly volatile and highly reactive, exposure is often to the headspace of a small quantity of the derivatizing reagent, rather than directly to the reagent itself. Important considerations in the derivatization step include choice of the derivatizing reagent, reaction time, and temperature. Finally, the fiber, containing the derivatized analytes, is exposed in the inlet of the gas chromatograph as usual for an SPME injection, with consideration to injection liner volume, deactivation and temperature, splitless time, purge flow and initial column conditions.30 While the bulk of SPME-GC applications have involved volatile analytes, it was recognized relatively early on that the use of SPME could be significantly widened if derivatization, either before or after the extraction, could be performed, and the first reports of derivatization in combination with SPME for drug analysis were made in the 1990s, for the analysis of estrogens.31 Especially in drug analysis, ­applications such as in vivo sampling combined with the practical difficulties with SPME-HPLC raise interest in post-extraction derivatization of the extracted analytes on the SPME fiber. In the analysis of drugs from biological and environmental samples, post-­extraction derivatization is preferred over pre-extraction fiber doping with the derivatizing reagents, or over derivatizing directly in the sample matrix, as it allows the use of water sensitive derivatizing reagents and limits reaction of the derivatizing reagents with unwanted matrix components. Much of the work using post-extraction derivatization on the fiber has focused on the analysis of estrogens, steroids, and other endocrine disrupting compounds. We note that estrogens are excellent model compounds for testing pharmaceutical analysis techniques because both the derivatives and the parent compounds are ­detectable by gas chromatography. Several authors have described analyses of estrogens and steroids employing SPME with post-extraction on-fiber silylation using BSTFA as the derivatizing reagent.32 Most recently, Yang, Luan, and Lan demonstrated an SPME-post-extraction derivatization-GC-MS determination of several estrogens

382

Advances in Chromatography: Volume 48

and steroid hormones that provides a useful example of the method development process and typical analytical figures of merit.33 Figure 9.6 shows typical GC-MS data for a mixture of eight analytes, spiked in water at 1–100 µg/L. Herbicides are another family of suspected endocrine disruptors that have received continuing attention in the literature. SPME with post-extraction on-­fiber derivatization has been used to determine several acidic herbicides in water.34 Following extraction with a PA fiber, extracted herbicides were derivatized using MTBSFA (N-methyl-N-(tert-butyldimethylsilyl)-trifluoroacetamide) to form silyl derivatives. This derivatization was performed by exposing the extracted fiber to the vapor of the derivatizing reagent at room temperature in a sealed vial for 10 minutes. Figure 9.7 shows typical chromatograms for several herbicides. As seen in the figure, the extracted ion chromatograms are relatively interference free, showing both the selectivity of the extraction process and the additional selectivity provided by the onfiber derivatization, which, as observed by several authors, appears to provide more regioselective reactions than traditional solution derivatization methods. They also found that the SPME-based derivatization method compared favorably with more traditional SPE and LLE based methods. Bisphenol A, one of the most widely used monomers in polymer synthesis and a common contaminant in finished polymers, has received much recent attention in the press due to concerns about toxicity and wide exposure to the public.35 Post-extraction derivatization on the fiber has been used to determine bisphenol A in landfill leachates in China.36 Landfill leachate samples were adjusted to pH 2 and extracted using a PA fiber. Following extraction, derivatization was performed on the fiber by exposing it to BSTFA (bis-trimethylsilyl trifluoroacetamide) vapor in a sealed vial for 5 minutes at 25°C. As observed by other authors for drug analysis, increased derivatization time and temperature led to reduced analyte signals, probably due to evaporative loss of the derivative. A linear range of 0.09–200 ng/mL was observed.

9.6 Post-extraction Derivatization in the Inlet Derivatization may also be performed directly in the gas chromatographic inlet. Typically this involves performing the SPME procedure as usual and then desorbing the analytes into a gas chromatographic inlet that has been pre-saturated with derivatization reagent, typically by injecting an aliquot of reagent into the inlet prior to desorption. Derivatization in the gas chromatographic inlet following SPME extraction was originally proposed by Pan and Pawliszyn in their early work on fatty acid methyl esters.21 They formed the more volatile methyl esters from the parent fatty acids by predoping the glass sleeve within the inlet with the derivatizing reagent. Alzaga and colleagues used derivatization in the inlet to propose SPME as a tool for analysis of anionic surfactants in water.37 Linear alkylbenzenesulfonates were first combined with an ion pairing reagent (tetrabutylammonium) in the sample matrix. The resulting low polarity ion pairs were then extracted using PDMS fibers, followed by reaction (again with the same ion pairing reagent) upon heating in the GC inlet to form sulfonated butyl esters. Thus both ion pair extraction and derivatization were performed with the addition of a single reagent to the original samples. Figure 9.8 shows selected ion chromatograms of standard water, waste water, and sea surface waters obtained by this method.

Abundance

1

2 7 6

8

500000

1500000

2500000

3500000

4500000

5500000

2 3a 3b

4

5

7 6

8

Time

7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00

1

(b)

4 5 3a 3b 200000 0 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00 Time

600000

1000000

1400000

1800000

2200000

2600000

(a)

3a

3a

2

2

3b

3b

Figure 9.6  SPME-headspace silylation-GC-MS full-scan chromatogram (a) and SIM chromatogram (b) of estrogens and steroids. Peak identification: 1: octylphenol, 2: nonylphenol, 3a, 3b: diethylstilbestrol, 4: dehydroisoandrosterone, 5: estrone, 6: 17-β-estradiol, 7: testosterone, 8: pregnenolone. Spike level: DES at 1 µgL −1, T and PREG at 100 µgL −1, other compounds at 10 µgL −1. (From Yang, L., Luan, T., and Lan, C., J. Chromatogr. A, 1104, 23–32, 2006. Copyright 2006, Elsevier Science. With permission.)

Abundance

3000000

SPME with Derivatization 383

16

1

2

17

4

5

18

3

19 20 Minutes

21

22

23

m/z: 245+247

m/z: 211+213

m/z: 277+279

m/z: 225+227

24

70 60 50 40 30 20 10 0 50 40 30 20 10 0

10.0 7.5 5.0 2.5 0.0

0

10

20

30

16

17

m/z: 219+221

m/z: 199+201

m/z: 283+285

m/z: 279+281

18

19 20 Minutes

6

21

7

22

8

23

9

24

Figure 9.7  SPME-post-extraction on-fiber derivatization-GC-MS extracted ion chromatograms of herbicides for a spiked water sample (0.1 ng/ml per compound). Peak identification: 1: 2-(4-chloro-2-methylphenoxy) propanoic acid (Mecoprop), 2: 3,6-dichloro-2-methoxybenzoic acid (dicamba), 3:  2,4-diphenoxyacetic acid (2,4-D), 4: 4-chloro-2-methylphenoxy acetic acid (MPCA), 5: 2-(2,4-dichlorophenoxy) propanoic acid (2,4-DP), 6: 2-(2,4,5trichlorophenoxy) propanoic acid (2,4,5-TP), 7: 2,4,5-trichlorophenoxyacetic acid (2,4,5-T), 8: 4-(4-chloro-2-methylphenoxy) butanoic acid (MCPB), 9:  4-(2,4-dichlorophenoxy) butanoic acid (2,4-DB). Note selectivity gained by combining derivatization with selective detection. (From Rodriguez,  I., Rubi, E., Gonzalez, R., Quintana, J., and Cela, R., Anal. Chim. Acta., 537, 259–266, 2005. Copyright 2005, Elsevier Science. With permission.)

0 50 40 30 20 10 0

2.5

5.0

7.5

30 25 20 15 10 5 0 15.0 12.5 10.0 7.5 5.0 2.5 0.0 10.0

KCouns

KCouns

KCouns

KCouns

KCouns KCouns KCouns KCouns

384 Advances in Chromatography: Volume 48

Relative abundance

17.0

I.S.

18.0

φ–5 I.S. φ–4 φ–3

I.S.

C10 LAS

φ–5+6

φ–5+6

19.0

φ–4

φ–2

φ–4

φ–2

φ–6

C12 LAS

φ–6+7 φ–5 φ–2 φ–4 φ–3

C13 LAS

21.0

22.0

φ–6+7 φ–5 φ–5 φ–4 φ–2 φ–2 φ–4 φ–3 φ–3

φ–2 φ–4 φ–3

φ–6 φ–5

20.0 Time (min)

φ–3

φ–3

C11 LAS

23.0

φ–2

φ–2

Figure 9.8  SPME–in-port derivatization–GC–MS reconstructed selected ion chromatograms from m/z 171 + 185 + 271 showing isomeric separation of (a) LAS standard, (b) urban waste water, and (c) sea surface microlayer samples. The surrogate n-C8 -LAS is indicated (I.S.). The notation φ-x means the carbon number where phenyl is substituted. Note the group separation by carbon number. (From Alzaga, R., Pena, A., Ortiz, L., and Bayona, J., J. Chromatogr. A, 999, 51–60, 2003. Copyright 2003, Elsevier Science. With permission.)

0

20

40

60

80

(c) 100

20

40

60

80

(b) 100

20

40

60

80

(a) 100

SPME with Derivatization 385

386

Advances in Chromatography: Volume 48

9.7 Summary and Conclusions The development of SPME over the past 20 years, and the more recent wide availability of automated SPME systems, has generated renewed interest in derivatization with gas chromatography for the analysis of highly polar or labile analytes. By using an SPME fiber as the extracting phase, the derivatization process is straightforward and readily automated. From anecdotal examination of the literature, it appears that the fiber may offer benefits over traditional liquid phase reactions in that control over the selectivity of the reaction for the desired product and fewer interfering byproducts may be more readily achieved. Nearly all authors have reported that SPMEbased derivatization methods provide equivalent or superior analytical performance to their traditional counterparts. While attention to derivatization in general has waned among gas chromatographers, the advent of procedures in combination with SPME makes derivatization again attractive and worthy of consideration in analytical method development.

Acknowledgments The author gratefully acknowledges Sanofi-Aventis for providing funding for the Center for Academic Industry Partnership and Supelco (Sigma-Aldrich), which has provided nearly all of the SPME fibers used by the author over the years.

References

1. Belardi, R., Pawliszyn, J. The application of chemically modified fused silica fibers in the extraction of organics from water matrix samples and their rapid transfer to capillary columns. Water Pollution Res. J. Canada 24 179–91 1989. 2. Arthur, C., Pawliszyn, J. Solid phase microextraction with thermal desorption using fused silica optical fibers, Anal. Chem. 62(19) 2145–8 1990. 3. Pawliszyn, J. Solid Phase Microextraction: Theory and Practice, John Wiley and Sons, New York, 1997. 4. Lord, H., Pawliszyn, J., Evolution of solid-phase microextraction technology, J. Chromatogr. A, 885 153–193 2000. 5. Ouyang, G., Pawliszyn, J. Recent developments in SPME for on-site analysis and monitoring Trends. Anal. Chem. 25 692–703 2006. 6. Mustaeta, F., Pawliszyn, J., Bioanalytical applications of solid-phase microextraction, Trends Anal. Chem. 26 36–45 2007. 7. Louch, D., Matlagh, S., Pawliszyn, J., Dynamics of extraction on coated fused silica Fibers, Anal. Chem. 64 1187–1192 1992. 8. Zhang, Z., Pawliszyn, J., Headspace solid-phase microextraction, Anal. Chem. 65 1843–52 1993. 9. Abdel-Rehim M., Bielenstein M., Arvidsson T., Evaluation of solid-phase microextraction in combination with gas chromatography (SPME-GC) as a tool for quantitative bioanalysis, J. Microcolumn Sep. 12(5) 308–315 2000. 10. Theodoridis, G., Koster, E., de Jong, G., Solid-phase Microextraction for the Analysis of Biological Samples, J. Chromatogr. B 745 49 2000. 11. Vas G., Vekey K. Solid-phase microextraction: a powerful sample preparation tool prior to mass spectrometric analysis J. Mass Spectrom. 39 233–254 2004.

SPME with Derivatization

387

12. Witkamp, B.L., Tilotta, D.3, Determination of BTEX compounds in water by solidphase microextraction and raman spectroscopy, Anal. Chem. 67 600–605 1995. 13. http://www.sigmaaldrich.com/analytical-chromatography/sample-preparation/learningcenter/spme.html, accessed in October, 2008. 14. Sci Finder ScholarTM, American Chemical Society, Accessed October, 2008. 15. SPME CD, 7th Edition, Supelco, Bellefonte, PA, 2008. 16. Pawliszyn, J. (ed), Applications of SPME, Royal Society of Chemistry Monographs, London, 1999. 17. Wercinski, S., Solid Phase Microextraction: A Practical Guide, Marcel Dekker, New York, 1999. 18. Knapp, D., Handbook of Analytical Derivatization Reactions, John Wiley and Sons, New York, 1979. 19. Wells, R., Recent advances in non-silylation derivatization for gas chromatography, J. Chromatogr. A 843 1 1999. 20. Kaal, E., Janssen, H-G., Extending the molecular application range of gas chromatography, J. Chromatogr. A 1184 43–60 2008. 21. Pan, L., Pawliszyn, J., Derivatization/solid-phase microextraction: new approach to polar analytes, Anal. Chem. 69 196–205 1997. 22. Stashenko, E., Martinez, J., Derivatization and solid-phase microextraction, Trends Anal. Chem., 23(8) 553–561 2006. 23. Liu, Z, Hara, K., Kashimura, S., Liu, J., Fujii, H., Kashiwagi, M., Miyoshi, A.,Yanai, T., Kageura, M., Two simple methods for enantiomeric analysis of urinary amphetamines by GC/MS using deuterium-labeled L-amphetamines as internal standards, Forensic Toxicol. 24(1) 2–7 2006. 24. Liu, J., Hara, K., Kashimura, S., Kashiwagi, M., Kageura, M., New method of derivatization and headspace solid-phase microextraction for gas chromatographic–mass spectrometric analysis of amphetamines in hair J. Chromatogr. B 758 95–101 2001. 25. Sarrion, M., Santos, F., Galceran, M., In situ derivatization/solid-phase microextraction for the determination of haloacetic acids in water, Anal. Chem. 72 4865–4873 2000. 26. Stashenko, E., Mora, A., Cervantes, M., Martinez, J., HS-SPME determination of volatile carbonyl and carboxylic compounds in different matrices, J. Chromatogr. Sci. 44 347–353 2006. 27. Cancho, B., Ventura, F., Galceran, M., Determination of aldehydes in drinking water using pentafluorobenzylhydroxylamine derivatization and solid-phase microextraction J. Chromatogr. A 943 1 2002. 28. Pan, L., Adams, M., Pawliszyn, J., Determination of fatty acids using solid phase microextraction Anal. Chem. 67(23) 4396–4403 1995. 29. Koziel, J., Noah, J., Pawliszyn, J., Field Sampling and determination of formaldehyde in indoor air with solid-phase microextraction and on-fiber derivatization, Environ. Sci. Technol. 35 1481–1486 2001. 30. Okeyo, P., Snow, N., Optimizing SPME-GC Injections, LC-GC, 15(12), 1130–1136 1997. 31. Okeyo, P., Rentz, S., Snow, N., Analysis of steroids from human serum by SPME with headspace derivatization and GC/MS, J. High Res. Chromatogr., 20, 171–173 1997. 32 Kawaguchi, M., Ito, R., Sakui, Okanouchi, N., Saito, K., Nakazawa, H., Dual derivatization–stir bar sorptive extraction–thermal desorption–gas chromatography–mass ­spectrometry for determination of 17β-estradiol in water sample, J. Chromatogr. A 1105 140–147 2006. 33. Yang, L., Luan, T., Lan, C., Solid-phase microextraction with on-fiber silylation for simultaneous determinations of endocrine disrupting chemicals and steroid hormones by gas chromatography-mass spectrometry, J. Chromatogr A. 1104 23–32 2006.

388

Advances in Chromatography: Volume 48

34. Rodriguez, I., Rubi, E., Gonzalez, R., Quintana, J., Cela, R., On-fibre silylation following solid-phase microextraction for the determination of acidic herbicides in water samples by gas chromatography, Anal. Chim. Acta 537 259–266 2005. 35. Parker-Pope, T., Panel faults FDA in stance that chemical found in plastic is safe, New York Times, October 30, 2008. 36. Xiangli, L., Li, L., Shichun, A., Chongyu, L, Tiangang, L., Determination of bisphenol A in landfill leachate by solid phase microextraction with headspace derivatization and gas chromatography-mass spectrometry, Chin J. Anal. Chem. 34(3) 325–328 2006. 37. Alzaga, R., Pena, A., Ortiz, L., Bayona, J., Determination of linear ­alkylbenzenesulfonates in aqueous matrices by ion-pair solid-phase microextraction-in port derivatization-gas chromatography-mass spectrometry, J. Chromatogr. A 999 51–60 2003.

Index A Abraham solute descriptors, 199; see also Packed column SFC (pSFC) Absorbance detectors, 158 Accelerated solvent extraction (ASE), 346 Acetaminophen separation, 267 Acetonitrile–water mixture studies, 12 computed incremental transfer free energy diagrams, 13 K(z) profile for, 22 n-octadecane chain solvated in, 21 Acetylcholine separation, 268 ACQUITY UPLC®, 101, 103 gradient separation of, 123 Activity coefficient models, 96 Acylation, 353, 355 Adiabatic column, temperature and pressure change in, 110 Adsorbates excess properties, 93 Adsorption equilibria association equilibrium, 89 in protein chromatography, 88 standard Gibbs energy change of association, 89 stoichiometric coefficient, 89 thermodynamic equilibrium constant, 89 Adsorption isotherms asymmetric activity coefficients of proteins, 62 estimation of parameters, 62 partition coefficients, 62–63 ion-exchange, 63–64 β-lactoglobulins A and B, 64–68 hydrophobic interactions, 70–72 in mixed solvents, 68–70 self-association isotherm, 62 thermodynamic equilibrium constants, 62 Advanced molecular simulation methods, 3 Affinity chromatography, 164–165; see also Biointeraction affinity chromatography kinetic measurements methods, comparison of, 167 linear elution methods peak profiling, 171–173 plate height measurements, 168–171 practical considerations, 174

non-linear elution methods combined assay methods, 183 frontal analysis, 176–178 non-linear peak fitting, 174–176 peak decay method, 181–182 practical considerations, 183–184 split-peak method, 178–181 principles, 166 zonal elution in, 148 applications of, 150 in binding and competition studies, 151–153 temperature and solvent studies, 153–155 binding sites, characterization of, 155–156 7-hydroxycoumarin binding to HSA, 149, 150 practical considerations for, 156–158 principles, 150–151 Affinity ligand, 146; see also Biointeraction affinity chromatography biological interaction study by, 147 AGP, see Albumin/α1-acid glycoprotein columns (AGP) Albumin/α1-acid glycoprotein columns (AGP), 147 Alcohol incremental free energy, 13 Alkane incremental free energy, 13 Alkylbenzene distribution coefficient, 9 incremental transfer free energy diagram, 10 Allosteric effects, between analyte and competing agent, 152–154 studies on drug–protein systems, 153–154 Amino-caproic acid separation, 268 6-Aminoquinolyl N-hydroxysuccinimidyl carbamate (AQC reagent), 125 Ammonium dihydrogen fluoride as etching agent, 282 Ammonium sulfate activity coefficients, 77 Amphetamine recovery from SPME, 378 Amylbenzene plate height and velocity curves, 105 reduced plate height and reduced velocity curves, 106 Analyte–ligand systems, zonal elution studies on, 149–151

389

390 Analytical affinity chromatography, see Biointeraction affinity chromatography ANP, see Aqueous normal phase (ANP) of silica hydride-based columns Antituberculosis tablets and ultra-fast UPLC-UV method, 136 Apolar/polar column combination, 297 AQC reagent, see 6-Aminoquinolyl N-hydroxysuccinimidyl carbamate (AQC reagent) Aqueous normal phase (ANP) of silica hydride-based columns, 260 properties, 267–270 retention capabilities in, 271 Aromatic hydrocarbons gradient separation, 267–268 ASE, see Accelerated solvent extraction (ASE) Asymmetric activity coefficients, 87 defined, 88 at infinite dilution, 88 of proteins, 62 reference state potential, 88 ATD, see Automated thermal desorption (ATD) devices Automated thermal desorption (ATD) devices, 335 AUX, see Advanced pressure control unit (AUX) Advanced pressure control unit (AUX), 296 Avogadro’s number, 86

B Band broadening method, see Plate height method Band entrapment/focusing, 305 Benzene, 352–353 Betaine identification, 271 BFR, see Brominated flame retardants (BFRs) Biointeraction affinity chromatography, 146 advantages of, 146–147 applications, 150 equilibrium and thermodynamic measurements by, 147–148 frontal analysis (see Frontal analysis uses) zonal elution (see Zonal elution, in affinity chromatography) and kinetic measurements (see Kinetic studies, in affinity chromatography) Bioseparations oligonucleotides, 131–137 peptide mixtures chromatogram, 128, 130 gain analysis time, 127–128

Index resolution of, 128 tryptic digest of bovine hemoglobin, 131 proteins, 128 Bisphenol A, 382 Bis-trimethylsilyl trifluoroacetamide (BSTFA), 382 as derivatizing reagent, 381 BNP II, see Bovine neurophysin II (BNP II) Boltzmann weight, 39–40 Born potentials, 86 Bovine hemoglobin tryptic digest, 131 Bovine neurophysin II (BNP II), 171 Bovine serum albumin (BSA), 147 Brominated flame retardants (BFRs), 330 in fish samples, 349 Brown alga Dictyopteris membrana, volatile metabolites determination, 361 BSA, see Bovine serum albumin (BSA) BSTFA, see Bis-trimethylsilyl trifluoroacetamide (BSTFA) Bulk molfraction, 46 n-Butane distribution coefficient profiles, 23

C Capillary electrophoresis, 100 Capillary flow technology, 307 Carbohydrate structural isomer separation, 262 Carbonic anhydrase electrochromatograms for, 285 Carbowax carbowax-polyethylene glycol (PEG) coating, 341 carbowax/templated resin (CW/TPR) coating, 341 Carboxen/poly(dimethylsiloxane) (CAR/PDMS) coating, 341 Catecholamines separation on hydride C18 column, 266 Chenodeoxycholic acid identification, 271 Chinese medicines and natural products, analysis by UPLC®, 137 4-Chloro-7-nitrobenz-2,1,3-oxadiazole (NBD chloride) fluorescent derivative, 355 Choline separation, 268 choline identification, 271 2DChromatograms, 322–323 expansion relative to, 322–323 generation of, 317 Classical contribution, 86 Classical isotherm model, 62 Cold-trapping, 295 Column choice in zonal elution studies, 156–157 Combined assay methods for kinetics measurements, 183

Index Complex environmental samples chromatographic techniques, 329 interlaboratory studies, 330 preparation procedure, 329–330 sample and matrices, 330–331 preparation technique, 330–331 processing and pretreatment, 330 Comprehensive two-dimensional chromatography application, 292 COMSOL software, 111–112 Configurational-bias Monte Carlo (CBMC), 38 application of, 42 Boltzmann weight, 39–40 probability, 39, 41 regrowth/insertion, 39 Rosenbluth weight, 39–40 Contaminated sediment extract, 354 Continuum solvation models, 31 Convex ion-exchange, 61 convex isotherms expression for, 95 HIC and PRC, 95 IEC, 95 Coulombic potentials, 31 Counterion activity coefficients, 93 Creatine and creatinine separation by gradient method, 271 with 4-hydroxyproline in human urine, 272 Cyclodextrins, 158 Cytidine separation, 278

D Database for pSFC, 228 method development with solvation parameter model, 244–247 ODS phases, 240–244 predictive capability of models, 247–248 variation of system constants among stationary phases, 228–232 visual representation of, 232–240 DBP, see Dibutylphthalates (DBPs) DDT, see Dichlorodiphenyltrichloroethane (DDT) Deans switching principle, 298 Deans switch-type flow modulator in injection state, 307 schemes, 296 Debye–Hückel potential, 86 Decanophenone plate height and velocity curves, 105 reduced plate height and reduced velocity curves, 106

391 Derivatization and gas chromatography SPME, 374–375 GC-MS applications, 376 modes with sample matrix, 376 Diamond hydride column with mass spectroscopy detection ONP conditions and, 275 separation and identification of metabolites, 273 Dibenzofurans (PXDFs), 352–353 Dibutylphthalates (DBPs), 331 Dichlorodiphenyltrichloroethane (DDT), 331 Dichloromethane fractions, 352–353 Differential material balance model of unit volume of column, 59 ion-exchange chromatography, 61 ordinary and eddy diffusion coefficient, 60 phase ratio, 60–61 plug flow in axial direction, 60 residence time, 60 timescale of flow through, 60 Diffuse reflectance infrared Fourier transform (DRIFT) silica hydride analysis, 258 3,4-Dihydroxyphenylacetic acid for retention, 271 5-Dimethyl aminonaphthalene-1-sulfonyl chloride fluorescent derivatives, 355 Diphenylethers, 352 Dispersive liquid-liquid microextraction (DLLME), 342 Dissociation equilibrium constant, 162 rate constant, 169, 171–173, 175–177, 181, 182 Divinylbenzene/carboxen/ poly(dimethylsiloxane)(DVB/CAR/ PDMS) coating, 341 DLLME, see Dispersive liquid-liquid microextraction (DLLME) Drugs of abuse analysis for UPLC, 137 DSAE, see Dynamic sonication-assisted extraction (DSAE) Dual-column stop-flow instrument, 300 Dual mode retention of silica hydride phases capabilities, 263, 272 chromatographic behavior for, 273 Dual separation modes, hydride-based etched capillaries electrophoretic format, 282 liquid crystals, 282 methanol in, 281 synthetic peptide sample, electrochromatogram, 280

392 Dynamic headspace (DHS), 335 Dynamic sonication-assisted extraction (DSAE), 347

E Eddy diffusion, 61 EE, see Extraction efficiency (EE) Electrovalve (EV), 296 EF, see Enrichment factor (EF) Electroneutrality, 90 Electronic pressure controllers, 298 Electrozone sensing technique, 105 Embedded polar groups (EPGs), 7 Enrichment factor (EF), 342 Environmental applications of LPME, 342 of MASE, 344 samples with novel extraction techniques methods used for, 358–359 EPGs, see Embedded polar groups (EPGs) Epinephrine for retention, 271 Etched chemically modified capillaries, 282; see also Hydride-based etched capillaries analysis peptide, 283–284 proteins, 284–285 cytochrome c tryptic digest, separation of, 284 EOF characteristics, changes result, 282–283 PEGylated compounds, analysis, 285 surface modification, 283 tricyclic antidepressants and tetracyclines, 283 2-Ethylpyridine phase, 196 EV, see Electrovalve (EV) Excess potentials, 87 Exchange process, 63 Exclusion factors, 63 Extract cleanup derivatization GC analysis, 353, 355 liquid chromatography (LC), 355 fractionation automated fractionation procedure by HPLC, 354 cleanup systems, 353 ethanolic KOH solutions, 351 and GPC, 351 HPLC, 351, 353 liquid–liquid partitioning, 351 POPs, adsorbents used in, 351–352 PXDD and PXDF, 352 Soxhlet extract of sewage sludge, purification, 352–353 lipid removal from biological extracts, 350–351

Index online techniques MMLLE-GC system, 355–356 sample preparation methods, selection for analysis of selected groups of POPS in, 358–359 HD, fMAE and SFE techniques for, 357, 362 liquid of fluid-based extraction methods, 357–358 LPME techniques, 363 MAE and PLE used for, 363 merit of extraction methods, relevant figures, 360 PCB extraction, comparison of, 364 PCDD/PCDFs, extraction of, 365–366 Soxhlet in extraction of PAHs in sludge sample, comparison, 364–365 volatile and semivolatile compounds, extraction, 362 volatile compounds, analysis with, 356–357 sulfur removal from sediment extracts, 351 treatment with concentrated sulfuric acid, 351 Extracted ion chromatogram (EIC) for creatine and creatinine, 271 Extraction efficiency (EE), 342 Extraction procedures brown alga Dictyopteris membrana, volatile metabolites comparison of extraction methods, 361 cleanup of extract fractionation, 351–353 lipid removal from biological extracts, 350–351 matrix compounds, 350 sulfur removal from sediment extracts, 351 treatment with concentrated sulfuric acid, 351 liquid environmental samples ionic liquids (IL), 336 liquid-phase microextraction (LPME), 341–342 LPME and membrane extraction techniques, 335–336 membrane assisted solvent extraction (MASE), 343–344 microextraction techniques, basics, 336 solid phase extraction (SPE), 336–338 SPME and SBSE, 338–335, 341–336 solid and semisolid samples and ASE, 346 MAE and SAE, 345 matrix solid phase dispersion (MSPD), 349 pressurized liquid extraction (PLE), 346

393

Index SAE and MAE, 347–348 SFE, 344–345, 347 and Soxhlet extraction, 345–346 supercritical fluid extraction (SFE), 344–345 vapor-phase extraction dynamic headspace (DHS), 335 HS and GC techniques, 334 thermal extraction, 335

F FAME, see Methyl ester-derivatized plasma fatty acids (FAMEs) Faraday’s number, 86 Fiber coating, 374 FID primary-column analysis, test-mixture chromatograms, 302 Flavanone O-diglycosides with UPLC®ESI-MS method, 137 Florisil, 349, 351 Flory–Huggins parameters, 8 Fluidized-bed extraction technique, 345 fMAE, see Focused microwave assisted hydrodistillation (fMAE) FME, see Focused microwave (FME) system Focused microwave assisted hydrodistillation (fMAE), 357 Focused microwave (FME) system, 347 methods, 363 Force field parameters, 32 Formaldehyde analysis, 381 Free energy terms for solvophobic theory, 5–6 Frontal affinity chromatography-mass spectrometry (FAC-MS), 161 Frontal analysis uses, 148, 158 advantages of, 158–159 applications of, 159 in binding studies, 159–161 in competition studies, 161–162 in temperature and solvent studies, 163–164 general principles of, 158 practical considerations for, 164 Fructus aurantii (Zhi qiao) with UPLCESI-MS method, 137 Fumaric acid separation, 270 Fused-silica restriction, 296

G Gas chromatography (GC), 289, 334 and derivatization reagents, 381 SPME, 374–375

experts, 291 preliminary monodimensional applications, 292 resolution, 290, 291 Gauche defects, 16–17 average fraction of, 19 statistics, 45 GC, see Gas chromatography (GC) GDS, see Gibbs dividing surface (GDS) Gel permeation chromatography (GPC), 350–351 Gibbs dividing surface (GDS), 20 Gibbs energy models, 96 Gibbs ensemble method, 36 acceptance probability, 37–38 types of, 37 Glyburide elution characteristics, 263 GPC, see Gel permeation chromatography (GPC) Gradient elution method, 270 Group contribution methods, 8–9

H Haloacetic acids analysis of, 377–378 Harmonic approximation, 31 HCB, see Hexachlorobenzene (HCB) HCH, see Hexachlorocyclohexane (HCH) HD, see Hydrodistillation (HD) Headspace-SPME-GC-MS total-ion chromatogram and single ion chromatograms of methyl haloacetates, 379 Headspace techniques (HCC-HS), 334–335 Heart-cutting two-dimensional gas chromatography, 293 heart-cutting zone compression-band injection, 305 pressure switching and capillary columns, use, 294 Heat shock protein 90, 175 Herbicides, 382 Heterocyclic aryl compounds separation, 275 Heterogeneity in system composition, 46 Hexachlorobenzene (HCB), 331 Hexachlorocyclohexane (HCH), 331 n-Hexadecane as model for retentive phase material, 10 solvent distribution coefficient measurements, 9 Hexane fractions, 352–353 Hexane/THF mobile phase, 276 High performance affinity chromatography (HPAC), 146 High performance liquid chromatographic separation (HPLC)

394 detector, 350 fractionation method and derivatization, 353 POPs fractionation in extract of contaminated sediment, 354 High-performance liquid chromatography (HPLC), 100 High-speed solubility screening assay, 136 Hofmeister series, 64 HPAC, see High-performance affinity chromatography (HPAC) HPLC, see High-performance liquid chromatography (HPLC) HPLC hydrophilic interaction chromatography (HILIC) separations, 136 HSA, see Human serum albumin (HSA) Human serum albumin (HSA), 147; see also Biointeraction affinity chromatography and human IgG, separation on etched C5 capillary, 285–286 Hydride-based etched capillaries capillary properties, characterization electrophoretic behavior, 279–280 EOF and pH, plot of, 279 tryptic digest, electropherogram of, 280 dual separation modes electrophoretic format, 282 liquid crystals, 282 methanol in, 281 synthetic peptide sample, electrochromatogram, 280 etched chemically modified capillaries, 282 cytochrome c tryptic digest, separation of, 284 EOF characteristics, changes result, 282–283 PEGylated compounds, analysis, 285 peptide analysis, 283–284 proteins analysis, 284–285 surface modification, 283 tricyclic antidepressants and tetracyclines, 283 fabrication nitrogen and fluoride, 279 OTCEC, 278 Hydrodistillation (HD), 357 Hydrophobic and hydrophilic compounds retention map, 274 Hydrophobic isotherms, 61 hydrophobic associations material balance, 91 scheme of, 91 standard Gibbs energy change, 91 interactions, 70 linking HIC and solubility data, 74–77 modulator effect, 71

Index reversed-phase chromatography, 72 salt-induced hydrophobic chromatography, 73–74 Hydrophobic ligand activity coefficient, 93 Hydrosilation process, 257–258 5-Hydroxy-3-indole acetic acid for retention, 271 Hydroxyl increment, 14 Hypoxanthine identification, 271

I Ideal mixture activity coefficients, 87 behavior, 87 chemical potentials, 86 Ideal vapor reference state in RPLC, 5 IL, see Ionic liquids (IL) Industrial applications of thermodynamic modeling of chromatographic separation, 79–81 Intermolecular potential energy function, 31 Interstitial porosity, 63 Ion-exchange chromatography (IEC), 61 electroneutrality, 90 ion-exchange scheme, 90 mole fraction and activity coefficient, 90 Ionic liquids (IL), 336 Ionizable analytes, 117 Ion-pair reversed-phase liquid chromatography (IP RP LC), 131 IP RP LC, see Ion-pair reversed-phase liquid chromatography (IP RP LC) Isobaric acids separation, 270 Isotherm compressibility coefficient, 108 and stoichiometric coefficient, 80

K Kidney cancer in urine samples, markers search, 136 Kinetic studies, in affinity chromatography, 164–165 kinetic measurements methods, comparison of, 167 linear elution methods peak profiling, 171–173 plate height measurements, 168–171 practical considerations, 174 non-linear elution methods combined assay methods, 183 frontal analysis, 176–178 non-linear peak fitting, 174–176 peak decay method, 181–182 practical considerations, 183–184 split-peak method, 178–181 principles, 166

395

Index Kohn–Sham density functional theory, 31 Kozeny–Carman equation, 102–103, 119

L β-Lactoglobulins A and B association constant of, 66 double-logarithmic plots of isocratic retention volumes of, 64–65 isocratic elution chromatograms, 66 modeled, 67 isotherms adsorption, 68 parameters for, 66 slope and partition coefficient, 68 non-Langmuirian behavior, 65–66 Landfill leachate samples, 382 Langmuir adsorption models, 176–177 Lattice theories ordering of chains, 8 solvent properties, 8 LC, see Liquid chromatography (LC) Lennard-Jones potentials, 31 portion of, 32 Linear alkylbenzene sulfonates, 382 Linear elution methods peak profiling, 171–173 plate height measurements, 168–171 practical considerations, 174 Linear retention indices (LRI), 318 Linear solvation energy relationship (LSER), 199 Lipophilic interaction for nonpolar molecules in RPLC, 14 Liquid chromatography (LC) UV chromophores and fluorophores, 355 Liquid crystal problems, 8 stationary phase, 282 Liquid environmental samples ionic liquids (IL), 336 liquid-phase microextraction (LPME), 341–342 LPME and membrane extraction techniques, 335–336 membrane assisted solvent extraction (MASE), 343–344 microextraction techniques, basics, 336 solid phase extraction (SPE), 336–338 SPME and SBSE, 335–336, 338–341 techniques, 363 Liquid–liquid extraction (LLE), 330, 365 Liquid-phase microextraction (LPME) DLLME, 342 hydrophilic compounds, extraction of, 342 optimization, critical parameters in, 342 two-phase and three-phase systems, 341–342

Lithium aluminium hydride, 256 LLE, see Liquid–liquid extraction (LLE) LMCS, see Longitudinally modulated cryogenic system (LMCS) Local solubility model, 25 Longitudinally modulated cryogenic system (LMCS), 305 GC×GC experiments ToF MS and rapid-scanning qMS systems in, 319 system, 318 Lorentz–Berthelot combining rules, 32 Low thermal mass (LTM) benefits, 298 capillary flow technology, 299 LPME, see Liquid-phase microextraction (LPME) LRI, see Linear retention indices (LRI) LSER, see Linear solvation energy relationship (LSER) LTM, see Low thermal mass (LTM) Lumped constant thermodynamic model, 4 Lysozyme binding to Cibacron Blue 3GA, 176–177

M MAE, see Microwave-assisted extraction (MAE) Maleic acid separation, 270 Many-body systems, 5 Martin equation for methylene increment, 14 Mass spectrometry (MS) mass spectrometer spectra, 318 system, 313 specialists, 291 Mass-transfer, 77 boundary condition, 78–79 coefficients of, 78 rate of, 79 Matrix solid phase dispersion (MSPD) technique, 366 C8-and C18-bonded silica and Florisil, 349 McGowan’s molecular volume, 226 MDGC, see Multidimensional gas chromatographic (MDGC) system Membrane assisted solvent extraction (MASE) extraction in, 343–344 forms, 344 MMLLE membrane, 343 types of, 343 Membrane extraction with sorbent interface (MESI), 335

396 Membrane liquid–liquid extraction (MMLLE), 343 MEPS, see Microextraction in packed pipette/ syringe (MEPS) technique Metabonomics, 136 Metal ferrules, 298 Metanephrine normetanephrine separation, 268 Metformin elution characteristics, 263 Methanol–water mixtures isothermal compressibility, 110 studies, 12 computed incremental transfer free energy diagrams, 13 n-octadecane chain solvated in, 21 Methotrexate separation, 268 Methylene incremental free energies of transfer, 24 increment and Martin equation, 14 lipophilic interactions in, 24 Methyl ester-derivatized plasma fatty acids (FAMEs), 319, 322 Methyl haloacetates headspace-SPME-GC-MS total-ion chromatogram and single ion chromatograms of, 379 Microcolumn HPLC with hydride-based C18 stationary phase, 277 Microextraction techniques, 337 basics of, 336 microextraction in packed pipette/syringe (MEPS) technique, 337 online connection with LC/GC, 338 parameters in, 338 Microwave-assisted extraction (MAE), 345, 347, 364 optimization, critical parameters in, 348 Midpoint gas flow regulation pneumatically activating valve, 300 MIP, see Molecularly imprinted polymers (MIPs) Mixed-mode sorbents, 337 MMLLE, see Membrane liquid–liquid extraction (MMLLE) MMLLE-GC system, 355 POPs in water sample, analysis, 357 Model parameter sensitivity analysis, 59 Modulation period, 292 Modulators, 292 capillary flow technology, 307 Deans switch-type flow modulator in injection state, 307 LMCS, 305 longitudinally modulated cryogenic system, 306 low-cost pneumatic GC×GC, 307–308 pneumatic-modulated GC×GC, 305, 307

Index short microbore column segment, 308–309 single-stage air-cooled and electrically heated thermal modulator, GC×GC instrument with, 308 thermal sweeper, 305–306 tube, 308 Molecular dynamics (MD) simulation technique, 29 Molecularly imprinted polymers (MIPs), 336–337 Molecular simulations of RPLC Gibbs ensemble method, 36–38 heterogeneity in system composition, 46 Monte Carlo methods for, 34–36 order parameter, 45–46 solute distribution coefficients and transfer free energies, 46–47 transferable potentials for phase equilibria force field, 31–34 Monte Carlo method algorithms for particle-based simulations, 12 for molecular simulation, 34–35 application of, 36 phase equilibria, computation of, 30 MS, see Mass spectrometry (MS) MSPD, see Matrix solid phase dispersion (MSPD) MTBSFA, see N-methyl-N-(tertbutyldimethylsilyl)trifluoroacetamide (MTBSFA) Multidimensional gas chromatographic (MDGC) system, 291 and applications, 295 AUX and EV, 296 first-dimension FID chromatogram of, 297 instrument, 296–297 LTM, capillary flow technology, 299 microfluidic transfer system, 298 midpoint gas flow regulation, 300 second-dimension MDGC-MS TIC chromatogram of, 298 series-coupled column system with stop-flow operation, scheme, 301 spectral purity, 297 stop-flow methodology, 301 target compounds, 298, 300 test-mixture chromatograms, 302 transfer system, schemes of, 296 classical heartcutting, 293 advantage, 292 cold-trapping, 295 primordial valve-based, scheme of, 294 twin-oven configuration, 294–295 Murine monoclonal antibody separation, 132

Index N NBD chloride, see 4-Chloro-7-nitrobenz2,1,3-oxadiazole (NBD chloride) fluorescent derivative Neutral hydrophobic analytes, 117 Nicotinic acetylcholine receptor (nAChR), 175 Nitrophenols separation, 239 N-methyl-N-(tert-butyldimethylsilyl)trifluoroacetamide (MTBSFA), 382 Non-competitive peak decay method, 182 Non-crystalline systems, 16 Nonequilibrium process, 30 Non-Langmuirian behavior of β-lactoglobulins A and B, 65–66 Non-linear elution methods combined assay methods, 183 frontal analysis, 176–178 non-linear peak fitting, 174–176 peak decay method, 181–182 practical considerations, 183–184 split-peak method, 178–181 Non-linear peak fitting method, 174–176 Non-steroidal anti-inflammatory drugs chromatograms of, 243 elution orders of, 242 structures of, 242

O OcCTA, see Octachlorothianthrene (OcCTA) Octachlorothianthrene (OcCTA), 353 Octadecyl silane (ODS) phase backbone orientation, 20 EPG phases and, 25–26 grafting densities for, 16 hydrogen bonds in, 15 K(z) profile for n-butane, 22 organic modifier, chain solvation by, 21 in RPLC mechanism, 12 solute-ODS interaction, 14 solvent penetration in, 11 transfer free energy, 14 Octadecylsiloxane-bonded silica (ODS) phases, 196 Octanol-water distribution coefficient (Ko/w), 338–339 ODS, see Octadecylsiloxane-bonded silica (ODS) phases Oligodeoxythymidines peak capacity, 134 Oligonucleotides separation BEH columns, 133 and column peak capacity, 133 experimental data, 135 flow rate gradient duration, 136 gains in resolution, 133–134 liquid chromatography, 131

397 retention pattern, 133 simulated chromatograms, 135 UPLC with mass spectrometry detection, 135–136 On-coating derivatization, 340 Online techniques MMLLE-GC system, 355–356 online coupled MMLLE-GC-FID, 357 online extraction chromatography environmental samples, analysis of, 356 ONP, see Organic-normal phase (ONP) separations Open tubular capillary electrochromatography (OTCEC), 278, 280–281, 283 human IgG and serum albumin separation on etched C5 capillary, 285–286 PEGylated proteins, analysis of, 285 reproducibility studies, 286 Open-tubular capillary (OTC) column, 289–290 Operational parameters applications Arabica coffee volatiles, headspace SPME-GC×GC-qMS result, 322 cigarette smoke, TIC GC×GC-ToF MS contour plot, 315, 317 2D chromatograms, 316, 317, 322–323 2D GC combined with mass spectrometry, 315 diesel oil, GC×GC-FID chromatograms, 326 expansion relative to, 325 fatty acid matrices, GC-MS analysis of, 322 GC experiments, 314 GC×GC plasma FAMEs, identification, 324 GC×GC-qMS experiment, 317 GC×GC, series, 314–315 GC-MS software use, 317 intra-class separation, quality, 325 LMCS system and LRI, 318 major and minor perfume constituents, analysis, 318 orthogonal combination, 323, 325 peak identification, GC-qMS result for, 321 plasma FAMEs, GC×GC-FID result, 323 quadrupole mass spectrometry, 318 2-second elution range, 315, 317 single perfume peak, GC-qMS result, 320 spatial order and enhanced sensitivity, 319–325 TIC GC×GC-MS result for perfume, 318–319 ToF MS and rapid-scanning qMS systems in, 319

398 column selectivities degree of correlation, 310 detection bidimensional methodologies, 313 GC×GC-FID contour plot, 313 GC-MS experiments, 313 GC peak base widths, 312–313 time-of-flight mass spectrometry (ToF MS), 313–314 gas flows GC×GC experiment, 311–312 ideal linear velocity, 311–312 stop-flow GC×GC, 312 temperature gradient degree of correlation, 310–311 disadvantages, 311 primary capillary temperature ramps, 311 second-dimension separations, 311 Order parameter, 45–46 Organic acids separation, 265 Organic modifier molfraction enhancement, 21 Organic-normal phase (ONP) hydride-based separation materials and, 275 phenols separation by, 275 separations, 259 Organo-chlorine pesticides (OCPs) in fish samples, 349 Organosilanes in aqueous solutions stability, 258 Orthogonal column configuration, 310 OTCEC, see Open tubular capillary electrochromatography (OTCEC) Oxime, 381

P Packed column SFC (pSFC), 195–196 choice of stationary phase, 196–197 chromatographic systems, 196 characterization of, 198 database for, 228 method development with solvation parameter model, 244–247 ODS phases, 240–244 predictive capability of models, 247–248 variation of system constants among stationary phases, 228–232 visual representation of, 232–240 mobile phase effects with stationary phase, comparison of, 197 phases, stationary of, 196 solvation parameter model uses in, 199–201 chemical structures, of stationary phases, 206 choice of solvation descriptors, 225–228 chromatographic system, 201–202

Index covariance matrix for solute set, 216 data analysis, 218–225 descriptor values among test set, distribution of, 210, 215 normalized residuals, plot of, 223 ODS phases, selected in study, 206–209 operating conditions, choice of, 202–203 plot of S and E descriptor for solutes, 216 reduced test set for ODS phases, 217 selection of columns, 203–206 set of test solutes, selection of, 206, 210–218 solutes in final set with Abraham descriptors, 211–214 stationary phases, characterized in study, 204–205 system constants for, 219–222 Partial molar excess Gibbs energies, 86–87 Particle-based simulation methodology, 3 PBB, see Polybrominated biphenyls (PBBs) PBDE, see Polybrominated diphenyl ethers (PBDEs) PCA, see Principal component analysis (PCA) PCB, see Polychlorinated biphenyls (PCBs) PCDD, see Polychlorinated dibenzo-p-dioxins (PCDDs) PCDF, see Polychlorinated dibenzofurans (PCDFs) p-CEC, see Pressurized capillary electrochromatography (p-CEC) PCN, see Polychlorinated naphthalenes (PCNs) PDMS, see Polydimethylsiloxane (PDMS) Peak decay method, 181–182 Peak profiling method, 171–173 multi-column peak profiling method, 172 Péclet number, 60 O-(2,3,4, 5,6-Pentafluorobenzyl) hydroxyamine hydrochloride (PFBHA), 378 Peptide mixtures separation chromatogram, 128, 130 gain analysis time, 127–128 resolution of, 128 tryptic digest of bovine hemoglobin, 131 Perfluorooctanoic acid (PFOA), 331 Perfluorooctanoic sulfates (PFOS), 331 Persistent organic pollutants (POPs) levels of, 331 PFBHA, see O-(2,3,4, 5,6-Pentafluorobenzyl) hydroxyamine hydrochloride (PFBHA) PFC, see Polyfluorinated chemicals (PFCs) PFOA, see Perfluorooctanoic acid (PFOA) PFOS, see Perfluorooctanoic sulfates (PFOS) PGC, see Porous graphitic carbon (PGC) Phase ratio, 61 Phase retention on silica hydride-based column, 261

399

Index Phenols phase separation, 262 by ONP mode, 275 on silica hydride-based C18 column, 276 Phenylalanine separation, 260 chromatogram for, 269 Phenylglycine separation chromatogram for, 269 Phosphorylase b MassPREP™ Digestion Standard, 129 Physiological amino acids separation by UPLC, 125–126 Plate height method, 168–171 PLE, Pressurized liquid extraction (PLE) Pneumatic-modulated GC×GC, 305, 307 Poly(acrylate) (PA) coating, 341 Polybrominated biphenyls (PBBs), 331 Polybrominated diphenyl ethers (PBDEs), 331 Polychlorinated biphenyls (PCBs), 331 Polychlorinated dibenzofurans (PCDFs), 331 Polychlorinated dibenzo-p-dioxins (PCDDs), 331 Polychlorinated naphthalenes (PCNs), 331 Polycyclic aromatic hydrocarbons separation on silica-hydride-based C18 stationary phase, 266 Polydimethylsiloxane (PDMS) coating, 341, 374 Polyfluorinated chemicals (PFCs), 330 Polyhalogenated dibenzo-p-dioxins (PXDDs), 352–353 Polymeric adsorbents, 337 POP, see Persistent organic pollutants (POPs) Porous graphitic carbon (PGC), 201, 206, 210, 218, 221, 228–230 Positive free energy, 14 Post-extraction on-fiber derivatization estrogens and steroids, analyses of, 381–382 gas chromatographic derivatizing reagents, 381 herbicides and bisphenol A, 382 SPME-GC applications, 381 Practical equilibrium constants, 93 hydrophobic adsorbents asymmetric activity coefficients of proteins, 94 ion-exchange adsorbents asymmetric activity coefficients of proteins, 94 self-association asymmetric activity coefficients of proteins, 94 Pre-extraction derivatization amphetamines analysis, 376–377 haloacetic acids analysis, 377–378 headspace-SPME-GC-MS total-ion chromatogram and single ion chromatograms of methyl haloacetates, 378

(O-(2,3,4, 5,6-pentafluorobenzyl) hydroxyamine hydrochloride) (PFBHA), 378 PFBHA–headspace-SPME–GC–ECD chromatogram of water sample, 380 Pressure switching, 294 Pressurized capillary electrochromatography (p-CEC), 277 elution of charged compounds, 278 Pressurized liquid extraction (PLE), 330, 346–347, 364–365 optimization, critical parameters in, 346 quantitative recoveries for, 347 SWE, 346–347 Primary capillary temperature ramps, 311 Primordial modulators, 305 Principal component analysis (PCA), 198 1-Propanol distribution coefficient profiles, 23 Protein concentrations, estimation of, 62–63 separation by UPLC chromatogram, 128 gain analysis time, 127–128 resolution of, 128 solute, activity coefficients of, 71 pSFC, see Packed column SFC (pSFC) Purge-and-trap technique, 335 PXDD, see Polyhalogenated dibenzo-p-dioxins (PXDDs) PXDF, see Dibenzofurans (PXDFs)

Q qMS, see Quadrupole mass spectrometer (qMS) QSARs, see Quantitative-structure activity relationships (QSARs) QSRRs, see Quantitative structure-retention relationships (QSRRs) Quadrupole mass spectrometry (qMS) rapid-scanning qMS systems, 318 spectrometer, 313 instrument, 319 Quantitative affinity chromatography, see Biointeraction affinity chromatography Quantitative structure activity relationships (QSARs), 175 Quantitative structure-retention relationships (QSRRs), 198

R RAM, see Restrictive access materials (RAM) Rapid UPLC electrospray ionization mass spectrometry (UPLC-ESI-MS) method flavanone O-diglycosides analysis, 137

400 Fructus aurantii (Zhi qiao) analysis, 137 Trollius ledibouri Reichb, constituents analysis, 137 Rat urine, total ion chromatogram, 124 Real mixtures chemical potentials, 86 Restrictive access materials (RAM), 336–337 Retention factor, 151 Retention mechanism for RPLC, 11; see also Reversed-phase liquid chromatography (RPLC) average stationary phase structural and interfacial properties, 19 bonded-phase–solvent–solute environment, 27–28 density profiles, 18 driving forces for, 12–15 embedded polar groups, effects of, 25–26 partition/adsorption, 21–25 phase volumes, determination of, 26–27 pressure and pore curvature effects, 27 simulation approach, 12 simulation snapshots of, 16–17 solvent penetration, 20 Reversed-phase adsorbents, 70–71 Reversed-phase chromatography of hydrophobic isotherms, 72 thermal expansion coefficients, 111 Reversed-phase liquid chromatography (RPLC), 2 ideal vapor reference state in, 5 molecular simulations of, 29–30 analysis and presentation of data, 45 configurational-bias Monte Carlo, 38–45 gauche defect statistics, 45 Gibbs ensemble method, 36–38 heterogeneity in system composition, 46 Monte Carlo methods for, 34–36 order parameter, 45–46 solute distribution coefficients and transfer free energies, 46–47 transferable potentials for phase equilibria force field, 31–34 retention mechanism, 11 average stationary phase structural and interfacial properties, 19 bonded-phase–solvent–solute environment, 27–28 density profiles, 18 driving forces for, 12–15 embedded polar groups, effects of, 25–26 partition/adsorption, 21–25 phase volumes, determination of, 26–27 pressure and pore curvature effects, 27 simulation approach, 12

Index simulation snapshots of, 16–17 solvent penetration, 20 thermodynamic-based models of, 3 Reversed-phase silica hydride-based columns C18 and C8 hydride-based stationary phases, 264–265 properties of, 261 verification, 264 Robustness analysis, 80–81 Robust thermal modulator steel tube, 308 Rosenbluth weight, 39–40 RPLC, see Reversed-phase liquid chromatography (RPLC)

S SAE, see Sonication-assisted extraction (SAE) SAFE-CBMC algorithm, 41 Salt induced hydrophobic retention, 71 retention volume, 63 salting-out effect, 71 Sample preparation extraction (SPE) drying and homogenization of solid samples, 333 procedure, 331, 333 liquid environmental samples, 335–344 solid and semisolid samples, 344–349 vapor-phase extraction, 334–335 schemes, 332 SBSE, see Stir-bar sorptive extraction (SBSE) SCFA, see Self consistent field theory for adsorption (SCFA ) Scheibel equation, 104, 118 Self-association isotherm (SAS), 62, 95 Self-association equilibria association scheme, 91 electroneutrality balance, 92 equilibrium constant, 91 exclusion factors, 92 material balance, 92 mole fraction and activity coefficient, 92 monomer adsorbate concentration, 92 total adsorbate concentration, 92 Self consistent field theory for adsorption (SCFA), 8 Self-potentials, see Born potentials Sequential frontal analysis system, 177 Serotonin for retention, 271 SFE, see Supercritical fluid extraction (SFE) Silanol, 256 Silica columns for biointeraction studies, 156, 158 Silica hydride based etched capillaries application, 283–286

Index capillary properties, characterization of, 279–280 dual separation modes, 280–282 fabrication, 278–279 formats, 282–283 based stationary phases for HPLC applications, 264–278 chromatographic properties of, 259–264 materials, stability of, 258–259 synthesis and characterization, 257–258 C8-and C18-bonded silica, 349 C18 column, ANP mode on, 268 chemical surface structures, 256 difference between ordinary silica and, 256 as separation medium, 255 synthesis of, 256–257 Silylation, 353 Simplifying assumptions, 93 Simulation methodology and theory of liquids, 28–29 Simultaneous extraction and derivatization fiber and fatty acids, 380 formaldehyde, analysis of, 381 Sinanogˇlu’s theory, 5 Single component systems, 32 Single ion chromatograms of methyl haloacetates, 379 SLE, see Solid-liquid extraction (SLE) SLM, see Supported liquid membrane extraction (SLM) Small particles separation, 100–101 analysis time, 101–102 instrument performance, 104 Kozeny–Carman equation, 102 linear velocity, 103 molecular diffusion coefficient, 103 particle size and column length, 102 resolution, 102 scaling linear velocity, 102 Scheibel equation, 104 use of, 102 van Deemter curve, 102–104 Wilke–Chang equation, 104 Solenoid valve, 307 Sol-gel coatings, 341 Solid and semisolid samples and ASE, 346 MAE and SAE, 345 matrix solid phase dispersion (MSPD) C8-and C18-bonded silica and Florisil, 349 pressurized liquid extraction (PLE) optimization, critical parameters in, 346 quantitative recoveries for, 347 SWE, 346–347

401 SAE and MAE, 347 optimization, critical parameters in, 348 SLE, 345 and Soxhlet extraction, 345–346 supercritical fluid extraction (SFE), 344–345, 347, 366 optimization, critical parameters in, 349 POP bioavailability studies, 348 Solid–liquid extraction (SLE), 345 Solid-phase extraction (SPE), 330 disk, 337 MIP and RAM, 336 molecularly imprinted polymers (MIPs), 336–337 polymeric adsorbents, 337 Solid-phase microextraction (SPME), 330, 335, 373 amphetamine recovery, comparison of, 378 analyte molecular weight, polarity and analysis technique, relationship between, 375 derivatization in gas chromatographic inlet, 382 modes of, 376 procedures, 377 reactions for, 375 environmental pollutants, analysis, 340–341 extraction efficiency, 340 fiber material in environmental analysis, choice, 341 gas chromatography, 374–375 headspace silylation-GC-MS full-scan chromatogram, 383 history and timeline sorptive microextraction, 374 in-port derivatization–GC–MS reconstructed selected ion chromatograms from, 385 instrumentation of, 362 maximum extraction yield, 339 parameters in, 340 polar substituent groups with nonpolar substituents, replacement, 375–376 polydimethylsiloxane, volume, 338–339 post-extraction on-fiber derivatization, 381–382 GC-MS extracted ion chromatograms of herbicides for spiked water sample, 384 pre-doping technique, 381 pre-extraction derivatization hair amphetamines, analysis of, 376–377 simultaneous extraction and derivatization of fiber, 380 sorptive extraction, 338–340 Solute distribution coefficients and transfer free energies, 46–47

402 Solute retention process, 5 Solvation parameter model, use in pSFC, 199–201 choice of solvation descriptors, 225–228 experimental conditions chemical structures, stationary phases of, 206 chromatographic system, 201–202 covariance matrix for solute set, 216 data analysis, 218–225 descriptor values among test set, distribution of, 210, 215 normalized residuals, plot of, 223 ODS phases, selected in study, 206–209 operating conditions, choice of, 202–203 plot of S and E descriptor for solutes, 216 reduced test set for ODS phases, 217 selection of columns, 203–206 set of test solutes, selection of, 206, 210–218 solutes in final set with Abraham descriptors, 211–214 stationary phases, characterized in study, 204–205 system constants for chromatographic system, 219–222 Solvation process, 6 Solvent box, 43 extraction techniques, 364 saturated ODS bonded phase, 15 Solvophobic theory, 3 deficiencies of, 7–8 distribution coefficient, 7 free energy terms, 5–6 molecular surface areas, 7 Sinanogˇlu’s theory, 5 solute retention process, 5 solvation process, 6 thermodynamic model for, 4–5 Sonication-assisted extraction (SAE), 347, 364–365 optimization, critical parameters in, 348 Sorptive extraction, 338–340 Sorptive microextraction solid-phase microextraction (SPME), 374 Soxhlet extraction, 330, 345–346 Soxhlet extract of sewage sludge purification, 352–353 Spatial order and enhanced sensitivity FAMEs, analysis, 319, 322 SPE, see Solid-phase extraction (SPE) SPE pipette tip (SPEPT) extraction techniques, 335–336 SPEPT, see SPE pipette tip (SPEPT) Spider diagram, plotting in SFC, 232, 234–241 Split-peak method, 178–181

Index SPME, see Solid-phase microextraction (SPME) SPME-GC applications, 381 Standard Gibbs energy change of association, 89–90 Stationary phase of silica hydride-based column hydrophobic and hydrophilic compounds, retention map for, 274 Steroids separation on cholesterol column, 264–265 by pCEC on column, 277 Stir-bar sorptive extraction (SBSE), 338–341 extraction efficiency, 340 instrumentation of, 362 maximum extraction yield, 339 parameters in, 340 polydimethylsiloxane, volume, 338–339 sorptive extraction, 338–340 Stir-bar sorptive extraction (SBSE) techniques, 335 Subcritical water extraction (SWE), 346 Sulfonamide separation, 267 Sunscreen molecules set for developing separation method, 244, 246–247 chromatograms, 246 elution orders of, 245 structures of, 245 Supercritical fluid, 202 supercritical fluid chromatography (SFC), 195 Supercritical fluid extraction (SFE), 344–345, 357, 366 optimization, critical parameters in, 349 POP bioavailability studies, 348 Supported liquid membrane extraction (SLM), 330 SWE, see Subcritical water extraction (SWE) Symmetric activity coefficient model, 96 Synthetic peptide sample migration, pH effect on, 281

T Thermal effect adiabatic column at column outlet, 109 changes in temperature, magnitude of, 109 for incompressible fluids, 110 isothermal compressibility coefficient, 108 liquid volume, 107 mechanical energy, change in, 108 principle, 106 quasi-static transformation, 107 second principle of thermodynamics and, 107 small heat change, 107 specific heat capacity, 108 temperature change, 109 thermal expansion coefficient, 107

403

Index Thermal sweeper, 305–306 Thermodynamic-based models of RPLC, 3 Thermodynamic cycle for solute retention, 5 Thermodynamic modeling of chromatographic separation, 58 industrial applications, 79–81 Three-box GEMC setup, 42 Time-of-flight mass spectrometry (ToF MS), 300, 313–314 TMS, see Trimethylsilyl (TMS) group Tobramycin separation, 268–269 ToF MS, see Time-of-flight mass spectrometry (ToF MS) Transferable potentials for phase equilibria (TraPPE-UA) force field, 31 intramolecular potential, 32 parameterization philosophy, 33 validation of, 34 Trans-3-hydroxycinnamic acid identification, 271 TraPPE-UA, see Transferable potentials for phase equilibria (TraPPE-UA) force field Triethoxysilane and silica, condensation reaction between, 256 Trimethylsilyl (TMS) group, 353 Trollius ledibouri Reichb, constituents analysis by UPLC-ESI-MS method, 137 Tryptic digest, 129 of bovine hemoglobin, 131 Two-dimensional gas chromatography, 302 advantages of, 304–305 dual-stage modulation process, schematic of, 303 GC×GC chromatogram, 304 modulation time-window wraparound, 303–304 modulator capillary flow technology, 307 cutting stage, 310 Deans switch-type flow modulator in injection state, 307 integrated MDGC instrument, scheme, 309 LMCS, 305 longitudinally modulated cryogenic system, 306 low-cost pneumatic GC×GC, 307–308 pneumatic-modulated GC×GC, 305, 307 short microbore column segment, 308–309 single-stage air-cooled and electrically heated thermal modulator, GC×GC instrument with, 308 thermal sweeper, 305–306 tube, 308

operational parameters column selectivities, 310 detection, 312–314 gas flows, 311–312 GC×GC applications, series, 314–326 temperature gradient, 310–311

U Ultra-fast UPLC-UV method for antituberculosis tablets, 136 Ultra performance liquid chromatography (UPLC instrumentation), 100 applications of bioseparations, 127–137 higher performance, 124–125 higher speed, 122–124 physiological amino acids, 125–126 column performance, effect of pressure and temperature on combined effect of, 121 frictional heating, 117 Kozeny–Carman equation, 119 local HETP inside, 118 modeling studies, 121 pressure-induced and temperaturedependent changes, 118–119 pressure-induced influences, 120 Scheibel equation, 118 van Deemter equation, 118 viscosity, 118 Wilke–Chang equation, 118 retention, pressure and thermal effects on, 111–117 adiabatic column, temperature and pressure profiles for, 112 benzene and phenylpropanol, resolution between, 115 changes in, 117 column performance, deterioration in, 115 degree of, 116 ionized analytes, 116 migration velocity, 113 mobile phase, frictional heating of, 111 radial temperature differences, 113 selectivity of, 113 stationary and mobile phase, 115 temperature and superficial velocity, radial profiles of, 114 van’t Hoff equation, 113 volume changes, 116 small particles for separation analysis time, 101–102 instrument performance, 104 Kozeny–Carman equation, 102 linear velocity, 103

404 molecular diffusion coefficient, 103 particle size and column length, 102 resolution, 102 scaling linear velocity, 102 Scheibel equation, 104 use of, 102 van Deemter curve, 102–104 Wilke–Chang equation, 104 thermal effect adiabatic column at column outlet, 109 changes in temperature, magnitude of, 109 for incompressible fluids, 110 isothermal compressibility coefficient, 108 liquid volume, 107 mechanical energy, change in, 108 principle, 106 quasi-static transformation, 107 second principle of thermodynamics and, 107 small heat change, 107 specific heat capacity, 108 temperature change, 109 thermal expansion coefficient, 107 Ultrasound, as in sonication-assisted extraction (SAE), 345 UNIFAC, see Universal Functional Activity Coefficient (UNIFAC) model Universal Functional Activity Coefficient (UNIFAC) model, 8–9 UPLC instrumentation, see Ultra Performance Liquid Chromatography (UPLC instrumentation)

V van Deemter curve, 102–104, 118, 120 van’t Hoff equation, 113 Vapor–liquid coexistence curves (VLCCs), 32 Vapor–liquid equilibrium (VLE) simulations of molecule, 33

Index Vapor-phase extraction dynamic headspace (DHS), 335 HS and GC techniques, 334 thermal extraction, 335 VLCCs, see Vapor-liquid coexistence curves (VLCCs)

W Water thermal expansion coefficient, 109 water–ethanol mixtures molar density, 71–72 relative permittivity of, 70 Waters ACQUITY UPLC with tunable UV detector, 101 Whelk-O1 stationary phase, 30 Wilke–Chang equation, 104, 118 Wilson’s local composition activity coefficient model symmetric activity coefficient, 97

X XBridge®, 101 XL Stat software, 218

Z Zonal chromatography, 148 Zonal elution, in affinity chromatography, 148 applications of, 150 in binding and competition studies, 151–153 temperature and solvent studies, 153–155 characterization of binding sites by, 155–156 7-hydroxycoumarin binding to HSA, 149, 150 practical considerations for, 156–158 principles, 150–151

0

12

z (Å)

24

0

12

z (Å)

24

36

Figure 1.4  Simulation snapshots of stationary phase configurations taken from simulations of various RPLC systems. The left column, from top to bottom, shows snapshots for systems ODS-2.9/WAT, ODS-2.9/33M, ODS-2.9/67M, and ODS-2.9/33A. The right column, from top to bottom, shows snapshots for systems ODS-1.6/50M, ODS-4.2/50M, Amide2.9/33M, and Ether-2.9/33M. The silica substrate and grafted alkyl chains are shown as tubes with oxygen in orange, silica in yellow, and CHx groups in gray. Methanol, acetonitrile, and water are shown in the ball and stick representation with oxygen in red, hydrogen in white, nitrogen in green, and methyl groups in blue. Solutes are shown as large spheres with CHx groups in cyan, oxygen in red, and hydrogen in white.

1.2

ODS-2.9/WAT

ODS-1.6/50M

ODS-2.9/33M

ODS-4.2/50M

ODS-2.9/67M

Amide-2.9/33M

ODS-2.9/33A

Ether-2.9/33M

0.8 0.4 0 0.8

ρ (z) (g/mL)

0.4 0 0.8 0.4 0 0.8 0.4 0

0

12

z (Å)

24

0

12

z (Å)

24

36

Figure 1.5  The density profiles for the grafted chains (black), water (red), methanol (blue), and acetonitrile (green) in the retentive phase. The eight panels depict ensemble averages for the same eight systems shown in Figure 1.4. The total system density, computed as the sum of bonded phase and solvent densities, is shown in purple. The interfacial region (all panels), defined by the range where the total solvent density falls between 10 and 90% of its bulk value, is shaded in gray while the Gibbs dividing surface fitted to the total solvent density is shown by the dashed orange vertical line. The location of z = 0 Å corresponds to the silica surface.

120

15

1600

80

10

2

800

40

5

1

0

0

60

4

24

2

30

2

12

1

0

0

8

2

30

4

4

1

15

2

0

0

20

2

12

2

10

1

6

1

0 36

0

ODS-2.9/WAT

K (z)

0

ODS-2.9/33M

0

ODS-2.9/67M

0

0

ODS-2.9/33A

0

12

z (Å)

24

3

ODS-1.6/50M

ODS-4.2/50M

Amide-2.9/33M

Ether-2.9/33M

0

12

z (Å)

24

0

0

K (z)

2400

0

0 36

Figure 1.8  The distribution coefficient profiles for n-butane (blue) and 1-propanol (red). The eight panels depict ensemble averages for the same eight systems shown in Figure 1.4. The interfacial region (all panels) is shaded in gray while the Gibbs dividing surface is shown by the dashed orange vertical lines. The blue axis labels correspond to n-butane and the red to 1-propanol.

8.0E – 02 7.0E – 02

UV 280 nm

6.0E – 02 5.0E – 02 4.0E – 02 3.0E – 02 2.0E – 02 1.0E – 02 0.0E + 00

0

2

4

6

8 10 12 Volume (CV)

14

16

18

20

Figure 2.4  Modeled isocratic elutions of β-lactoglobulins A (right) and B (left) on a Source 30Q adsorbent at pH 7. The salt concentration is 157mM sodium chloride, and the loads are 5, 10, 20, 50, 100, 200, 350, and 500 µL, respectively, at a concentration of 10g/L. The column volume (CV) is 7.8 mL.

8.0E – 01 7.0E – 01

UV 280 nm

6.0E – 01 5.0E – 01 4.0E – 01 3.0E – 01 2.0E – 01 1.0E – 01 0.0E + 00

0

2

4

6

8

10 12 Volume (CV)

14

16

18

20

Figure 2.5  Modeled isocratic elutions of β-lactoglobulins A (right) and B (left) on a Source 30Q adsorbent at pH 7. The salt concentration is 157mM sodium chloride, and the loads are 350, 500, 1000, 2000, 3000, 4000, and 5000 µL, respectively, at a concentration of 10g/L. The column volume (CV) is 7.8 mL. The scale of the ordinate is 10 times the scale of Figure 2.4.

(a)

60

Peak capacity

40 20

2

(cm

gth

en

nl

4

8

lum

50 40 Grad ient t 30 20 ime ( minu 10 tes)

6

Co

60

14 12 10

)

0

(b)

Peak capacity

60

40 20

50 40 Grad ient t 30 20 ime ( 10 minu tes)

2

4

6

(cm )

8

Co lum

60

14 12 10

nl en gth

0

Figure 3.19  Representation of peak capacity for 25–30 nt oligodeoxythymidines varying gradient and column length. Peak capacity calculated from Equation 11 in reference [69] for columns packed with (a) 1.7 µm or (b) 3.5 µm sorbent. Flow rate: 0.2 mL/min, separation temperature: 60°C, gradient start: 17.5% MeOH, difference in solvent composition Δc: 0.05. Calculated for a 15 mM TEA – 400 mM HFIP aqueous ion-pairing system. Experimental constants for the model were obtained using XBridge and ACQUITY BEH C18 packed columns.

6 5

2nd dimension (s)

5 4 3

4 3 2 1 0

1000 1500 2000 2500 3000 3500

2 1 0

750

1000

1250

1500 1750 1st dimension (s)

2000

2250

2500

Figure 7.15  TIC GC × GC–ToF MS contour plot of cigarette smoke showing the firstdimension range between 500 and 2600 seconds. (From Dallüge, J., van Stee, L. L. P., Xu, X., Williams, J., Beens, J., Vreuls, R. J. J., and Brinkman, UATh, J. Chromatogr. A, 974, 169–184, 2002. With permission.)

585.0

585.5

(a)

0.25

0.50

0.75

1.00

1.25

1.50

1.75

2.00

563

575 588 600 613 1st dimension (s)

625

1.14 s 0.80 s 0.60 s 0.51 s 0.36 s 0.27 s

Butanenitrile 4-Penten-2-one Thiophene Benzene Cyclohexadien or Heptatriene-isomer Heptadiene-isomer Heptane+heplene-isomer 0.20 s

0.24 s

1.58 s

Butenenitrile

2t R

1-Chloro-2-propanone

Analyte

98

20 30 40 50 60 70 80 90 100110

32

67

200

400

600

800

55 67 95 20 30 40 50 60 70 80 90 100 110

15

27

39

(d) Library hit-similarity 837, “1,5-Hexadiene, 2-methl-” 81 1000

200

400

600

(c) Peak true-sample “drum cigarette smoke ZOEX interface, manual opertion” 39 55 1000 81 800

Figure 7.16  (a) expansion of the 2D chromatogram shown in Figure 7.15; the vertical line (at 583 second) indicates the raw second-dimension ­chromatogram, shown separately in (b). In (b) the horizontal lines indicate the positions where compounds were located by the deconvolution software. (c) the deconvoluted mass spectrum of the peak at 0.24 sec. (d) the corresponding library spectrum. (From Dallüge, J., van Stee, L. L. P., Xu, X., Williams, J., Beens, J., Vreuls, R. J. J., and Brinkman, UATh, J. Chromatogr. A, 974, 169–184, 2002. With permission.)

583.5

584.0

584.5

Total retention time (s)

(b)

2nd dimensions (s)

7.00 6.00

Sec

5.00 4.00 3.00 2.00 1.00 0.00

4

14

24

34

44 54 Min

64

74

84

Figure 7.17  TIC GC × GC-MS result for perfume. (From Mondello, L., Casilli, A., Tranchida, P. Q., Dugo, G., and Dugo, P., J. Chromatogr. A, 1067, 235, 2005. With permission.)

5.4 4.7

Sec

4.1 3.4 2.7 2.1 1.4 0.7 8.3 12.6 16.7 21.2 25.6 29.9 34.2 38.5 42.8 47.1 51.5 55.8 60.1 64.4 68.7 73.0 77.3 81.7 86.0 90.3 Min

Figure 7.20  Headspace SPME-GC × GC-qMS result for Arabica coffee volatiles. (From Mondello, L., Tranchida, P. Q., Dugo, P., and Dugo, G., Mass Spectrom. Rev., 27, 101, 2008. With permission.)

6.01 5.34 4.68

Sec

4.01 3.34 2.68 2.01 1.35 0.68

ω6 ω3 ω6 ω3 65 ω1 64 ω3 ω6 63 60 62 ω6 ω3 59 61 56 55 53 ω6 ω3 54 50 57 58 46 51 52 39 48 49 47 DB6 41 42 44 45 37 38 40 32 43 DB5 31 24 29 36 34 35 23 30 22 DB4 33 21 27 28 20 DB3 18 19 26 14 16 17 DB2 25 15 13 DB1 9 10 11 12 g 67 8 f DB0 IS e 5 d c 34 b 12a

0.02 4.4

C14 C15 7.54

10.67

C16 13.81

C17

C18

16.95

C19

20.08 Min

C20

C21

23.22

26.36

C22

C23 29.49

C24 32.63

Figure 7.21  GC × GC-FID result for plasma FAMEs. Refer to Table 7.2 for peak identification. (From Tranchida, P. Q., Costa, R., Donato, P., Sciarrone, D., Ragonese, C., Dugo, P., Dugo, G., and Mondello, L., J. Sep. Sci., 31, 3347, 2008. With permission.) 6.01 5.34 4.67 4.01

ω6

ω3

ω1

62 55 49

45

Sec

3.34

56 50 36

2.67

46 37 30

20

DB6 DB5 DB4 DB3 DB2 DB1 DB0

2.01 1.34 0.67 0.01 20.2

C20 20.72

21.23

21.75

22.26

22.78 Min

23.29

23.81

24.33

24.84

Figure 7.22  Expansion relative to the 2D chromatogram shown in Figure 7.21, illustrating the C20 FAMEs group (see Table 7.2 for peak assignment). (From Tranchida, P. Q., Costa, R., Donato, P., Sciarrone, D., Ragonese, C., Dugo, P., Dugo, G., and Mondello, L., J. Sep. Sci., 31, 3347, 2008. With permission.)

BP20,2tR (s)

7.5

5.0

Di-aromatics

2.5 Mono-aromatics Alkanes

0.0

13

25

63 38 50 DB-1, 1tR (min)

88

Alkanes

5 BPX-35,2 tR (s)

75

4 3 2

Mono-aromatics

Di-aromatics

1 0

0

10

20

30 40 BP21, 1tR (min)

50

60

70

Figure 7.23  GC × GC–FID chromatograms of diesel oil obtained on two different ­column sets, namely, apolar–polar (top) and polar–­medium polarity. (From Adahchour, M., Beens, J., Vreuls, R. J. J., Batenburg, A. M., and Brinkman UATh, J. Chromatogr. A, 1054, 47, 2004. With permission.)

Recovery (%)

100 80 60 40

SBSE SPME

20 0

0

2

4 Log Ko/w

6

Figure 8.3  Maximum extraction yield of SPME and SBSE with different logKo/w values.